Location of the configuration file

docassemble reads its configuration directives from a YAML file, which is located in /usr/share/docassemble/config/config.yml. If you are using the jhpyle/docassemble Docker image with S3, S3-compatible object storage, or Azure blob storage, docassemble will attempt to copy the configuration file from your S3 bucket or Azure blob storage container before starting.

How to edit the configuration file

The configuration file can be edited through the web application by any user with admin privileges. The editing screen is located on the menu under “Configuration.” After the configuration YAML is saved, the web application is restarted.

Some Configuration directives do not fully take effect until the entire system is restarted.

You can also edit the configuration file directly on the file system. You may need to do so if the web application becomes inoperative due to a problem with the configuration.

If you are using the jhpyle/docassemble Docker image and you try to edit /usr/share/docassemble/config/config.yml on the file system, keep in mind that under some conditions the config.yml file may be overwritten by a file located in another place. If you are using S3 or Azure blob storage, the version of config.yml located in the cloud always overwrites the local copy, so you should edit config.yml using the cloud provider’s web interface while the container is stopped, and then start the container again. If you are not using S3 or Azure blob storage, the Docker container’s initialization script will copy config.yml from /usr/share/docassemble/backup/config.yml. If you are using a Docker volume for data storage, you can stop the container and edit this file in the volume, then start the container.

After editing the config.yml file manually, you can restart the web application by running supervisorctl start reset.

Sample configuration file

Here is an example of what a configuration file may look like:

debug: True
exitpage: https://docassemble.org
db:
  prefix: postgresql+psycopg2://
  name: docassemble
  user: docassemble
  password: abc123
  host: localhost
  port: null
  table prefix: null
secretkey: 28asflwjeifwlfjsd2fejfiefw3g4o87
default title: docassemble
default short title: doc
mail:
  username: null
  password: null
  server: localhost
  default sender: '"Administrator" <[email protected]>'
default interview: docassemble.demo:data/questions/default-interview.yml
language: en
locale: en_US.utf8
default admin account:
  nickname: admin
  email: [email protected]
  password: password
voicerss:
  enable: False
  key: ae8734983948ebc98239e9898f998432
  dialects:
    en: us
    es: mx
    fr: fr
  voices:
    en: Amy
    es: Juana
    fr: Axel
s3:
  enable: False
  access key id: FWIEJFIJIDGISEJFWOEF
  secret access key: RGERG34eeeg3agwetTR0+wewWAWEFererNRERERG
  bucket: yourbucketname
  region: us-west-2
azure:
  enable: False
  account name: example-com
  account key: 1TGSCr2P2uSw9/CLoucfNIAEFcqakAC7kiVwJsKLX65X3yugWwnNFRgQRfRHtenGVcc5KujusYUNnXXGXruDCA==
  container: yourcontainername
ec2: False
words:
  - docassemble.base:data/sources/us-words.yml

At a bare minimum, your configuration file should set the secretkey and default sender directives:

secretkey: 28asflwjeifwlfjsd2fejfiefw3g4o87
mail:
  default sender: '"Administrator" <[email protected]>'

Configuration directives

Development vs. production

Set the debug directive to True on development servers and False on production servers. The default is True.

debug: True

Setting debug to True enables the following features:

• The “Source” button in the web app, which shows the YAML source code used to generate the current question, an explanation of the path the interview took to get to the current question, and readability statistics for the question and the help text.
• Viewing Markdown source in document attachments.

This can be overridden using the debug feature.

Note that even if debug is False, the “Source” button will still be present when running interviews in the Playground. This is because the Playground is a development tool, not a means of deploying interviews in production. For more information on how to install your interviews as packages, see the Packages section.

Note that if you set debug to True, then the sample interviews that come packaged with docassemble in the docassemble.base and docassemble.demo packages will not be accessible to ordinary users unless you choose to set allow demo: True. Note that default default interview is initially set to docassemble.base:data/questions/default-interview.yml, so if you set debug: False, you should change default interview to one of your own interviews.

Whether error messages should appear in the web browser

By default, error messages that relate to the YAML source code display in the browser if the user is logged in and has admin or developer privileges, but otherwise they only display in the docassemble.log file. If you have a development server with debug set to True that is not accessible to the internet, you might wish for these error messages to display in the browser even if the user is anonymous or does not have admin or developer privileges. If so, you can set development site is protected to True, and then all users will see these informational error messages.

development site is protected: True

Allowing sample interviews in production mode

By default, when debug is False, users without the privileges of admin or developer cannot run the sample interviews in the docassemble.base and docassemble.demo packages. If you would like to allow users to run these interviews, you can set allow demo to True.

allow demo: True

Enabling the Playground

By default, the Playground is available, even when debug is false. You can disable the Playground by setting:

enable playground: False

When enable playground is false, administrators and developers will not be able to access the Playground or its associated features, and users will not be able to use Playground interviews.

Enable users to use each others’ Playgrounds

By default, administrators and developers have a private Playground that only they can access. If you want to enable administrators and developers to access each other’s playgrounds, you can set enable sharing playgrounds to true:

enable sharing playgrounds: True

The default is false.

When enabled, the dropdown menu on the Playground page that is used to access Playground “projects” will have an additional menu item, Browse Other Playgrounds, which can be used to enter the Playgrounds of other users who have admin or developer privileges.

When enable sharing playgrounds is true, all users with admin or developer privileges will have read and write access to other users’ Playgrounds. Therefore, this feature should only be used if the developers using the server trust each other.

Allow statistics collection

If you set collect statistics to True, then docassemble will use Redis to keep track of the number of interview sessions initiated.

collect statistics: True

The default is False.

Allowing the viewing of logs

By default, a user with admin or developer privileges can view log files by going to Logs from the menu. To disable this feature, set allow log viewing to False.

allow log viewing: False

Sending log messages to standard error

By default, web application log messages are written to docassemble.log and celery worker log messages are written to worker.log. By setting log to std to True, these messages will be written to STDERR instead.

log to std: True

Allowing use of the Monitor page

By default, a user with admin or advocate privileges can go to the Monitor from the menu. To disable this feature, set enable monitor to False.

enable monitor: False

Allowing use of the Train page

By default, a user with admin, developer, or trainer privileges can go to Train from the menu and manage the training data of machine learning models. To disable this feature, set enable training to False.

enable training: False

Whether users with developer accounts can install packages

By default, any user with admin or developer privileges can use the Package Management page and install Playground packages on the system. If you want those features to be available only to users with admin privileges, set developer can install to False.

developer can install: False

Whether files on the server are owned by root

Applicable only to the jhpyle/docassemble Docker image.

By default, the docassemble Configuration and packages can be updated through the web interface. Accordingly, the Configuration file and the Python packages are owned by the www-data user.

However, if you set root owned to True, and you also set allow upgrading, enable playground, and allow configuration editing , then as extra security, files that no longer need to be editable by www-data will be owned by root instead of www-data.

root owned: True

Allowing developers to install the same package

By default, users with developer privileges have ownership over the packages they directly install, and other developers cannot create packages by the same name. If you want to allow developers to work on the same packages, set the package protection directive to False.

package protection: False

Path to web application

Set the root directive if you have configured docassemble to run from a sub-location on your web server. The default is /.

root: /da/

The value of root depends on how you configured your web server during installation. If the WSGI application runs from the URL https://docassemble.example.com/da/, set root to /da/. Always use a trailing slash.

If the WSGI application runs from the root of your web server (e.g., https://docassemble.example.com/), do not set this directive, or set it to /.

URL of application

The directive url root indicates the part of the URL that comes before the root.

url root: http://example.com

It is normally not necessary for docassemble to know how it is being accessed from the web, because the URL is provided to the web application with every HTTP request.

However, there are some circumstances when docassemble code runs outside the context of an HTTP request. For example, if you have a scheduled task that uses send_sms() to send a text message with a media attachment, the URL for the media attachment will be unknown unless url root is set correctly in the configuration.

You need to make sure url root is set correctly if you use Auth0 logins or you use any of the live help features.

If you correctly specified DAHOSTNAME and BEHINDHTTPSLOADBALANCER (if your server is behind a proxy that provides SSL termination) when you first started docassemble, then url root should be correct.

url root should match up with what you see in the location bar when you access the server.

Redirecting the root of the server

By default, if the user accesses the root URL of the server (which is / unless configured differently using root), the user is redirected to:

• The URL indicated by root redirect url, if it is defined in the Configuration, but if it is not defined then the user is redirected to
• Their current interview, if they have started one, but if they have not started an interview already, then they are directed to
• The default interview, if it is defined, but if it is not defined then the user is redirected to
• A list of available interviews, if dispatch is set up, but if not then the user is redirected to
• A factory default interview (docassemble.base:data/questions/default-interview.yml).

If you want the user to be directed to some other URL, you can define the root redirect url directive:

root redirect url: https://example.com/site/intro.html

In this example, if the user access the root path, they will be redirected to a web site.

The root redirect url can also be a relative path:

root redirect url: /site/intro.html

This can be useful if you want to operate a static web site on the same domain as your docassemble server. To do this, you will need to configure your web server to handle the /site/ path, or operate docassemble on a non-standard port and set up a reverse proxy that redirects traffic to the non-standard port.

Interview shortcuts

The dispatch directive allows users to start interviews with user-friendly URLs like https://example.com/start/legal.

dispatch:
  legal: docassemble.demo:data/questions/questions.yml
  madlibs: docassemble.base:data/questions/examples/madlibs.yml

A dispatch directive is a lookup table; it must be structured as a YAML dictionary where the keys are shortcut names and the values are interview names.

In the above example, the URL shortcuts /start/legal and /start/madlibs are enabled.

In addition, when you define the dispatch directive, your users can see a list of available interviews by going to the URL /list on the site. They will see a page like this:

If you want to take advantage of the /start/ shortcuts but you do not want the interview listed in the interview list, set unlisted: True in the metadata of the interview.

The /list URL accepts a URL parameter tag. If tag=estates, then /list will only list interviews where estates is one of the tags in the interview metadata.

You can also control whether an interview appears in the /list by adding the required privileges specifier to the interview metadata.

If you would like to embed the list of interviews into another web site, you can send a GET request to /list?embedded=1 to obtain a snippet of HTML. For example, the HTML snippet might look like this:

<ul class="dastart">
  <li><a href="/interview?i=docassemble.demo%3Adata%2Fquestions%2Fquestions.yml">Demonstration interview</a></li>
  <li><a href="/interview?i=docassemble.base%3Adata%2Fquestions%2Fexamples%2Fmadlibs.yml">Mad libs</a></li>
</ul>

On your web site, you can embed this into the body of a page. You can use the dastart CSS class to apply styles to this list.

Here is an example of a complete page of HTML that demonstrates how you can embed the result of /list?embedded=1 into a page:

<!DOCTYPE html>
<html lang="en">
  <head>
    <meta charset="utf-8">
    <title>Page of a web site</title>
    <style>
      ul.dastart {
        list-style-type: circle;
      }
      ul.dastart li {
        font-size: 12px;
      }
      ul.dastart a {
        text-decoration: none;
      }
      ul.dastart a:hover {
        text-decoration: underline;
      }
    </style>
  </head>
  <body>
    <h1>Hello, world!</h1>

    <p>I thought you might be interested in some interviews.</p>

    <div id="interviews"></div>

    <p>Goodbye.</p>

    <script src="https://cdnjs.cloudflare.com/ajax/libs/jquery/3.2.1/jquery.min.js"></script>
    <script type="text/javascript">
      $.get("https://demo.docassemble.org/list?embedded=1", function(data){
        $("#interviews").html(data);
      }).fail(function(){
        $("#interviews").html('<span style="color: red">Error: unable to get list of interviews.</span>');
      });
    </script>
  </body>
</html>

You can try this out here.

On your web site, you may need to edit files in different places in order to control the various parts of the HTML response. Some things to note about this example:

• The place where you want the interview list to appear is designated by <div id="interviews"></div>.
• The part between <style> and </style> is optional and simply demonstrates how the list of interviews can be styled. The CSS commands could also be put in a separate CSS file.
• The part that retrieves the interview list is the JavaScript call to $.get(), which is between <script type="text/javascript"> and </script>. The part that plugs the HTML into the screen is the call to .html().
• jQuery needs to be loaded before the call to $.get(). jQuery is very common on web sites, so it may already be loaded on your site.
• The JavaScript code is wrapped in a call to $( document ).ready. This may not be necessary on your site, but it can help avoid the potential problem where call to .html() takes place before the <div id="interviews"></div> HTML even exists on the screen.

You can also call /list?json=1 to obtain a JSON version of the available interviews.

By default, the main menu does not contain a link to the /list page, but you can add it to the main menu using the show dispatch link directive.

Customization of column widths

docassemble uses the Bootstrap grid system for its mobile responsive layout. The widths of columns that are used in various parts of the application can be customized using the grid classes directive.

The following example shows the default values.

grid classes:
  user: offset-lg-3 col-lg-6 offset-md-2 col-md-8 offset-sm-1 col-sm-10
  admin: col-md-7 col-lg-6
  admin wide: col-sm-10
  vertical navigation:
    bar: offset-xl-1 col-xl-2 col-lg-3 col-md-3
    body: col-lg-6 col-md-9
    right: d-none d-lg-block col-lg-3 col-xl-2
    right small screen: d-block d-lg-none
  flush left:
    body: offset-xxl-1 col-xxl-5 col-lg-6 col-md-8
    right: d-none d-lg-block col-xxl-5 col-lg-6
    right small screen: d-block d-lg-none
  centered:
    body: offset-lg-3 col-lg-6 offset-md-2 col-md-8
    right: d-none d-lg-block col-lg-3
    right small screen: d-block d-lg-none
  label width: md-4
  field width: md-8
  grid breakpoint: md
  item grid breakpoint: md

user refers to administrative screens that end users see, like the login screen. admin refers to administrative screens that only administrators can see, like the User List. admin wide refers to administrative screens that are wider than others, like the Configuration.

The vertical navigation, flush left, and centered directives refer to modes in an interview when different column widths must be used. The default mode is centered. If you set centered: False in the features, then the flush left directive is used. However, if you have a vertical navigation bar enabled in your interview, the vertical navigation directive is used.

Under these directives, body refers to the width of the main part of the screen. right refers to the width of the right screen part that is positioned to the right of the body. On small screens, the right part is invisible and the same contents appear below the body, between the under part (if any) and the post part (if any). The right small screen directive refers to the size of this version of the right screen part. Under vertical navigation, the bar refers to the width of the vertical navigation bar.

The label width, field width, grid breakpoint, and item grid breakpoint are not classes, but rather parts of classes. The actual classes used will be prefixed with col- or offset-, depending on the context. The default values of md-4 and md-8 mean that when labels and fields appear side-by-side, the width of the labels is half of the width of the fields, and below the md breakpoint, the labels appear above the fields. The grid breakpoint is a default value for the grid field modifier. It indicates the breakpoint in Bootstrap’s grid system when the side-by-side arrangement should be shown. The default is md. Likewise, the item grid breakpoint is a default value for the item grid field modifier.

Customization of alerts

When the front end needs to alert the user about something, it places a Bootstrap alert at the top of the screen. The HTML of these alerts can be customized using alert html an alert container html.

alert html: |
  <div class="da-alert alert alert-%s alert-dismissible fade show" role="alert">%s<button type="button" class="btn-close" data-bs-dismiss="alert" aria-label="Close"></button></div>
alert container html: |
  <div class="datopcenter col-sm-7 col-md-6 col-lg-5" id="daflash">%s</div>

In alert html, the first %s represents the Bootstrap color of the alert, while the second %s represents the HTML of the content of the message. In alert container html, the %s represents one or more alert html <div>s.

Customization of administrative pages

When the user is interacting with an interview, the page can be customized with a features block. If you want to customize the appearance of other pages on the site, you can do so in the Configuration using directives like login page heading or profile page title.

The names of these directives have two parts: the first part is the name of the page, and the second is the name of the aspect of the page you wish to customize.

The available page names are:

• start page: the start page that shows a list of available interviews.
• register page: the page where the user registers for an account.
• login page: the page where the user logs in.
• interview page: the page that shows a list of saved interview sessions.
• profile page: the page where the user can edit his or her name.
• change password page: the page where users can change their password.
• forgot password page: the page where users can type in their e-mail address and receive an e-mail with a link to reset their password.
• reset password page: the page where users are directed when they forget their password.
• 404 page: the page users see when they try to visit a URL that does not exist on the server.
• error page: the page users see when Flask traps a Python exception.
• main page: the pages of interviews. (This one is special; see below.)

The aspects of the page that can be manipulated are:

• tab title: the HTML title, which is displayed in the web browser tab.
• title: the title of the page in the navigation bar.
• navigation bar html: raw HTML tags that you can insert into the navigation bar. The contents will be placed inside of a <ul> in a <div>: <div class="collapse navbar-collapse" id="danavbar-collapse"><ul class="navbar-nav ms-auto">. For best results, use <li class="nav-item"> elements containing <a class="nav-link" href="#"> elements. On large screens, the items will appear in the navigation bar, and on small screens, the items will appear as menu items.
• heading: the large font heading that displays the title of the page in the main body of the page.
• pre: text that appears after the heading but before the rest of the content of a page. This is empty by default.
• submit: text that appears before the submit buttons on a page that has one or more submit buttons. This is empty by default.
• post: text that appears after the content of a page. This is empty by default.
• footer: text that appears at the bottom of the screen, in a 60px tall box with the Bootstrap light color.
• extra css: raw HTML tags that you can insert into the <head> of the page, usually to incorporate CSS files. This is empty by default.
• extra javascript: raw HTML tags that you can insert at the end of the <body> of the page, usually to include Javascript in <script> tags. This is empty by default.

All of the directives accept HTML, except you will not want to use HTML in the tab title, since it will not be displayed as HTML.

Here are some examples of using these directives:

start page title: Missouri interviews
start page heading: Interviews available to Missouri residents
interview page title: Your interviews
interview page heading: Interviews you have started
interview page extra css: |
  <link href="https://example.com/css/mystyles.css" rel="stylesheet">
login page pre: |
  <p>Unauthorized access to this site is <strong>strictly</strong>
  prohibited!</p>
register page extra javascript: |
  <script>
    console.log("We are registering.")
  </script>

You can also use these directives to provide different language to be used depending on the user’s language.

start page title:
  en: Available tools
  es: Herramientas disponibles

If you want to include the same CSS or JavaScript files in every page of the site, including administrative pages and interview pages, you can use the global css and global javascript directives.

If you want to set a footer that is available on all pages, you can set global footer instead of start page footer and interview page footer, etc. In the context of an interview, you can set the footer screen part to off in order to turn off the global footer or the main page footer. You can also set the main page footer to off if you have a global footer that you do not want to appear during interviews.

The main page directive is special because it relates to actual interviews, which are different from administrative pages. The directives that work with the main page prefix are:

• main page back button label
• main page continue button label
• main page corner back button label
• main page exit label
• main page exit link
• main page footer
• main page help label
• main page logo
• main page navigation bar html
• main page post
• main page pre
• main page resume button label
• main page right
• main page short logo
• main page short title
• main page submit
• main page subtitle
• main page title url opens in other window
• main page title url
• main page title
• main page under

You can set these when you want special content to be set by default for various parts of the screen for all interviews running on your server.

main page post: |
  <p>This interview was sponsored in part by a grant from the Example Foundation.</p>

You can set the directives to a dictionary where the keys are languages and the values are the text.

main page post:
  en: |
    <p class="smallprint">This interview was sponsored in part by a grant from the Example Foundation.</p>
  es: |
    <p class="smallprint">Esta entrevista fue patrocinada en parte por una beca de la Fundación Ejemplo.</p>

For information about other ways to set defaults for different parts of the screens during interviews, and what these screen parts mean, see the screen parts section.

There are two pages that you can customize even more extensively:

• The start page that shows a list of available interviews.
• The list of saved interview sessions.

For these pages, you can specify an HTML template to use. Use the start page template and interview page template directives to point to files in the “templates” directory of a package.

start page template: docassemble.missouri:data/templates/my_start_page.html
interview page template: docassemble.missouri:data/templates/my_interview_page.html

To see what the standard templates look like, see:

• start.html
• interviews.html

You might want to use these as a starting point. Note that docassemble uses Flask, which uses Jinja2 HTML templates.

These templates use Jinja2 to load a standard HTML framework. If you want, you can replace the entire body of the page using these templates. However, if you do so, you may wish to include {{ extra_css }} in the <head> and {{ extra_js }} at the end of the <body>. On the interviews page, there is a bit of JavaScript that asks the user “are you sure?” before deleting all of the interview. The start page does not use JavaScript.

You can also customize these pages using CSS. The HTML elements in the standard template use some classes do nothing; they are just placeholders for customization. For example, to customize start.html and interviews.html, you could include the following in a CSS file that you include with global css (which is discussed in the next section):

h1.dastartpage {
  font-size: 15px;
}
ul.dastartpage {
  background-color: red;
}
ul.dastartpage li {
  font-weight: bold;
}
h3.dainterviewpage {
  font-decoration: underline;
}
p.dainterviewpage {
  color: #aaaaaa;
}
table.dainterviewpage td {
  border-style: solid;
  border-width: 1px;
}

Replacing built-in pages with interviews

Instead of customizing the start page and the interview page, you can replace them with interviews.

session list interview: docassemble.base:data/questions/examples/session-interview.yml
dispatch interview: docassemble.base:data/questions/examples/list-interview.yml

The interview_list() function allows you write an interview that accesses information about the interview sessions of a logged-in user. Here is an example interview, docassemble.base:data/questions/examples/session-interview.yml, which mimics the style of the standard list of interview sessions.

metadata:
  require login: True
  temporary session: True
  hidden: True
---
mandatory: True
code: |
  current_page = 0
  next_id_of_page = [None]
---
reconsider: True
code: |
  current_info = user_info()
  list_of_interviews = []
  (items, next_id) = interview_list(next_id=next_id_of_page[current_page])
  if len(next_id_of_page) == current_page + 1 and next_id is not None:
    next_id_of_page.append(next_id)
  del next_id
  for info in items:
    if not info['metadata'].get('hidden', False):
      list_of_interviews.append(info)
    del info
  if len(list_of_interviews) > 0:
    list_of_interviews += [dict()]
  set_save_status('overwrite')
---
table: session_table
rows: |
  list_of_interviews
columns:
  - Title: |
      '<a href="' + interview_url(i=row_item['filename'], session=row_item['session']) + '">' + row_item['title'] + '</a>' if 'title' in row_item else ''
  - Modified: |
      row_item['modtime'] if 'modtime' in row_item else ''
  - Delete: |
      action_button_html(url_action('delete', filename=row_item['filename'], session=row_item['session']), color='warning', label='Delete') if 'title' in row_item else action_button_html(url_action('delete_all'), color='danger', label='Delete All')
---
table css class: table table-striped
mandatory: True
question: |
  Resume an interview
subquestion: |
  % if len(list_of_interviews) > 0:
  Click interview title to resume an interview
  you already started.

  ${ session_table }
  % else:
  You do not have any sessions.
  % endif

  % if current_page > 0:
  ${ action_button_html(url_action('change_page', direction=-1), label="Prev", icon='chevron-left', size='md', color='secondary') }
  % endif
  % if current_page + 1 < len(next_id_of_page):
  ${ action_button_html(url_action('change_page', direction=1), label="Next", icon='chevron-right', size='md', color='secondary') }
  % endif
---
event: change_page
code: |
  current_page += int(action_argument('direction'))
---
event: delete
code: |
  if interview_list(
      action='delete',
      filename=action_argument('filename'),
      session=action_argument('session')):
    log("Deleted interview", "success")

GitHub

The interview_menu() function allows you write an interview that accesses information about the interviews available on the start page. Here is an example interview, docassemble.base:data/questions/examples/list-interview.yml, which mimics the style of the standard start page:

code: |
  available_interviews = interview_menu()
---
prevent going back: True
mandatory: True
question: |
  Available interviews
subquestion: |
  % if len(available_interviews) > 0:
  % for interview in available_interviews:
  * [${ interview['title'] }](${ interview['link'] })
  % endfor
  % else:
  There are no available interviews.
  % endif

GitHub

Requiring login for all interviews

If you want to require users to be logged in before they can run any interviews, set allow anonymous access to False.

allow anonymous access: False

The default is True. You can also forbid anonymous access on an interview-by-interview basis using required privileges.

Administrative interviews

The administrative interviews directive adds hyperlinks to interviews (or other things) to the main menu.

Suppose you have two interviews, inspect.yml and manage.yml, that you use frequently. You can list them in the Configuration as follows:

administrative interviews:
  - docassemble.missouri:data/questions/inspect.yml
  - docassemble.missouri:data/questions/manage.yml

Then, there will be hyperlinks in the menu to:

• /interview?i=docassemble.missouri:data/questions/inspect.yml&new_session=1
• /interview?i=docassemble.missouri:data/questions/manage.yml&new_session=1

The labels for these hyperlinks will be taken from the metadata of the interview. The label will be the short title as specified in the metadata, or if that is not defined the label will be the title.

These interview links will always be shown in the main menu; however, a link to an interview will not be shown if the user is currently in a session of that interview.

If you want the links in the menu to appear only for users in a given role, you can set the required privileges for listing specifier in the metadata to a list of privileges. In that case, only users who have one of the listed privileges will see the link in the menu. If a required privileges for listing specifier does not exist, the required privileges specifier will be used to determine if the link in the menu is shown.

The titles and privileges that will be used for purposes of the menu can be overridden by specifying an item in the administrative interviews list in the form of a YAML dictionary instead of the name of the interview file. For example:

administrative interviews:
  - interview: docassemble.missouri:data/questions/inspect.yml
    title: Inspect
    required privileges:
      - admin
      - advocate
  - interview: docassemble.missouri:data/questions/manage.yml
    title:
      en: Inspect
      es: Inspeccionar
    required privileges:
      - admin

When listing an item in dictionary form, only the interview key is required. The title key can be set to a single name, or to a YAML dictionary where the keys are languages and the values are the titles to be used for that language.

The hyperlinks from the menu will contain the &new_session=1 URL parameter, which means that a new session will be started every time the user clicks the interview. In many situations, you will not want the clicking of the link from the menu to result in a proliferation of interview sessions. You may want to set up the metadata of your administrative interviews as follows:

metadata:
  title: Manager
  required privileges:
    - admin
  sessions are unique: True
  hidden: True

The sessions are unique specifier means that there will only ever be one session per user. This means that if the user already has a session in the interview, the &new_session=1 directive will not result in a new session being started; instead, the original session will be resumed.

The hidden specifier means that sessions in this interview will be hidden on the My Interviews page.

It is a good practice to make your “administrative” interviews simple so that they feel more like control panels than interviews. For example, you might want to have a single screen with buttons created using action_button_html(). Or if you need multiple screens, you can have a variable to keep track of the current screen, and then use menu_items and url_action() to provide the user with a menu for navigating from screen to screen.

The administrative interviews directive can also be used to add URLs of your own choosing to the menu. Provide a dictionary with keys for url and title:

administrative interviews:
  - url: https://secret.example.com/
    title: Secret Site

As with interview entries, you can specify the title in different languages, and you can limit the appearance of the menu item based on the user’s privileges. For example:

administrative interviews:
  - url: https://portal.example.com/
    title:
      en: Portal
      es: El Portal
    required privileges:
      - admin

Always go back to the current interview after logging in

If a user starts an interview as an anonymous user and then registers or logs in, they generally be taken to the My Interviews page. However, if the interview session they started is their only interview session, then the My Interviews page is skipped and the user is sent right back into the interview they had started.

The My Interviews page is shown to users with more than one session because it resolves confusion for users who may have visited the interview for purposes of resuming an existing interview session, but finding themselves not logged in, they log in. Some users may expect to resume the session they just started; other users may expect to resume their original session. By sending users to the My Interviews page, docassemble lets users know that they have two sessions and can make a choice.

If you want users to always be redirected to the session they just started, you can set resume interview after login to True.

resume interview after login: True

The default value is False.

If resume interview after login is set to True, users who start out conducting an interview as an anonymous user and then log in will not be sent to the My Interviews page, but will be redirected back into the interview they just started.

Automatically resuming or starting an interview

The My Interviews page at /interviews normally displays a list of the user’s interview sessions. By default, users are directed to /interviews when they log in.

If your server only offers a single interview, for example, docassemble.eviction:data/questions/defend.yml, you might not want your users to look at a list of sessions; you might simply want them to resume their existing session, if they have one, or start a new session in that interview.

To accomplish this, you can set the following:

auto resume interview: docassemble.eviction:data/questions/defend.yml

When this is set, then whenever a user logs in, the user will be redirected to their first interview session involving the docassemble.eviction:data/questions/defend.yml interview. Or, if they have no such session, a new session will

If you set an auto resume interview, you may wish to set show interviews link to False if the My Interviews menu is not useful on your server.

Customizing the page that appears after login

By default, the user is directed to the My Interviews page after logging in or registering. This can be customized with the page after login directive:

page after login: profile

In this example, the user will be directed to the “Profile” page after logging in or registering. See the documentation for the url_of() function to see what names to use for each page of the web site. (The name you give is effectively passed to url_of().) Instead of passing the name of a page of a web site, you can also pass a URL beginning with http:// or https://, or a path beginning with /, such as /start/taxes, in which case the user will be taken to that page on the server.

You can also set up different pages for users with different privileges:

page after login:
  - admin: config
  - developer: playground

In this example, users with privileges of admin will be directed to the Configuration page, while users with privileges of developer will be directed to the Playground.

If a user has privileges of both admin and developer, the user will be directed to config. Each item in page after login is processed in order, and the first match is used. If no items match, the user is directed to “My Interviews.”

If a next parameter is present on the login or register page, the user will be redirected to that location instead of the page after login.

JavaScript loading

The javascript defer setting should be set to True for faster JavaScript loading. This causes the <script> tags to have the defer attribute set.

javascript defer: True

For backwards compatibility, it is set to False by default, in case you have raw global javascript that depends on jQuery and that will fail if scripts are loaded with defer.

CSS and Javascript customization

You can use the javascript features setting and the css features setting to modify the JavaScript and CSS for a particular interview.

By using global javascript and global css, you can apply JavaScript and CSS on a more global level. These directives allow you to include JavaScript and CSS files in every interview and also in the start page and the page showing the list of interviews. The directive should refer to files located in the “static” directory of a package:

global css: docassemble.missouri:data/static/missouri.css
global javascript: docassemble.missouri:data/static/missouri.js

If you have more than one CSS file or more than one JavaScript file, you can use these directives to refer to a list of files:

global css:
  - docassemble.missouri:data/static/missouri.css
  - docassemble.midwest:data/static/look_and_feel.css

You can also refer to files on the internet:

global css: https://example.com/css/site_style.css
global javascript: https://example.com/js/tracker.js

These JavaScript and CSS files are loaded after the other JavaScript and CSS files on the page.

The global css and global javascript directives can only refer to file names; if you want to write raw <link>, <meta>, or <script> content, you can use the raw global css and raw global javascript directives:

raw global css: |
  <meta property="og:title" content="Child Custody Interview" />
raw global javascript: |
  <script defer>
    (function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
    (i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
    m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
    })(window,document,'script','https://www.google-analytics.com/analytics.js','ga');
    ga('create', 'UA-77777777-1', 'auto');
  </script>

The raw global css content is inserted into the <head> of the page, while the raw global javascript is inserted into the end of the <body> of the page. There is no requirement that the raw global css content relate to CSS, or that that the raw global javascript content relate to JavaScript; the names simply refer to the conventions about where CSS and JavaScript are typically placed in the HTML page.

You should set the defer attribute in any <script> tags you include in raw global javascript if javascript defer (see the previous section) is set to True. If you don’t set the defer attribute, your code may run in an unpredictable order relative to the loading of other files.

Bootstrap theme

Using the bootstrap theme directive, you can change the look and feel of your site’s web interface by substituting a non-standard CSS file in place of the standard CSS file used by Bootstrap.

bootstrap theme: docassemble.demo:data/static/lumen.min.css

You can also refer to files on the internet:

bootstrap theme: https://bootswatch.com/5/minty/bootstrap.min.css

There are many alternative Bootstrap themes available on Bootswatch. (Use Bootstrap 5 themes.) Note that you may need to use apply additional CSS changes in order to get a Bootstrap theme to work with docassemble. For example, by default docassemble assumes that the navigation bar is 56 pixels in height. But if your theme makes the navigation bar 66 pixels tall, you will need to add the following styles:

.da-pad-for-navbar {
    padding-top: 76px;
}
@media (min-width: 576px) {
    body.dasignature.da-pad-for-navbar {
	    padding-top: 76px;
    }
}
.da-top-for-navbar {
    top: 76px;
}

Also, some of the colors that docassemble uses in the navigation bar might not work well with every theme. You can override these colors with CSS like the following:

span.daactivetext       { color: yellow;         }
span.daactivetext:hover { color: #ffff99;        }
i.da-chat-active           { color: #45bf41;        }
i.da-chat-active:hover     { color: #55cf51;        }
i.da-chat-inactive         { color: #4f4f4f;        }
i.da-chat-inactive:hover   { color: #5f5f5f;        }

These styles refer to the yellow “Help” button that appears when question-specific help is available, the green icons that appear in the navigation bar when live help features are available, and the grey color that these icons become when live help is no longer available.

You can also control the Bootstrap theme on a per-interview basis with the bootstrap theme feature. You can also alter CSS on a per-interview basis using the css features setting.

Inverted Bootstrap navbar

By default, docassemble uses Bootstrap’s “dark” (formerly known as “inverted”) style of navigation bar so that the navigation bar stands out from the white background. If you do not want to use the inverted navbar, set the inverse navbar to False.

inverse navbar: False

You can also control this on a per-interview basis with the inverse navbar feature.

Automatic color scheme switching

By default, if the OS or browser indicates that the user is using a “dark” color scheme, docassemble will activate the “dark” color scheme in Bootstrap by adding data-bs-theme="dark" to the <html> element. (This does not apply if the interview is embedded in a <div>.)

If you do not want the “dark” color scheme to be activated automatically, set auto color scheme to False. The default value is True.

auto color scheme: False

Button color

The colors of buttons in the docassemble user interface can be customized using Bootstrap colors.

You can choose a different Bootstrap color than the default by setting one or more items under the button colors directive:

button colors:
  "continue": primary
  "answer yes": primary
  "answer no": secondary
  "answer maybe": warning
  "clear": warning
  "back": link
  "register": primary
  "new session": primary
  "leave": warning
  "url": link
  "restart": warning
  "refresh": success
  "signin": info
  "exit": danger
  "logout": danger
  "send": primary
  "download": primary
  "review": secondary
  "add": secondary
  "edit": secondary
  "delete": danger
  "undelete": info
  "reorder": info
  "navigation bar login": primary
  "help": info
  "question help": info
  "back to question": info
  "labelauty": primary
  "labelauty nota": primary
  "labelauty aota": primary

The above example lists the default colors. The meanings of the keys in this dictionary are as follows:

• continue refers to Continue buttons and buttons that act like Continue buttons.
• answer yes refers to the “Yes” button in a yesno question.
• answer no refers to the “No” button in a yesno question.
• answer maybe refers to the “Maybe” button in a yesnomaybe question.
• clear refers to the “Clear” button in a signature question.
• back refers to the “Back” button when the question back button feature is used.
• register refers to the button that is generated when the value of an item in a buttons list is register. refers to register
• new session refers to the button that is generated when the value of an item in a buttons list is register.
• leave refers to the button that is generated when the value of an item in a buttons list is register.
• restart refers to the button that is generated when the value of an item in a buttons list is restart.
• refresh refers to the button that is generated when the value of an item in a buttons list is refresh.
• signin refers to the button that is generated when the value of an item in a buttons list is signin.
• exit refers to the button that is generated when the value of an item in a buttons list is exit.
• logout refers to the button that is generated when the value of an item in a buttons list is logout.
• send refers to the button for sending an e-mail from the attachment user interface.
• download refers to the button for sending an e-mail from the attachment user interface.
• review refers to the buttons that are generated as part of a review screen. This is also customizable with review button color under features.
• add refers to the button that is generated by the .add_action() method of a DAList.
• edit refers to the button for editing an item in a table.
• delete refers to the button for deleting an item in a table or list collect.
• undelete refers to the button for un-deleting an item in a list collect.
• reorder refers to the buttons for changing the order of an item in a table.
• navigation bar login refers to the “Sign in” button that appears in the upper right corner of the screen if login link style is set to button.
• help: refers to the question mark icons that appear when there is help text related to an option in a list of options.
• question help: refers to the button that is available when the question help button feature is being used and there is help text associated with the current question.
• back to question: refers to the “Back to question” button that appears when the user is viewing the “help” tab on a question screen.
• labelauty: refers to the default color of a selected item in a radio button or checkbox list. Labelauty is the name of the JQuery add-on that is used to style radio buttons and checkboxes.
• labelauty nota: refers to the default color of a selected “None of the above” item in a checkbox list.
• labelauty aota: refers to the default color of a selected “All of the above” item in a checkbox list.

One reason to set button colors is to avoid biasing the user toward clicking a particular button. For example you might want to set no: primary so that the “No” selection is the same colors as the “Yes” selection. Or you might want to set exit: secondary because the default color of danger might discourage users from clicking a button to leave the interview.

If you simply want different colors, then before editing button colors, you may wish to consider creating your own Bootstrap theme in which the concepts of primary, warning, danger, success, etc. are associated with different colors than blue, yellow, red, and green.

Button style

There are two button styles: normal and outline. The normal style uses Bootstrap styles like btn btn-success. The outline style uses styles like btn btn-outline-success. The default is normal.

button style: outline

Button size

By default, most buttons in the docassemble user interface are regular size Bootstrap buttons. If you want the buttons to be “large” size, set the following in your configuration:

button size: large

If you want the buttons to be “small” size, set the following in your configuration:

button size: small

The default value is medium.

Default CSS class for the footer

By default, if you use a footer, the footer has the CSS class bg-secondary-subtle, in addition to the class dafooter. If you would like to use a different class instead of bg-secondary-subtle, you can set this class using footer css class.

footer css class: bg-dark text-white

Default CSS class for Markdown tables

By default, when a Markdown table is converted to HTML, <table> elements will be given the CSS class table table-striped. This can be varied by setting the table css class specifier on the question (or using the default screen parts or metadata). If you want to set a global default for this setting, you can set table css class in the Configuration:

table css class: table table-bordered

You can also use table css class to specify a class for the <thead> within the table. If you set table css class to table table-bordered, thead-dark, then the <table> will have the class table table-bordered and the <thead> will have the class thead-dark.

Using standard sets of icons

Font Awesome is enabled on all pages. docassemble includes the free version Font Awesome. This allows you to refer to Font Awesome icons using raw HTML.

In addition, docassemble allows you to use Google’s Material Icons in your raw HTML. Since the Material Icons are just a font, they are enabled by default and there is no configuration directive to enable the use of Material Icons.

You can also use Font Awesome or Material Icons for decorations and inline icons, without writing any raw HTML.

By default, when you include decorations, or you include inline icons, you can only refer to image files that you have predefined. An inline image reference that does not have a corresponding predefined image will appear literally (e.g., :cow:). However, if you set the default icons configuration directive to font awesome, then references to any images that you have not predefined will be treated as references to icons in the Font Awesome collection.

default icons: font awesome

Then you can have an interview like this:

mandatory: True
decoration: chart-bar
question: |
  Third quarter metrics
subquestion: |
  We are making more money
  :far-fa-money-bill-alt: than
  we did in the second quarter.

  So you can sleep well tonight! :bed:

GitHub

The “solid” style of the icons (CSS class fas) is used by default.

If you want to use icons in a different style, you can add a special prefix to your icon reference, writing, e.g., :far-fa-circle: instead of :circle:. This will result in <i class="far fa-circle"></i> instead of <i class="fas fa-circle"></i>.

To use a different Font Awesome style by default, set the font awesome prefix directive. For example, to use the “brand” style by default, you can use:

font awesome prefix: fab

Alternatively, you can use the Material Icons for icon references that are not predefined image files:

default icons: material icons

Note that when you set default icons, no validation is performed to ensure that the image you reference actually exists in Font Awesome or the Material Icons.

Custom favicon in browser tab

The icon in the browser tab is known as a favicon. This is the icon associated with the web application. By default, docassemble uses the docassemble logo for this icon.

If users “pin” your application to their device’s main menu, this icon will be used to create the resulting icon. Microsoft, Apple, and Google have their own conventions for doing this.

In order to “brand” your application in a custom way, create a square logo and go to https://realfavicongenerator.net in order to generate favicon files from it. Upload a square image that is at least 512x512 in size. In addition to uploading your file, you will need to make some choices about different colors that should be used in different circumstances. At the end, you will download a Zip file containing a number of graphics files and other files:

• android-chrome-192x192.png
• android-chrome-512x512.png
• apple-touch-icon.png
• browserconfig.xml
• favicon-16x16.png
• favicon-32x32.png
• favicon.ico
• mstile-150x150.png
• safari-pinned-tab.svg
• site.webmanifest

If any of the above files is missing, your favicon will probably still work in most circumstances. The most important file is favicon.ico, a special graphics file in Microsoft’s ICO format.

To instruct docassemble to use these files for the Favicon, you need to set a directive in the configuration like this:

favicon: docassemble.abcincorporated:data/static

This indicates that docassemble should use the Favicon files located in the data/static folder of the docassemble add-on package called docassemble.abcincorporated, which is installed on your server. For more information about how docassemble uses packages, see the Packages section.

To create a package (called, for example, docassemble.abcincorporated), you can use the Playground. First, upload the Favicon files to the “Static” folder of the Playground. Then go to the “Packages” folder of the Playground and create (“Add”) a new package called abcincorporated. Under “Static files,” select all of the Favicon files you uploaded. Click “Save” to save the package definition, then click the “Install” button at the bottom of the screen to install the package on your server.

If you are creating package outside of the Playground, you can store your Favicon files in a subfolder under data/static. For example, you can create a subfolder called favicon. Then your Configuration would contain:

favicon: docassemble.abcincorporated:data/static/favicon

In addition to providing you with a Zip file containing the above files, the above web site will instruct you to place particular <link> and <meta> tags in the HTML of your site. docassemble does this for you automatically, so you can ignore most of this. However, two of the lines are important. The will look like this:

<link rel="mask-icon" href="/safari-pinned-tab.svg?v=2" color="#698aa7">
<meta name="msapplication-TileColor" content="#da532c">
<meta name="theme-color" content="#83b3dd">

Note the color in the first line and the content in the second and third lines. Then add the following three lines to your configuration, corresponding to the above three lines:

favicon mask color: "#698aa7"
favicon tile color: "#da532c"
favicon theme color: "#83b3dd"

If you do not specify favicon mask color, favicon tile color, or favicon theme color, your custom favicon will still work, but docassemble will choose colors for you. (The colors above are the defaults.)

Note that most web browsers will store a cache of the Favicon, so that when you make a change to your app’s Favicon, it may seem that your change is not working when it actually is. The way that web browsers cache the Favicon is different from the way they cache other resources, so even if you do a forced reload of the page, you will still see the cached Favicon.

docassemble tries to mitigate this problem by appending ?v= followed by the version number of the package referenced in favicon to the URLs for the Favicon files. That way, when you install a new version of your package, web browsers will think that the Favicon has changed, and they will download a new version. (The server actually ignores the v URL parameter, but browsers assume it is significant.) docassemble assumes the version number is the value of the variable __version__ defined in the __init__.py file of the package. The Playground sets this variable when it creates a package, but if you are creating packages on your own, you will need to update this variable manually.

You can also manually specify a v parameter by specifying favicon version:

favicon version: 2

You can then change this number every time you want the web browser to download a new Favicon.

Note that https://realfavicongenerator.net also has an option for setting the version number of your files. You may need to use this as well to force applications to retrieve new icons for your files. This adds, for example, ?v=2 to URLs referenced in the browserconfig.xml and site.webmanifest URLs, much in the same way the above favicon version directive will result in ?v=2 being appended to URLs in the HTML source.

Because of caching, it can be difficult to test changes to your Favicon. One way to check to see if the server is delivering the correct Favicon is to manually visit the Favicon file by entering the /favicon.ico address in your browser location bar, and then forcing a page reload. If you see the correct image in the browser, then the server is delivering the right file, but you are just not seeing it because of a caching issue.

Note that if you have set the root directive (if your docassemble site at https://example.com/something/ instead of https://example.com/), you will need to account for this when you generate the favicon files using the above web site. This is important because the manifest.json and browserconfig.xml files contain URL references. If your site is at /something/, then manifest.json will need to refer to "/something/android-chrome-192x192.png" instead of "/android-chrome-192x192.png".

Controlling the robots.txt file

By default, docassemble discourages web crawlers by returning the following in response to a request for the /robots.txt file:

User-agent: *
Disallow: /

If you want to set a permissive /robots.txt file, for example if you want social meta tags to work, you can set allow robots to True.

allow robots: True

Then, /robots.txt will return:

User-agent: *
Disallow:

If you want to customize the /robots.txt file, leave allow robots unset and set robots to a reference to a file in a package.

robots: docassemble.missouri:data/static/robots.txt

Then the contents of the referenced file will be returned in response to a request for the robots.txt file.

Social meta tags

You can control the meta tags on docassemble pages by setting the social directive.

social:
  description: |
    A site that will solve your legal problems
  image: docassemble.missouri:data/static/logo.png
  twitter:
    description: |
      The greatest legal services site in Missouri.
    site: "@missourilaw"
    creator: "@missourilaw"
    image: docassemble.missouri:data/static/logofortwitter.png
  fb:
    app_id: "1447731792112118"
    admins: "556880902, 547009392"
  og:
    description: |
      The greatest legal services site in Missouri.
    image: https://missourilaw.org/images/missourilaw.jpg

The resulting HTML will include the following in the <head>:

<meta name="description" content="A site that will solve your legal problems">
<meta name="image" content="https://docassemble.missourilaw.org/packagestatic/docassemble.missouri/logo.png?v=1.0.2">
<meta itemprop="name" content="Missouri Law: Interviews">
<meta itemprop="description" content="A site that will solve your legal problems">
<meta itemprop="image" content="https://docassemble.missourilaw.org/packagestatic/docassemble.missouri/logo.png?v=1.0.2">
<meta name="twitter:card" content="summary">
<meta name="twitter:description" content="The greatest legal services site in Missouri.">
<meta name="twitter:site" content="@missourilaw">
<meta name="twitter:creator" content="@missourilaw">
<meta name="twitter:image" content="https://docassemble.missourilaw.org/packagestatic/docassemble.missouri/logofortwitter.png?v=1.0.2">
<meta name="twitter:title" content="Missouri Law: Interviews">
<meta name="og:title" content="Missouri Law: Interviews">
<meta name="og:description" content="The greatest legal services site in Missouri.">
<meta name="og:image" content="https://missourilaw.org/images/missourilaw.jpg">
<meta name="og:url" content="https://docassemble.missourilaw.org/list">
<meta name="og:locale" content="en_US">
<meta name="og:site_name" content="Missouri Law">
<meta name="og:type" content="website">
<meta name="fb:app_id" content="1447731792112118">
<meta name="fb:admins" content="556880902, 547009392">

In this example, the server is https://docassemble.missourilaw.org, the brandname is Missouri Law, the version of the docassemble.missouri package is 1.0.2, and the page that was accessed was the list of available interviews (/list).

The image references are special because if you set them to a reference to a static file in a package, they will be replaced with a full URL to that file. Alternatively, you can provide a full URL.

Note that the itemprop="name", twitter:title, and og:title fields were populated automatically. Since the title of the page varies from page to page, you can’t set this value in the Configuration. For administrative pages, these fields are set to the brandname followed by a colon, followed by the name of the page. Note that these page names are customizable. For example, to change the name from “Interviews” to “Get Started,” you could set start page title: Get Started.

Likewise, og:url field is automatically set to the URL of the page, and this cannot be changed.

By default, twitter:card is set to summary, og:site_name is set to the value of brandname, og:locale is determined from the locale, and og:type is set to website. These can be specifically overridden.

The social directive sets server-wide defaults. Administrative pages will reflect these tags. On interview pages, these defaults can be overridden using the social specifier under the interview metadata.

Note that by default, the docassemble server disallows web crawling by returning a restrictive /robots.txt file. That means that as a practical matter, sites will not be able to consume your meta tags. The /robots.txt file can be customized so that your meta tags are accessible.

Icon size

The size of icons in decorations and in image buttons is not specified in CSS. To change these sizes, you can edit these configuration directives:

decoration size: 2.0
decoration units: em
button icon size: 4.0
button icon units: em

Exit page

The exitpage directive contains the default URL to which the user should be directed after clicking a button that runs an exit or leave command. (See the Setting Variables section.)

For example:

exitpage: https://example.com/pages/thankyou.html

Page after logout

The logoutpage directive contains the default URL to which a logged-in user should be directed after logging out. By default, the user is redirected to the login page after logging out.

logoutpage: https://example.com/pages/comeagain.html

SQL database

The db section of the Configuration indicates where docassemble can find the SQL database in which to store users’ answers, login information, and other information.

The db directives should be treated as read-only. The primary way you would adjust your database configuration is when starting a new server for the first time. At that time, you can specify that a particular SQL database should be used. For example, you would change the default configuration if you were using a cloud-based managed database system like RDS.

The password should only contain alphanumeric characters. The special characters _, ~, /, -, ^, *, ?, and ! may be used.

Changing the db directives after you already have a working system can easily lead to an unbootable system. You should not change the values using the web-based front end unless you really know what you are doing.

db:
  prefix: postgresql+psycopg2://
  name: docassemble
  user: docassemble
  password: abc123
  host: localhost
  port: 5432
  table prefix: null
  backup: True

The prefix is a SQLAlchemy prefix. If you use a database other than PostgreSQL, change this.

docassemble will connect to the SQL database at the hostname host on the port port, and will authenticate with the user and password. It will connect to the database called name. If you want separate docassemble systems to share the same database, you can set a table prefix.

The value of backup is only applicable if you are using the jhpyle/docassemble Docker image in a multi-server configuration and your SQL database is external. If backup is true (which is the default), then the SQL database will be backed up on a daily basis to the rolling backup. You will want to set it to false if backing up the SQL database could lead to the exhaustion of hard drive space. If you are using a managed SQL server that has its own backup feature, you will probably want to set backup to false.

If you want to connect to a PostgreSQL server using an SSL certificate, you can use the ssl mode, ssl cert, ssl key, and/or ssl root cert:

db:
  prefix: postgresql+psycopg2://
  name: docassemble
  user: docassemble
  password: abc123
  host: someserver.example.com
  port: 5432
  ssl mode: require
  ssl cert: postgresql.crt
  ssl key: postgresql.key
  ssl root cert: postgresql_root.crt

The possible values of ssl mode are disable, allow, prefer,require, verify-ca, and verify-full. This corresponds with the sslmode parameter of libpq.

The ssl cert, ssl key, and ssl root cert parameters correspond with the sslcert, sslkey, and sslrootcert parameters of libpq. These parameters must refer to the names of files present in the docassemble certificates directory. If you are using S3 or Azure Blob Storage, these files are stored under certs/; otherwise, the files must be present in /usr/share/docassemble/certs/ at the time the system initializes (from there, they will be copied into other directories). For instructions on loading your own certificates when using the jhpyle/docassemble Docker image, see managing certificates with Docker. See also the certs Configuration directive.

After changing the db configuration, you need to restart the server in order for the change to take effect.

Avoiding SQL errors

If your docassemble server runs in an environment in which persistent SQL connections will periodically be severed, you can set sql ping to True in order to avoid errors.

sql ping: True

There is an overhead cost to using this, so only enable this if you get SQL errors when trying to connect after a period of inactivity.

SQL database for storing snapshots of interview answers

You can use the store_variables_snapshot() function to save unencrypted interview answers in JSON format to rows in a SQL database. By default, the SQL database that store_variables_snapshot() uses is the same as the database configured with db. However, you can use a separate database by specifying a variables snapshot db in your Configuration.

variables snapshot db:
  name: snapshots
  user: docassemble
  password: abc123

The format is the same as that of db. It is recommended that you use PostgreSQL for the database that stores variable snapshots, because if you do, then docassemble will use the JSONB data type for the JSON version of the interview answers, which will enable fast and convenient querying of the data.

Branding

The appname and brandname directives control the name of the application in various contexts.

appname: Legal Helper
brandname: Legal Helper Application

The appname is used in e-mails that are generated by the user login system.

The appname defaults to docassemble. The brandname will default to the appname if brandname is not specified.

Default name in navigation bar

The default title and default short title directives control the names that are displayed in the web browser tab and the navigation bar of interviews that do not specify these titles in their metadata.

The “short” title appears on mobile devices, while the long title appears on larger screens.

default title: Abramsom Baker and Calloway, LLP
default short title: ABC LLP

If not specified, the default title will default to the brandname.

Uploads directory

The uploads directive indicates the directory in which uploaded files are stored.

uploads: /netmount/files/docassemble/uploads

If you are using a multi-server arrangement and not using S3, S3-compatible object storage, or Azure blob storage, this needs to point to a central network drive.

The default value is /usr/share/docassemble/files.

Python directory

The packages directive indicates the base directory into which docassemble extension packages will be installed. The PYTHONUSERBASE environment variable is set to this value and pip is called with --install-option=--user. When Python looks for packages, it will look here. Python’s default value is ~/.local, but docassemble changes it to the value of this directive, or /usr/share/docassemble/local if the directive is not defined.

packages: /netmount/files/docassemble/local

If you are using a multi-server arrangement and not using S3, S3-compatible object storage, or Azure blob storage, this needs to point to a central network drive.

Path to WSGI application

The webapp directive indicates the path to the WSGI file loaded by the web server.

webapp: /netmount/files/docassemble/webapp/docassemble.wsgi

docassemble needs to know this filename because the server needs to reset itself after an add-on package is updated. The web server is reset by updating the modification time of the WSGI file.

The default value is /usr/share/docassemble/webapp/docassemble.wsgi.

Path to SSL certificates

The certs directive indicates a central location where SSL certificates can be found. The location can be a file path, an S3 path (beginning with s3://), or an Azure blob storage path (beginning with blob://). In most cases, you should leave this directive alone.

You might keep your certificates on a network drive:

certs: /netmount/files/docassemble/certs

When the server is running, it expects certificates to reside in /etc/ssl/docassemble. For example, by default, the NGINX HTTPS configuration contains:

ssl_certificate /etc/ssl/docassemble/nginx.crt;
ssl_certificate_key /etc/ssl/docassemble/nginx.key;

The supervisor process executes the docassemble.webapp.install_certs module during the startup process. This module copies certificates from the location indicated by certs to /etc/ssl/docassemble before starting the server.

The value of certs can be a file path, an S3 URI (e.g., s3://exampledotcom/certs), or an Azure blob storage URI (e.g., blob://youraccountname/yourcontainername/certs). The contents of the directory are copied to /etc/ssl/docassemble.

If you leave the certs setting undefined (which is recommended), docassemble will look in /usr/share/docassemble/certs if the s3 and azure settings are not enabled.

If you are using the jhpyle/docassemble Docker image and you are not using S3 or Azure blob storage, you can create a volume, copy your certificates there, and then mount that volume as /usr/share/docassemble/certs when you execute docker run. For more information on how to do this, see managing certificates with Docker.

If s3 is enabled, docassemble will look for S3 keys with the prefix certs/ in the bucket defined in the s3 configuration.

If azure is enabled, it will look for Azure blob storage objects with the prefix certs/ in the container defined in the azure configuration.

You can upload certificate files to S3 or Azure blob storage using the web interface. It is also possible to copy certificate files to S3 from the command line. For example, if you are using Linux locally, install s4cmd (if you have not done so already):

apt-get install s4cmd

Then do:

s4cmd --access_key=YOURACCESSKEY --secret_key=YOURSECRETKEY put yourserver.crt s3://yourbucket/certs/nginx.crt
s4cmd --access_key=YOURACCESSKEY --secret_key=YOURSECRETKEY put yourserver.key s3://yourbucket/certs/nginx.key

If your s3 configuration has bucket: yourbucket, then you do not need to set a certs configuration, because docassemble will by default look in s3://yourbucket/certs. However, if the certificates are stored in another location, you can specify a different location:

certs: s3://otherbucket/certificates

E-mail configuration

docassemble needs to send e-mail, for example to reset people’s passwords, or to let users of a multi-user interview know that it is their turn to start answering questions.

Before you set up your e-mail configuration, make sure that you are not using the default fake e-mail address of [email protected] as the e-mail address of the administrative user. If you forget your password, you will want to be able to reset it using the “forgot my password” feature, which relies on e-mailing an address you can access.

If you are going to send mail, you should at least set the default sender directive, which sets the return address on any e-mails generated from docassemble:

mail:
  default sender: '"Administrator" <[email protected]>'

In order to send e-mail, you will need to sign up with an e-mail sending service. You can use an external SMTP server, or you can use an e-mail service with an HTTP-based API. The latter is recommended because SMTP is slow.

Using an external SMTP server

To use another SMTP server as the mail server, do something like:

mail:
  default sender: '"Administrator" <[email protected]>'
  username: mailuser
  password: abc123
  server: smtp.example.com
  port: 25
  use ssl: False
  use tls: True

If you are hosting docassemble in the cloud, you will probably have to use a separate SMTP server in order to send e-mail.

A free option is Mailgun. You can sign up with Mailgun, provide a credit card, configure some DNS entries, and then set your configuration to something like this:

mail:
  default sender: '"Example, Inc." <[email protected]>'
  username: '[email protected]'
  password: '5a4a0c5f3da35f3bc10f0462364c26dd'
  server: 'smtp.mailgun.org'

Using the Mailgun API

Another way to send e-mail using Mailgun is through its API. If you use the API, mail is sent to Mailgun using HTTP, rather than SMTP. This may be more reliable than SMTP, since internet service providers may slow down or block SMTP traffic as a way of protecting against spam.

To use this method, obtain an API key from Mailgun and set it as the mailgun api key. Also set your Mailgun domain as the mailgun domain.

mail:
  default sender: '"Example, Inc." <[email protected]>'
  mailgun api key: key-b21b28f6e29be1f463478238d172813e
  mailgun domain: mg.example.com

If necessary, you can also configure the URL used by the Mailgun API. The default is https://api.mailgun.net/v3/%s/messages.mime.

mailgun api url: https://api.mailgun.net/v3/%s/messages.mime

(The mailgun domain will be substituted in place of the %s.)

Mailgun domains created in their EU region have a different base URL. To cater for this, you can configure the URL used by the Mailgun API as follows:

mailgun api url: https://api.eu.mailgun.net/v3/%s/messages.mime

Using the SendGrid API

You can send e-mail using the SendGrid API. First, sign up for SendGrid and obtain an API key. Under mail in your Configuration, set sendgrid api key to the API key.

mail:
  default sender: '"Example, Inc." <[email protected]>'
  sendgrid api key: s8_MK.S.GmtjGD6krnOl4FkkPGTYe3I8fHqIi2NGZ057k7S-ZhuQfOF2ItWYBW97w6Kzu

Using the Graph API on Microsoft Azure

You can send email using the Microsoft Graph API on Microsoft Azure.

• Log into the Azure Portal.
• Go to App Registrations.
• Create a New Registration. Give it a name like “Email sending” or whatever you want. Setting up a Redirect URI is not necessary.
• Note the “Application (client) ID.” This will need to go into your mail configuration as the azure client id.
• Note the “Directory (tenant) ID.” This will need to go into your mail configuration as the azure tenant id.
• Go into API Permissions and add a permission. Select Microsoft Graph, then Application Permissions, then Mail.Send. Click “Grant admin consent.”
• Go into Certificates & Secrets. Add a new client secret. Call it “Email sending secret” or whatever you want. Copy the “Value.” This will need to go into your mail configuration as the azure client secret.

Then edit your Configuration and add the codes you obtained. For example:

mail:
  default sender: '"Example, Inc." <[email protected]>'
  azure client id: "72f3829c-f887-8273-b716-9827e727b91a"
  azure tenant id: "82392784-bced-7ee5-94fa-7326347873c4"
  azure client secret: "oj238~2jo23rj02Y3r72y9~u203r9u27yr-82082"
  default sender: '"Philadelphia Legal Assistance" <[email protected]>'

You will only be able to send mail from senders whose mailboxes your App Registration has privileges of accessing.

Using multiple mail configurations

You can configure multiple mail configurations and then choose which configuration you want to use for a particular interview or when you call send_email().

For example, suppose that you wanted to send e-mail from an account belonging to ABC, Inc. in one interview and send e-mail from an account belonging to DEF, Inc. in another interview.

You could set your configuration as follows:

mail:
  - default sender: '"ABC, Inc." <[email protected]>'
    sendgrid api key: s8_MK.S.GmtjGD6krnOl4FkkPGTYe3I8fHqIi2NGZ057k7S-ZhuQfOF2ItWYBW97w6Kzu
    name: abc
  - default sender: '"DEF, Inc." <[email protected]>'
    mailgun api key: key-b21b28f6e29be1f463478238d172813e
    mailgun domain: mg.example.com
    name: def

Then, in the interview that should use ABC, Inc.’s server, you would set the email config to abc

metadata:
  email config: abc

In the interview that should use DEF, Inc.’s server, you would set the email config to def

metadata:
  email config: def

You could also specify the e-mail configuration in a call to send_email():

send_email(to=recipient, sender=sales_associate, template=offer_template, config='def')

When you are using multiple e-mail configurations, keep in mind that docassemble sometimes sends e-mail outside of the interview context, such as user invitation e-mails, error notifications, forgot password e-mails, etc. In these situations, docassemble will use the default configuration, which is the configuration tagged with name: default, or if no configuration has default as the name, the first configuration will be used.

Default interview

If no interview is specified in the URL when the web browser first connects to the docassemble server, the interview indicated by default interview will be used. The interview needs to be specified in “package name:relative file path” format. For example:

default interview: docassemble.base:data/questions/examples/attachment.yml

If this directive is not set, docassemble will redirect to a page showing a list of available interviews determined by the dispatch directive. However, if the dispatch directive is not defined, users will be directed to a placeholder interview, docassemble.base:data/questions/default-interview.yml.

Log format

Log messages written to the log files will by default contain the IP address, interview file, session ID, and user e-mail address, in addition to the log message itself.

The format of the log message is determined by the log format, the default value of which is:

log format: docassemble: ip=%(clientip)s i=%(yamlfile)s uid=%(session)s user=%(user)s %(message)s

If you would like to shorten the log messages, you could change it to something like:

log format: docassemble: i=%(yamlfile)s %(message)s

Flask log file location

docassemble uses the Flask web framework. The flask log and celery flask log directives contain the paths to the Flask log files for the web application and the Celery background workers, respectively. Most errors write to the docassemble.log file located in /usr/share/docassemble/log (or the directory indicated by log, or to standard web server error logs, but there are some errors, typically those that are Flask-related, that will only write to these log files.

flask log: /tmp/docassemble-log/flask.log
celery flask log: /tmp/docassemble-log/celery-flask.log

The default locates for the Flask log files are /tmp/flask.log and /tmp/celery-flask.log.

Log messages during the startup process

If the web application is slow to startup after a restart (i.e., after the Configuration is saved, or after a package is installed, you can set debug startup process to True in order to see log messages in the uwsgi.log log file showing the timing of various parts of the web application startup process.

debug startup process: True

The log should look something like this:

0.000s server: starting
0.071s config: load complete
4.179s backend: starting
4.179s backend: getting email configuration
4.179s backend: finished getting email configuration
4.179s backend: configuring common functions
4.179s backend: finished configuring common functions
4.179s backend: processing translations
4.840s backend: finished processing translations
4.840s backend: obtaining cloud object
4.840s backend: finished obtaining cloud object
4.840s backend: completed
4.892s server: done importing modules
4.896s server: creating session store
4.896s server: setting up Flask
4.903s server: finished setting up Flask
4.903s server: setting up logging
4.903s server: finished setting up logging
4.903s server: getting page parts from configuration
4.903s server: finished getting page parts from configuration
5.776s server: making directories that do not already exist
5.776s server: finished making directories that do not already exist
5.776s server: building documentation
6.218s server: finished building documentation
6.218s server: entering app context
6.218s server: entering request context
6.235s server: loading preloaded interviews
6.458s server: finished loading preloaded interviews
6.458s server: copying playground modules
6.504s server: finished copying playground modules
6.504s server: deleting LibreOffice macro file if necessary
6.504s server: fixing API keys
6.504s server: starting importing add-on modules
7.490s server: finished importing add-on modules
7.490s server: running app

The period of four seconds at the start is taken up by loading a large number of necessary Python modules. If your system takes much longer than four seconds to reach backend: starting, your system may have insufficient CPU or memory.

The time between processing translations and finished processing translations can be reduced by removing unnecessary lines from the words configuration.

The time between finished getting page parts from configuration and making directories that do not already exist consists of loading necessary Python functions.

The time between starting importing add-on modules and finished importing add-on modules can be reduced by uninstalling unnecessary add-on packages, adding # do not pre-load to the top of module files that should not load when the system starts, and removing files from the Modules folders of users’ Playgrounds.

If you want to see logs of the initial system startup process (the process that happens when you do docker run, see the log files mentioned in the Troubleshooting section, in particular the log files at /var/log/supervisor/initialize*.

Check the worker.log file for information about package installation.

Default language, locale, dialect, and voice

These directives set the default language and locale settings for the docassemble server.

language: en
locale: en_US.utf8
dialect: us
voice: Amy

The language needs to be a lowercase ISO-639-1 or ISO-639-3 code.

language controls the default language, which is the language that will be used when:

• the user’s browser requests a language that the docassemble server does not support, or
• docassemble code is running in a context where it has not been told what language to use.

The languages that docassemble supports are determined by the words configuration. docassemble always supports English; whether other languages are supported depends on whether words files are included that provide translations for system phrases. So if you change the language to es, you will also need to edit your words configuration so that it includes a translation file for Spanish words.

You can control the language used in an interview session by calling the set_language() function in an initial block. You can write interviews in such a way that they support multiple languages. For more information, see the the language support section of the documentation.

The locale needs to be a locale name that will be accepted by the locale library. The locale is primarily used for determining the formatting of numbers and currency values. For example, some locales use . to indicate decimals and some use ,.

The locale can be changed later in Python code by calling set_locale() followed by update_locale(). However, changing the locale is is not thread-safe; this means that if two users are using your system simultaneously, and one is in Japan and expects currency to be in Yen, and the other is in France and expects currency to be in Euros, the code that runs update_locale() for the French user could cause the Japanese user to see the Euro symbol in place of the Yen symbol in a question.

The dialect and voice are only relevant for the text-to-speech feature, which is controlled by the special variable speak_text. See the voicerss configuration for more information about configuring this feature. The default dialect and default voice are only used as a fallback, in case the dialect and voice cannot be determined any other way. It is better to set the dialect and voice using set_language() or the voicerss configuration.

Default country

The country directive sets the default country for docassemble.

country: US

The country is primarily relevant for interpreting telephone numbers. The code needs to be a valid country code accepted by the phonenumbers library (ISO 3166-1 alpha-2 format).

Operating system locale

Applicable only to the jhpyle/docassemble Docker image.

If you are using the jhpyle/docassemble Docker image, the os locale directive will set the default operating system locale.

os locale: so_SO.UTF-8 UTF-8

The os locale must exactly match one of the locale values that Ubuntu/Debian recognizes. The default value is en_US.UTF-8 UTF-8, which is appropriate for servers in the United States.

When the server starts, the value of os locale is appended to /etc/locale.gen (if not already there in uncommented form) and locale-gen and update-locale are run.

After changing os locale, you will need to restart your container (docker stop -t 600 <container ID> followed by docker start <container ID>).

Other available locales

Applicable only to the jhpyle/docassemble Docker image.

If your interviews use locale and language settings that your operating system does not support, you will get an error.

If you are using the jhpyle/docassemble Docker image, you can enable locales other than the os locale in the operating system by providing a list of locales to the other os locales directive:

other os locales:
  - en_GB.UTF-8 UTF-8
  - es_MX.UTF-8 UTF-8

Each other os locales value must exactly match one of the locale values that Ubuntu/Debian recognizes (listed in the /etc/locale.gen file on an Ubuntu/Debian system).

When the server starts, each of the other os locales is appended to /etc/locale.gen (if not already there in uncommented form) and locale-gen and update-locale are run.

After changing other os locales, you will need to restart your container (docker stop -t 600 <container ID> followed by docker start <container ID>).

E-mail address to which error messages shall be sent

If error notification email is set to an e-mail address or a list of e-mail addresses, then if any user sees an error message, the server will try to send an e-mail notification to this e-mail address (or addresses). If possible, the e-mail will contain the error message.

error notification email:
  - [email protected]
  - [email protected]

Information about errors is also available in the Logs.

Suppressing notifications about certain types of errors

If you receive error notifications that are not helpful, you can suppress errors by the class of error by defining the suppress error notificiations directive:

suppress error notificiations:
  - ZeroDivisionError
  - ReadTimeout

This means that if the name of the class of error is ZeroDivisionError or ReadTimeout, no notification will be sent.

Redirecting user in case of session error

If the user tries to access a non-existent session, the default behavior is to start a new session in the interview and display an error message to the user. If you would rather redirect the user to a different URL, you can set session error redirect url to the URL to which you want to redirect users in this circumstance.

session error redirect url: https://example.com/launch_page.html

You can set the session error redirect url to a template into which the named parameters i and error can be inserted. The i parameter refers to the URI-encoded interview filename, and the error refers to an error message (also safe for inclusion in a URL parameter), which will be either answers_fetch_fail (in case there was an error retrieving the interview answers, such as a decryption error) or answers_missing (in case no session could be found).

session error redirect url: https://example.com/error_page?filename={i:s}&error_code={error:s}

Whether error messages contain details

By default, when the user sees an error on the screen, information about the error is shown. To turn this off, set verbose error messages to False.

verbose error messages: False

If you set verbose error messages, you should also set error help in your interviews or set the error help directive.

Whether to include interview variables in error notification

If error notification variables is set to true, then when an error notification e-mail is sent, the interview variables will be attached as a JSON file.

The default value is the value of debug; that is, the variables will be included if the server is a development server, but not included if the server is a production server.

Putting the interviews variables into a file and e-mailing them lowers security, especially when server-side encryption is being used.

error notification variables: True

In the case of some errors, obtaining the interview variables is not possible; in that case, the variables will not be attached.

Help text to display when there is an error

When an error happens, the user will see the error message on the screen, which can be confusing. You can add some additional content to this screen so that the user sees something other than the error message:

error help: |
  We are sorry, our computer has malfunctioned. Please call
  202-555-1515 for assistance, or visit our [web site](https://example.com).

The format of the error help is Markdown.

You can also set this on a per-interview basis using error help inside the metadata.

ubuntu packages to install

Applicable only to the jhpyle/docassemble Docker image.

When using the jhpyle/docassemble Docker image, you can ensure that particular Ubuntu packages are installed by providing a list of packages to the ubuntu packages directive:

ubuntu packages:
  - r-base
  - r-cran-effects

These packages will be installed when the Docker container starts.

For backwards compatibility, the directive debian packages works the same way.

Non-idempotent question check

You can set allow non-idempotent questions: False in order to enforce strict idempotency of questions with generic objects, iterators, or multiple choice selections. If an interview has non-idempotent logic, there is a risk that variables will be set improperly.

The allow non-idempotent questions directive currently defaults to True, but in the future, it will default to False, which may break some interviews.

To make sure you are ready for this change, set the following on your development server:

allow non-idempotent questions: False

Then test your interviews to see if this message is shown in the screen at any point:

Input not processed because the question changed. Please continue.

A message will also be written to the logs beginning with index: not the same question name.

If these messages appear, take a close look at the logic of your interview.

Users may see this message if:

• The interview logic is non-idempotent, meaning that the user sees one question, but if they immediately refresh the screen, they see a different question.
• They have the interview open in more than one device, browser, or browser tab, and they proceed with the interview in one tab and then try to proceed with the interview in the other;
• You have a multi-user interview and two users have the interview open at the same time, and the actions of one user change the interview logic for the other user.

If your interviews work with allow non-idempotent questions set to False, ist is recommended that you set allow non-idempotent questions to False on your production server.

If a user sees this message, they can usually proceed with the interview normally. The input they submitted before seeing the message is discarded, but the problem should not be a permanent problem. However, if the user continues to see the message, and it prevents the user from proceeding, there is a flaw in the interview logic that needs to be fixed; an immediate solution would be to set allow non-idempotent questions to True, but the long-term solution is to fix the logic in the interview.

Restricting the browser from setting arbitrary variables

By default, the docassemble front end is designed to be flexible for JavaScript developers. A POST request to /interview is capable of setting arbitrary variable names in the interview answers; the field names are simply base64-encoded variable names. This allows JavaScript developers to build their own front ends or heavily customize the standard front end.

However, when this kind of flexibility exists on the JavaScript side of the application, you need to trust that your users will not try to manipulate their interview answers. For example, if you have a variable in your interview called user_has_paid, you wouldn’t want users to manipulate their web browser and set user_has_paid to True.

If you enable restrict input variables, then the web browser will only be able to set variables that are present on the current question. What question counts as the current question depends on the interview logic. When the browser submits input, then before any variables are set, the interview logic is evaluated and the current question is determined. If the incoming variables are not part of the current question, then the input is rejected.

restrict input variables: True

If you enable restrict input variables, then your interview logic must be idempotent for all questions. Note that the allow non-idempotent questions directive (despite its general-sounding name) only enforces idempotency for questions with generic objects, iterators, or multiple choice selections. The restrict input variables is more strict; it enforces idempotency for all questions and also prevents the browser from altering the variable names or data types.

Enabling restrict input variables: True is generally a good practice. If you want to add additional fields to a question using JavaScript, you can use the allowed to set modifier to enable those field names.

Python packages to install

Applicable only to the jhpyle/docassemble Docker image.

When using the jhpyle/docassemble Docker image, you can ensure that particular Python packages are installed by providing a list of packages to the python packages directive:

python packages:
  - stripe
  - slackclient

These packages will be installed when the Docker container starts.

The python packages directive should not be used to control the installation of packages on a machine that is part of an existing setup (i.e., it is connecting to a populated SQL database). If you have an existing system, the way to control package installation is through the Package Management interface or the API.

Default administrator account credentials

When a docassemble SQL database is first initialized, an administrative user is created. By default, this user has the e-mail address [email protected] and the password password. As soon as the web server comes on-line, you can log in and change the e-mail address and password to something more secure.

However, you can also use the default admin account directive in the configuration so that a more secure e-mail address and password are used during the setup process.

default admin account:
  email: [email protected]
  password: 67Gh_Secret_2jx

You can also specify the first and last name of the admin user:

default admin account:
  email: [email protected]
  password: 67Gh_Secret_2jx
  first_name: Roger
  last_name: Jones

The settings are only used by the docassemble.webapp.create_tables module during the initial setup process. Using the information defined here, that script sets up a single account in the user login system with “admin” privileges.

You can also use default admin account to initialize an API key:

default admin account:
  email: [email protected]
  password: 67Gh_Secret_2jx
  api key: Dbah81njFzdMLw07hnU2TBYA4eClcRw1

The API key must be exactly 32 characters long. It will be owned by the default admin user and will have no constraints on IP address or Referer.

After create_tables runs for the first time, you can (and should) delete the default admin account information from the configuration file.

Whether the administrator can delete user accounts

By default, administrators can edit user accounts and mark them inactive. If an account is inactive, the user cannot log in and the user cannot register using the same e-mail address. Marking an account as inactive does not delete the user’s data.

When administrators edit user accounts, they can also delete the user’s account. Since the deletion of the information is permanent, you might want to turn off this feature for safety reasons. To do so, set admin can delete account to False:

admin can delete account: False

Whether users can delete

their accounts

By default, users have the power to delete their own accounts and all of their data. To disable this, you can set user can delete account to False:

user can delete account: False

If users have joined multi-user interviews with other people, it is ambiguous who “owns” the data in the interview. By default, shared interviews are not deleted from the server when users delete their data. However, if you want to give users the power to delete any session of a multi-user interview, even if other people used it, you can set delete account deletes shared to True.

delete account deletes shared: True

Whether users can request developer accounts

By default, if enable playground is not set to False, then users can submit a request for a developer account on the user profile page. To prevent users from requesting developer accounts, you can set user can request developer account to False:

user can request developer account: False

Database table upgrades

By default, docassemble uses alembic to upgrade the SQL database from one version to another. To turn this off, you can add:

use alembic: False

Secret key for Flask

The Flask web framework needs a secret key in order to manage session information, provide protection against cross-site request forgery, and to help encrypt passwords. Set the secretkey to a random value that cannot be guessed.

secretkey: CnFUCzajSgjVZKD1xFfMQdFW8o9JxnBL

The startup process on the jhpyle/docassemble Docker image sets the secretkey to a random value.

After you already have a working system, you should not change the secretkey. If you do, none of your passwords or API keys will work.

Number of days of backups to keep

Applicable only to the jhpyle/docassemble Docker image.

The Docker setup creates 14 days of daily backups. The number of days can be changed using the backup days directive.

backup days: 7

If you set backup days to 0, the daily backup process will be disabled.

Whether to backup file storage

Applicable only to the jhpyle/docassemble Docker image.

By default, files that users upload and files that are generated by the system (e.g., uploaded files, DAFile files, document assembly results, etc.) are backed up in daily snapshots. However, this can take up a lot of space, and having snapshots of files may not be important to you. To exclude these files from the daily backups, set backup file storage to false.

backup file storage: false

Secret key for passwords

The password secretkey is used in the process of encrypting interview answers for users who log in using third-party logins. It defaults to the same value as secretkey. If this value changes, users who log in through third-party logins will not be able to resume stored interviews.

Allowing referer blocking

By default, when HTTPS is used, docassemble uses a form of CSRF protection that checks whether the referer header matches the host. For privacy reasons, some users set their browsers to disable the sending of the referer header. If the referer header is missing, the CSRF protection mechanism will generate an error. To bypass this security feature, set:

require referer: False

The default setting is True.

Limiting size of HTTP requests

By default, as a security precaution, the web application rejects any HTTP request larger than 16 megabytes. If you encounter this limit, you may see an error like “413 Request Entity Too Large.”

This limit can be extended by setting maximum content length to a different number of bytes.

maximum content length: 26214400

This sets the limit to 25 megabytes.

To have no limit, set maximum content length to None.

The maximum content length applies at the Python level and the web server level. If you are using the jhpyle/docassemble Docker image, then after changing this value, you will need to do a complete restart of the system for the change to take effect.

Scaling factor for thickness of signature line

If you think the line thickness of users’ signatures is either too thick or too thin, you can set the signature pen thickness scaling factor. If you want a 10% thicker line, set signature pen thickness scaling factor to 1.1. If you want a 50% thinner line, set it to 0.5.

signature pen thickness scaling factor: 0.5

Image conversion resolution

When users supply PDF files and docassemble includes those files within a document, the PDF pages are converted to PNG images in order to be included within DOCX and RTF files. The png resolution directive defines the dots per inch to be used during this conversion.

png resolution: 500

PDF files are also converted to PNG for previewing within the web app, but at a lower resolution. The png screen resolution directive defines the dots per inch to be used for conversion of PDF pages to PNG files for display in the web browser.

png screen resolution: 72

When converting page images, the actual DPI used will be scaled based on the page size. If the long edge of the page is 11 inches and the png resolution is 500, then the DPI will be 500. However if the long edge of the page is 22 inches, the DPI will be 250.

OCR resolution

If you use the ocr_file() function, the pages of the PDF file will be converted to images before being read by the OCR engine. By default, the resolution used is 300 dpi. To change this to something else, set the ocr dpi directive:

ocr dpi: 500

The actual DPI used will be scaled based on the page size. If the long edge of the page is 11 inches and the ocr dpi is 500, then the DPI will be 500. However if the long edge of the page is 22 inches, the DPI will be 250.

Controlling whether registering users need to retype their passwords

By default, users when registering must type in their passwords twice. To allow users to register after only typing the password once, you can set the retype password directive to False.

retype password: False

Controlling password complexity

By default, users when registering must select a password that is at least six characters long and that contains at least one lower case letter, at least one upper case letter, and at least one digit character. You can alter these requirements using the password complexity directive.

password complexity:
  length: 8
  uppercase: 0
  lowercase: 1
  digits: 2
  punctuation: 1

In this example, passwords must be eight characters in length, must contain at least one lower case character, must contain at least two digit characters, and must contain at least one punctuation character.

If you leave out any one of these directives, the default value for that setting will be used.

Changing the password complexity directive has no effect on users who have already set their passwords.

If a user tries to set a password that does not meet the password complexity requirements, an error message will be displayed that describes what the password complexity requirements are. You can manually override this error message by setting the error message directive within the password complexity directive to what you want the error message to be.

Changing the style of the login link

By default, when a user is not logged in, there is a link in the upper right corner labeled “Sign in or sign up to save answers.” You can change the style of this by setting login link style to button:

login link style: button

Then, the user will instead see a “Sign up” link and a “Sign in” button. The Bootstrap color of the “Sign in” button can be changed using button colors.

Hiding the login link

If the show login directive is set to False, users will not see a “Sign in or sign up to save answers” link in the upper-right-hand corner of the web app.

show login: False

The default behavior is to show the “Sign in or sign up to save answers” link. If show login is False, users will see an “Exit” option. The appearance and behavior of the “Exit” option is configurable in the interview metadata using the directives exit link and exit label. There is also a show login metadata directive that contols this feature on a per-interview basis.

Hiding the profile link

If the show profile link directive is set to False, logged-in users will not see a “Profile” link in the web app menu. Instead, they will see a “Change Password” link (unless allow changing password is false).

show profile link: False

Note that users with admin or developer privileges will always see the “Profile” link.

Allowing users to change their passwords

If allow changing password is set to False, then only users with admin or developer privileges will be able to change their passwords.

allow changing password: False

Fields available to users when registering

By default, logged-in users without privileges of admin or developer can only edit their first and last names on their user profiles, and when they register they can only set their e-mail address and password.

If you would like to expand the fields available to the user on the registration page and in the user profile, you can set user profile fields to a list of fields that you wish to be editable.

user profile fields:
  - first_name
  - last_name
  - country
  - subdivisionfirst
  - subdivisionsecond
  - subdivisionthird
  - organization
  - timezone
  - language

Although you can expose the language field to the user for editing, this is not recommended because the language field is a special code. The field is not a multiple-choice list; developers can use any code of their choosing to represent a language or a dialect of a language. You can use an interview to ask the user for their language and then set the field in the user profile using set_user_info().

Hiding the “my interviews” link

If the show interviews link directive is set to False, logged-in users will not see a My Interviews link in the web app menu.

show interviews link: False

Note that users with admin privileges will always see the “My Interviews” link.

Showing the “available interviews” link

If the show dispatch link directive is set to True, and you have configured the dispatch directive, then logged-in users will see an “Available Interviews” link in the web app menu, which will direct them to the /list page of your site.

show dispatch link: True

By default, this is not shown in the web app menu.

Invitation-only registration

If the allow registration directive is set to False, users will not be allowed to register to become users of the site unless they are invited by an administrator.

allow registration: False

The default behavior is to allow any user to register.

Password reset

By default, there is a “Forgot your password?” link on the login page that allows a user to reset their password after receiving an e-mail at their e-mail address. You can disable this feature with the allow forgot password directive:

allow forgot password: False

Note that by default, interview answers are encrypted with the user’s password. If the user forgets their password and resets it with the “Forgot your password?” feature, they will not be able to access their past interview sessions (unless multi_user is set to True in the interview answers, which disables encryption). By contrast, when a user changes their password, their sessions are re-encrypted with the new password, and the user does not lose access to their sessions.

Restricting e-mail/password registration to particular domains

The authorized registration domains directive can be used to restrict registraion to one or more e-mail domains.

authorized registration domains:
  - abclegal.org
  - missourilegalhelp.org

This directive only applies to username/password registration.

E-mail confirmation after registration

By default, users who register can start using the site right away. To require e-mail validation first, set confirm registration to True:

confirm registration: True

This feature enables enhanced privacy. If someone tries to register using the e-mail of someone who has already registered, the fact that the e-mail address is already registered is not revealed.

E-mail confirmation for certain privileges

If you are using confirm registration but you want to verify e-mail addresses only for specific user privileges, you can set the email confirmation privileges directive to a list of privileges for which e-mail confirmation should be a requirement of logging in.

confirm registration: True
email confirmation privileges:
  - admin
  - developer
  - advocate

By default, e-mail confirmation is not required for any privileges (i.e., the email confirmation privileges is an empty list. Also, the default administrative user is confirmed when initially created, so you will never need to confirm that user’s e-mail.

Before you enable this feature, make sure you have a working email configuration.

Suppressing alerts during login and logout

By default, the Flask-User package flashes an alert saying “You have signed in successfully,” “You have signed out successfully,” and “You have registered successfully.” To suppress these messages, set suppress login alerts to true.

suppress login alerts: True

LDAP login

If you want to connect the docassemble login system to a local LDAP (Active Directory) server, add an ldap login section to the Configuration.

At a minimum, the following directives must be set:

ldap login:
  enable: True
  server: 192.168.1.124

If ldap login is enabled, then when a user tries to log in, the user’s e-mail address and password will be checked against the LDAP server. If authentication succeeds, docassemble will check to see if a user with that e-mail address already exists in the docassemble login system. If so, the user will be logged in as that user. If the user does not already exist, then a user with that e-mail address will be created in the docassemble login system and will be logged in as that user.

If the password does not authenticate, then the login process will proceed as normal; if a user already exists in the docassemble login system with that username and password, the user will be logged in as that user.

Thus, the LDAP login system can coexist with the standard docassemble login system.

You can also set the base dn, bind email, and bind password directives:

ldap login:
  enable: True
  server: 192.168.1.124
  base dn: "ou=users,dc=example,dc=com"
  bind email: "[email protected]"
  bind password: "xxsecretxx"

If these additional directives are set, then when a user tries to register on the docassemble system, an error message will be generated if the e-mail address exists on the LDAP server. The bind email and bind password are necessary so that docassemble can connect to the LDAP server and run a search for the e-mail address in question. When it does so, it searches within the base dn for an entry with the attribute mail. If one or more entries are found, the error message is raised. If your server uses a different attribute for the e-mail address, you can set the directive search pattern under ldap login. By default, search pattern is set to mail=%s. (The %s is replaced with the e-mail address when the search is run.)

You may wish to disable registration entirely when using ldap login. In that case, it is not necessary to set the base dn, bind email, and bind password directives.

Configuring the websockets server

Applicable only to the jhpyle/docassemble Docker image.

The live help features depend on a WebSocket server. In the jhpyle/docassemble Docker image, there is a supervisor service called websockets that runs the docassemble.webapp.socketserver module. By default, the server runs on port 5000 on the localhost network (127.0.0.1). This can be configured:

websockets ip: 75.34.2.14
websockets port: 5001

The most common reason you might want the server to run on an IP address other than 127.0.0.1 is if you are running docassemble in a Docker container and want to connect to the server from outside the container. This is necessary in order to run a Docker container behind a reverse proxy. If you set:

expose websockets: True

then the server will attempt to connect to the first network that is not the local network. Typically, inside a Docker container, there are two networks, lo and eth0. Setting expose websockets to true will cause the server to connect to whatever address is associated with the eth0 network.

Alternative port for HTTP

Applicable only to the jhpyle/docassemble Docker image.

By default, if you are not using HTTPS, the docassemble web application runs on port 80. When running Docker, you can map any port on the host to port 80 in the container. However, if you are using a system like Heroku which expects HTTP to be available at a specific port other than port 80, you can set http port to the port number.

Directory for log files

docassemble writes messages to a log files stored in the directory indicated by the log directive. These messages are helpful for debugging problems with interviews.

log: /tmp/docassemble-log

The default directory is /usr/share/docassemble/log.

If a log server is set, docassemble will write messages to TCP port 514 on that server, and will not write to the log directory.

Days of inactivity before interview deletion

When the scheduled tasks feature is enabled on the server, docassemble will delete interviews after 90 days of inactivity. To change the number of days, set the interview delete days directive in the configuration. For example:

interview delete days: 180

If interview delete days is set to 0, interviews will never be deleted through scheduled tasks.

If you want to override interview delete days for particular interviews, you can set interview delete days by filename to a dictionary in which the keys are interview filenames and the values are days in integers.

interview delete days by filename:
  docassemble.missouri:data/questions/custody.yml: 10
  docassemble.missouri:data/questions/support.yml: 20

Automatic user account deletion

While interview delete days and interview delete days by filename result in sessions being deleted, the user accounts associated with inactive sessions are unaffected, and users are still able to log in even if their sessions have been deleted.

If you want user accounts to be automatically deleted if a user does not log in after a period of time, you can enable this with user auto delete.

user auto delete:
  inactivity days: 90

The inactivity days refers to the number of days of inactivity after which the user account will be deleted.

Additional sub-directives that can appear under user auto delete include enable, privileges, and delete shared.

user auto delete:
  enable: True
  inactivity days: 90
  privileges:
    - user
    - advocate
  delete shared: True

The enable sub-directive can be set to False if you want to disable auto deletion but keep user auto delete in the Configuration.

The privileges sub-directive refers to a list of user privileges for which auto-deletion should apply. The default is that only users with the privilege of user can be auto-deleted; users with admin, advocate or developer privileges (who do not also have the user privilege) will not have their accounts deleted.

The delete shared directive refers to the type of user deletion that should occur. If True, then all sessions accessed by the user will be deleted, even if the session is a multi-user session and other users of the session still have accounts. The default is False.

If inactivity days is 0, this has the effect of enable: False.

Inactivity is measured based on when the user last logged in. Use of the API counts as a log in. If a user logged in more than inactivity days ago but an interview session to which the user has access was modified within the inactivity days period, then the user’s account will not be deleted.

Flask session lifetime

By default, Flask remembers sessions for 31 days. To set this period to a different amount, set session lifetime seconds to a number of seconds after which the sessions should expire.

For example, to expire a session after 24 hours, include:

session lifetime seconds: 86400

Polling frequency

By default, the user interface polls the server every six seconds to “check in” to see if anything has happened that it needs to know about, and to let the server know that the user is still active.

You can change this interval by setting the checkin interval directive. The number refers to the milliseconds between each call to the server.

checkin interval: 6000

If you set the checkin interval to 0, this will turn off the check-in mechanism altogether. This might be useful if you do not need the check-in feature and you want to prevent unnecessary traffic and CPU usage.

The features that rely on the checkin interval being greater than zero include:

• Notifications of the results of background tasks.
• The check in feature.
• The live help features.

Other than setting checkin interval to 0, it is probably not a good idea to reduce this value below 6000. If requests to modify a single interview come in too frequently, problems can occur.

Using the bugfix-only version

By default, when you press “Upgrade” on the Package Management page, you will download the latest version of docassemble. If you want to stay on the 1.0.x series, which includes bug fixes only and no feature changes or new features, set stable version to True.

stable version: True

Alternative GitHub location for docassemble

The Packages feature updates the docassemble software directly from GitHub. If you need to change the location of this repository to your own fork of docassemble, you can add this directive.

docassemble git url: https://github.com/yourusername/docassemble

The default value is 'https://github.com/jhpyle/docassemble'.

This directive only has an effect during initial database setup.

Pre-defined variables for all interviews

If you would like to pass variable definitions from the configuration into the interviews, you can set values of the initial dict:

initial dict:
  host_company: Example, Inc.
  host_url: http://example.com

Then, in all of the interviews running on the server, you can include things like:

---
sets: splash_screen
question: Welcome to the interview.
subquestion: |
  Web hosting has been graciously provided by ${ host_company }.
---

Translations of words and phrases

If your server will be used by people who speak languages other than English, you will want to make sure that built-in words and phrases used within docassemble, such as “Continue” and “Sign in,” are translated into the user’s language.

The words directive loads one or more YAML, XLSX, or XLIFF files in order:

words:
  - docassemble.mypackage:data/sources/es-words.yml

Each YAML file listed under words must be in the form of a dictionary in which the keys are languages (ISO-639-1 or ISO-639-3 codes) and the values are dictionaries with the translations of words or phrases.

Assuming the following is the content of the data/sources/es-words.yml file in docassemble.mypackage:

es:
  Continue: Continuar
  Help: ¡Ayuda!

then if interview calls set_language('es') (Spanish) and docassemble code subsequently calls word('Help'), the result will be ¡Ayuda!.

When a user visits a page on the docassemble server, the user’s browser tells docassemble the user’s preferred language. This language is in the browser settings. If the user’s preferred language is Spanish (es), then docassemble will try to provide Spanish translations if they are available in any of the files listed under words. If a translation of a given phrase is not available, docassemble will try to provide a translation of the phrase for whatever language is specified in language in the Configuration. If no translation is available for that language, docassemble will use an English version of the phrase.

When you are logged in as a developer or administrator, you can go to the “Utilities” page from the main menu, where you will find a utility for generating a draft YAML, XLSX, or XLIFF file for translating all of the words and phrases that docassemble uses and that might be presented to the user. If you have set up a Google API key, it will use the Google Cloud Translation API to prepare “first draft” translations for any ISO-639-1 language you designate.

When you have prepared a translation file and you want to use it on your server, you will need to put the file into the “sources” folder of a package and install that package on your server using Package Management. For more information about how to create packages, docassemble, see the Packages section.

Users of docassemble have contributed translations of built-in system phrases. These are available in the docassemble.base package, which is part of the core docassemble code. To see a list of translation YAML files that are available, see the list of files on GitHub. For example, you can use the Italian translation file by including docassemble.base:data/sources/it-words.yml in your words directive. Note that these user-contributed files have been added at various times and may not be 100% complete; the words-es.yml file, for example, contains minimal translations, while the words-es.xlf file contains translations for most phrases used by the system.

The words feature with language en can be used to translate the default English phrases into alternative phrases, if you want to change the default phrases that are used in docassemble.

If docassemble is not able to read any of the files listed under words, errors will be written to the uwsgi.log file, which you can find in Logs. If you find that your translations are not being used, make sure to check uwsgi.log for errors.

For more information about how docassemble handles different languages, see the language and locale settings section and the functions section (specifically the functions set_language() and word()).

Currency symbol

You can set a default currency symbol if the symbol generated by the locale is not what you want:

currency symbol: €

This symbol will be used in the user interface when a field has the datatype of currency. It will also be used as the return value of the currency_symbol() function.

Setting currency symbol in the configuration has the same effect as running the following from a Python module:

import docassemble.base.functions
docassemble.base.functions.update_language_function('*', 'currency_symbol', lambda: '€')

This will also affect the behavior of the currency() function; instead of using the locale’s method of formatting a currency value, the currency symbol will be printed, followed by the numerical value.

There are other ways to customize currency features. In an interview, you can configure the currency symbol using the set_locale() function with the keyword parameter currency_symbol. In addition, you can set a field-specific currency symbol using the currency symbol field modifier. See also the section on customizing based on language and locale.

URL to central file server

If you are using a multi-server arrangement you can reduce bandwidth on your web server(s) by setting fileserver to a URL path to a dedicated file server:

fileserver: https://files.example.com/da/

Always use a trailing slash.

If this directive is not set, the value of root will be used to create URLs to uploaded files and static files.

Note that if you are using Azure blob storage, S3-compatible object storage S3, the URLs to files will point directly to files stored in the cloud, so there would be no reason for a docassemble file server.

Geocoding

The .geocode() and .normalize() methods of Address depend on an API for geocoding. By default, the Google Maps Geocoding API is used. You can use the Azure Maps API instead by setting geocoder service to azure maps:

geocoder service: azure maps

The azure maps setting requires you to set the primary key directive under the azure maps directive. See the Azure Maps API key section for more information.

The default value is google maps:

geocoder service: google maps

The google maps setting requires you to set the api key directive under the google directive. See the Google API key section for more information.

To disable geocoding, set geocoder service to null.

geocoder service: null

Geocoding is performed by the GeoPy package, which supports a number of other geocoders besides Google Maps and Azure Maps. The docassemble.base.geocode module can be extended to include support other geocoders. Make a request on the Slack if you need support for a different geocoding service.

Azure Maps API key

If you set geocoder service to azure maps, you need to enable the Azure Maps API on your Azure account and obtain a primary key. Then plug this primary key into your Configuration as follows:

azure maps:
  primary key: AS9W5FS2HS5EFAEFafASFARefa

Google API key

If you want to use the .geocode() method, which uses the Google Maps Geocoding API, or the feature in Utilities for translating system words into other languages, which uses the Google Cloud Translation API, you need to obtain a Google API key and plug it into your Configuration:

google:
  api key: UIJGeyD-23aSdgSE34gEGRg3GDRGdrge9z-YUia

The api key under google is used when the server sends requests to Google; thus, in the Google Cloud Console, you should add security to the API key based on the IP address of the machine making the request. If you migrate to a different server with a different IP address, remember to add your new IP address to the list of allowable IP addresses.

docassemble also has features, including address autocomplete and the map_of() function, which cause the user’s web browser to call the Google API. For these features, you need to go into Google Cloud Console, enable the Google Places API (the “New” version), and create an API key for accessing this API. When you obtain the API key, place it under the google directive as the google maps api key:

google:
  google maps api key: YyFeyuE-36grDgEE34jETRy3WDjGerye0y-wrRb
  use places api new: True

Setting use places api new: True enables the use of the “New” Google Places API, which all users should use or transition to in 2025.

The address autocomplete and map_of() features require sharing the API key with the web browser, which means your API key is not secret. Google allows you to add security based on the Referer header, so that the API key can only be used when it is called from a particular web site.

Therefore, if you use both the server-side features and the client-side features, obtain two separate API keys from Google, and lock down the api key based on the IP address of your server, and lock down the google maps api key based on the URL of your site.

If you use the Google Cloud Translation API for the feature in Utilities that translates system phrases into other languages, you can control how many phrases are translated in a single request to the API.

google translate words at a time: 40

The default number is 20.

Google service account credentials

If you want to integrate with Google Docs or Google Drive, you can add service account credentials, which is a JSON object you obtain from the Google Developers Console.

To do this, go on the Google Developers Console and create a “project” (or use a project you have already created). Under “Enabled APIs and Services,” enable the Google Sheets API, the Google Drive API, and any other APIs you might want to use. Within the project, create a service account. When you create the service account, you will be provided with “credentials.” Download the JSON (not p12) credential file for the service account.

Then, in the configuration, set service account credentials under google to the literal contents of the JSON file you downloaded. (Make sure to provide the necessary indentation so that the YAML is valid.)

For example:

google:
  service account credentials: |
    {
      "type": "service_account",
      "project_id": "redacted",
      "private_key_id": "redacted",
      "private_key": "-----BEGIN PRIVATE KEY-----REDACTED-----END PRIVATE KEY-----\n",
      "client_email": "[email protected]",
      "client_id": "redacted",
      "auth_uri": "https://accounts.google.com/o/oauth2/auth",
      "token_uri": "https://accounts.google.com/o/oauth2/token",
      "auth_provider_x509_cert_url": "https://www.googleapis.com/oauth2/v1/certs",
      "client_x509_cert_url": "https://www.googleapis.com/robot/v1/metadata/x509/googledocs%40redacted.iam.gserviceaccount.com"
    }

Google Analytics ID

If you want to use Google Analytics on the screens in your interviews, you can set analytics id within the google directive:

google:
  analytics id: G-9876543210

When this is configured, JavaScript for Google Analytics will be inserted into your interview pages, and pageview events will be sent to Google Analytics for any question that has an id defined.

If you are using Google Analytics 4, set the analytics id to the GA4 “measurement ID” (beginning with G-). If you are using Google’s older system, set the analytics id to the “tracking ID” (beginning with UA-).

It is possible to use both a UA- and G- ID at the same time if analytics id refers to a list of IDs.

google:
  analytics id:
    - G-9876543210
    - UA-77777777-1

The pageview events that docassemble sends will record a view on an artificial path. For example, suppose you have the following question:

id: lead certification
question: |
  Does your landlord have a valid lead certification?
yesno: lead_certification_exists

Suppose this question is part of the interview answer.yml in the package docassemble.eviction. In that case, the pageview will record a view of the following pseudo-URL on your site:

/eviction/answer/lead_certification

Since the id is also used as a unique identifier for a question, you might to use a different identifier for purposes of Google Analytics. If so, tag your question with ga id instead of an id:

id: lead certification
ga id: landlordLeadCert
question: |
  Does your landlord have a valid lead certification?
yesno: lead_certification_exists

Segment ID

Segment is an analytics aggregator that feeds data about user behavior to other tools, such as Amplitude, Customer.io, FullStory, and Google Analytics.

If you set a segment id in the Configuration, JavaScript will be included that initializes Segment. When the user arrives at a question with an id, a Segment event will be fired with the id as a name.

segment id: ZAfFSeR6BTWiEfGnT8YpujspehLtswJP

Since the id is also used as a unique identifier for a question, you might to use a different identifier for purposes of Segment. If so, tag your question with segment id modifier (not to be confused with the segment id Configuration directive) instead of an id:

id: lead certification
segment id: landlordLeadCert
question: |
  Does your landlord have a valid lead certification?
yesno: lead_certification_exists

You can also send Segment messages with arguments, using the segment specifier pointing to a dictionary with keys for id and arguments:

id: lead certification
segment:
  id: landlord
  arguments:
    certification: lead
question: |
  Does your landlord have a valid lead certification?
yesno: lead_certification_exists

VoiceRSS API key

If the special variable speak_text is set to True, docassemble will present the user with an audio control that will convert the text of the question to speech. This relies on the VoiceRSS web service. You will need to obtain an API key from VoiceRSS and set the configuration below in order for this feature to function. (The service allows 350 free requests per day.)

voicerss:
  enable: True
  key: 347593849874e7454b9872948a87987d
  dialects:
    en: us
    es: mx
    fr: fr

The enable key must be set to True in order for the text-to-speech feature to work. The key is the VoiceRSS API key. The dialects key refers to a dictionary that associates languages with the dialect to be used with that language. For more information about dialects, see the VoiceRSS documentation.

Sometimes, the language code you use in your interview is not the same as the language code that VoiceRSS uses for that language. In this case, you can add a language map to your VoiceRSS configuration to remap your language code so that it is compatible with VoiceRSS.

For example, suppose the language code used in your interview is zho for Chinese. But the language that VoiceRSS recognizes for Chinese is zh, and the dialects that VoiceRSS recognizes for zh include cn, hk, and tw. Suppose you want to use cn as the dialect. You would set your voicerss configuration as follows:

voicerss:
  enable: True
  key: 347593849874e7454b9872948a87987d
  language map:
    zho: zh
  dialects:
    zh: cn

You can set the default “voice” for a given language:

voicerss:
  enable: True
  key: 347593849874e7454b9872948a87987d
  voices:
    it: Pietro

You can set the default “voice” for a given dialect of a given language:

voicerss:
  enable: True
  key: 347593849874e7454b9872948a87987d
  voices:
    en:
      us: Amy
      au: Jack
    fr:
      ca: Logan

You can set the audio format for VoiceRSS audio files:

voicerss:
  enable: True
  key: 347593849874e7454b9872948a87987d
  format: 44khz_16bit_stereo

The default format is 16khz_16bit_stereo.

Translation of dates

Many of the date functions rely on the babel.dates package. This package does not support every language. For example, if you run set_language('ht'), functions that use babel.dates will return an error. You can set a different language to be used by babel.dates as an alternative by setting the babel dates map directive to a dictionary that translate language codes into alternative language codes. For example:

babel dates map:
  ht: fr

OCR language settings

The ocr_file() function uses the Tesseract optical character recognition (OCR) application to extract text from image files and PDF files. One of the options for the OCR process is the language being recognized. The codes for language that Tesseract may be different from those that docassemble uses. If you are using ISO-639-1 for your language codes, the code that Tesseract uses is the ISO-639-3 code that corresponds to the ISO-639-1 code that docassemble uses, and docassemble can make this conversion automatically. However, in some cases this does not work, so there is an override, which is controlled by the ocr languages directive. The default value maps Chinese to Traditional Chinese:

ocr languages:
  zh: chi-tra

If all of the Chinese documents that you want to OCR are written in Simplified Chinese, and all the Uzbek documents are written in Cyrillic, you could set the following:

ocr languages:
  zh: chi-sim
  uz: uzb-cyrl

For more information, see the documentation of the ocr_file() function.

s3

If you are using Amazon S3 or an S3-compatible object storage service to store shared files, your access keys, bucket name, and region name are in stored in the Configuration as follows:

s3:
  enable: True
  access key id: FWIEJFIJIDGISEJFWOEF
  secret access key: RGERG34eeeg3agwetTR0+wewWAWEFererNRERERG
  bucket: yourbucketname
  region: us-west-1

There is also an option under s3 called endpoint url that you can set if you are using an S3-compatible object storage service. (e.g., endpoint url: https://mys3service.com). By default, Amazon S3 is used. If you are using an S3-compatible object storage service, the region directive may not be necessary.

You will need to create the bucket before using it; docassemble will not create it for you.

If your S3 bucket uses encryption, you can specify your server-side encryption information under server side encryption. E.g.,

s3:
  enable: True
  access key id: FWIEJFIJIDGISEJFWOEF
  secret access key: RGERG34eeeg3agwetTR0+wewWAWEFererNRERERG
  bucket: yourbucketname
  region: us-west-1
  server side encryption:
    algorithm: null
    customer algorithm: AES256
    customer key: 91F8F26A0783345D16C4249990B54B6E
    KMS key ID: null

If you are using the jhpyle/docassemble Docker image, you should not define an s3 directive in the Configuration using the web application when you already have a server running. The S3 configuration is used throughout the boot process and the shutdown process. If you want to start using S3, you should start a new container using docker run with the S3BUCKET and other S3 environment variables set. For more information, see the Docker section.

azure

If you are using Azure blob storage to store shared files, enter your account name, account key, and container name as follows:

azure:
  enable: False
  account name: example-com
  account key: 1TGSCr2P2uSw9/CLoucfNIAEFcqakAC7kiVwJsKLX65X3yugWwnNFRgQRfRHtenGVcc5KujusYUNnXXGXruDCA==
  container: yourcontainername

You will need to create the container before using it; docassemble will not create it for you.

ec2

If you are running docassemble from within an Amazon EC2 or Lightsail instance, set this to true:

ec2: True

This is necessary because when docassemble runs in a multi-server arrangement, each docassemble web server instance needs to allow other docassemble web instances to send messages to it through supervisor. Each web server instance advertises the hostname or IP address through which its supervisor can be accessed. Normally, this can be obtained using the computer’s hostname, but within an EC2 or Lightsail instance, this hostname is not one that other web servers can resolve. If ec2 is set to True, then docassemble will determine the hostname by calling http://169.254.169.254/latest/meta-data/local-ipv4.

ec2 ip url

If ec2 is set to True, docassemble will determine the hostname by calling http://169.254.169.254/latest/meta-data/local-ipv4. If this URL does not work for some reason, but a different URL would work, you can change the URL that docassemble uses by setting the ec2 ip url configuration item.

ec2 ip url: http://169.254.169.254/latest/meta-data/local-ipv4

kubernetes

If you are running docassemble from within a Kubernetes deployment, add this to your Configuration

kubernetes: True

This is necessary because when docassemble runs in a multi-server arrangement, each docassemble web server instance needs to allow other docassemble web instances to send messages to it through supervisor. Each web server instance advertises the hostname or IP address through which its supervisor can be accessed. Normally, this can be obtained using the computer’s hostname, but within a Kubernetes cluster, this hostname is not one that other web servers can resolve. If kubernetes is set to True, then docassemble will use the IP address of the Pod as the hostname.

API usage

By default, only users with privileges of admin or developer may obtain an API key in order to use the docassemble API. You can change this by editing the api privileges.

api privileges:
  - admin
  - developer
  - user

Password login

If password login is set to False, then users will not see an option for entering their username and password on the login screen; they will only see buttons for other login methods that you have enabled. The default value is True.

Since there is a possibility that your other login methods will fail, there is a “back door” that bypasses this setting, so that you as the administrator can still log in using a username and password. Instead of going to /user/sign-in in the web browser, go to /user/sign-in?admin=1. The username and password fields will be shown, regardless of the password login setting.

Phone number login

If phone login is set to True, then docassemble will allow users to log in with a phone number. For this system to work, the twilio configuration must be set up. The user needs to put in his or her phone number, and then a random verification code will be sent to that phone number via SMS. The user needs to type in that code in order to log in. As with the external authentication methods, registration happens automatically upon the first login. Subsequent logins will also require entering a random verification code.

The details of how this login system functions, such as the number of digits in the verification code and the number of seconds that the code remains valid, can be configured.

Two-factor authentication

For added security, you can allow users who log in with passwords to enable two-factor authentication on their accounts. The two-factor authentication system can use Google Authenticator, Authy, or another compatible app. Or, if you have set up a twilio configuration, the system can send a verification code to the user in an SMS message.

To give your users the option of using two-factor authentication, set two factor authentication as follows:

two factor authentication:
  enable: True

Logged-in users will then see an option on their “Profile” page for configuring two-factor authentication. By default, only administrators and developers see an option on their user profile to configure second-factor authentication. To configure which privileges have the option of using second factor authentication, set the allow for subdirective to the full list of privileges for which you want the feature to be available.

two factor authentication:
  allowed for:
    - admin
    - developer
    - user
    - advocate

Regardless of these settings, two-factor authentication is not available to users who sign in with external authentication methods or who are using the phone login feature. It is only available when the first authentication method is a standard e-mail and password combination.

If you want to enable of the two methods and disable the other, you can use the allow sms and allow app subdirectives.

two factor authentication:
  allow sms: False
  allow app: True

By default, allow sms and allow app are both True.

If you want users with certain privileges to be required to use two-factor authentication, use the required for subdirective:

two factor authentication:
  required for:
    - developer
    - admin

By default, two-factor authentication is optional for all users. As usual, two-factor authentication is not compatible with external authentication methods or the phone login feature, so you will want to disable those features if you want to enforce two-factor authentication.

The default value of enable is True, so you can omit the enable line. You can write two factor authentication: True to enable two-factor authentication with all of the default options.

Automatic login

The auto login directive causes users to automatically redirect to a third-party login service when they arrive at the /user/sign-in endpoint.

If you have only one alternative login method, you can password login to False and set auto login to True:

auto login: True
password login: False

If you have multiple alternative login methods, and you want one of them to be automatic, you can set auto login to the name of the login method:

auto login: google
password login: False

Valid values are phone, google, facebook, auth0, keycloak, authentik, zitadel, miniorange, and azure.

If you wish to prevent the user from being automatically redirected when they visit /user/sign-in, you can add from_logout=1 to the URL parameters. (This is how the logout process avoids an immediate login when logoutpage is not defined and the user is directed back to the login page after they log out.)

List of examples in the Playground

As a development aid, the Playground contains an examples area with a list of short sample interviews that demonstrate particular features. The interviews shown here are controlled by the file docassemble.base:data/questions/example-list.yml. If you would rather use a list of your own, you can define playground examples:

playground examples: docassemble.michigan:data/questions/examples.yml

This will replace all of the example interviews with the examples you specify.

You can also set playground examples to a list of files:

playground examples:
  - docassemble.base:data/questions/example-list.yml
  - docassemble.michigan:data/questions/examples.yml

In this example, the standard list (docassemble.base:data/questions/example-list.yml) will be used first, and then the example categories and interviews from the docassemble.michigan package will be included after.

For example, suppose the file docassemble.michigan:data/questions/examples.yml contained:

- Slack:
    - slack-example
    - slack-example-2

This would assume that there was an interview docassemble.michigan:data/questions/slack-example.yml and an interview docassemble.michigan:data/questions/slack-example-2.yml.

It would also assume that there were PNG screenshots for this interviews available in the data/static folder of the docassemble.michigan package called slack-example.png and slack-example-2.png.

There are some metadata headers specific to example interviews. For example, suppose the metadata of slack-example.yml was:

metadata:
  title: Test Slack Posting
  short title: Test Slack
  documentation: "https://michigan.example.com/docs/using_slack.html"
  example start: 1
  example end: 2
---
question: |
  What do you want to post on Slack?
fields:
  - Message: the_message
---
code: |
  post_to_slack(the_message)
  message_posted

The short title, “Test Slack,” will be used in the navigation bar of the Playground examples section. The URL under documentation will be the URL behind the “View documentation” link in the Playground examples section. The blocks of the interview that will be shown on the screen are based on the example start and example end items. In this example, block “1” is the question block, and block “2” is the code block. (Block “0” is the metadata block.) So the question block and the code block will be shown, but not the metadata block.

Changing the mode of the editor in the Playground

If the keymap directive is set to vim, emacs, or sublime, then the in-browser text editors in the Playground will emulate Vim, Emacs, or Sublime Text, respectively. This uses the Vim bindings option, Emacs bindings option, and Sublime Text bindings option of CodeMirror.

For Vim:

keymap: vim

For Emacs:

keymap: emacs

For Sublime Text:

keymap: sublime

URL to the site

The external hostname is the hostname by which users will access docassemble. If users will see https://docassemble.example.com in the location bar on the web browser, external hostname will be docassemble.example.com.

external hostname: docassemble.example.com

This setting is used at the Python level and also at the web server level. If you are using the jhpyle/docassemble image, and you change external hostname, you need to do a complete restart of the system for the change to take effect. The web server configuration needs to know the correct hostname.

When behind a proxy

Set the behind https load balancer directive to True if you are running docassemble in a configuration where docassemble itself is running HTTP, but requests are being forwarded to it by a server running HTTPS. This might be your configuration if you are using a load balancer or you are running Docker in a context where a web server or web service forwards HTTPS requests to Docker on port 80 or a non-standard HTTP port.

This variable is typically set using the Docker environment variable BEHINDHTTPSLOADBALANCER.

This is important because if docassemble thinks that it is serving requests on port 80, it won’t set the secure flag on cookies. However, if the user’s browser actually is accessing the site over HTTPS, then users will get cookies that lack the secure flag. The mismatch between the use of HTTPS and the flag on the cookies can cause site to reject the cookies. The symptom of the browser rejecting cookies can be the continual reloading of the screen or the Continue button acting like a reload button. Cookies are necessary for docassemble’s session system, so the site will not function unless cookies work. Thus it is very important to set behind https load balancer if you are using a load balancer or proxy.

Also make sure that the proxy server you are using sets the X-Forwarded-* headers. docassemble needs these headers in certain circumstances in order to form correct URLs. Specifically, the proxy server should set the following headers:

• X-Forwarded-For
• X-Forwarded-Host
• X-Forwarded-Port
• X-Forwarded-Proto

You can test whether your proxy server is sending these headers to your docassemble server by visiting /headers on your site.

Using Let’s Encrypt

Applicable only to the jhpyle/docassemble Docker image.

If you are using the jhpyle/docassemble Docker image and you want your server to use HTTPS set up through Let’s Encrypt, you can set the following in your Configuration.

external hostname: docassemble.example.com
use https: True
use lets encrypt: True
lets encrypt email: [email protected]

These variables are typically set through the environment variables DAHOSTNAME, USEHTTPS, USELETSENCRYPT, and LETSENCRYPTEMAIL when the docker run command is run for the first time. You can change the use https, use lets encrypt, and lets encrypt email variables on a running server, but they will only be effective if you restart the system using docker stop -t 600 <container ID> followed by docker start <container ID>.

SSL protocols

Applicable only to the jhpyle/docassemble Docker image.

The nginx ssl protocols directive indicates the SSL protocols that NGINX should accept (assuming you are using NGINX). You might want to set it to TLSv1 TLSv1.1 TLSv1.2 TLSv1.3 if you need to support older browsers.

nginx ssl protocols: TLSv1 TLSv1.1 TLSv1.2 TLSv1.3

The value is passed directly to the NGINX directive ssl_protocols. The default is TLSv1.2 TLSv1.3.

This variable is typically set through the environment variable DASSLPROTOCOLS. You can change nginx ssl protocols on a running server, but the change will only be effective if you restart the system using docker stop -t 600 <container ID> followed by docker start <container ID>.

SSL ciphers

Applicable only to the jhpyle/docassemble Docker image.

The nginx ssl ciphers directive indicates the SSL ciphers that NGINX should accept (assuming you are using NGINX).

nginx ssl ciphers: ECDHE-ECDSA-AES128-GCM-SHA256:ECDHE-RSA-AES128-GCM-SHA256:ECDHE-ECDSA-AES256-GCM-SHA384:ECDHE-RSA-AES256-GCM-SHA384:ECDHE-ECDSA-CHACHA20-POLY1305:ECDHE-RSA-CHACHA20-POLY1305:DHE-RSA-AES128-GCM-SHA256:DHE-RSA-AES256-GCM-SHA384:DHE-RSA-CHACHA20-POLY1305;

The value is passed directly to the NGINX directive ssl_ciphers. The default is HIGH:!aNULL:!MD5.

This variable is typically set through the environment variable DASSLCIPHERS. You can change nginx ssl ciphers on a running server, but the change will only be effective if you restart the system using docker stop -t 600 <container ID> followed by docker start <container ID>.

Cross-Origin Resource Sharing (CORS)

Set the cross site domains directive if you want the web server to send Access-Control-Allow-Origin headers in order to permit Cross-Origin Resource Sharing.

cross site domains:
  - https://example.com
  - https://help.example.com

It is important to specify the protocol as well as the domain.

See also allow embedding.

E-mail domain of the site

If you use the e-mail receiving feature, set the incoming mail domain to whatever e-mail domain will direct e-mail to the docassemble server configured to perform the all or mail services.

incoming mail domain: mail.example.com

The interview_email() function will then return an e-mail address with @mail.example.com at the end.

If incoming mail domain is not specified, the value of external hostname will be used.

Whether software updating is allowed

By default, a user with privileges of administrator or developer can install Python packages on the server, or update existing Python packages. To disable updating of packages through the user interface and the API, set allow updates to false.

allow updates: false

Note that the Playground also allows users to change the code on the system, so if you set allow updates: false, you will probably also want to set enable playground to false.

Whether software is updated on start

Applicable only to the jhpyle/docassemble Docker image.

By default, when the Docker container starts, whether it is starting for the first time during a docker run process, or restarting during a docker start process, one of the steps taken during the container initialization process is the updating of Python packages.

This process is necessary in order to install any custom Python packages you have, such as interviews you have created. If you replace a container with docker stop -t 600, docker rm, and docker run, you will want the new container to have the appropriate software on it. When you use Package Management or the packages folder of the Playground to install a package on your system, the packages that get installed are tracked in the SQL database. When you start up a new Docker container that uses that SQL database, the update process will install the packages listed in the database.

This is the main reason why launching a Docker container can take a long time.

By default, the update process happens any time the container starts, which is both during the docker run process and also during the docker start process.

If you do not want your container to go through this update process when the container starts, you can set:

update on start: false

Alternatively, if you want the update process to run during the docker run process but not subsequently, you can set:

update on start: initial

Whether the configuration can be edited

If you want to prevent administrators from editing the configuration through the web interface or the API, you can set allow configuration editing to False.

allow configuration editing: False

Log server hostname

If the log server directive is set, docassemble will write log messages to TCP port 514 on the hostname indicated by log server instead of writing them to /usr/share/docassemble/log/docassemble.log (or whatever other directory is set in log).

Redis server location

The redis directive indicates where docassemble can find the Redis server that it should used for storing Flask session state, API keys, background task information, and other data that are stored in Redis.

The redis directive should be treated as read-only. The primary way you would adjust your Redis configuration is when starting a new server for the first time. For example, you would change this setting if you were using a cloud-based managed Redis system like ElastiCache for Redis.

Changing the redis directive after you already have a working system can lead to major problems, such as API keys not working. You should not change the redis value using the web-based front end unless you really know what you are doing.

In the default configuration of the jhpyle/docassemble Docker deployment, redis directive is set to null.

redis: null

This indicates that the Redis server is located on the same server as the web server.

If redis is set to a something else, then an external redis server is used.

redis: redis://192.168.0.2

The redis directive needs to be written in the form of a redis URI. To specify a non-standard port, use the form redis://192.168.0.2:7000. To specify a password, use the form redis://:[email protected] or redis://192.168.0.2?password=xxsecretxx. You can specify a username along with the password (redis://jsmith:[email protected]), but Redis does not use usernames, so this may not work.

To use SSL, use the prefix rediss:// instead of redis://. You can provide your own certificates to Redis if necessary. If you are using S3 or Azure blob storage, you need to copy the certificate files to certs/ in the bucket or container. Otherwise, you need to make sure the certificate files are in /usr/share/docassemble/certs when the server initializes (from there, the files are copied into working directories). For more information, see managing certificates with Docker and certs. The certificate files must have the following names, which correspond with particular parameters for connecting to Redis with Python.

• redis_ca.crt - for the ssl_ca_certs file (certificate authority certificate)
• redis.crt - for the ssl_certfile file (certificate)
• redis.key - for the ssl_keyfile (certificate key)

If the URL begins with rediss://, the connection is initialized with the parameter ssl=True. If any of the above files are present, the ssl_ca_certs, ssl_certfile, and/or ssl_keyfile parameters will be used as well.

docassemble uses three Redis “databases.” By default, it uses databases 0, 1, and 2. If you want it to use different database numbers, you can set redis database offset to a number.

redis database offset: 3

In this case, docassemble will use databases 3, 4, and 5 instead of 0, 1, and 2.

If you reference a database number in your URI (e.g., using redis://192.168.0.2?db=1 or redis://192.168.0.2/1), then the referenced database will be used as the redis database offset. However, if you also include redis database offset in your Configuration, the value of redis database offset will override whatever is specified in the URI.

RabbitMQ server location

By default, docassemble assumes that the RabbitMQ server is located on the same server as the web server. You can designate a different RabbitMQ server by setting the rabbitmq directive:

rabbitmq: pyamqp://[email protected]//

The rabbitmq directive needs to be written in the form of an AMQP URI.

Image conversion

By default, docassemble assumes that you have ImageMagick and pdftoppm installed on your system, and that they are accessible through the commands convert and pdftoppm, respectively. If you do not have these applications on your system, you need to set the configuration variables to null:

imagemagick: null
pdftoppm: null

If you have the applications, but you want to specify a particular path, you can set the path using the configuration variables:

imagemagick: /usr/local/bin/convert
pdftoppm: /usr/local/bin/pdftoppm

Sound file conversion

By default, docassemble assumes that you have pacpl (the Perl Audio Converter and/or ffmpeg installed on your system, and that they are accessible through the commands pacpl and ffmpeg, respectively. If you do not have these applications on your system, you need to set the configuration variables to null:

pacpl: null
ffmpeg: null

You can also set these variables to tell docassemble to use a particular path on your system to run these applications.

Document conversion

docassemble requires that you have Pandoc installed on your system. It assumes that it can be run using the command pandoc. If you need to specify a different path, you can do so in the configuration:

pandoc: /opt/pandoc/bin/pandoc

By default, docassemble uses LibreOffice through unoconv to convert DOCX templates to PDF format. If your docassemble server was created many years ago, you may need to enable unoconv with the enable unoconv directive; otherwise, docassemble will call LibreOffice directly.

Instead of using unoconv, you can use an external Gotenberg server for DOCX to PDF conversion. If you have a Gotenberg server that is accessible from within the docassemble container on hostname mygotenberg, port 3000, you would disable unoconv and set the gotenberg url to the following:

enable unoconv: false
gotenberg:
  url: http://mygotenberg:3000

Then, all DOCX-to-PDF conversion will use this server.

There are many ways to deploy a Gotenberg server alongside a docassemble container. Gotenberg is available as a Docker container, so you can run a Gotenberg container on the same server where you are hosting a docassemble container. In order for one Docker container running on a host to talk to another Docker container, you need to make sure that docassemble container can reach the Gotenberg container; this can be moderately difficult because of the way networking works with Docker. One way is to create a bridge network and run both the Gotenberg and docassemble containers on that network:

docker network create --driver bridge docassemble-net
docker run -d --name gotenberg \
--restart always \
--network docassemble-net \
-p 3000:3000 \
gotenberg/gotenberg:7
docker run -d --name docassemble \
--restart always \
--network docassemble-net \
-v dabackup:/usr/share/docassemble/backup \
-p 80:80 \
jhpyle/docassemble

Then you can set the gotenberg url as follows:

enable unoconv: false
gotenberg:
  url: http://gotenberg:3000

You can also start gotenberg on a publically visible server with basic authentication enabled, so that it can only be used if a username and password are supplied. These can also be set in the config.

enable unoconv: false
gotenberg:
  url: https://gotenberg.example.com
  username: myusername
  password: abc-makethismuchlonger-123

The gotenberg url can be prepopulated in the docassemble configuration using the GOTENBERGURL environment variable. The [GOTENBERGUSERNAME] and [GOTENBERGPASSWORD] environment variables can also be set to connect to a gotenberg server with basic auth.

Disabling unoconv when using Gotenberg is important not only to save memory but to prevent any conflicts between Gotenberg’s use of LibreOffice and the docassemble’s use of LibreOffice, if the containers are running on the same host. After you change enable unoconv, you need to do a docker stop/docker start for the change to fully take effect.

Another third party service that you can use is ConvertAPI, a web service for file conversion. To enable this, sign up with ConvertAPI and set the convertapi secret to the “Secret” that ConvertAPI generates for you.

convertapi secret: SrWh4XfuOje5jYx1

You can also use CloudConvert. Sign up with CloudConvert, go to the Dashboard, go to Authorization, API keys under “API V2,” and click “Create New API key.” Copy the API key into your Configuration as the cloudconvert secret:

cloudconvert secret: afwefhwfperowewERQWrasfASEf234Frthe33t4ge4rgf3evn3480w9ejfornv9n40923nvqiniQWeeorgmkbwyeghgpojrgbu

Enabling ConvertAPI or CloudConvert can be useful if you are using docx template file to generate PDF files, but the standard LibreOffice conversion does not work properly due to incompatibilities between Microsoft Word and LibreOffice.

Pandoc’s LaTeX engine

By default, Pandoc uses pdflatex to generate PDF files from Markdown. If you would like to use a different engine, you can specify it using the pandoc engine directive:

pandoc engine: xelatex

Producing PDF/A files

If you want the PDF files produced by interviews on your server to be in PDF/A format, you can set this as a default:

pdf/a: True

The default is False. The setting can also be made on a per-interview basis by setting the pdf/a features setting.

When using docx template file, you also have the option of creating a “tagged PDF,” which is similar to PDF/A. You can set this as a server-wide default:

tagged pdf: True

This setting can also be made on a per-interview basis by setting the tagged pdf features setting.

Changing the default font for rendering PDF fields

When pdf template file is used with editable: False, pdftk is used to convert form fields to rendered text. The default font that pdftk uses may not have support for characters that you want to use. You can specify a different font for a particular attachment using rendering font.

To specify a font that should be used by default instead of pdftk’s Arial font, you can specify a default rendering font in the Configuration.

default rendering font: /usr/share/fonts/truetype/msttcorefonts/Arial.ttf

For more information about what font file to use, see rendering font.

Using unoconv instead of LibreOffice for processing DOCX files

Applicable only to the jhpyle/docassemble Docker image.

Starting with system version 1.3.18, a unoconv listener is available in the jhpyle/docassemble Docker image, which keeps LibreOffice in memory and uses it to convert files with a client/server model. docassemble uses LibreOffice to convert DOCX to PDF and other formats, as well as to update references within a DOCX after docx template file is used to assemble a document.

The unoconv listener will run only if enable unoconv is set to True in your Configuration. If enable unoconv is missing or is not set to True, unoconv will not be used. If you started using docassemble before 1.3.18 and you want to use unoconv, set the following in your Configuration, and then do a system upgrade.

enable unoconv: True

This is part of the default configuration if you created your server since version 1.3.18.

After changing enable unoconv, you need to restart your system with docker stop/docker start.

See also the ENABLEUNOCONV environment variable.

Unfortunately, unoconv is no longer under active development. Using Gotenberg is recommended.

Limiting size of uploaded images

If your users upload digital photos into your interviews, the uploads may take a long time. Using the maximum image size field modifier or the maximum image size interview feature, you can cause your users’ web browsers to reduce the size of your images before uploading them. For example, if maximum image size is set to 1200, then the image will be reduced in size if it is taller or wider than 1200 pixels.

Note that the image file type of the uploaded file may be changed to PNG during the conversion process. Different browsers behave differently.

If you would like to set a site-wide default value for the maximum image size, you can use the maximum image size configuration directive:

maximum image size: 1200

This is just a default value; it can be overridden for a given interview using the maximum image size interview feature and it can be overridden for a given field using the maximum image size field modifier. To override the setting in order to allow image uploads of any resolution, use None as the override value.

Converting the format of uploaded images

If you are using maximum image size, you can also cause images to be converted to PNG, JPEG, or BMP by the browser during the upload process by setting the image upload type to png, jpeg, or bmp.

image upload type: jpeg

Number of concurrent background tasks

docassemble uses Celery to execute background tasks. The Celery system is able to execute multiple tasks concurrently. The number of concurrent workers is set by default to the number of CPU cores on the machine, except that the number of workers will not be larger than the number of gigabytes of total memory divided by 2. For example:

• If a machine has 4 CPU cores and 16GB of RAM, four Celery workers will be started (CPU limited)
• If a machine has 32 CPU cores and 16GB of RAM, eight Celery workers will be started (memory limited).

The number of CPUs is determined by calling nproc --all. Note that the result may be the same or may be different from the number of vCPUs that a virtual machine has.

If you want to manually override this formula, set the celery processes directive.

celery processes: 15

This will cause 15 Celery workers to be spawned.

Note that there are two Celery systems: one called celerysingle with a single worker, and one called celery with one or more workers. The celerysingle system is used for background processes that will perform parallel processing and use every CPU in the machine; it would be dangerous to run more than one such process at a time. The celery system is used for all other tasks, which Celery may run in parallel on all the available worker processes. Thus, if the celery processes is 15, the worker_concurrency will be 1 for celerysingle and 14 for celery. You can see this if you do ps ax | grep celery on the server.

If you want the number of Celery processes to scale with CPU but you think your system can handle more concurrency than the standard formula would provide, you can set a higher maximum:

max celery processes: 10

This setting will mean that:

• If a machine has 8 CPUs and 4GB of RAM, the number of celery processes will be 8 (1 for celerysingle and 7 for celery).
• If a machine has 32 CPUs and 4GB of RAM, the number of celery processes will be 10 (1 for celerysingle and 9 for celery).

It is important not to spawn too many Celery workers because each of them loads software into its own memory space, which could cause your memory to be exhausted.

Using oauthlib on http servers

If you use the oauthlib library and users access your server using http:// rather than https://, set run oauthlib on http to True.

run oauthlib on http: True

This will set the environment variable OAUTHLIB_INSECURE_TRANSPORT to 1.

Microsoft Word sidebar

You will need to set office addin url if you are running a Microsoft Word sidebar from a location other than your own server.

As a security measure, when a Microsoft Word sidebar communicates with your server, your docassemble server will only allow communications that come from a particular hostname on the internet.

The office addin url directive in the Configuration specifies which hostname is allowed.

office addin url: https://sidebar.example.com

This means that your server will only allow Microsoft Word sidebar communications that originate from https://sidebar.example.com.

If you don’t set office addin url, the value of url root will be used, and if url root is not set, the URL will be discerned from the HTTP request. However, if your server is behind a proxy, this latter method may not be accurate, so should url root in order for the Microsoft Word sidebar to work correctly.

Note that a Word sidebar is just a web page that consists of an HTML file with some CSS and JavaScript files associated with it. These files can be hosted on any site on the internet as long as HTTPS is in use. (Microsoft requires HTTPS.) When your Microsoft Word application accesses this web page, you can click the “Use a different server” button to instruct the sidebar to access the Playground on any server to which you have access. Then the JavaScript on the page will communicate directly with that server and pass information back and forth between Microsoft Word and the Playground on the server.

Your docassemble server can serve as the host for the web page. If your server’s hostname is https://interviews.example.com, then the sidebar will be available at https://interviews.example.com/officetaskpane. If you create an XML manifest file for the sidebar from the Utilities page of your server, you will see this URL embedded in the XML.

Setting the time zone

Applicable only to the jhpyle/docassemble Docker image.

Functions like as_datetime() that deal with dates will use a default time zone if an explicit timezone is not supplied. If you set the timezone directive to something like America/Los_Angeles, this will be used as the default time zone. Otherwise, the default time zone will be set to the time zone of the server.

timezone: America/Los_Angeles

For a list of valid time zones, see the tz database.

If you change the timezone, you need to do a docker stop and docker start so that the operating system recognizes the change.

Sharing packages on PyPI

If you want users of your server with developer privleges to be able to publish packages to PyPI from the packages folder of the Playground, set pypi to True:

pypi: True

Then, a user with developer privileges can configure a PyPI username and password on their Profile. Typically, the username is __token__ and the password is a long security token beginning with pypi-. This is available from PyPI when you create an account and create an API token. If using twine from the command line, you would set up this username and password in a .pypirc file. docassemble runs twine to upload to PyPI and it passes this username and password to twine.

You can also tweak the operation of docassemble’s interaction with PyPI by setting the following optional directives (which you should never need to do):

pypi url: https://pypi.python.org/pypi
pypirc path: /var/www/.pypirc

The pypi url directive refers to the repository to use for publishing. It should not have a trailing slash.

The pypirc path directive refers to the file where the repository URL will be stored. You may need to edit this if you run docassemble on a non-standard operating system.

Facebook, Google, Auth0, Keycloak, Authentik, Zitadel, miniOrange, and Azure login

If you want to enable logging in with Facebook, Google, Auth0, KeyCloak, Authentik, Zitadel, miniOrange, or Microsoft Azure, you will need to tell docassemble your OAuth2 keys for these services:

oauth:
  facebook:
    enable: True
    id: 423759825983740
    secret: 34993a09909c0909b9000a090d09f099
  google:
    enable: True
    id: 23123018240-32239fj28fj4fuhf394h3984eurhfurh.apps.googleusercontent.com
    secret: DGE34gdgerg3GDG545tgdfRf
  auth0:
    enable: True
    id: ceyd6imTYxTNmuj2CYpwH_cEhdWt93e6
    secret: LLGEBDrrgDFDfBsFjErsdSdsntkrtAbfa4ee3ss_adfdSDFEWEsfgHTerjNsd3dD
    domain: example.auth0.com
  keycloak:
    enable: True
    id: example
    secret: ifYzXtUv7WlSkWjMbHwgvgk0aACoXTcF
    domain: "keycloak.mysite.com"
    protocol: "https://"
    realm: master
  authentik:
    enable: True
    id: 29zWxVegWUEaIh75WTOAg9jUnbpAOiuc683JmL2s
    secret: 6dg8kgi8I87xcYxaOu2LSamj7cWE4j6VxC1bPcGTIl4ddMXr3Xaq6IBLo3TupITFylydPciPLULOKxIff3LmT7FxIOmEcnAsUdQf7OG3334S4FlMzEXhGFgFOARnYcVo
    application slug: myapplication-authentik
    domain: myserver:9000
    protocol: "http://"
  zitadel:
    enable: True
    id: 168422810448705515@yourprojectname
    domain: your-instance-name-edy3i2.zitadel.cloud
  miniorange:
    enable: True
    id: TQBwKWUcsECMUIkCctjxLQjPFXdIsnYU
    secret: lTUyixltiJmMMKhXYulCBrKWbFbXPrdy
    domain: myserver.com
  azure:
    enable: True
    id: e378beb1-0bfb-45bc-9b4b-604dcf640c87
    secret: UwHEze20pzSO+lvywJOjcos3v7Kd6Y7tsaCYTG7Panc=

You can disable these login methods by setting enable to False or by removing the configuration entirely.

For more information about how to obtain these keys, see the installation page’s sections on Facebook, Google, Auth0, Keycloak, Authentik, Zitadel, miniOrange and Azure.

Note that in YAML, dictionary keys must be unique. So you can only have one ouath: line in your configuration. Put all of your OAuth2 configuration details (for Google logins, Google Drive integration, OneDrive integration, etc.) within a single oauth directive.

If an external authentication method is used, the decryption key for server-side encryption is based on information sent back by the external authentication provider. Encryption still happens (e.g., when multi_user is not set to True), but the encryption key is not a secret The protection offered by server-side encryption will be less, because a resourceful hacker could figure out the encryption key for decrypting interview answers, based on information stored unencrypted on the server.

By default, users who log with an authentication mechanism other than username/password cannot have their privileges elevated to admin. However, if you want to take the risk that administrators could be locked out of their accounts if the external authentication mechanism fails, you can set allow external auth with admin accounts to True:

allow external auth with admin accounts: True

By default, a use who signs in for the first time (registers) with one external authentication method cannot then log in using a different external authentication method that is connected with the same email address.

However, if you want to allow users to do this, set allow external auth with multiple methods to True:

allow external auth with multiple methods: True

The default is False.

Note that if you enable this feature, users may experience an apparent loss in access to sessions. While they are allowed to log in to the same account, because it has the same email address, the encryption key will be different because the authentication method is different.

Google Drive configuration

To enable the Google Drive synchronization feature, add a googledrive entry to your oauth configuration with your OAuth2 keys for Google Drive.

oauth:
  googledrive:
    enable: True
    id: 23123018240-32239fj28fj4fuhf394h3984eurhfurh.apps.googleusercontent.com
    secret: DGE34gdgerg3GDG545tgdfRf

For more information about obtaining these keys, see the Google Drive section of the installation page.

OneDrive configuration

To enable the OneDrive synchronization feature, add a onedrive entry to your oauth configuration with your OAuth2 keys for OneDrive.

oauth:
  onedrive:
    enable: True
    id: 7df57ca4-d736-84b3-9aec-746c63b8e88d
    secret: "xcyTW522[bip)BIWXY92YsT"

For more information about obtaining these keys, see the OneDrive section of the installation page.

GitHub configuration

To enable the GitHub integration feature, add a github entry to your oauth configuration with your OAuth2 keys for GitHub.

oauth:
  github:
    enable: True
    id: 78df98b2573io5d78a1b
    secret: 75dc98473847a7c938f8beeee87f322fcb980232

For more information about obtaining these keys, see the GitHub integration section of the installation page.

Twilio configuration

There are several features of docassemble that involve integration with the Twilio service, including the send_sms() function for sending text messages, the text messaging interface for interacting with interviewees through text messaging, and the call forwarding feature for connecting interviewees with operators over the phone.

These features are enabled using a twilio configuration directive. Here is an example:

twilio:
  sms: True
  voice: True
  account sid: ACfad8e668d876f5473fb232a311243b58
  auth token: 87559c7a427c25e34e20c654e8b05234
  number: "+12762410114"
  whatsapp number: "+12762268342"
  mms attachments: False
  dispatch:
    color: docassemble.base:data/questions/examples/buttons-code-color.yml
    doors: docassemble.base:data/questions/examples/doors.yml
  default interview: docassemble.base:data/questions/default-interview.yml

The sms: True line tells docassemble that you intend to use the text messaging features. This should be set to True even if you only intend to use WhatsApp.

The voice: True line tells docassemble that you intend to use the call forwarding feature.

The account sid is a value you copy and paste from your Twilio account dashboard.

The auth token is another value you copy and paste from your Twilio account dashboard. This is only necessary if you intend to use the send_sms() function or the phone login feature.

The number is the phone number you purchased. The phone number must be written in E.164 format. This is the phone number with which your users will exchange SMS messages.

The whatsapp number (optional) is the phone number you use for sending WhatsApp messages. The number must be written in E.164 format.

The mms attachments (optional) option allows you to indicate whether MMS should be used to deliver documents to the user when a question has attachments, attachment, or attachment code. When mms attachments is True (which is the default), docassemble will only deliver PDF attachments (because Twilio blocks other content types). If you set mms attachments: False, then instead of using MMS, docassemble will include hyperlinks to PDF, RTF, and DOCX attachments.

The dispatch configuration allows you to direct users to different interviews. For example, with the above configuration, you can tell your prospective users to “text ‘color’ to 276-241-0114.” Users who initiate a conversation by sending the SMS message “help” to the Twilio phone number will be started into the docassemble.base:data/questions/examples/sms.yml interview.

The default interview configuration allows you to set an interview that will be used in case the user’s initial message does not match up with a dispatch entry. If you do not set a default interview, the global default interview will be used. If you want unknown messages to be ignored, set default interview to null.

Multiple Twilio configurations

You can use multiple Twilio configurations on the same server. You might wish to do this if you want to advertise more than one Twilio number to your users. You can do this by specifying the twilio directive as a list of dictionaries, and giving each dictionary a name. In this example, there are two configurations, one named default, and one named bankruptcy:

twilio:
  - name: default
    sms: True
    voice: True
    account sid: ACfad8e668d876f5473fb232a311243b58
    auth token: auth token: 87559c7a427c25e34e20c654e8b05234
    number: "+12762410114"
    default interview: docassemble.base:data/questions/examples/sms.yml
    dispatch:
      color: docassemble.base:data/questions/examples/buttons-code-color.yml
      doors: docassemble.base:data/questions/examples/doors.yml
    default interview: docassemble.base:data/questions/default-interview.yml
  - name: bankruptcy
    sms: True
    voice: False
    account sid: ACfad8e668d876f5473fb232a311243b58
    auth token: auth token: 87559c7a427c25e34e20c654e8b05234
    number: "+12768571217"
    default interview: docassemble.bankruptcy:data/questions/bankruptcy.yml
    dispatch:
      adversary: docassemble.base:data/questions/adversary-case.yml
      chapter7: docassemble.base:data/questions/bankruptcy.yml

When you call send_sms(), you can indicate which configuration should be used:

send_sms(to='202-943-0949', body='Hi there!', config='bankruptcy')

This will cause the message to be sent from 276-857-1217.

If no configuration is named default, the first configuration will be used as the default. The call forwarding feature uses the default configuration.

Fax configuration

The fax sending feature was previously provided by Twilio, but Twilio discontinued support for sending faxes in 2021. If you want to use the send_fax() function, you will need to create an account with ClickSend or Telnyx.

When using ClickSend or Telnyx, set fax provider to either clicksend or telnyx. The default value is clicksend.

ClickSend configuration

Sending faxes through ClickSend is enabled by setting fax provider to clicksend and specifying a clicksend configuration directive.

Here is an example:

fax provider: clicksend
clicksend:
  api username: "[email protected]"
  api key: "F8BE02C1-D95E-D3A9-297B-0028EFB6163B"
  number: "+12762510247"
  from email: "[email protected]"

The api username is the username that you use to log into ClickSend. This is usually the e-mail address that you used when you created the account.

The api key is a code that you can find in the ClickSend dashboard under “API credentials.” You may need to generate the API Key.

The number is the fax phone number you purchased. You can find the phone number under Numbers, Fax in the ClickSend dashboard. The phone number must be written in E.164 format.

The from email is a field that ClickSend uses in its Fax API. The documentation describes it as “An email address where the reply should be emailed to.”

After setting up your phone number, you need to go to Messaging Settings, then go to the Fax tab, then go to the Delivery Reports tab. Then click “Add new rule.” Add a rule called “post fax delivery report” with “Match For” set to “All reports.” Set the “Action” to “URL,” and under “URL,” type in the address of your docassemble server followed by /clicksend_fax_callback. For example, if your server is at https://docassemble.example.com, you would write https://docassemble.example.com/clicksend_fax_callback. Then make sure that the rule is set to “enabled.”

Multiple ClickSend configurations

You can use multiple ClickSend configurations on the same server. You might wish to do this if you want to use more than one ClickSend fax number and/or account to send faxes. You can do this by specifying the clicksend directive as a list of dictionaries, and giving each dictionary a name. In this example, there are two configurations, one named default, and one named bankruptcy:

clicksend:
  - name: default
    api username: "[email protected]"
    api key: "F8BE02C1-D95E-D3A9-297B-0028EFB6163B"
    number: "+12762510247"
    from email: "[email protected]"
  - name: bankruptcy
    api username: "[email protected]"
    api key: "E7CE2453-253F-E3C1-3562-0129EDB727BC"
    number: "+13012503257"
    from email: "[email protected]"

When you call send_fax(), you can indicate which configuration should be used:

send_fax(to='202-943-0949', welcome_fax, config='bankruptcy')

This will cause the fax to be sent from 301-250-3257.

If no configuration is named default, the first configuration will be used as the default.

Telnyx configuration

Sending faxes through Telnyx is enabled by setting fax provider to telnyx and specifying a telnyx configuration directive.

Here is an example:

fax provider: telnyx
telnyx:
  number: "+16052519139"
  api key: "KEY017D2E7B3D5C992F50E721E560AD262F_pYbe5lWyefduFhfUv2eUdd"
  app id: "1823943157766443451"

The api key is a code that you can find in the Telnyx portal under “API Keys.” You may need to generate the API Key.

The app id is the “App ID” of the “Fax Application” you want to use.

The number is the fax phone number you purchased. You can find the phone number under Numbers in the Telnyx portal. The phone number must be written in E.164 format.

To set up Telnyx for use with docassemble, go to Telnyx and sign up for an account. You will need to provide your credit card information and add some money to your Balance.

Go to “Numbers” and then “Search & Buy Numbers” to set up a phone number to use for faxing. Make a note of the phone number; you will need to put the phone number into your telnyx configuration as the number, in E.164 format (e.g., +12025551515).

Go to “API Keys” and create an API key for use with “Telnyx API v2.” Make a note of the API key; you will need to put the API key into your telnyx configuration as the api key.

Go to “Programmable Fax” and click “Add new App.” Give your “Fax Application” a name like “docassemble fax” (or whatever you want). Make a note of the “App ID”; this is what you will need to set in your telnyx configuration as the app id. Under “Send a webhook to the URL,” put in the URL of your server, followed by /telnyx_fax_callback. For example, if your server is at https://docassemble.example.com, you would write https://docassemble.example.com/telnyx_fax_callback. Then, under “Outbound Settings,” set the “Outbound Voice Profile” to “Default” or to whatever other Outbound Voice Profile you want to use. You can click “Outbound Voice Profile” in the navigation menu to see the Outbound Voice Profiles that are configured on your account.

Multiple Telnyx configurations

You can use multiple Telnyx configurations on the same server. You might wish to do this if you want to use more than one Telnyx fax number and/or account to send faxes. You can do this by specifying the telnyx directive as a list of dictionaries, and giving each dictionary a name. In this example, there are two configurations, one named default, and one named bankruptcy:

telnyx:
  - name: default
    number: "+16052519139"
    api key: "KEY017D2E7B3D5C992F50E721E560AD262F_pYbe5lWyefduFhfUv2eUdd"
    app id: "1823943157766443451"
  - name: bankruptcy
    number: "+12025552033"
    api key: "KEY017D2F7B5D5C992Y51E731E860BD365E_qYZbf7lUyeRdrFjfJv3eYfe"
    app id: "1924953657746434242"

When you call send_fax(), you can indicate which configuration should be used:

send_fax(to='202-943-0949', welcome_fax, config='bankruptcy')

This will cause the fax to be sent from 301-250-3257.

If no configuration is named default, the first configuration will be used as the default.

User agent for downloading files

Functions such as path_and_mimetype() will download files from the internet. When the server downloads the file, it reports its user agent as curl/7.64.0 by default. This might cause problems for your interviews because some web sites may not want to serve files to a user that identifies itself as a “robot.”

To use a different user agent, you can set the user agent configuration directive. For example:

user agent: Mozilla/5.0 (X11; Linux x86_64; rv:59.0) Gecko/20100101 Firefox/59.0

Protecting against multiple login attempts

By default, users who unsuccessfully log in are blocked after 10 failed attempts. This is a form of security protection against bots that try to guess passwords by repeatedly logging in.

The block is based on IP address, so if users share an outgoing WAN IP address, all of those users are blocked.

The threshold of 10 failed attempts can be configured with attempt limit.

The time period for blocking defaults to 86,400 seconds (one day). The number of seconds of blocking can be configured with ban period.

attempt limit: 5
ban period: 3600

Restarting the server will clear out all IP address bans. The server can be restarted through the API or by going to the Configuration and pressing Save (without making any changes).

This feature can be turned off by setting ip address ban enabled to false.

ip address ban enabled: False

Phone login verification codes

When you use the phone login feature, the user needs to enter a 5 digit code that they receive via SMS. The number of digits in this code can be configured with verification code digits. The code is only valid for a limited period of time. This period of time defaults to 180 seconds and is configurable with verification code timeout.

verification code digits: 6
verification code timeout: 360

Infinite loop protection

Since docassemble allows you to write code in a declarative fashion, it needs to do a lot of looping and recursion. If you make a mistake in your interview logic and use circular reasoning, you may send docassemble into an infinite loop. Sometimes these infinite loops can be detected and a warning raised, but other times they are hard to detect because even deliberate code could loop or recurse for a long time. Thus, docassemble has hard limits on the amount of looping and recursion it will do. By default, these limits are set to 500 loops and 500 recursions. If you get an error that “there appears to be an infinite loop” or “there appears to be a circularity,” then these limits were exceeded. If you want to change these limits on a global level in your system, you can include use the loop limit and recursion limit directives:

loop limit: 600
recursion limit: 600

It is important that some reasonable limit be in place, because if the server is in development mode, an infinite loop could result in the memory of the machine being exceeded, which could cause the system to crash.

You can also change these limits on a per-interview basis with the loop limit and recursion limit features

Editable file types

By default, the only file types that can be edited in the folders of the Playground are text MIME types and file types that are used in docassemble, like YAML, CSS, JavaScript, and HTML. Other types are presumed to be non-editable, and there is no “Edit” button next to the file in the Playground folder.

You can make additional file types editable using editable mimetypes or editable extensions. For example, if you wanted to be able to edit RTF files in the Playground folders, you could add:

editable mimetypes:
  - application/rtf

Or if you had a text file format that was not associated with a MIME type and that ended in .law, you could add:

editable extensions:
  - law

Inserting Markdown templates in Word files

Originally, when inserting the results of a template in DOCX file that you assemble using docx template file, you would include the template as follows:

{{r the_template }}

The template text would be inserted as a single “run,” meaning that it cannot contain any proper paragraph breaks, only manual line breaks.

There is a newer system of inserting Markdown templates into docx template files, which can handle multiple paragraphs and a wider variety of Markdown. However, this new system is not backwards compatible with the original system, because under the new system you need to include the template as follows:

{{p the_template }}

To upgrade to the new system today, set the following in your Configuration:

new markdown to docx: True

The new system will become the default at some point in the future, so when you have time, you should adapt your DOCX files if you are using {{r ... }} to insert templates.

Choosing Apache instead of NGINX

Applicable only to the jhpyle/docassemble Docker image.

By default, the web server used with Docker is NGINX. You can change this to Apache by adding the following to your Configuration:

web server: apache

The default is nginx. After changing this, you will need to do a docker stop -t 600 followed by a docker start. This feature requires system version 0.5.0 or later.

See also the DAWEBSERVER environment variable.

Support for Apache is deprecated and will be removed in a future version.

Using NGINX to serve files

The X-Accel-Redirect feature of NGINX allows Python, running inside of uwsgi, to serve a file in response to an HTTP request by returning an empty response instead of a response with the content of the file in the body. Python tells NGINX the file to be served by setting the X-Accel-Redirect header of the empty response to the file system path. NGINX intercepts the response and serves the file to the user. This increases efficiency because Python code does not have to process the contents of the file.

The value of use nginx to serve files is False unless set to True in the Configuration.

use nginx to serve files: True

If you use NGINX, the value of this should be True unless the X-Accel-Redirect feature causes problems for some reason. The default value is False only for backwards compatibility with system versions prior to 1.7.8; a change to the NGINX configuration is required to enable this feature to work. In the default Configuration for new systems, use nginx to serve files is True.

See also the USENGINXTOSERVEFILES environment variable.