Docutils Transforms

Author:	David Goodger
Contact:	docutils-develop@lists.sourceforge.net
Revision:	10166
Date:	2025-06-16
Copyright:	This document has been placed in the public domain.

Contents

Introduction
Transforms Listed in Priority Order
- Transform Priority Range Categories
Transforms Added by Components
Transforms Reference
Appendix
- PreBibliographic Elements

Introduction

Transforms change the document tree in-place, add to the tree, or prune it. Transforms resolve references and footnote numbers, process interpreted text, and do other context-sensitive processing. Each transform is a subclass of docutils.transforms.Transform.

Most transforms are added by components, others (e.g. parts.Contents) are added by the parser if a corresponding directive is found in the document.

To add a transform, components (objects inheriting from docutils.Component like Readers, Parsers, Writers, Input, Output) overwrite the get_transforms() method of their base class. After the Reader has finished processing, the Publisher calls Transformer.populate_from_components() with a list of components and all transforms returned by the component's get_transforms() method are stored in a Transformer object attached to the document tree.

For more info about transforms and the Transformer object, see also PEP 258. [1]

[1]	The default_transforms() attribute of component classes mentioned in PEP 258 is deprecated. Use the get_transforms() method instead.

Transforms Listed in Priority Order

Transform classes each have a default_priority attribute which is used by the Transformer to apply transforms in order (low to high). The default priority can be overridden when adding transforms to the Transformer object.

Transform: module.Class	Added By	Priority
misc.ClassAttribute	"class" (d/p)	210
references.Substitutions	standalone (r), pep (r)	220
references.PropagateTargets	standalone (r), pep (r)	260
frontmatter.DocTitle	standalone (r)	320
frontmatter.DocInfo	standalone (r)	340
frontmatter.SectSubTitle	standalone (r)	350
peps.Headers	pep (r)	360
peps.Contents	pep (r)	380
universal.StripClassesAndElements	writers (w)	420
references.AnonymousHyperlinks	standalone (r), pep (r)	440
references.IndirectHyperlinks	standalone (r), pep (r)	460
peps.TargetNotes	pep (r)	520
references.TargetNotes	peps.TargetNotes (t/p)	0
misc.CallBack	peps.TargetNotes (t/p)	1
references.TargetNotes	"target-notes" (d/p)	540
references.Footnotes	standalone (r), pep (r)	620
references.ExternalTargets	standalone (r), pep (r)	640
references.InternalTargets	standalone (r), pep (r)	660
parts.SectNum	"sectnum" (d/p)	710
parts.Contents	"contents" (d/p), peps.Contents (t/p)	720
universal.StripComments	readers (r)	740
peps.PEPZero	peps.Headers (t/p)	760
references.CitationReferences	latex2e (w)	770
components.Filter	not used	780
universal.Decorations	readers (r)	820
misc.Transitions	standalone (r), pep (r)	830
universal.Validate	parsers (p)	835
universal.ExposeInternals	readers (r)	840
references.DanglingReferences	standalone (r), pep (r), deprecated [2]	850
universal.SmartQuotes	rst (p)	855
universal.Messages	writers (w)	860
universal.FilterMessages	writers (w)	870
universal.TestMessages	DocutilsTestSupport	880
writer_aux.Compound	not used, to be removed	910
writer_aux.Admonitions	_html_base (w), latex2e (w)	920
misc.CallBack	n/a	990

Key:

(r): Reader
(w): Writer
(d): Directive
(t): Transform
(/p): Via a <pending> element

[2]	references.DanglingReferences will be replaced by separate transforms for resolving refnames to refids and for reporting unresolved references to make space for reference-resolving transforms added by extensions or applications (cf. Release Notes).

Transform Priority Range Categories

Priority
From	To	Category
0	99	immediate execution (added by another transform)
100	199	very early (non-standard)
200	299	very early
300	399	early
400	699	main
700	799	late
800	899	very late
900	999	very late (non-standard)

Transforms Added by Components

readers.Reader:: universal.StripComments (740)

universal.Decorations (820)

universal.ExposeInternals (840)
readers.ReReader:: None
readers.standalone.Reader:: references.Substitutions (220)

references.PropagateTargets (260)

frontmatter.DocTitle (320)

frontmatter.DocInfo (340)

frontmatter.SectSubTitle (350)

references.AnonymousHyperlinks (440)

references.IndirectHyperlinks (460)

references.Footnotes (620)

references.ExternalTargets (640)

references.InternalTargets (660)

misc.Transitions (830)

references.DanglingReferences (850)
readers.pep.Reader:: references.Substitutions (220)

references.PropagateTargets (260)

peps.Headers (360)

peps.Contents (380)

references.AnonymousHyperlinks (440)

references.IndirectHyperlinks (460)

peps.TargetNotes (520)

references.Footnotes (620)

references.ExternalTargets (640)

references.InternalTargets (660)

misc.Transitions (830)

references.DanglingReferences (850)
parsers.Parser: universal.Validate (835)
parsers.rst.Parser: universal.SmartQuotes (855)
writers.Writer:: universal.StripClassesAndElements (420)

universal.Messages (860)

universal.FilterMessages (870)
writers.UnfilteredWriter: None
writers.latex2e.Writer: writer_aux.Admonitions (920)

references.CitationReferences (770)
writers._html_base.Writer:: writer_aux.Admonitions (920)
writers.odf_odt.Writer:: removes references.DanglingReferences

Transforms Reference

Note

This section is still incomplete.

See also Transforms Listed in Priority Order and the sources in docutils/transforms.

DocInfo

Default priority:
Module:	frontmatter
Added by:	standalone Reader
	340
Configuration setting:
	docinfo_xform (default: True)

Given a document starting [3] with a field list, the DocInfo transform converts fields with registered bibliographic field names to the corresponding document tree elements becoming child elements of the <docinfo> element (except for "dedication" and "abstract", which become <topic> elements after <docinfo>).

[3]	A document title and subtitle, header and footer, as well as body elements that do not show up in the output before the field list don't prevent the DocInfo transformation. See the DocTitle examples and PreBibliographic Elements in the Appendix for details.

For example, this document fragment:

<document>
    <title>
        Document Title
    <field_list>
        <field>
            <field_name>
                Author
            <field_body>
                <paragraph>
                    A. Name
        <field>
            <field_name>
                Status
            <field_body>
                <paragraph>
                    $RCSfile$
    ...

will be transformed to:

<document>
    <title>
        Document Title
    <docinfo>
        <author>
            A. Name
        <status>
            frontmatter.py
    ...

DocTitle

Default priority:
Module:	frontmatter
Added by:	standalone Reader
	320
Configuration setting:
	doctitle_xform (default: True)

Under the conditions explained below, the DocTitle transform converts the document's first section title(s) to a document title and subtitle.

If the document contains a single top-level section as its first body element, [4] the top-level section's title is used as document title and default metadata title. The top-level section's contents become the document's immediate contents.
If step 1 successfully determines the document title, the transform checks for a subtitle: If the lone top-level section itself contains a single second-level section as its first element, [4] that section's title is promoted to the document's subtitle, and that section's contents become the document's immediate contents.

The transform can be disabled with the doctitle_xform configuration setting or the corresponding --no-doc-title command line option.

[4]	(1, 2, 3) Header and footer as well as body elements that do not show up in the output before the section don't stop the transformation of the section title to document title or subtitle. See PreBibliographic Elements in the Appendix for details.

Examples

The input text

Top-Level Title
===============
A paragraph.

is parsed to the following document tree:

<document>
    <section names="top-level\ title">
        <title>
            Top-Level Title
        <paragraph>
            A paragraph.

The DocTitle transform converts it to:

<document names="top-level\ title" title="Top-Level Title">
    <title>
        Top-Level Title
    <paragraph>
        A paragraph.

Given this input with a lone section and sub-section:

Top-Level Title
===============
Second-Level Title
~~~~~~~~~~~~~~~~~~

A paragraph.

the result after parsing and running the DocTitle transform is:

<document names="top-level\ title">
    <title>
        Top-Level Title
    <subtitle names="second-level\ title">
        Second-Level Title
    <paragraph>
        A paragraph.

(Note that the implicit hyperlink target generated by the "Second-Level Title" is preserved on the "subtitle" element itself.)

The following examples do not comply with the conditions:

More than one top-level section:
```
Top-Level Title
===============
A paragraph.

Another Top-Level Title
=======================
Another paragraph.
```
The DocTitle transform will leave the document tree as-is. The document has no title. It is recommended to set the metadata title with the "title" directive.
More than one second-level section:
```
Top-Level Title
===============
Second-Level Title
~~~~~~~~~~~~~~~~~~
A paragraph.

Another Second-Level Title
~~~~~~~~~~~~~~~~~~~~~~~~~~
Another paragraph.
```
Step 2 is skipped. "Top-Level Title" becomes the document title; "Second-Level Title" and "Another Second-Level Title" become titles of top-level sections.

This is what you normally want in a document with title but no subtitle.

Body elements before the first section: [4]

.. note:: This element would not stop the transform, if it were
   nested in a header.

Top-Level Title
===============
A paragraph.

The DocTitle transform will leave the document tree as-is. The document has no title.

A configuration file entry "doctitle_xform: False" or converting with the CLI command
```
docutils --no-doc-title example.rst > example.html
```
The DocTitle transform is skipped. The document has no title.

SectSubTitle

Default priority:
Module:	frontmatter
Added by:	standalone Reader
	350
Configuration setting:
	sectsubtitle_xform (default: False)

The SectSubTitle transform works like step 2 of the DocTitle transform, but for sections.

For example,

<section>
    <title>
        Title
    <section>
        <title>
            Subtitle
        ...

is transformed into

<section>
    <title>
        Title
    <subtitle>
        Subtitle
    ...

This transform is disabled by default.

Appendix

PreBibliographic Elements

The document tree elements <comment>, <decoration>, <footer>, <header>, <meta>, <pending>, <raw>, <substitution_definition>, <subtitle>, <system_message>, <target>, and <title> are ignored when the DocTitle and DocInfo transforms check for elements before frontmatter candidates.

This means that in the reStructuredText source, comments, hyperlink targets, substitution definitions, and the directives "class", "default-role", "footer", "header", "meta", "raw", "sectnum", "sectnum", "target-notes", and "title" may be placed before the document title or bibliographic fields.

The DocTitle transform inserts document title and subtitle before these elements. DocInfo inserts the bibliographic fields before <comment>, <raw>, <substitution_definition>, and <target>.

For example, the document source:

.. a comment
.. |today| date::
.. _hypertarget: http://example.org
.. class:: spam
.. role:: ham
.. default-role:: ham
.. footer:: footer text
.. header:: header text
.. meta:: :test: value
.. raw:: html

   raw html text
.. sectnum::
.. target-notes::
.. title:: metadata document title

Top-Level Title
===============
Second-Level Title
~~~~~~~~~~~~~~~~~~

:date: |today|

is parsed and transformed to:

<document ids="top-level-title" names="top-level\ title"
          title="metadata document title">
    <title>
        Top-Level Title
    <subtitle ids="second-level-title" names="second-level\ title">
        Second-Level Title
    <meta content="value" name="test">
    <decoration>
        <header>
            <paragraph>
                header text
        <footer>
            <paragraph>
                footer text
    <docinfo>
        <date>
            2024-03-01
    <comment xml:space="preserve">
        a comment
    <substitution_definition names="today">
        2024-03-01
    <target ids="hypertarget" names="hypertarget" refuri="http://example.org">
    <raw classes="spam" format="html" xml:space="preserve">
        raw html text