Add an 'include-read' event (#11657)

Author: AA-Turner

Diff for: CHANGES

@@ -28,6 +28,10 @@ Bugs fixed
   when an object claims to be an instance of ``type``,
   but is not a class.
   Patch by  James Braza.
+* 11620: Cease emitting :event:`source-read` events for files read via
+  the :dudir:`include` directive.
+* 11620: Add a new :event:`include-read` for observing and transforming
+  the content of included files via the :dudir:`include` directive.
 
 Testing
 -------
@@ -143,7 +147,7 @@ Features added
 * #11572: Improve ``debug`` logging of reasons why files are detected as out of
   date.
   Patch by Eric Larson.
-* #10678: Emit "source-read" events for files read via
+* #10678: Emit :event:`source-read` events for files read via
   the :dudir:`include` directive.
   Patch by Halldor Fannar.
 * #11570: Use short names when using :pep:`585` built-in generics.

Diff for: doc/extdev/appapi.rst

@@ -260,6 +260,23 @@ Here is a more detailed list of these events.
 
    .. versionadded:: 0.5
 
+.. event:: include-read (app, relative_path, parent_docname, content)
+
+   Emitted when a file has been read with the :dudir:`include` directive.
+   The *relative_path* argument is a :py:class:`~pathlib.Path` object representing
+   the relative path of the included file from the :term:`source directory`.
+   The *parent_docname* argument is the name of the document that
+   contains the :dudir:`include` directive.
+   The *source* argument is a list whose single element is
+   the contents of the included file.
+   You can process the contents and replace this item
+   to transform the included content,
+   as with the :event:`source-read` event.
+
+   .. versionadded:: 7.2.5
+
+   .. seealso:: The :dudir:`include` directive and the :event:`source-read` event.
+
 .. event:: object-description-transform (app, domain, objtype, contentnode)
 
    Emitted when an object description directive has run.  The *domain* and

Diff for: sphinx/directives/other.py

@@ -1,7 +1,8 @@
 from __future__ import annotations
 
 import re
-from os.path import abspath
+from os.path import abspath, relpath
+from pathlib import Path
 from typing import TYPE_CHECKING, Any, cast
 
 from docutils import nodes
@@ -19,7 +20,6 @@
 from sphinx.util.docutils import SphinxDirective
 from sphinx.util.matching import Matcher, patfilter
 from sphinx.util.nodes import explicit_title_re
-from sphinx.util.osutil import os_path
 
 if TYPE_CHECKING:
     from docutils.nodes import Element, Node
@@ -373,24 +373,25 @@ class Include(BaseInclude, SphinxDirective):
 
     def run(self) -> list[Node]:
 
-        # To properly emit "source-read" events from included RST text,
+        # To properly emit "include-read" events from included RST text,
         # we must patch the ``StateMachine.insert_input()`` method.
         # In the future, docutils will hopefully offer a way for Sphinx
         # to provide the RST parser to use
         # when parsing RST text that comes in via Include directive.
         def _insert_input(include_lines, source):
             # First, we need to combine the lines back into text so that
-            # we can send it with the source-read event.
+            # we can send it with the include-read event.
             # In docutils 0.18 and later, there are two lines at the end
             # that act as markers.
-            # We must preserve them and leave them out of the source-read event:
+            # We must preserve them and leave them out of the include-read event:
             text = "\n".join(include_lines[:-2])
 
-            # The docname to pass into the source-read event
-            docname = self.env.path2doc(abspath(os_path(source)))
-            # Emit the "source-read" event
+            path = Path(relpath(abspath(source), start=self.env.srcdir))
+            docname = self.env.docname
+
+            # Emit the "include-read" event
             arg = [text]
-            self.env.app.events.emit("source-read", docname, arg)
+            self.env.app.events.emit('include-read', path, docname, arg)
             text = arg[0]
 
             # Split back into lines and reattach the two marker lines
@@ -401,8 +402,8 @@ def _insert_input(include_lines, source):
             # the *Instance* method and this call is to the *Class* method.
             return StateMachine.insert_input(self.state_machine, include_lines, source)
 
-        # Only enable this patch if there are listeners for 'source-read'.
-        if self.env.app.events.listeners.get('source-read'):
+        # Only enable this patch if there are listeners for 'include-read'.
+        if self.env.app.events.listeners.get('include-read'):
             # See https://github.com/python/mypy/issues/2427 for details on the mypy issue
             self.state_machine.insert_input = _insert_input  # type: ignore[method-assign]
 

Diff for: sphinx/events.py

@@ -38,6 +38,7 @@ class EventListener(NamedTuple):
     'env-before-read-docs': 'env, docnames',
     'env-check-consistency': 'env',
     'source-read': 'docname, source text',
+    'include-read': 'relative path, parent docname, source text',
     'doctree-read': 'the doctree before being pickled',
     'env-merge-info': 'env, read docnames, other env instance',
     'missing-reference': 'env, node, contnode',

Diff for: tests/roots/test-directive-include/bar.txt

@@ -0,0 +1 @@
+Text from :file:`bar.txt`.

Diff for: tests/roots/test-directive-include/baz/baz.rst

@@ -3,4 +3,4 @@ Baz
 
 .. include:: foo.rst
 
-Baz was here.
+Baz was here.

Diff for: tests/test_directive_other.py

@@ -1,4 +1,5 @@
 """Test the other directives."""
+from pathlib import Path
 
 import pytest
 from docutils import nodes
@@ -151,34 +152,41 @@ def test_toctree_twice(app):
 
 
 @pytest.mark.sphinx(testroot='directive-include')
-def test_include_source_read_event(app):
-    sources_reported = {}
-
-    def source_read_handler(app, doc, source):
-        sources_reported[doc] = source[0]
-
-    app.connect("source-read", source_read_handler)
-    text = (".. include:: baz/baz.rst\n"
-            "   :start-line: 4\n\n"
-            ".. include:: text.txt\n"
-            "   :literal:    \n")
+def test_include_include_read_event(app):
+    sources_reported = []
+
+    def source_read_handler(_app, relative_path, parent_docname, source):
+        sources_reported.append((relative_path, parent_docname, source[0]))
+
+    app.connect("include-read", source_read_handler)
+    text = """\
+.. include:: baz/baz.rst
+   :start-line: 4
+.. include:: text.txt
+   :literal:
+.. include:: bar.txt
+"""
     app.env.find_files(app.config, app.builder)
     restructuredtext.parse(app, text, 'index')
-    assert "index" in sources_reported
-    assert "text.txt" not in sources_reported  # text was included as literal, no rst parsing
-    assert "baz/baz" in sources_reported
-    assert sources_reported["baz/baz"] == "\nBaz was here."
+
+    included_files = {filename.as_posix()
+                      for filename, p, s in sources_reported}
+    assert 'index.rst' not in included_files  # sources don't emit 'include-read'
+    assert 'baz/baz.rst' in included_files
+    assert 'text.txt' not in included_files  # text was included as literal, no rst parsing
+    assert 'bar.txt' in included_files  # suffix not in source-suffixes
+    assert (Path('baz/baz.rst'), 'index', '\nBaz was here.') in sources_reported
 
 
 @pytest.mark.sphinx(testroot='directive-include')
-def test_include_source_read_event_nested_includes(app):
+def test_include_include_read_event_nested_includes(app):
 
-    def source_read_handler(app, doc, source):
+    def source_read_handler(_app, _relative_path, _parent_docname, source):
         text = source[0].replace("#magical", "amazing")
         source[0] = text
 
-    app.connect("source-read", source_read_handler)
-    text = (".. include:: baz/baz.rst\n")
+    app.connect("include-read", source_read_handler)
+    text = ".. include:: baz/baz.rst\n"
     app.env.find_files(app.config, app.builder)
     doctree = restructuredtext.parse(app, text, 'index')
     assert_node(doctree, addnodes.document)