[3.15] gh-142714: Exclude stripped PEG grammar text from translation catalogs (GH-159305)

Doc/tools/extensions/lexers/__init__.py

@@ -1,12 +1,34 @@
+from docutils import nodes
+from sphinx.transforms import SphinxTransform
+
 from .asdl_lexer import ASDLLexer
 from .peg_lexer import PEGLexer
 
 
+class StripPEGBlocks(SphinxTransform):
+    # Strip PEG blocks like PEGLexer does when highlighting them, so that
+    # other builders (gettext in particular) get the text the page shows.
+    # Must run before the i18n Locale transform (20).
+    default_priority = 10
+
+    def apply(self):
+        lexer = PEGLexer()
+        for node in self.document.findall(nodes.literal_block):
+            if node.get("language") != "peg":
+                continue
+            tokens = lexer.get_tokens_unprocessed(node.rawsource)
+            text = "".join(value for _, _, value in tokens)
+            node.rawsource = text
+            node[:] = [nodes.Text(text)]
+
+
 def setup(app):
     # Used for highlighting Parser/Python.asdl in library/ast.rst
     app.add_lexer("asdl", ASDLLexer)
     # Used for highlighting Grammar/python.gram in reference/grammar.rst
     app.add_lexer("peg", PEGLexer)
+    # Strips hidden grammar parts before translation catalogs are built
+    app.add_transform(StripPEGBlocks)
 
     return {
         "version": "1.0",