Remove llms.txt mention

Author: mnocon

build_package_docs.py

@@ -53,6 +53,10 @@
 # Same fence detection as llmstxt_preprocess.renumber_ordered_lists.
 _FENCE_OPEN_RE = re.compile(r"^(\s*)(`{3,}|~{3,})")
 
+_LLMS_TXT_POINTER_RE = re.compile(
+    r"(?m)^> For the complete documentation index, see \[llms\.txt\]\([^)]*\)\.\n\n?"
+)
+
 
 class DocSet(NamedTuple):
     """One documentation set shipped in the package."""
@@ -187,6 +191,11 @@ def _replace(match):
     return _MD_LINK_RE.sub(_replace, line)
 
 
+def strip_llms_txt_pointer(content):
+    """Drop the llmstxt plugin's "see llms.txt" pointer line (see _LLMS_TXT_POINTER_RE)."""
+    return _LLMS_TXT_POINTER_RE.sub("", content)
+
+
 def rewrite_page(content, page_path, docsets, class_paths):
     """Rewrite a page's links, leaving fenced code blocks untouched."""
     lines = content.split("\n")
@@ -289,12 +298,8 @@ def build(dev_site, dev_plugins, user_site, user_plugins, class_map_path, versio
     for docset, site in docsets:
         for page_rel in sorted(docset.pages):
             page_path = f"{docset.root}/{page_rel}"
-            rewritten[page_path] = rewrite_page(
-                (site / page_rel).read_text(encoding="utf-8"),
-                page_path,
-                (developer, user),
-                class_paths,
-            )
+            content = strip_llms_txt_pointer((site / page_rel).read_text(encoding="utf-8"))
+            rewritten[page_path] = rewrite_page(content, page_path, (developer, user), class_paths)
         llms[docset.root] = rewrite_llms_txt(
             (site / "llms.txt").read_text(encoding="utf-8"), docset
         )

tests/python/test_build_package_docs.py

@@ -3,6 +3,7 @@
     check_relative_doc_links,
     rewrite_llms_txt,
     rewrite_page,
+    strip_llms_txt_pointer,
 )
 
 DEVELOPER = DocSet(
@@ -219,6 +220,53 @@ def test_unknown_page_stays_absolute(self):
         assert rewrite_llms_txt(content, DEVELOPER) == content
 
 
+class TestLlmsTxtPointer:
+    """Mirrors the shapes inject_page_metadata() (tools/llms_txt/llmstxt_preprocess.py) produces."""
+
+    def test_pointer_is_removed_without_leaving_a_double_blank_line(self):
+        content = (
+            "# Getting started\n"
+            "\n"
+            "> For the complete documentation index, see [llms.txt](https://doc.ibexa.co/en/5.0/llms.txt).\n"
+            "\n"
+            "Body text here.\n"
+        )
+        assert strip_llms_txt_pointer(content) == "# Getting started\n\nBody text here.\n"
+
+    def test_description_and_editions_lines_are_kept(self):
+        content = (
+            "# Heading\n"
+            "\n"
+            "> For the complete documentation index, see [llms.txt](https://doc.ibexa.co/en/5.0/llms.txt).\n"
+            "\n"
+            "Some page description.\n"
+            "\n"
+            "Editions: Content, Experience\n"
+        )
+        assert strip_llms_txt_pointer(content) == (
+            "# Heading\n\nSome page description.\n\nEditions: Content, Experience\n"
+        )
+
+    def test_nested_project_url_is_also_stripped(self):
+        content = (
+            "# Edit images\n"
+            "\n"
+            "> For the complete documentation index, see "
+            "[llms.txt](https://doc.ibexa.co/projects/userguide/en/5.0/llms.txt).\n"
+            "\n"
+            "Body.\n"
+        )
+        assert strip_llms_txt_pointer(content) == "# Edit images\n\nBody.\n"
+
+    def test_pointer_without_a_heading_is_stripped_cleanly(self):
+        content = (
+            "> For the complete documentation index, see [llms.txt](https://doc.ibexa.co/en/5.0/llms.txt).\n"
+            "\n"
+            "Just body text.\n"
+        )
+        assert strip_llms_txt_pointer(content) == "Just body text.\n"
+
+
 class TestSelfCheck:
     def test_valid_links_pass(self):
         pages = {