Extracted Markdown converter to a separate package (#3292)

Author: mnocon

.gitignore

@@ -19,3 +19,5 @@ lychee.toml
 lychee-report.md
 .phpunit.result.cache
 code_samples/_inline_php/
+/build/
+*.egg-info/

hooks.py

@@ -1,96 +1,5 @@
-"""
-MkDocs hooks for Ibexa developer documentation.
+"""MkDocs hooks entry point — delegates to the installable llms_txt package."""
 
-- Keeps the llmstxt plugin's ``sections`` config in sync with the ``nav``
-  defined in ``mkdocs.yml``.
-- Post-processes the Markdown generated by the llmstxt plugin.
+from llms_txt.hooks import on_config, on_page_content
 
-All content transformations live in ``llmstxt_preprocess.py``; this module is
-only MkDocs event glue.
-"""
-
-from __future__ import annotations
-
-import sys
-from pathlib import Path, PurePosixPath
-from typing import TYPE_CHECKING
-
-_here = Path(__file__).parent
-if str(_here) not in sys.path:
-    sys.path.insert(0, str(_here))
-
-from llmstxt_preprocess import (
-    absolutize_image_urls,
-    editions_from_frontmatter,
-    expand_macros,
-    inject_page_metadata,
-    renumber_ordered_lists,
-)
-from update_llmstxt_config import convert_nav_to_llmstxt_sections
-
-if TYPE_CHECKING:
-    from mkdocs.config.defaults import MkDocsConfig
-    from mkdocs.structure.pages import Page
-
-
-def on_config(config: "MkDocsConfig") -> None:
-    """Populate llmstxt sections from nav before the build starts.
-
-    The llmstxt plugin reads ``config.sections`` in its ``on_files`` event,
-    which fires after ``on_config``, so injecting here is the right place.
-    """
-    nav = config.get("nav")
-    if not nav:
-        return
-
-    llmstxt = config["plugins"].get("llmstxt")
-    if llmstxt is None:
-        return
-
-    docs_dir = Path(config["docs_dir"])
-    llmstxt.config.sections = convert_nav_to_llmstxt_sections(nav, docs_dir)
-
-
-def on_page_content(html: str, *, page: "Page", config: "MkDocsConfig", **kwargs) -> None:
-    """Reformat the Markdown content generated by the llmstxt plugin for this page.
-
-    Hooks run after plugins for every event, so at this point the llmstxt
-    plugin has already generated Markdown and stored it in ``_md_pages``.
-    We reformat here so the plugin writes the corrected content in its own
-    ``on_post_build`` (which also runs before ours, for the same reason).
-    """
-    llmstxt = config["plugins"].get("llmstxt")
-    if llmstxt is None:
-        return
-
-    src_uri = page.file.src_uri
-    page_info = llmstxt._md_pages.get(src_uri)
-    if page_info is None:
-        return
-
-    frontmatter = _read_frontmatter(page, config)
-    editions = editions_from_frontmatter(frontmatter)
-    description = expand_macros(str(frontmatter.get("description") or ""), config.get("extra") or {})
-    if "[[=" in description:
-        # Unresolved macros must not leak into the output.
-        description = ""
-    content = inject_page_metadata(page_info.content, description, editions)
-    content = renumber_ordered_lists(content)
-    # Same base URL and page directory the plugin uses for making link hrefs absolute.
-    page_dir = PurePosixPath(page.file.dest_uri).parent.as_posix()
-    content = absolutize_image_urls(content, llmstxt._base_url, page_dir)
-    if content != page_info.content:
-        llmstxt._md_pages[src_uri] = page_info._replace(content=content)
-
-
-def _read_frontmatter(page: "Page", config: "MkDocsConfig") -> dict:
-    """Read the page's YAML frontmatter from its source file."""
-    src_path = Path(config["docs_dir"]) / page.file.src_path
-    try:
-        from mkdocs.utils import meta as mkdocs_meta
-        with open(src_path, encoding="utf-8") as f:
-            raw = f.read()
-        _, frontmatter = mkdocs_meta.get_data(raw)
-        return frontmatter
-    except Exception:
-        return {}
+__all__ = ["on_config", "on_page_content"]