From 72a7de61468804482eefd781353441ec4a64066a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=D0=A1=D0=B5=D1=80=D0=B3=D0=B5=D0=B9=20=D0=9C=D0=B0=D1=80?=
 =?UTF-8?q?=D0=B8=D0=BD=D0=BA=D0=B5=D0=B2=D0=B8=D1=87?=
 <sergey.marinkevich@eltex.loc>
Date: Wed, 15 Jul 2026 18:14:08 +0700
Subject: [PATCH] Include parent sections in chapter filenames

Preserve outline ancestry when generating chapter slugs so split files retain their section context.

Co-authored-by: Cursor <cursoragent@cursor.com>
---
 extract.py | 29 ++++++++++++++++++++++++-----
 1 file changed, 24 insertions(+), 5 deletions(-)

diff --git a/extract.py b/extract.py
index e76137f..14c3192 100644
--- a/extract.py
+++ b/extract.py
@@ -192,10 +192,17 @@ def render_page_png(
 
 
 def toc_sections(doc: fitz.Document, level: int) -> list[dict]:
-    """Return unique page boundaries from the PDF outline at one level."""
+    """Return unique page boundaries and ancestor titles from the PDF outline."""
     sections: list[dict] = []
     seen_pages: set[int] = set()
+    parents: list[str] = []
     for toc_level, title, page in doc.get_toc():
+        title = title.strip()
+        if not title:
+            continue
+        parents = parents[: toc_level - 1]
+        ancestors = parents.copy()
+        parents.append(title)
         if toc_level != level or not title.strip():
             continue
         page = max(1, min(page, doc.page_count))
@@ -205,12 +212,15 @@ def toc_sections(doc: fitz.Document, level: int) -> list[dict]:
         if page in seen_pages:
             continue
         seen_pages.add(page)
-        sections.append({"title": title.strip(), "start_page": page})
+        sections.append(
+            {"title": title, "parents": ancestors, "start_page": page}
+        )
     if sections:
         return sections
 
     # Keep PDFs without bookmarks usable.  This fallback only recognizes
     # numbered headings and is deliberately limited to one boundary per page.
+    heading_path: list[str] = []
     for page_index, page in enumerate(doc, start=1):
         text = clean_page_text(page.get_text())
         if is_toc_page(text):
@@ -220,14 +230,23 @@ def toc_sections(doc: fitz.Document, level: int) -> list[dict]:
             if not match:
                 continue
             number, title = match.groups()
-            if number.count(".") + 1 != level:
+            heading_level = number.count(".") + 1
+            heading_path = heading_path[: heading_level - 1]
+            heading_path.append(title.strip())
+            if heading_level != level:
                 continue
             title = title.strip()
             if not title or not title[0].isalpha() or TOC_LINE.search(title):
                 continue
             if page_index not in seen_pages:
                 seen_pages.add(page_index)
-                sections.append({"title": title, "start_page": page_index})
+                sections.append(
+                    {
+                        "title": title,
+                        "parents": heading_path[:-1],
+                        "start_page": page_index,
+                    }
+                )
             break
     return sections
 
@@ -322,7 +341,7 @@ def convert(
             {
                 "num": f"{index:02d}",
                 "title": entry["title"],
-                "slug": f"{index:02d}-{slugify(entry['title'])}",
+                "slug": f"{index:02d}-{slugify(' - '.join((*entry['parents'], entry['title'])))}",
                 "pages": pages,
             }
         )