Fix mdBook navigation with absolute GitHub Pages URLs.

Rewrite internal href/src to /deepseek-tech-notes/ + encoded paths so
sidebar and content links never escape the project prefix.

Author: fooSynaptic

scripts/fix_mdbook_paths.py

@@ -1,82 +1,145 @@
 #!/usr/bin/env python3
 """Post-process mdBook HTML for GitHub Pages project sites.
 
-1. Replace root index.html with redirect to the real first chapter (avoids ``../``
-   resolving outside ``/deepseek-tech-notes/``).
-2. Percent-encode non-ASCII path segments in relative href/src (GitHub Pages 400).
+GitHub Pages serves this book at ``/deepseek-tech-notes/``.  mdBook emits
+relative ``../`` links that escape the project prefix on root/index pages, and
+unencoded CJK path segments return 400.  Rewrite internal links to absolute,
+percent-encoded URLs under site-url and fix search index paths to match.
 """
 from __future__ import annotations
 
+import json
 import re
-from pathlib import Path
-from urllib.parse import unquote, quote
+from pathlib import Path, PurePosixPath
+from urllib.parse import quote, unquote
 
 REPO = Path(__file__).resolve().parents[1]
 OUT = REPO / "mdbook-out"
-FIRST_CHAPTER = "00-%E5%89%8D%E8%A8%80/02-%E4%B8%AD%E6%96%87%E5%AF%BC%E8%AF%B2.html"
+BOOK_TOML = REPO / "book.toml"
 
-SKIP_PREFIXES = ("http://", "https://", "//", "#", "mailto:", "javascript:", "data:")
+SKIP_PREFIXES = ("http://", "https://", "//", "mailto:", "javascript:", "data:")
 ATTR_RE = re.compile(r'(href|src)="([^"]+)"')
+PATH_TO_ROOT_RE = re.compile(r'var path_to_root = "[^"]*";')
 
 
-def encode_path(path: str) -> str:
-    if not path or path.startswith(SKIP_PREFIXES):
-        return path
+def read_site_url() -> str:
+    text = BOOK_TOML.read_text(encoding="utf-8")
+    m = re.search(r'^site-url\s*=\s*"([^"]+)"', text, re.MULTILINE)
+    if not m:
+        raise SystemExit("book.toml missing output.html site-url")
+    url = m.group(1)
+    return url if url.endswith("/") else url + "/"
+
+
+SITE_URL = read_site_url()
+
+
+def encode_segment(part: str) -> str:
+    if part in (".", "..", ""):
+        return part
+    return quote(unquote(part), safe="")
+
+
+def encode_book_rel(rel: str) -> str:
     frag = ""
-    if "#" in path:
-        path, frag = path.split("#", 1)
+    if "#" in rel:
+        rel, frag = rel.split("#", 1)
         frag = "#" + frag
-    parts = path.split("/")
-    encoded = "/".join(
-        quote(unquote(part), safe="") if part not in (".", "..", "") else part
-        for part in parts
-    )
+    encoded = "/".join(encode_segment(p) for p in rel.split("/") if p)
     return encoded + frag
 
 
+def site_href(rel: str) -> str:
+    return SITE_URL + encode_book_rel(rel)
+
+
+def resolve_relative(href: str, page_dir: PurePosixPath) -> str | None:
+    if not href or href.startswith(SKIP_PREFIXES):
+        return None
+    if href.startswith(SITE_URL):
+        return href
+    frag = ""
+    if "#" in href:
+        href, frag = href.split("#", 1)
+        frag = "#" + frag
+    if href.startswith("/"):
+        rel = href.lstrip("/")
+    else:
+        rel = PurePosixPath(page_dir, href).as_posix()
+    parts: list[str] = []
+    for part in rel.split("/"):
+        if part == "..":
+            if parts:
+                parts.pop()
+        elif part not in (".", ""):
+            parts.append(part)
+    if not parts:
+        return None
+    return site_href("/".join(parts)) + frag
+
+
+def patch_html(path: Path) -> None:
+    page_dir = PurePosixPath(path.relative_to(OUT).parent.as_posix())
+    if page_dir.parts == (".",):
+        page_dir = PurePosixPath()
+
+    def repl(m: re.Match[str]) -> str:
+        attr, val = m.group(1), m.group(2)
+        new = resolve_relative(val, page_dir)
+        return m.group(0) if new is None else f'{attr}="{new}"'
+
+    text = path.read_text(encoding="utf-8")
+    text = ATTR_RE.sub(repl, text)
+    text = PATH_TO_ROOT_RE.sub(f'var path_to_root = "{SITE_URL}";', text)
+    path.write_text(text, encoding="utf-8")
+
+
 def write_redirect_index() -> None:
+    target = site_href("00-前言/02-中文导读.html")
     html = f"""<!DOCTYPE html>
 <html lang="zh-Hans">
 <head>
   <meta charset="utf-8">
   <title>Redirecting…</title>
-  <link rel="canonical" href="{FIRST_CHAPTER}">
-  <meta http-equiv="refresh" content="0; url={FIRST_CHAPTER}">
-  <script>location.replace("{FIRST_CHAPTER}");</script>
+  <link rel="canonical" href="{target}">
+  <meta http-equiv="refresh" content="0; url={target}">
+  <script>location.replace("{target}");</script>
 </head>
 <body>
-  <p><a href="{FIRST_CHAPTER}">中文导读</a></p>
+  <p><a href="{target}">中文导读</a></p>
 </body>
 </html>
 """
     (OUT / "index.html").write_text(html, encoding="utf-8")
 
 
-def patch_html(path: Path) -> int:
-    text = path.read_text(encoding="utf-8")
-
-    def repl(m: re.Match[str]) -> str:
-        attr, val = m.group(1), m.group(2)
-        new = encode_path(val)
-        return m.group(0) if new == val else f'{attr}="{new}"'
-
-    patched = ATTR_RE.sub(repl, text)
-    if patched != text:
-        path.write_text(patched, encoding="utf-8")
-    return text.count('href="') + text.count('src="')
+def patch_searchindex() -> None:
+    path = OUT / "searchindex.json"
+    if not path.is_file():
+        return
+    data = json.loads(path.read_text(encoding="utf-8"))
+    data["doc_urls"] = [encode_book_rel(u) for u in data["doc_urls"]]
+    path.write_text(json.dumps(data, ensure_ascii=False), encoding="utf-8")
+    js = OUT / "searchindex.js"
+    if js.is_file():
+        js.write_text(
+            "Object.assign(window.search, " + json.dumps(data, ensure_ascii=False) + ");",
+            encoding="utf-8",
+        )
 
 
 def main() -> None:
     if not OUT.is_dir():
         raise SystemExit(f"missing {OUT.relative_to(REPO)} — run mdbook build first")
-    write_redirect_index()
     n = 0
     for html in OUT.rglob("*.html"):
         if html.parent == OUT and html.name == "index.html":
             continue
         patch_html(html)
         n += 1
-    print(f"OK fix_mdbook_paths: redirect index + encoded paths in {n} html files")
+    write_redirect_index()
+    patch_searchindex()
+    print(f"OK fix_mdbook_paths: absolute URLs under {SITE_URL} ({n} chapter html files)")
 
 
 if __name__ == "__main__":