Make catalog PDF links host-relative and add a hostname rewrite script.

New builds write path-relative PDF annotations so the tree can be served
from docs.verae-time.net (or any host at the site root) without a rebuild.
scripts/rewrite-docs-host.py retargets an already-built site when absolute
URLs are preferred. Live service doors stay on their own hostnames.

scripts/check-doc-links.py

@@ -5,7 +5,6 @@ from __future__ import annotations
 import sys
 from html.parser import HTMLParser
 from pathlib import Path
-from urllib.parse import unquote, urljoin, urlparse
 
 ROOT = Path(__file__).resolve().parents[1]
 SITE = ROOT / "site"
@@ -24,23 +23,46 @@ class Anchors(HTMLParser):
                 self.hrefs.append(d["href"] or "")
 
 
+def _exists(dest: Path) -> bool:
+    if dest.is_file():
+        return True
+    if dest.is_dir() and any(
+        (dest / n).exists() for n in ("index.html", "index.pdf", "README.html", "README.pdf")
+    ):
+        return True
+    return False
+
+
 def local_ok(url: str) -> bool:
-    if url.startswith(WEB):
-        rel = url[len(WEB) :].split("#", 1)[0]
-        dest = SITE / rel
-        if dest.is_file():
-            return True
-        if dest.is_dir() and any(
-            (dest / n).exists() for n in ("index.html", "index.pdf", "README.html", "README.pdf")
-        ):
-            return True
-        # yaml/svg/png on homepage
-        return False
+    if url.startswith(WEB) or url.startswith("http://zapier.georgelambert.org/"):
+        rel = url.split("://", 1)[-1].split("/", 1)[-1].split("#", 1)[0]
+        return _exists(SITE / rel)
     if url.startswith("file:"):
         return False
     return True  # http(s) off-site; homepage checker uses GET separately
 
 
+def pdf_dest_ok(pdf: Path, raw: str) -> bool:
+    raw = raw.split("#", 1)[0]
+    if raw.startswith("file:"):
+        return False
+    if raw.startswith("mailto:") or raw.startswith("data:"):
+        return True
+    if raw.startswith("http://") or raw.startswith("https://"):
+        if raw.startswith(WEB) or raw.startswith("http://zapier.georgelambert.org/"):
+            return local_ok(raw)
+        return True
+    if raw.startswith("/"):
+        dest = SITE / raw.lstrip("/")
+    else:
+        dest = (pdf.parent / raw).resolve()
+        try:
+            dest.relative_to(SITE.resolve())
+        except ValueError:
+            return False
+    return _exists(dest)
+
+
 def main() -> int:
     idx = SITE / "index.html"
     if not idx.exists():
@@ -91,12 +113,10 @@ def main() -> int:
                     n_file += 1
                     broken.append(f"file URI {pdf.relative_to(SITE)} -> {raw[:120]}")
                     continue
-                if raw.startswith(WEB):
-                    path = raw.split("#", 1)[0]
-                    if "/research/" in path:
-                        continue
-                    if not local_ok(path):
-                        broken.append(f"pdf missing {pdf.relative_to(SITE)} -> {raw}")
+                if "/research/" in raw:
+                    continue
+                if not pdf_dest_ok(pdf, raw):
+                    broken.append(f"pdf missing {pdf.relative_to(SITE)} -> {raw}")
 
     print(f"file:// leftover {n_file}")
     print(f"broken {len(broken)}")