Track processed candidates in session and use OAuth for API calls

- Record skips, saves, and no-match results in session["skipped"] so
  revisiting an article resumes past already-checked candidates
- Filter self-links (case-insensitive first letter) from hit list
- Use OAuth session for all API reads when logged in for higher rate limits
- Add "for" template to exclusion list to avoid bad edits
- Improve API error handling with HTTP status codes logged to stderr

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

add_links/api.py

@@ -1,4 +1,5 @@
 import re
+import sys
 import typing
 
 import requests
@@ -72,18 +73,30 @@ webpage_error = (
 )
 
 
+def _get_active_session() -> requests.sessions.Session:
+    """Return OAuth session if one is available in Flask context, else plain session."""
+    try:
+        from flask import g
+        if hasattr(g, "oauth_session") and g.oauth_session is not None:
+            return g.oauth_session  # type: ignore[return-value]
+    except RuntimeError:
+        pass
+    return get_session()
+
+
 def api_get(params: StrDict) -> StrDict:
     """Make call to Wikipedia API."""
-    s = get_session()
+    s = _get_active_session()
 
     r = s.get(get_query_url(), params=params)
     try:
         ret: StrDict = r.json()
     except JSONDecodeError:
+        print(f"API request failed: HTTP {r.status_code}", file=sys.stderr)
+        print(f"Response body: {r.text!r}", file=sys.stderr)
         if webpage_error in r.text:
             raise MediawikiError(webpage_error)
-        else:
-            raise MediawikiError("unknown error")
+        raise MediawikiError(f"HTTP {r.status_code}: unexpected response from Wikipedia API")
     check_for_error(ret)
     return ret
 
@@ -271,7 +284,7 @@ def call_get_diff(title: str, section_num: int, section_text: str) -> str:
         "rvdifftotext": section_text.strip(),
     }
 
-    s = get_session()
+    s = _get_active_session()
     r = s.post(get_query_url(), data=data)
     try:
         ret = r.json()

add_links/match.py

@@ -78,7 +78,7 @@ re_cite = re.compile(
     re.I | re.S,
 )
 
-re_cite_template_start = re.compile(r"\{\{(?:cite|citation|short description|gli|defn|annotated link|excerpt|main|see)\b", re.I)
+re_cite_template_start = re.compile(r"\{\{(?:cite|citation|short description|gli|defn|annotated link|excerpt|main|see|for)\b", re.I)
 re_no_param_template = re.compile(r"\{\{[^|{}]+\}\}")
 re_external_link = re.compile(r"\[https?://[^\]]+\]")
 # Italic text (work titles in bibliographies). Handles apostrophes in content

add_links/mediawiki_oauth.py

@@ -99,6 +99,27 @@ def userinfo_call() -> typing.Mapping[str, typing.Any]:
     return api_request(params)
 
 
+def get_oauth_session() -> OAuth1Session | None:
+    """Return an OAuth1Session for the current user, or None if not logged in."""
+    if "owner_key" not in session or "owner_secret" not in session:
+        return None
+    app = current_app
+    client_key = app.config["CLIENT_KEY"]
+    client_secret = app.config["CLIENT_SECRET"]
+    oauth = OAuth1Session(
+        client_key,
+        client_secret=client_secret,
+        resource_owner_key=session["owner_key"],
+        resource_owner_secret=session["owner_secret"],
+    )
+    oauth.headers.update({"User-Agent": ua})
+    oauth.params = typing.cast(
+        dict[str, str | int],
+        {"format": "json", "action": "query", "formatversion": 2},
+    )
+    return oauth
+
+
 def get_username() -> None | str:
     """Get the username or None if not logged in."""
     if "owner_key" not in session: