lib/attribution.py

@@ -15,6 +15,7 @@ Epimetheus owns this module. Leo reviews changes.
 
 import logging
 import re
+import sqlite3
 from pathlib import Path
 
 logger = logging.getLogger("pipeline.attribution")
@@ -130,8 +131,11 @@ def is_publisher_handle(handle: str, conn) -> int | None:
         ).fetchone()
         if row:
             return row["id"] if hasattr(row, "keys") else row[0]
-    except Exception:
+    except sqlite3.OperationalError:
-        logger.debug("is_publisher_handle: lookup failed for %r", h, exc_info=True)
+        # Pre-v26 DB: publishers table doesn't exist yet. Fall through to None
+        # so writer behaves as before. Any other exception class is real signal
+        # (programming error, lock contention, corruption) — let it propagate.
+        logger.debug("is_publisher_handle: publishers table not present (pre-v26?)", exc_info=True)
     return None
 
 

lib/contributor.py

@@ -428,6 +428,13 @@ def upsert_contributor(
     # Schema v26 gate: orgs/citations live in publishers table, not contributors.
     # Skip without writing so the v26 classifier cleanup isn't undone by every
     # merge that has `sourcer: cnbc` (or similar) in claim frontmatter.
+    #
+    # Note: bare normalization (lower + lstrip @), no alias resolution. This is
+    # consistent with the existing `SELECT handle FROM contributors WHERE handle = ?`
+    # below — both look up by canonical-form-as-stored. Today's classifier produces
+    # one publisher row per canonical handle, so bare lookup hits. Branch 3 will
+    # normalize alias→canonical at writer entry points (extract.py, post_extract);
+    # at that point this gate auto-tightens because callers pass canonical handles.
     canonical_handle = handle.strip().lower().lstrip("@") if handle else ""
     if canonical_handle and is_publisher_handle(canonical_handle, conn) is not None:
         logger.debug("upsert_contributor: %r is a publisher — skipping contributor row", canonical_handle)