irs 990 doc prarsers and some web stuffHEADmaster

1 files changed, 10 insertions, 1 deletions

diff --git a/scripts/common/filing.py b/scripts/common/filing.py
index 44bcabc..3c7aaf3 100644
--- a/scripts/common/filing.py
+++ b/scripts/common/filing.py
@@ -5,7 +5,7 @@ Provides upsert/record operations for raw.filing and raw.filing_source.
 Form-specific child row operations live in the parsers themselves.
 """
 
-from scripts.common.db import execute_scalar
+from scripts.common.db import execute_all, execute_scalar
 
 
 def _execute_scalar(conn, sql, params=None):
@@ -91,3 +91,12 @@ def record_raw_filing_source(raw_filing_id, ingest_run_id, source_archive, sourc
             f"already mapped to raw_filing_id={existing_filing_id}, "
             f"but trying to map to {raw_filing_id}"
         )
+
+
+def get_seen_source_paths(source_archive):
+    """Return the set of source_path values already committed for an archive."""
+    rows = execute_all(
+        "SELECT source_path FROM raw.filing_source WHERE source_archive = %s",
+        (source_archive,),
+    )
+    return {row[0] for row in rows}