fix: merge Apec duplicate metadata

src/job_research/apec/dedupe.py

@@ -1,23 +1,74 @@
 from job_research.models import ApecListing
 
 
+_MERGEABLE_FIELDS = (
+    "source_job_id",
+    "title",
+    "company",
+    "location",
+    "contract_type",
+    "description_text",
+    "published_at",
+)
+
+
+def _merge_listing_metadata(survivor: ApecListing, source: ApecListing) -> None:
+    for field_name in _MERGEABLE_FIELDS:
+        if getattr(survivor, field_name) is None:
+            value = getattr(source, field_name)
+            if value is not None:
+                setattr(survivor, field_name, value)
+
+    for warning in source.warnings:
+        if warning not in survivor.warnings:
+            survivor.warnings.append(warning)
+
+
+def _register_listing(
+    url_to_listing: dict[str, ApecListing],
+    source_job_id_to_listing: dict[str, ApecListing],
+    listing: ApecListing,
+    survivor: ApecListing,
+) -> None:
+    url_to_listing[listing.url] = survivor
+
+    if listing.source_job_id is not None:
+        source_job_id_to_listing[listing.source_job_id] = survivor
+
+
 def dedupe_apec_listings(listings: list[ApecListing]) -> list[ApecListing]:
-    seen_urls: set[str] = set()
-    seen_source_job_ids: set[str] = set()
+    url_to_listing: dict[str, ApecListing] = {}
+    source_job_id_to_listing: dict[str, ApecListing] = {}
     deduped: list[ApecListing] = []
 
     for listing in listings:
         source_job_id = listing.source_job_id
-        is_duplicate = listing.url in seen_urls or (
-            source_job_id is not None and source_job_id in seen_source_job_ids
-        )
+        matches: list[ApecListing] = []
+
+        url_match = url_to_listing.get(listing.url)
+        if url_match is not None:
+            matches.append(url_match)
 
-        seen_urls.add(listing.url)
         if source_job_id is not None:
-            seen_source_job_ids.add(source_job_id)
+            source_job_id_match = source_job_id_to_listing.get(source_job_id)
+            if source_job_id_match is not None and source_job_id_match not in matches:
+                matches.append(source_job_id_match)
 
-        if is_duplicate:
+        if not matches:
+            deduped.append(listing)
+            _register_listing(url_to_listing, source_job_id_to_listing, listing, listing)
             continue
-        deduped.append(listing)
+
+        survivor = matches[0]
+        for other in matches[1:]:
+            if other is survivor:
+                continue
+
+            _merge_listing_metadata(survivor, other)
+            deduped[:] = [item for item in deduped if item is not other]
+            _register_listing(url_to_listing, source_job_id_to_listing, other, survivor)
+
+        _merge_listing_metadata(survivor, listing)
+        _register_listing(url_to_listing, source_job_id_to_listing, listing, survivor)
 
     return deduped

tests/apec/test_dedupe.py

@@ -80,3 +80,27 @@ def test_dedupe_apec_listings_keeps_secondary_ids_from_skipped_rows() -> None:
     deduped = dedupe_apec_listings([first, second, third])
 
     assert deduped == [first]
+
+
+def test_dedupe_apec_listings_merges_metadata_from_duplicate_rows() -> None:
+    first = ApecListing(
+        source="apec",
+        source_job_id=None,
+        published_at=None,
+        url="url1",
+        fetched_at="2026-06-01T10:00:00Z",
+    )
+    second = ApecListing(
+        source="apec",
+        source_job_id="job-123",
+        published_at="2026-06-01",
+        url="url1",
+        fetched_at="2026-06-01T10:01:00Z",
+    )
+
+    deduped = dedupe_apec_listings([first, second])
+
+    assert len(deduped) == 1
+    assert deduped[0].url == "url1"
+    assert deduped[0].source_job_id == "job-123"
+    assert deduped[0].published_at == "2026-06-01"