From 3baec1794b8c4535d692f83b197a1b2fe05a840b Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Fri, 9 Oct 2026 22:45:57 +0200
Subject: [PATCH] fix: match a loose dump by crc32 and size

---
 scripts/provenance_report.py |  8 ++++++++
 tests/test_provenance.py     | 14 ++++++++++++++
 2 files changed, 22 insertions(+)

diff --git a/scripts/provenance_report.py b/scripts/provenance_report.py
index bb067b6c..6100a231 100644
--- a/scripts/provenance_report.py
+++ b/scripts/provenance_report.py
@@ -88,6 +88,13 @@ def build_report(db: dict, snapshots: dict, members: dict | None = None) -> dict
     by_md5_size = {
         (entry.get("md5", ""), entry.get("size", 0)) for entry in by_sha1.values()
     }
+    # Redump lists some dumps by crc32 alone (ps2-0101jd-20030110, DTL-H10100):
+    # collected as a loose file it would stay MISSING for good without this.
+    by_crc32_size = {
+        (str(entry.get("crc32", "")).lower(), entry.get("size", 0))
+        for entry in by_sha1.values()
+        if entry.get("crc32")
+    }
 
     report = {}
     for source, snapshot in sorted(snapshots.items()):
@@ -98,6 +105,7 @@ def build_report(db: dict, snapshots: dict, members: dict | None = None) -> dict
             if (
                 entry.get("sha1") in by_sha1
                 or (entry.get("md5"), entry.get("size")) in by_md5_size
+                or (str(entry.get("crc32") or "").lower(), entry.get("size")) in by_crc32_size
                 or _held_in_archive(entry, members)
             ):
                 matched += 1
diff --git a/tests/test_provenance.py b/tests/test_provenance.py
index f8eb24c5..4ef006b8 100644
--- a/tests/test_provenance.py
+++ b/tests/test_provenance.py
@@ -353,6 +353,20 @@ class TestProvenanceReport(unittest.TestCase):
             report["redump"]["covered_dats"], ["Sony - PlayStation - BIOS Images"]
         )
 
+    def test_a_loose_file_matches_by_crc32_and_size(self):
+        """Redump lists DTL-H10100's image by crc32 alone; collected as a
+        loose file it stayed MISSING for good."""
+        db = {"files": {"f" * 40: {"name": "ps2-0101jd.bin", "size": 4194304,
+                                   "md5": "e" * 32, "crc32": "E9836A0C"}}}
+        snapshots = {"redump": {"source": "redump", "imported_at": "2026-10-09",
+                                "dats": {"PS2 BIOS": "2026-10-01"},
+                                "entries": [{"name": "ps2-0101jd.bin", "size": 4194304,
+                                             "crc32": "e9836a0c", "md5": "", "sha1": "",
+                                             "dat": "PS2 BIOS"}]}}
+        report = build_report(db, snapshots)
+        self.assertEqual(report["redump"]["matched"], 1)
+        self.assertEqual(report["redump"]["missing"], [])
+
     def test_uncovered_dat_entries_are_out_of_scope(self):
         """Entries from a DAT the collection never matches are not targets.