fix: apply code-review findings to import_fruits script and tests

- find_images: verify exact fruit_id prefix (stem-parse before _s0.) to
  prevent cross-fruit image contamination (e.g. mitschurins steals
  mitschurins_fruchtbare images)
- DATA_DIR uses os.path.abspath to handle symlinks / relative invocation
- counts["images"] derived from IMAGE_DIRS constant, not hardcoded keys
- counts["warnings"] incremented on missing-id/name skip (was missing)
- UPSERT preserves existing comment (comment = fruits.comment, not NULL)
- Print summary and rollback moved inside try/except to avoid NameError
- Remove dead XML_PATH constant
- Tests: add tearDown to clean up tmpdir; fix vacuous _tn exclusion test
  to include positive-control _s0 file; add prefix-collision regression test

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

scripts/import_fruits.py

@@ -17,8 +17,7 @@ from datetime import date, datetime
 
 import psycopg2
 
-DATA_DIR = os.path.join(os.path.dirname(__file__), "..", "03-data")
-XML_PATH = os.path.join(DATA_DIR, "obstsorten.xml")
+DATA_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "..", "03-data")
 
 # typ values that are aggregate categories — coerced with a logged warning
 AGGREGATE_TYPS = {"x", "bq", "aprpfi", "mirren", "pflzwe", "bo", "k"}
@@ -86,13 +85,23 @@ def find_images(fruit_id: str, data_dir: str) -> list:
     """
     results = []
     for subdir, image_type in IMAGE_DIRS:
-        pattern = os.path.join(data_dir, subdir, f"{fruit_id}_*_s0.*")
+        pattern = os.path.join(data_dir, subdir, f"{glob.escape(fruit_id)}_*_s0.*")
         for path in sorted(glob.glob(pattern)):
+            basename = os.path.basename(path)
+            # Verify the fruit_id is the exact prefix, not just a string prefix of
+            # another fruit's id (e.g. "mitschurins" must not steal
+            # "mitschurins_fruchtbare_nda_s0.jpg").
+            # Convention: basename = {id}_{type_code}_s0.{ext}
+            # So stripping the last _-segment before _s0 gives the exact id.
+            stem = basename.split("_s0.")[0]
+            actual_id = "_".join(stem.split("_")[:-1])
+            if actual_id != fruit_id:
+                continue
             with open(path, "rb") as f:
                 data = f.read()
             results.append(
                 {
-                    "filename": os.path.basename(path),
+                    "filename": basename,
                     "data": data,
                     "image_type": image_type,
                 }
@@ -112,7 +121,7 @@ def import_fruits(conn, data_dir: str = None) -> dict:
     tree = ET.parse(xml_path)
     root = tree.getroot()
 
-    counts = {"fruits": 0, "synonyms": 0, "images": {"fruit": 0, "flower": 0, "tree": 0}, "warnings": 0, "skipped": 0}
+    counts = {"fruits": 0, "synonyms": 0, "images": {img_type: 0 for _, img_type in IMAGE_DIRS}, "warnings": 0, "skipped": 0}
     warnings = []
 
     with conn.cursor() as cur:
@@ -125,6 +134,7 @@ def import_fruits(conn, data_dir: str = None) -> dict:
 
             if not osdb_number or not name:
                 counts["skipped"] += 1
+                counts["warnings"] += 1
                 warnings.append(f"SKIP: missing id or name (id={osdb_number!r})")
                 continue
 
@@ -151,7 +161,7 @@ def import_fruits(conn, data_dir: str = None) -> dict:
                     ON CONFLICT (osdb_number) DO UPDATE
                       SET name = EXCLUDED.name,
                           fruit_type = EXCLUDED.fruit_type,
-                          comment = NULL,
+                          comment = fruits.comment,
                           updated_at = NOW()
                     RETURNING id
                     """,
@@ -165,7 +175,7 @@ def import_fruits(conn, data_dir: str = None) -> dict:
                     ON CONFLICT (osdb_number) DO UPDATE
                       SET name = EXCLUDED.name,
                           fruit_type = EXCLUDED.fruit_type,
-                          comment = NULL,
+                          comment = fruits.comment,
                           updated_at = NOW()
                     RETURNING id
                     """,
@@ -218,20 +228,22 @@ def main():
     try:
         print("Importing fruits...")
         counts = import_fruits(conn)
+        total_images = sum(counts["images"].values())
+        print(f"\nDone.")
+        print(f"  Fruits upserted:  {counts['fruits']}")
+        print(f"  Synonyms inserted: {counts['synonyms']}")
+        print(f"  Images inserted:  {total_images}")
+        for img_type, n in counts["images"].items():
+            print(f"    {img_type}: {n}")
+        print(f"  Warnings:         {counts['warnings']}")
+        print(f"  Skipped:          {counts['skipped']}")
+    except Exception as e:
+        print(f"Error: {e}", file=sys.stderr)
+        conn.rollback()
+        sys.exit(1)
     finally:
         conn.close()
 
-    total_images = sum(counts["images"].values())
-    print(f"\nDone.")
-    print(f"  Fruits upserted:  {counts['fruits']}")
-    print(f"  Synonyms inserted: {counts['synonyms']}")
-    print(f"  Images inserted:  {total_images}")
-    print(f"    fruit:  {counts['images']['fruit']}")
-    print(f"    flower: {counts['images']['flower']}")
-    print(f"    tree:   {counts['images']['tree']}")
-    print(f"  Warnings:         {counts['warnings']}")
-    print(f"  Skipped:          {counts['skipped']}")
-
 
 if __name__ == "__main__":
     main()