testing out OCR viewer of OCR'd text

scripts/check_eod.py

@@ -0,0 +1,33 @@
+import json, pathlib
+from collections import Counter, defaultdict
+
+data = json.loads(pathlib.Path('outputs/Calgary-Highlanders_Sep44_positions.json').read_text())
+
+cats = Counter(p['category'] for p in data)
+print('=== CATEGORIES ===')
+for k, v in sorted(cats.items()):
+    print(f'  {k}: {v}')
+
+eod = [p for p in data if p['is_end_of_day']]
+print(f'\n=== EOD entries: {len(eod)} ===')
+for p in eod:
+    print(f"  {str(p['date']):<16} cat={p['category']:<14} grid={str(p['grid']):<8} place={p['place_name']}")
+
+eod_by_date = defaultdict(list)
+for p in eod:
+    eod_by_date[p['date']].append(p)
+multi = {d: ps for d, ps in eod_by_date.items() if len(ps) > 1}
+if multi:
+    print('\nWARNING - multiple EOD on same date:')
+    for d, ps in multi.items():
+        print(f'  {d}: {len(ps)} entries')
+else:
+    print('\nOK: exactly one EOD per date')
+
+# List dates with no EOD
+all_dates = sorted({p['date'] for p in data if p['date']})
+eod_dates = set(eod_by_date.keys())
+missing = [d for d in all_dates if d not in eod_dates]
+if missing:
+    print(f'\nDates with NO EOD: {missing}')
+