Algo alternatif pour trouver les oublis

Squelette de l'algo

million/analyze/find_holes.py

@@ -71,3 +71,33 @@ def find_holes(messages: List[Message], accepted_max: int = 1_000_000) -> List[H
                 end_message=sequence.start_message
             ))
     return holes
+
+def find_holesV2(messages: List[Message]) -> List[Hole]:
+    current = 1
+    msg_idx = 0
+    threshold = 1000
+
+    result = []
+
+    while True:
+        msgCurrent = next((m for m in messages if m.get_counted_value() == current), None)
+        if msgCurrent:
+            if len(result) > 0 and result[-1].end < 1:
+                result[-1].end = current-1
+                result[-1].end_message = msgCurrent
+
+            messages.remove(msgCurrent)
+            current += 1
+        else:
+            if len(result) == 0 or result[-1].end > 0:
+                hole = Hole(
+                    start = current,
+                    end = 0,
+                    start_message = Message(sender_name='',timestamp_ms=0),
+                    end_message = Message(sender_name='',timestamp_ms=0)
+                    )
+                result.append(hole)
+            elif current - result[-1].start > threshold:
+                break
+            
+    return result

scripts/find_holesV2.py

@@ -0,0 +1,16 @@
+from million.analyze.find_holes import find_holesV2
+from million.analyze.retain_counts import retain_counts
+from million.parse.fb_exports import FacebookExportParser
+
+
+DATA_PATH = './data/'
+parser = FacebookExportParser()
+
+export = parser.parse(DATA_PATH)
+filtered = retain_counts(export.messages)
+
+holes = find_holesV2(filtered)
+
+for hole in holes:
+    print(f"{hole.start} - {hole.end} ({hole.end - hole.start})")
+    print(hole.end_message)