Z230

ECG/20 ECG test2.py

@@ -26,9 +26,22 @@ def ocr_page(page):
 
 
 def extract_rodne_cislo(text):
-    """Extract 10-digit rodné číslo (no slash)."""
-    m = re.search(r"\b\d{9,10}\b", text)
-    return m.group(0) if m else None
+    """
+    Extract rodné číslo in formats:
+    - 6 digits + slash + 4 digits  → 655527/1910
+    - 6 digits + slash + 3 digits  → 655527/910
+    - 10 digits without slash      → 6555271910
+
+    Always returns 10 digits without slash.
+    """
+    m = re.search(r"\b(\d{6})/?(\d{3,4})\b", text)
+    if not m:
+        return None
+
+    left = m.group(1)
+    right = m.group(2).zfill(4)  # ensure 4 digits
+
+    return left + right
 
 
 def extract_date(text):