Improve patient list OCR cleanup

患者列表处理/数据处理工作区/05_同步PostgreSQL单表.py

@@ -12,6 +12,7 @@ import argparse
 import csv
 import json
 import os
+import re
 import subprocess
 import tempfile
 from pathlib import Path
@@ -22,6 +23,7 @@ DEFAULT_RESULT_JSON = Path("数据处理结果区/合并_患者列表_结构化.
 DEFAULT_SCHEMA_SQL = Path("数据处理工作区/06_PostgreSQL建表结构.sql")
 
 TABLE_NAME = '"Patient_Lists"'
+DATE_TIME_RE = re.compile(r"^\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}$")
 
 PATIENT_FIELDS = [
     "batch_name",
@@ -75,6 +77,16 @@ def review_note_items(review: dict[str, Any]) -> list[str]:
     return note_items
 
 
+def admission_after_discharge(admission_time: Any, discharge_time: Any) -> bool:
+    admission = scalar(admission_time).strip()
+    discharge = scalar(discharge_time).strip()
+    if not admission or not discharge:
+        return False
+    if not DATE_TIME_RE.match(admission) or not DATE_TIME_RE.match(discharge):
+        return False
+    return admission > discharge
+
+
 def psql_quote(path: Path) -> str:
     return "'" + str(path.resolve()).replace("'", "''") + "'"
 
@@ -87,6 +99,17 @@ def flatten_patient_record(record: dict[str, Any]) -> dict[str, Any]:
     hospital_days, hospital_days_note = clean_integer(patient["住院天数"])
     if hospital_days_note:
         note_items.append(hospital_days_note)
+    review_status = review["状态"]
+    patient_name = scalar(patient["姓名"]).strip()
+    if not patient_name:
+        patient_name = "缺少姓名"
+        review_status = "需人工复核"
+        if "缺少姓名" not in note_items:
+            note_items.append("缺少姓名")
+    if admission_after_discharge(patient["入院时间"], patient["出院时间"]):
+        review_status = "需人工复核"
+        if "入院时间晚于出院时间" not in note_items:
+            note_items.append("入院时间晚于出院时间")
     review_notes = "；".join(note_items)
     return {
         "batch_name": record["处理批次"],
@@ -96,7 +119,7 @@ def flatten_patient_record(record: dict[str, Any]) -> dict[str, Any]:
         "image_path": image["图片路径"],
         "image_name": image["图片名"],
         "image_row_no": image["图片内行号"],
-        "patient_name": patient["姓名"],
+        "patient_name": patient_name,
         "gender": patient["性别"],
         "age": patient["年龄"],
         "inpatient_no": patient["住院号"],
@@ -106,7 +129,7 @@ def flatten_patient_record(record: dict[str, Any]) -> dict[str, Any]:
         "hospital_days": hospital_days,
         "discharge_time": patient["出院时间"],
         "postoperative_days": patient["手术后天数"],
-        "review_status": review["状态"],
+        "review_status": review_status,
         "review_notes": review_notes,
         "manual_corrected": bool(review.get("人工修正")),
     }