enhance

OCRPython/extractor/identitycard_extractor.py

@@ -74,9 +74,9 @@ class IdentityCardExtractor(Extractor):
                 "address": "",
                 "idNumber": ""}
             # 提取签发机关
-            issuing_authority = re.search(r"签发机关\n*(.+?)\n", text, re.DOTALL)
+            issuing_authority = re.search(r"(签发机关|签发机美)\n*(.+?)\n", text, re.DOTALL)
             if issuing_authority:
-                result["issuingAuthority"] = issuing_authority.group(1).strip()
+                result["issuingAuthority"] = issuing_authority.group(2).strip()
 
             # 提取有效期限
             valid_time = re.search(r"有效期限\n*(\d{4}\.\d{2}\.\d{2}-\S+)", text, re.DOTALL)
@@ -87,7 +87,7 @@ class IdentityCardExtractor(Extractor):
             name = re.search(r"姓名\s*(.*?)\n", text,re.DOTALL)
             if name:
                 tempName = name.group(1).strip()
-                if tempName in "性别男" or tempName in "性别女":
+                if tempName in "性别男" or tempName in "性别女" or "性别男" in tempName or "性别女"  in tempName:
                     name = re.search(r"(\S+)\s*姓名", text, re.DOTALL)
                     result["name"] = name.group(1).strip()
                 else:
@@ -107,9 +107,9 @@ class IdentityCardExtractor(Extractor):
                             result["name"] = name.group(1).strip()
 
             # 提取民族
-            ethnicity = re.search(r"民\s*族\s*(\S+)", text, re.DOTALL)
+            ethnicity = re.search(r"民\s*(族|旅)\s*(\S+)", text, re.DOTALL)
             if ethnicity:
-                result["ethnicity"] = ethnicity.group(1).strip()
+                result["ethnicity"] = ethnicity.group(2).strip()
 
             # 提取地址
             address = re.search(r"(住址|佳址)(.*?)公民身份号码", text, re.DOTALL)
@@ -302,6 +302,15 @@ class IdentityCardExtractor(Extractor):
 # 居民身份证
 # 签发机关巴林右旗公安局
 # 有效期限2004.10.27-2024.10.26"""
+# text ="""唐昌梅
+# 姓名
+# 性别女民族苗
+# 出生1975年8月15日
+# 住址
+# 湖南省常德市鼎城区双桥
+# 坪镇全家坪村3组
+# 公民身份号码
+# 433030197508150820"""
 # extractor = IdentityCardExtractor()
 # jsonstring = extractor.extract_textbyPaddle(text)
 # print(jsonstring)

OCRPython/maincopy.py

@@ -30,6 +30,8 @@ def extractIdCardInfoByPath(filePath1: str, filePath2: str)->str:
                 res = result[idx]
                 for line in res:
                     text += (line[1][0] + '\n')
+        else:
+            logger.error(f"{filePath1} doesn't exist")
 
         if os.path.exists(filePath2):
             result = ocr.ocr(filePath2, cls=False)
@@ -37,6 +39,9 @@ def extractIdCardInfoByPath(filePath1: str, filePath2: str)->str:
                 res = result[idx]
                 for line in res:
                     text += (line[1][0] + '\n')
+        else:
+            logger.error(f"{filePath2} doesn't exist")
+
     except Exception as e:
         logger.error(e)
         print(e)
@@ -49,6 +54,8 @@ def extractIdCardInfoByPath(filePath1: str, filePath2: str)->str:
         end_time = time.time()  # 记录结束时间
         execution_time = end_time - start_time  # 计算执行时间
         logger.info(f"extractIdCardInfoByBase64Data 耗时{execution_time}秒")
+    else:
+        logger.error(f"{filePath1},{filePath2} can't be recognized")
     return jsonString
 
 def extractIdCardInfoByBase64Data(base64data1:str, base64Data2: str)->str:
@@ -98,7 +105,7 @@ def extractIdCardInfoByBase64Data(base64data1:str, base64Data2: str)->str:
 # #
 # jsonString = extractIdCardInfoByPath("./images/han.jpg","")
 # logger.info(f"test")
-# jsonString = extractIdCardInfoByPath("/Users/wangvivi/Desktop/Code/ocrtest/images/2.jpg","/Users/wangvivi/Desktop/Code/ocrtest/images/1.jpg")
+# jsonString = extractIdCardInfoByPath("./images/43302919641130423X_202311061953_front.jpg","./images/43302919641130423X_202311061953_back.jpg")#, "./images/江六斤反.jpg，./images/han.jpg
 # print(jsonString)
 
 if __name__ == "__main__":
@@ -116,9 +123,6 @@ if __name__ == "__main__":
             logger.info(f"开始执行sys.stdin.read")
             input_data = sys.stdin.read()
             logger.info(f"len(input_data):{len(input_data)}")
-            # imageData = base64.b64decode(input_data)
-            # logger.info(f"image_data1:{image_data1}")
-            #logger.info(f"input_data:{input_data}")
             split_data = input_data.split(os.linesep)
             data1 = ""
             data2 = ""
@@ -128,8 +132,6 @@ if __name__ == "__main__":
                 data2 = split_data[1]
             elif 1 == len(split_data):
                 data1 = split_data[0]
-            # logger.info(data1)
-            # logger.info(data2)
             jsonString = extractIdCardInfo(int(sys.argv[1]), data1, data2)
         print(jsonString)
     except KeyboardInterrupt: