OCRPython/logs/ocr_reconginition.log

@@ -201,15 +201,3 @@
 2024-04-21 15:25:57,306 - main.py[line:119] - INFO: 0
 2024-04-21 15:25:57,483 - ocr.py[line:19] - INFO: from rapidocr_onnxruntime import RapidOCR
 2024-04-21 15:26:00,173 - main.py[line:42] - INFO: extractIdCardInfo 耗时2.866658926010132秒
-2024-04-22 11:01:10,204 - ocr.py[line:19] - INFO: from rapidocr_onnxruntime import RapidOCR
-2024-04-22 11:01:14,447 - main.py[line:42] - INFO: extractIdCardInfo 耗时4.689878702163696秒
-2024-04-22 11:01:54,639 - main.py[line:96] - INFO: main.py len of parameter: 2
-2024-04-22 11:01:54,640 - main.py[line:104] - INFO: python 脚本里的接收到的参数是：
-2024-04-22 11:01:54,640 - main.py[line:105] - INFO: 0
-2024-04-22 11:01:54,640 - main.py[line:106] - INFO: 开始执行sys.stdin.read
-2024-04-22 11:01:54,640 - main.py[line:108] - INFO: 
-2024-04-22 11:01:54,640 - main.py[line:109] - INFO: 55
-2024-04-22 11:01:54,640 - main.py[line:118] - INFO: 55
-2024-04-22 11:01:54,640 - main.py[line:119] - INFO: 0
-2024-04-22 11:01:54,819 - ocr.py[line:19] - INFO: from rapidocr_onnxruntime import RapidOCR
-2024-04-22 11:01:57,426 - main.py[line:42] - INFO: extractIdCardInfo 耗时2.785860061645508秒

OCRPython/main.py

@@ -95,12 +95,12 @@ if __name__ == "__main__":
     try:
         logger.info(f"main.py len of parameter: {len(sys.argv)}")
         jsonString = ""
-        if len(sys.argv) > 3:
+        if len(sys.argv) > 3: 
             logger.info(f"{sys.argv[1]}")
             logger.info(f"{sys.argv[2]}")
             logger.info(f"{sys.argv[3]}")
             jsonString = extractIdCardInfo(int(sys.argv[1]), sys.argv[2], sys.argv[3])
-        elif len(sys.argv) > 1:
+        elif len(sys.argv) > 1: 
             logger.info(f"python 脚本里的接收到的参数是：")
             logger.info(f"{sys.argv[1]}")
             logger.info(f"开始执行sys.stdin.read")
@@ -117,7 +117,7 @@ if __name__ == "__main__":
                 data1 = split_data[0]
             logger.info(f"{len(data1)}")
             logger.info(f"{len(data2)}")
-
+        
             jsonString = extractIdCardInfo(int(sys.argv[1]), data1, data2)
 
         print(jsonString)

OCRPython/maincopy.py

@@ -37,7 +37,6 @@ def extractIdCardInfoByPath(filePath1: str, filePath2: str)->str:
     except Exception as e:
         logger.error(e)
 
-    logger.info(f"text:{text}")
     extractor = IdentityCardExtractor()
     jsonString = extractor.extract_textbyPaddle(text)
     end_time = time.time()  # 记录结束时间
@@ -87,59 +86,59 @@ def extractIdCardInfoByBase64Data(base64data1:str, base64Data2: str)->str:
 
 # with open('/Users/wangvivi/Desktop/Code/ocrtest/images/id_card.JPG', 'rb') as image_file:
 #     base64_image_string = base64.b64encode(image_file.read()).decode('utf-8')
-#
+
 # jsonString = extractIdCardInfoByBase64Data(base64_image_string,"")
 # print(jsonString)
-#
+
-# jsonString = extractIdCardInfoByPath("/Users/wangvivi/Desktop/Code/ocrtest/images/id_card.JPG","")
+jsonString = extractIdCardInfoByPath("/Users/wangvivi/Desktop/Code/ocrtest/images/id_card.JPG","")
-# print(jsonString)
+print(jsonString)
 
 
-if __name__ == "__main__":
+# if __name__ == "__main__":
-    try:
+#     try:
-        logger.info(f"main.py len of parameter: {len(sys.argv)}")
+#         logger.info(f"main.py len of parameter: {len(sys.argv)}")
-        jsonString = ""
+#         jsonString = ""
-        if len(sys.argv) > 3:
+#         if len(sys.argv) > 3: 
-            logger.info(f"{sys.argv[1]}")
+#             logger.info(f"{sys.argv[1]}")
-            logger.info(f"{sys.argv[2]}")
+#             logger.info(f"{sys.argv[2]}")
-            logger.info(f"{sys.argv[3]}")
+#             logger.info(f"{sys.argv[3]}")
-            jsonString = extractIdCardInfo(int(sys.argv[1]), sys.argv[2], sys.argv[3])
+#             jsonString = extractIdCardInfo(int(sys.argv[1]), sys.argv[2], sys.argv[3])
-        elif len(sys.argv) > 1:
+#         elif len(sys.argv) > 1: 
-            logger.info(f"python 脚本里的接收到的参数是：")
+#             logger.info(f"python 脚本里的接收到的参数是：")
-            logger.info(f"{sys.argv[1]}")
+#             logger.info(f"{sys.argv[1]}")
-            logger.info(f"开始执行sys.stdin.read")
+#             logger.info(f"开始执行sys.stdin.read")
-            input_data = sys.stdin.read()
+#             input_data = sys.stdin.read()
-            logger.info("")
+#             logger.info("")
-            logger.info(f"len(input_data):{len(input_data)}")
+#             logger.info(f"{len(input_data)}")
-            split_data = input_data.split(os.linesep)
+#             split_data = input_data.split(os.linesep)
-            data1 = ""
+#             data1 = ""
-            data2 = ""
+#             data2 = ""
-            if 2 == len(split_data):
+#             if 2 == len(split_data):
-                data1 = split_data[0]
+#                 data1 = split_data[0]
-                data2 = split_data[1]
+#                 data2 = split_data[1]
-            elif 1 == len(split_data):
+#             elif 1 == len(split_data):
-                data1 = split_data[0]
+#                 data1 = split_data[0]
-            logger.info(f"{data1}")
+#             logger.info(f"{len(data1)}")
-            logger.info(f"{data2}")
+#             logger.info(f"{len(data2)}")
+        
+#             jsonString = extractIdCardInfo(int(sys.argv[1]), data1, data2)
+#         print(jsonString)
+#     except KeyboardInterrupt:
+#         logger.error("KeyboardInterrupt")
 
-            jsonString = extractIdCardInfo(int(sys.argv[1]), data1, data2)
+#     except IndexError:
-        print(jsonString)
+#     # 如果参数过长，捕获 IndexError 异常并进行处理
-    except KeyboardInterrupt:
+#         logger.error("参数过长，未指定足够的参数")
-        logger.error("KeyboardInterrupt")
 
-    except IndexError:
+#     except OverflowError:
-    # 如果参数过长，捕获 IndexError 异常并进行处理
+#     # 如果发生了 OverflowError 异常，捕获并进行相应的处理
-        logger.error("参数过长，未指定足够的参数")
+#         logger.error("命令行参数过长，导致溢出错误")
 
-    except OverflowError:
+#     except EOFError:
-    # 如果发生了 OverflowError 异常，捕获并进行相应的处理
+#     # 如果达到输入流的末尾，捕获 EOFError 异常并进行处理
-        logger.error("命令行参数过长，导致溢出错误")
+#         logger.error("已经到达输入流的末尾")
 
-    except EOFError:
+#     except Exception as e:
-    # 如果达到输入流的末尾，捕获 EOFError 异常并进行处理
+#     # 如果发生了其他异常，捕获并进行相应的处理
-        logger.error("已经到达输入流的末尾")
+#         logger.error("发生了异常:", e)
-
-    except Exception as e:
-    # 如果发生了其他异常，捕获并进行相应的处理
-        logger.error("发生了异常:", e)
 

OCRTool/.idea/.gitignore

@@ -0,0 +1,3 @@
+# Default ignored files
+/shelf/
+/workspace.xml

OCRTool/src/main/java/com/bonus/core/MySpringBootApplication.java

@@ -81,6 +81,7 @@ public class MySpringBootApplication {
             tesseract.setLanguage("chi_sim");
             // 从图像中提取文本
             String text = tesseract.doOCR(new File("/Users/wangvivi/Desktop/Code/ocrtest/images/id_card.JPG"));
+            // 打印识别的文本
             System.out.println("识别的文本：" + text);
             return text;
         } catch (TesseractException e) {

OCRTool/src/main/resources/application.yml

@@ -3,8 +3,8 @@ spring:
     active=dev:
 
 recognition:
-#  pythonEnv: /Users/wangvivi/Desktop/MySelf/myenv/bin/python
+  pythonEnv: /Users/wangvivi/Desktop/MySelf/myenv/bin/python
-#  scriptEnv: /Users/wangvivi/Desktop/Code/Component/OCRPython/main.py
+  scriptEnv: /Users/wangvivi/Desktop/Code/Component/OCR/main.py
-  pythonEnv: /Users/wangvivi/miniconda3/envs/ocr/bin/python
+  #pythonEnv: /Users/wangvivi/miniconda3/envs/ocr/bin/python
-  scriptEnv: /Users/wangvivi/Desktop/Code/Component/OCRPython/maincopy.py
+  #scriptEnv: /Users/wangvivi/Desktop/Code/Component/OCR/maincopy.py