Api improvements.

Author: vladd-bit

config.py

@@ -21,8 +21,8 @@
 # basic app settings
 OCR_SERVICE_PORT = os.environ.get("OCR_SERVICE_PORT", 8090)
 
-# Tesseract model path
-TESSDATA_PREFIX = os.environ.get("OCR_TESSDATA_PREFIX", "/usr/local/share/tessdata")
+# Tesseract model path: macos - /opt/homebrew/share/tessdata | linux - "/usr/local/share/tessdata"
+TESSDATA_PREFIX = os.environ.get("OCR_TESSDATA_PREFIX", "/opt/homebrew/share/tessdata")
 
 # Integer or Float - duration in seconds for the OCR processing, after which,
 #   tesseract will terminate and raise RuntimeError
@@ -68,6 +68,9 @@
 # dpi used for images in TESSERACT and other stuff
 OCR_IMAGE_DPI = int(os.environ.get("OCR_SERVICE_IMAGE_DPI", 200))
 
+# possible values: json (stringified output), dict (dict means no json.dumps() is applied to the output)
+OCR_SERVICE_RESPONSE_OUTPUT_TYPE: str = str(os.environ.get("OCR_SERVICE_RESPONSE_OUTPUT_TYPE", "json"))
+
 # LIBRE OFFICE SECTION
 
 # 60 seconds before terminating processes

env/ocr_service.env

@@ -16,7 +16,10 @@ OCR_WEB_SERVICE_WORKER_CLASS="sync"
 OCR_SERVICE_OPERATION_MODE=OCR
 
 # 50 - CRITICAL, 40 - ERROR, 30 - WARNING, 20 - INFO, 10 - DEBUG, 0 - NOTSET
-OCR_SERVICE_LOG_LEVEL=40
+OCR_SERVICE_LOG_LEVEL=10
+
+# possible values: json (stringified output), dict (dict means no json.dumps() is applied to the output)
+OCR_SERVICE_RESPONSE_OUTPUT_TYPE="dict"
 
 ######################################################################################################
 
@@ -27,7 +30,7 @@ OCR_SERVICE_TESSERACT_TIMEOUT=30
 # change this to whatever language you are trying to OCR, e.g. eng, deu, fra, ita, nld, ron, spa
 # please note that you need to have the corresponding language pack installed in the container
 # check Dockerfile_multilang for more information and look for tessaract-ocr-[lang] packages
-OCR_SERVICE_TESSERACT_LANG="ron"
+OCR_SERVICE_TESSERACT_LANG="eng"
 
 OCR_SERVICE_TESSERACT_NICE=-18
 OCR_SERVICE_TESSERACT_CUSTOM_CONFIG_FLAGS=""

env/ocr_service_text_only.env

@@ -19,4 +19,7 @@ OCR_SERVICE_OPERATION_MODE=NO_OCR
 # change this to whatever language you are trying to OCR, e.g. eng, deu, fra, ita, nld, ron, spa
 # please note that you need to have the corresponding language pack installed in the container
 # check Dockerfile_multilang for more information and look for tessaract-ocr-[lang] packages
-OCR_SERVICE_TESSERACT_LANG="eng"
+OCR_SERVICE_TESSERACT_LANG="eng"
+
+# possible values: json (stringified output), dict (dict means no json.dumps() is applied to the output)
+OCR_SERVICE_RESPONSE_OUTPUT_TYPE="json"

ocr_service/api/api.py

@@ -9,7 +9,7 @@
 
 from multiprocessing import Pool
 
-from config import CPU_THREADS, TESSERACT_TIMEOUT, LOG_LEVEL
+from config import CPU_THREADS, TESSERACT_TIMEOUT, LOG_LEVEL, OCR_SERVICE_RESPONSE_OUTPUT_TYPE
 from ocr_service.api.api_blueprint import ApiBlueprint
 from ocr_service.utils.utils import build_response, get_app_info, setup_logging
 
@@ -36,7 +36,7 @@ def process() -> Response:
         Response: json with the result of the OCR processing
     """
     stream = None
-    file_name: str = ""
+    file_name: str | None = ""
 
     footer = {}
 
@@ -48,7 +48,7 @@ def process() -> Response:
         stream = file.stream.read()
         file_name = file.filename
         del file
-        log.info("Processing file given via 'file' parameter, file name: " + file_name)
+        log.info("Processing file given via 'file' parameter, file name: " + str(file_name))
     else:
         # if it is sent as a data-binary
         log.info("Processing binary as data-binary, generating temporary file name...")
@@ -70,37 +70,48 @@ def process() -> Response:
 
             log.info("Stream contains valid JSON.")
         except json.JSONDecodeError:
-            log.warning("Stream does not contain valid JSON.")
+            log.error("Stream does not contain valid JSON.")
 
-    output_text, doc_metadata = api.processor.process_stream(stream=stream, file_name=file_name)
+    output_text, doc_metadata = api.processor.process_stream(stream=stream, file_name=file_name)  # type: ignore
 
+    success = False
+    code = 200
+    log_message = ""
     if len(output_text) > 0:
-        response = build_response(output_text, footer=footer, metadata=doc_metadata)
-        return Response(response=json.dumps({"result": response}),
-                        status=200,
-                        mimetype="application/json")
+        success = True
     else:
-        response = build_response(output_text,
-                                  metadata=doc_metadata,
-                                  success=False,
-                                  log_message="No text has been generated")
-
-        return Response(response=json.dumps({"result": response}),
-                        status=500,
-                        mimetype="application/json")
+        code = 500
+        log_message = "No text has been generated"
+
+    response = build_response(output_text,
+                              footer=footer,
+                              metadata=doc_metadata,
+                              success=success,
+                              log_message=log_message)
+
+    if OCR_SERVICE_RESPONSE_OUTPUT_TYPE == "json":
+        response = json.dumps({"result": response})
+    elif OCR_SERVICE_RESPONSE_OUTPUT_TYPE == "dict":
+        response = {"result": response}
+
+    return Response(response=response,
+                    status=code,
+                    mimetype="application/json")
 
 @api.route("/process_file", methods=["POST"])
 def process_file() -> Response:
     stream = None
-    file_name: str = ""
+    file_name: str | None = ""
+
+    global log
 
     # if it is sent via the file parameter (file keeps its original name)
     if len(request.files):
         file = list(request.files.values())[0]
         stream = file.stream.read()
         file_name = file.filename
         del file
-        log.info("Processing file given via 'file' parameter, file name: " + file_name)
+        log.info("Processing file given via 'file' parameter, file name: " + str(file_name))
     else:
         # if it is sent as a data-binary
         log.info("Processing binary as data-binary, generating temporary file name...")
@@ -109,7 +120,7 @@ def process_file() -> Response:
 
         stream = request.get_data(cache=False, as_text=False, parse_form_data=False)
 
-    output_text, doc_metadata = api.processor.process_stream(stream=stream, file_name=file_name)
+    output_text, doc_metadata = api.processor.process_stream(stream=stream, file_name=file_name)  # type: ignore
 
     if len(output_text) > 0:
         response = build_response(output_text, metadata=doc_metadata)