Add support for fpng-py, update Azure APIs

README.md

@@ -18,7 +18,7 @@ This has been tested with Python 3.11. Newer/older versions might work. It can b
 ## Cloud providers
 - Google Lens: Google Vision in disguise (no need for API keys!), however it needs to download a couple megabytes of data for each request. You need to install pyjson5 and requests (`pip install pyjson5 requests`) ("l" key)
 - Google Vision: you need a service account .json file named google_vision.json in `user directory/.config/` and installing google-cloud-vision (`pip install google-cloud-vision`) ("g" key)
-- Azure Computer Vision: you need to specify an api key and an endpoint in the config file (see below) and to install azure-cognitiveservices-vision-computervision (`pip install azure-cognitiveservices-vision-computervision`) ("v" key)
+- Azure Image Analysis: you need to specify an api key and an endpoint in the config file (see below) and to install azure-ai-vision-imageanalysis (`pip install azure-ai-vision-imageanalysis`) ("v" key)
 
 # Usage
 
@@ -31,6 +31,7 @@ However:
 - holding ctrl or cmd at any time will pause image processing temporarily
 - for systems where text can be copied to the clipboard at the same time as images, if `*ocr_ignore*` is copied with an image, the image will be ignored
 - optionally, notifications can be enabled in the config file to show the text with a native OS notification
+- optionally, you can speed up the online providers by installing fpng-py: `pip install fpng-py` (requires a developer environment on some operating systems/Python versions)
 - idle resource usage on macOS and Windows when reading from the clipboard has been eliminated using native OS polling
 - a config file (to be created in `user directory/.config/owocr_config.ini`, on Windows `user directory` is the `C:\Users\yourusername` folder) can be used to configure the script, as an example to limit providers (to reduce clutter/memory usage) as well as specifying provider settings such as api keys etc. A sample config file is provided [here](https://raw.githubusercontent.com/AuroraWright/owocr/master/owocr_config.ini)
 

owocr/ocr.py

@@ -33,10 +33,10 @@ except ImportError:
     pass
 
 try:
-    from azure.cognitiveservices.vision.computervision import ComputerVisionClient
+    from azure.ai.vision.imageanalysis import ImageAnalysisClient
-    from azure.cognitiveservices.vision.computervision.models import OperationStatusCodes
+    from azure.ai.vision.imageanalysis.models import VisualFeatures
-    from msrest.authentication import CognitiveServicesCredentials
+    from azure.core.credentials import AzureKeyCredential
-    from msrest.exceptions import ClientRequestError
+    from azure.core.exceptions import ServiceRequestError
 except ImportError:
     pass
 
@@ -66,6 +66,12 @@ try:
 except ImportError:
     pass
 
+try:
+    import fpng_py
+    optimized_png_encode = True
+except:
+    optimized_png_encode = False
+
 
 def empty_post_process(text):
     return text
@@ -79,6 +85,22 @@ def post_process(text):
     return text
 
 
+def pil_image_to_bytes(img, img_format='png', png_compression=6):
+    if img_format == 'png' and optimized_png_encode:
+        raw_data = img.convert('RGBA').tobytes()
+        width, height = img.size
+        image_bytes = fpng_py.fpng_encode_image_to_memory(raw_data, width, height)
+    else:
+        image_bytes = io.BytesIO()
+        img.save(image_bytes, format=img_format, compress_level=png_compression)
+        image_bytes = image_bytes.getvalue()
+    return image_bytes
+
+
+def pil_image_to_numpy_array(img):
+    return np.array(img.convert('RGBA'))
+
+
 class MangaOcr:
     name = 'mangaocr'
     readable_name = 'Manga OCR'
@@ -149,9 +171,7 @@ class GoogleVision:
         return x
 
     def _preprocess(self, img):
-        image_bytes = io.BytesIO()
+        return pil_image_to_bytes(img)
-        img.save(image_bytes, format='png')
-        return image_bytes.getvalue()
 
 class GoogleLens:
     name = 'glens'
@@ -216,9 +236,7 @@ class GoogleLens:
             new_h = int(new_w / aspect_ratio)
             img = img.resize((new_w, new_h), Image.LANCZOS)
 
-        image_bytes = io.BytesIO()
+        return pil_image_to_bytes(img)
-        img.save(image_bytes, format='png')
-        return image_bytes.getvalue()
 
 class AppleVision:
     name = 'avision'
@@ -268,9 +286,7 @@ class AppleVision:
             return x
 
     def _preprocess(self, img):
-        image_bytes = io.BytesIO()
+        return pil_image_to_bytes(img, 'tiff')
-        img.save(image_bytes, format='tiff')
-        return image_bytes.getvalue()
 
 class WinRTOCR:
     name = 'winrtocr'
@@ -326,27 +342,25 @@ class WinRTOCR:
         return x
 
     def _preprocess(self, img):
-        image_bytes = io.BytesIO()
+        return pil_image_to_bytes(img, png_compression=1)
-        img.save(image_bytes, format='png', compress_level=1)
-        return image_bytes.getvalue()
 
-class AzureComputerVision:
+class AzureImageAnalysis:
     name = 'azure'
-    readable_name = 'Azure Computer Vision'
+    readable_name = 'Azure Image Analysis'
     key = 'v'
     available = False
 
     def __init__(self, config={}):
-        if 'azure.cognitiveservices.vision.computervision' not in sys.modules:
+        if 'azure.ai.vision.imageanalysis' not in sys.modules:
-            logger.warning('azure-cognitiveservices-vision-computervision not available, Azure Computer Vision will not work!')
+            logger.warning('azure-ai-vision-imageanalysis not available, Azure Image Analysis will not work!')
         else:
             logger.info(f'Parsing Azure credentials')
             try:
-                self.client = ComputerVisionClient(config['endpoint'], CognitiveServicesCredentials(config['api_key']))
+                self.client = ImageAnalysisClient(config['endpoint'], AzureKeyCredential(config['api_key']))
                 self.available = True
-                logger.info('Azure Computer Vision ready')
+                logger.info('Azure Image Analysis ready')
             except:
-                logger.warning('Error parsing Azure credentials, Azure Computer Vision will not work!')
+                logger.warning('Error parsing Azure credentials, Azure Image Analysis will not work!')
 
     def __call__(self, img_or_path):
         if isinstance(img_or_path, str) or isinstance(img_or_path, Path):
@@ -356,29 +370,16 @@ class AzureComputerVision:
         else:
             raise ValueError(f'img_or_path must be a path or PIL.Image, instead got: {img_or_path}')
 
-        image_io = self._preprocess(img)
-        logging.getLogger('urllib3.connectionpool').disabled = True
-
         try:
-            read_response = self.client.read_in_stream(image_io, raw=True)
+            read_result = self.client.analyze(image_data=self._preprocess(img), visual_features=[VisualFeatures.READ])
-
+        except ServiceRequestError:
-            read_operation_location = read_response.headers['Operation-Location']
-            operation_id = read_operation_location.split('/')[-1]
-
-            while True:
-                read_result = self.client.get_read_result(operation_id)
-                if read_result.status.lower() not in [OperationStatusCodes.not_started, OperationStatusCodes.running]:
-                    break
-                time.sleep(0.3)
-        except ClientRequestError:
             return (False, 'Connection error!')
         except:
             return (False, 'Unknown error!')
 
         res = ''
-        if read_result.status == OperationStatusCodes.succeeded:
+        if read_result.read:
-            for text_result in read_result.analyze_result.read_results:
+            for line in read_result.read.blocks[0].lines:
-                for line in text_result.lines:
                 res += line.text + ' '
         else:
             return (False, 'Unknown error!')
@@ -387,10 +388,7 @@ class AzureComputerVision:
         return x
 
     def _preprocess(self, img):
-        image_io = io.BytesIO()
+        return pil_image_to_bytes(img)
-        img.save(image_io, format='png')
-        image_io.seek(0)
-        return image_io
 
 class EasyOCR:
     name = 'easyocr'
@@ -424,7 +422,7 @@ class EasyOCR:
         return x
 
     def _preprocess(self, img):
-        return np.array(img.convert('RGB'))
+        return pil_image_to_numpy_array(img)
 
 class RapidOCR:
     name = 'rapidocr'
@@ -450,6 +448,7 @@ class RapidOCR:
 
             logger.info('Loading RapidOCR model')
             self.model = ROCR(rec_model_path=rapidocr_model_file)
+            logging.getLogger().disabled = True
             self.available = True
             logger.info('RapidOCR ready')
 
@@ -461,7 +460,6 @@ class RapidOCR:
         else:
             raise ValueError(f'img_or_path must be a path or PIL.Image, instead got: {img_or_path}')
 
-        logging.getLogger().disabled = True
         res = ''
         read_results, elapsed = self.model(self._preprocess(img))
         if read_results:
@@ -472,4 +470,4 @@ class RapidOCR:
         return x
 
     def _preprocess(self, img):
-        return np.array(img.convert('RGB'))
+        return pil_image_to_numpy_array(img)