tesseract executor

Author: amprix

piper/base/backend/templates/fast-api-tsrct.j2

@@ -0,0 +1,34 @@
+import time
+
+from fastapi import FastAPI, Request, status, File, UploadFile
+from piper.envs import CurrentEnv
+from loguru import logger
+#logger = logging.getLogger(__name__)
+
+#logger.add("file.log", level="INFO", backtrace=True, diagnose=True, rotation='5 MB')
+
+{% for script_name in scripts.keys() %}
+from {{ script_name }} import *
+{% endfor %}
+
+app = FastAPI(debug=True)
+app.logger = logger
+logger.info(f'main here {time.time()}')
+logger.info(f'Tesseract executor')
+
+@app.post('/health_check', status_code = status.HTTP_200_OK)
+async def hl():
+    logger.info('health_check request')
+    return {"message": "health check"}
+
+with CurrentEnv():
+    logger.info(f'CurrentEnv')
+    service = {{ service_class }}( {% for k, v in service_kwargs.items() %} {{ k }}={{ v }}, {% endfor %} )
+    logger.info(f'service {service}')
+
+    @app.post('/{{ function_name }}')
+    async def {{ function_name }}(file: UploadFile = File(...)):
+        logger.info(f'recived file {file.filename}')
+        content = await file.read()
+        result = await service.{{ function_name }}(content)
+        return result

piper/base/docker/images/python-tesrct.j2

@@ -0,0 +1,16 @@
+FROM python:{{ python_docker_version }}
+
+RUN apt update && apt install -y software-properties-common
+RUN apt install -y tree cmake libgl1-mesa-glx poppler-utils tesseract-ocr \
+    libtesseract-dev libleptonica-dev tesseract-ocr-rus mc
+RUN pip3 install --upgrade pip
+
+WORKDIR /app
+
+COPY requirements.txt ./requirements.txt
+RUN pip3 install -r requirements.txt
+
+COPY ./ ./
+RUN chmod +x ./run.sh
+
+ENTRYPOINT ["{{ cmd }}"]

piper/services/ocr_data.jpg

@@ -0,0 +1,22 @@
+import cv2
+import pytesseract
+import numpy as np
+from loguru import logger
+
+
+def img_bytes_handler(img_bytes):
+    img = cv2.imdecode(np.asarray(bytearray(img_bytes), dtype=np.uint8), cv2.IMREAD_COLOR)
+    if img is not None:
+        logger.info(f'processing img with shape {img.shape}')
+        txt_dict = pytesseract.image_to_data(
+            img,
+            lang='rus',
+            config=r'--oem 1 --psm 11',
+            output_type=pytesseract.Output.DICT
+        )
+
+        logger.info(f'get text from image {txt_dict}')
+        return txt_dict
+
+    else:
+        logger.error('recive empty image or convertion failed')

piper/utils/tesrct_utils.py

@@ -0,0 +1,60 @@
+import docker
+import os
+import sys
+import asyncio
+import requests
+root_dir = os.path.join(os.path.realpath(os.path.pardir), 'piper')
+sys.path.insert(1, root_dir)
+
+from piper.utils import docker_utils as du
+from piper.envs import DockerEnv
+from piper.envs import is_docker_env
+from piper.configurations import get_configuration
+from piper.services import TesseractRecognizer, StringValue
+from pathlib import Path
+import os
+
+
+main_app_url = f'http://localhost:8788'
+
+# pytest -vs tests/tsrct_test.py::TestTesseract::test_recognizer
+class TestTesseract():
+
+    def test_tesseract_install(self):
+        #cfg = get_configuration()
+        # loop = asyncio.get_event_loop()
+        #tess = TesseractRecognizer(port=cfg.docker_app_port)
+        # result = loop.run_until_complete()
+
+        assert True
+
+    def test_recognizer(self):        
+        file_path = Path(__file__).parent
+        file_path = file_path.joinpath('ocr_data.jpg')
+        print(file_path)
+
+        url = f'{main_app_url}/recognize'
+        print(url)
+
+        multipart_form_data = {
+            'file': open(file_path, 'rb')
+            }
+
+        print(multipart_form_data)
+        print((multipart_form_data.get('file')))
+
+        # headers = {'Content-Type': 'multipart/form-data'}
+                
+        result = requests.post(url, files=multipart_form_data, verify=False)
+
+        data = result.json()
+
+        assert len(data) != 0
+
+        assert result.status_code == 200
+
+    def test_health_check(self):
+        url = f'{main_app_url}/health_check'
+        print(url)
+        result = requests.post(url)
+        assert result.status_code == 200