tesseract_integration

Dockerfile_base

@@ -8,7 +8,12 @@ ENV TZ=Europe/Moscow
 
 RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone
 
-RUN apt update && apt install -y libreoffice-writer libreoffice-impress default-jre
+RUN apt-get update &&  apt-get install -y \
+    libreoffice-writer \
+    libreoffice-impress \
+    default-jre \
+    tesseract-ocr \
+    tesseract-ocr-rus
 
 ADD requirements.txt .
 RUN python3 -m pip install -r requirements.txt --no-cache-dir

app/db/db_methods.py

@@ -37,16 +37,26 @@ def get_images(check_id):
     else:
         return None
 
-def save_image_to_db(check_id, image_data, caption, image_size):
+def save_image_to_db(check_id, image_data, caption, image_size, text=''):
     image = Image({
         'check_id': check_id,
         'image_data': image_data,
         'caption': caption,
-        'image_size': image_size
+        'image_size': image_size,
+        'text' : text
     })
-    images_collection.insert_one(image.pack())
-    print(str(check_id) + " " + str(caption))
+    result = images_collection.insert_one(image.pack())
+    return result.inserted_id 
 
+def update_image_text(image_id, new_text):
+    try:
+        result = images_collection.update_one(
+            {'_id': image_id},
+            {'$set': {'text': new_text}}
+        )
+        return result.matched_count > 0
+    except Exception:
+        return False
 
 # Returns user if user was created and None if already exists
 def add_user(username, password_hash='', is_LTI=False):

app/db/db_types.py

@@ -154,11 +154,13 @@ def __init__(self, dictionary=None):
         self.caption = dictionary.get('caption', '')  # Подпись к изображению
         self.image_data = dictionary.get('image_data')  # Файл изображения в формате bindata
         self.image_size = dictionary.get('image_size')  # Размер изображения в сантимерах
+        self.text = dictionary.get('text')
 
     def pack(self):
         package = super().pack()
         package['check_id'] = str(self.check_id)
         package['caption'] = self.caption
         package['image_data'] = self.image_data
         package['image_size'] = self.image_size
+        package['text'] = self.text
         return package

app/main/check_packs/pack_config.py

@@ -46,7 +46,6 @@
     ["theme_in_report_check"],
     ['key_words_report_check'],
     ["empty_task_page_check"],
-    ['image_quality_check'],
 ]
 
 DEFAULT_TYPE = 'pres'

app/main/checks/report_checks/__init__.py

@@ -25,5 +25,4 @@
 from .max_abstract_size_check import ReportMaxSizeOfAbstractCheck
 from .template_name import ReportTemplateNameCheck
 from .key_words_check import KeyWordsReportCheck
-from .empty_task_page_check import EmptyTaskPageCheck
-from .image_quality_check import ImageQualityCheck
+from .empty_task_page_check import EmptyTaskPageCheck

app/main/reports/docx_uploader/docx_uploader.py

@@ -244,6 +244,7 @@ def show_chapters(self, work_type):
 
     def extract_images_with_captions(self, check_id):
         from app.db.db_methods import save_image_to_db, get_images
+        from app.tesseract_tasks import tesseract_recognize
 
         emu_to_cm  = 360000
         image_found = False
@@ -293,13 +294,16 @@ def extract_images_with_captions(self, check_id):
                                 # Если параграф не содержит изображения и текст не пуст, то это подпись
                                 if not contains_image and next_paragraph_text:
                                     # Сохраняем изображение и его подпись
-                                    save_image_to_db(check_id, image_data, next_paragraph_text, (width_cm, height_cm))
+                                    image_id = save_image_to_db(check_id, image_data, next_paragraph_text, (width_cm, height_cm))
+                                    tesseract_recognize.delay(image_id, image_data)
                                     break
                                 else:
-                                    save_image_to_db(check_id, image_data, "picture without caption", (width_cm, height_cm))
+                                    image_id = save_image_to_db(check_id, image_data, "picture without caption", (width_cm, height_cm))
+                                    tesseract_recognize.delay(image_id, image_data)
                                     break
                         else:
-                            save_image_to_db(check_id, image_data, "picture without caption", (width_cm, height_cm))
+                            image_id = save_image_to_db(check_id, image_data, "picture without caption", (width_cm, height_cm))
+                            tesseract_recognize.delay(image_id, image_data)
 
                         image_found = False  # Сброс флага, чтобы искать следующее изображение
                         image_data = None  # Очистка данных изображения

app/tesseract_tasks.py

@@ -0,0 +1,49 @@
+import os
+from celery import Celery
+from celery.signals import worker_ready
+import pytesseract
+import cv2
+import numpy as np
+from db import db_methods
+from root_logger import get_root_logger
+
+TASK_RETRY_COUNTDOWN = 60
+logger = get_root_logger('tesseract_tasks')
+
+celery = Celery(__name__)
+celery.conf.broker_url = os.environ.get("CELERY_BROKER_URL", "redis://redis:6379")
+celery.conf.result_backend = os.environ.get("CELERY_RESULT_BACKEND", "redis://redis:6379")
+
+celery.conf.timezone = 'Europe/Moscow'
+
+TESSERACT_CONFIG = {
+    'lang': 'rus',
+    'config': '--psm 6',
+}
+
+@worker_ready.connect
+def at_start(sender, **k):
+    logger.info("Tesseract worker is ready!")
+
+
+@celery.task(name="tesseract_recognize", queue='tesseract-queue', bind=True)
+def tesseract_recognize(self, image_id, image_data):
+    try:
+        image_array = np.frombuffer(image_data, dtype=np.uint8)
+        img_cv = cv2.imdecode(image_array, cv2.IMREAD_COLOR)
+        if img_cv is None:
+            raise ValueError("Не удалось декодировать изображение из двоичных данных")
+        text = pytesseract.image_to_string(img_cv, **TESSERACT_CONFIG)
+        success = db_methods.update_image_text(image_id, text)
+        if not success:
+            logger.error(f"Не удалось записать текст для image_id: {image_id}")
+            raise Exception("Ошибка при обновлении текста изображения в базе данных")
+        logger.info(f"Текст успешно распознан и записан для image_id: {image_id}")
+        return text
+
+    except Exception as e:
+        logger.error(f"Ошибка при распознавании текста: {e}", exc_info=True)
+        if self.request.retries == self.max_retries:
+            logger.error(f"Достигнуто максимальное количество попыток для image_id: {image_id}")
+            return f"Ошибка: {e}"
+        self.retry(countdown=TASK_RETRY_COUNTDOWN)

docker-compose.yml

@@ -74,6 +74,23 @@ services:
     volumes:
       - ../slides_checker_mongo_data:/data/db
     cpuset: ${CONTAINER_CPU:-0-1}
+
+  tesseract_worker:
+    image: document_insight_system_image
+    restart: always
+    command: celery --app=app.tasks.celery worker -n tesseract@worker -Q tesseract-queue --loglevel=info
+    environment:
+      - CELERY_BROKER_URL=${REDIS_URL}
+      - CELERY_RESULT_BACKEND=${REDIS_URL}
+    depends_on:
+      - redis
+      - mongodb
+    volumes:
+      - presentation_files:/usr/src/project/files/
+      - "/etc/timezone:/etc/timezone:ro"
+      - "/etc/localtime:/etc/localtime:ro"
+    cpuset: ${CONTAINER_CPU:-0-1}
+    mem_limit: ${WORKER_MEMORY:-1G}
 
 volumes:
   flower_data:

requirements.txt

@@ -35,4 +35,5 @@ filetype==1.2.0
 language-tool-python==2.7.1
 markdown==3.4.4
 md2pdf==1.0.1
-opencv-python==4.5.5.64
+opencv-python==4.5.5.64
+pytesseract==0.3.10