Refactor OCR app. Removes document parsing. Moves OCR processing to

model manager. Add submit and finish events. Signed-off-by: Roberto Rosario <roberto.rosario.gonzalez@gmail.com>
2017-08-23 02:04:57 -04:00
parent 2052caada4
commit 317d07a355
20 changed files with 309 additions and 497 deletions
--- a/mayan/apps/ocr/managers.py
+++ b/mayan/apps/ocr/managers.py
@@ -0,0 +1,82 @@
+from __future__ import unicode_literals
+
+import logging
+import sys
+import traceback
+
+from django.apps import apps
+from django.conf import settings
+from django.db import models
+
+from documents.runtime import cache_storage_backend
+
+from .events import event_ocr_document_version_finish
+from .runtime import ocr_backend
+from .signals import post_document_version_ocr
+
+logger = logging.getLogger(__name__)
+
+
+class DocumentPageOCRContentManager(models.Manager):
+    def process_document_version(self, document_version):
+        logger.info('Starting OCR for document version: %s', document_version)
+        logger.debug('document version: %d', document_version.pk)
+
+        try:
+            for document_page in document_version.pages.all():
+                self.process_document_page(document_page=document_page)
+        except Exception as exception:
+            logger.error(
+                'OCR error for document version: %d; %s', document_version,
+                exception
+            )
+
+            if settings.DEBUG:
+                result = []
+                type, value, tb = sys.exc_info()
+                result.append('%s: %s' % (type.__name__, value))
+                result.extend(traceback.format_tb(tb))
+                document_version.ocr_errors.create(
+                    result='\n'.join(result)
+                )
+            else:
+                document_version.ocr_errors.create(result=exception)
+        else:
+            logger.info(
+                'OCR complete for document version: %s', document_version
+            )
+            document_version.ocr_errors.all().delete()
+
+            event_ocr_document_version_finish.commit(target=document_version)
+
+            post_document_version_ocr.send(
+                sender=document_version.__class__, instance=document_version
+            )
+
+    def process_document_page(self, document_page):
+        logger.info(
+            'Processing page: %d of document version: %s',
+            document_page.page_number, document_page.document_version
+        )
+
+        DocumentPageOCRContent = apps.get_model(
+            app_label='ocr', model_name='DocumentPageOCRContent'
+        )
+
+        # TODO: Call task and wait
+        cache_filename = document_page.generate_image()
+
+        with cache_storage_backend.open(cache_filename) as file_object:
+            document_page_content, created = DocumentPageOCRContent.objects.get_or_create(
+                document_page=document_page
+            )
+            document_page_content.content = ocr_backend.execute(
+                file_object=file_object,
+                language=document_page.document.language
+            )
+            document_page_content.save()
+
+        logger.info(
+            'Finished processing page: %d of document version: %s',
+            document_page.page_number, document_page.document_version
+        )