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'
)
task = task_generate_document_page_image.apply_async(
kwargs=dict(
document_page_id=document_page.pk
)
)
cache_filename = task.get(timeout=DOCUMENT_IMAGE_TASK_TIMEOUT)
with storage_documentimagecache.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
)
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'
)
task = task_generate_document_page_image.apply_async(
kwargs=dict(
document_page_id=document_page.pk
)
)
cache_filename = task.get(
timeout=DOCUMENT_IMAGE_TASK_TIMEOUT, disable_sync_subtasks=False
)
with document_page.cache_partition.get_file(filename=cache_filename).open() as file_object:
ocr_content = ocr_backend.execute(
file_object=file_object,
language=document_page.document.language
)
DocumentPageOCRContent.objects.update_or_create(
document_page=document_page, defaults={
'content': ocr_content
}
)
logger.info(
'Finished processing page: %d of document version: %s',
document_page.page_number, document_page.document_version
)