Unify the method to submit document for OCR, fix OCR error document re-queue view

mayan/apps/ocr/__init__.py

@@ -35,13 +35,17 @@ register_links(['ocr:queue_document_list'], [queue_document_list], menu_name='se
 register_maintenance_links([all_document_ocr_cleanup], namespace='ocr', title=_(u'OCR'))
 
 
+def document_ocr_submit(self):
+    task_do_ocr.apply_async(args=[self.pk], queue='ocr')
+
+
 @receiver(post_save, dispatch_uid='document_post_save', sender=DocumentVersion)
 def document_post_save(sender, instance, **kwargs):
     logger.debug('received post save signal')
     logger.debug('instance: %s' % instance)
     if kwargs.get('created', False):
         if AUTOMATIC_OCR:
-            task_do_ocr.apply_async(args=[instance.document.pk], queue='ocr')
+            instance.document.submit_for_ocr()
 
 
 @receiver(post_migrate, dispatch_uid='create_default_queue')
@@ -50,9 +54,11 @@ def create_default_queue_signal_handler(sender, **kwargs):
         DocumentQueue.objects.get_or_create(name='default')
 
 
-register_tool(ocr_tool_link)
+Document.add_to_class('submit_for_ocr', document_ocr_submit)
 
 class_permissions(Document, [PERMISSION_OCR_DOCUMENT])
 
 namespace = StatisticNamespace(name='ocr', label=_(u'OCR'))
 namespace.add_statistic(OCRStatistics(name='ocr_stats', label=_(u'OCR queue statistics')))
+
+register_tool(ocr_tool_link)

mayan/apps/ocr/exceptions.py

@@ -10,7 +10,3 @@ class UnpaperError(Exception):
     Raised by unpaper
     """
     pass
-
-
-class ReQueueError(Exception):
-    pass

mayan/apps/ocr/models.py

@@ -8,8 +8,6 @@ from django.utils.translation import ugettext_lazy as _
 
 from documents.models import Document
 
-from .exceptions import ReQueueError
-
 
 class DocumentQueue(models.Model):
     name = models.CharField(max_length=64, unique=True, verbose_name=_(u'Name'))
@@ -35,18 +33,6 @@ class QueueDocument(models.Model):
         verbose_name = _(u'Queue document')
         verbose_name_plural = _(u'Queue documents')
 
-    def requeue(self):
-        # TODO: Fix properly using Celery tasks
-        if self.state == QUEUEDOCUMENT_STATE_PROCESSING:
-            raise ReQueueError
-        else:
-            self.datetime_submitted = now()
-            self.state = QUEUEDOCUMENT_STATE_PENDING
-            self.delay = False
-            self.result = None
-            self.node_name = None
-            self.save()
-
     def __unicode__(self):
         try:
             return unicode(self.document)

mayan/apps/ocr/tasks.py

@@ -24,6 +24,8 @@ def task_do_ocr(document_pk):
     lock_id = u'task_do_ocr_doc-%d' % document_pk
     try:
         logger.debug('trying to acquire lock: %s' % lock_id)
+        # Acquire lock to avoid doing OCR on the same document more than once
+        # concurrently
         lock = Lock.acquire_lock(lock_id, LOCK_EXPIRE)
         logger.debug('acquired lock: %s' % lock_id)
         try:

mayan/apps/ocr/views.py

@@ -14,7 +14,6 @@ from documents.widgets import document_link, document_thumbnail
 from permissions.models import Permission
 
 from .api import clean_pages
-from .exceptions import ReQueueError
 from .models import DocumentQueue, QueueDocument
 from .permissions import (PERMISSION_OCR_CLEAN_ALL_PAGES,
                           PERMISSION_OCR_DOCUMENT,
@@ -120,7 +119,7 @@ def submit_document_to_queue(request, document, post_submit_redirect=None):
     This view is meant to be reusable
     """
 
-    task_do_ocr.apply_async(args=[document.pk], queue='ocr')
+    document.submit_for_ocr()
     messages.success(request, _(u'Document: %(document)s was added to the OCR queue.') % {
         'document': document}
     )
@@ -146,21 +145,15 @@ def re_queue_document(request, queue_document_id=None, queue_document_id_list=No
     if request.method == 'POST':
         for queue_document in queue_documents:
             try:
-                queue_document.requeue()
+                queue_document.document.submit_for_ocr()
                 messages.success(
                     request,
-                    _(u'Document: %(document)s was re-queued to the OCR queue: %(queue)s') % {
-                        'document': queue_document.document,
-                        'queue': queue_document.document_queue.label
+                    _(u'Document: %(document)s was re-queued for OCR.') % {
+                        'document': queue_document.document
                     }
                 )
             except Document.DoesNotExist:
                 messages.error(request, _(u'Document id#: %d, no longer exists.') % queue_document.document_id)
-            except ReQueueError:
-                messages.warning(
-                    request,
-                    _(u'Document: %s is already being processed and can\'t be re-queded.') % queue_document
-                )
         return HttpResponseRedirect(next)
 
     context = {