Migration updates

Squash version page migrations.
Add manual OCR and parsing migrations.
Fix tests.
Page search updates.

Signed-off-by: Roberto Rosario <roberto.rosario@mayan-edms.com>

mayan/apps/document_parsing/admin.py

@@ -3,13 +3,13 @@ from __future__ import unicode_literals
 from django.contrib import admin
 
 from .models import (
-    DocumentPageContent, DocumentVersionParseError
+    DocumentVersionPageContent, DocumentVersionParseError
 )
 
 
-@admin.register(DocumentPageContent)
+@admin.register(DocumentVersionPageContent)
-class DocumentPageContentAdmin(admin.ModelAdmin):
+class DocumentVersionPageContentAdmin(admin.ModelAdmin):
-    list_display = ('document_page',)
+    list_display = ('document_version_page',)
 
 
 @admin.register(DocumentVersionParseError)

mayan/apps/document_parsing/api_views.py

@@ -8,7 +8,7 @@ from rest_framework.response import Response
 from mayan.apps.documents.models import Document
 from mayan.apps.rest_api.permissions import MayanPermission
 
-from .models import DocumentPageContent
+from .models import DocumentVersionPageContent
 from .permissions import permission_content_view
 from .serializers import DocumentPageContentSerializer
 
@@ -41,8 +41,8 @@ class APIDocumentPageContentView(generics.RetrieveAPIView):
 
         try:
             content = instance.content
-        except DocumentPageContent.DoesNotExist:
+        except DocumentVersionPageContent.DoesNotExist:
-            content = DocumentPageContent.objects.none()
+            content = DocumentVersionPageContent.objects.none()
 
         serializer = self.get_serializer(content)
         return Response(serializer.data)

mayan/apps/document_parsing/apps.py

@@ -133,7 +133,7 @@ class DocumentParsingApp(MayanAppConfig):
         )
 
         document_search.add_model_field(
-            field='versions__pages__content__content', label=_('Content')
+            field='pages__content__content', label=_('Content')
         )
 
         document_page_search.add_model_field(

mayan/apps/document_parsing/forms.py

@@ -8,7 +8,7 @@ from django.utils.translation import ugettext_lazy as _, ugettext
 
 from mayan.apps.common.widgets import TextAreaDiv
 
-from .models import DocumentPageContent
+from .models import DocumentVersionPageContent
 
 
 class DocumentContentForm(forms.Form):
@@ -29,7 +29,7 @@ class DocumentContentForm(forms.Form):
         for page in document_pages:
             try:
                 page_content = page.content.content
-            except DocumentPageContent.DoesNotExist:
+            except DocumentVersionPageContent.DoesNotExist:
                 pass
             else:
                 content.append(conditional_escape(force_text(page_content)))
@@ -73,7 +73,7 @@ class DocumentPageContentForm(forms.Form):
 
         try:
             page_content = document_page.content.content
-        except DocumentPageContent.DoesNotExist:
+        except DocumentVersionPageContent.DoesNotExist:
             pass
         else:
             content = conditional_escape(force_text(page_content))

mayan/apps/document_parsing/migrations/0005_rename_page_content.py

@@ -0,0 +1,28 @@
+from __future__ import unicode_literals
+
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ('document_parsing', '0004_auto_20180917_0645'),
+        ('documents', '0052_rename_document_page'),
+    ]
+
+    operations = [
+        migrations.RenameModel(
+            'DocumentPageContent', 'DocumentVersionPageContent'
+        ),
+        migrations.AlterField(
+            model_name='documentversionpagecontent',
+            name='document_page',
+            field=models.OneToOneField(
+                on_delete=django.db.models.deletion.CASCADE,
+                name='document_version_page',
+                related_name='content',
+                to='documents.DocumentVersionPage',
+                verbose_name='Document version page'
+            ),
+        ),
+    ]

mayan/apps/document_parsing/models.py

@@ -5,20 +5,20 @@ from django.utils.encoding import force_text, python_2_unicode_compatible
 from django.utils.translation import ugettext_lazy as _
 
 from mayan.apps.documents.models import (
-    DocumentPage, DocumentType, DocumentVersion
+    DocumentPage, DocumentType, DocumentVersion, DocumentVersionPage
 )
 
 from .managers import DocumentPageContentManager, DocumentTypeSettingsManager
 
 
 @python_2_unicode_compatible
-class DocumentPageContent(models.Model):
+class DocumentVersionPageContent(models.Model):
     """
     This model store's the parsed content of a document page.
     """
-    document_page = models.OneToOneField(
+    document_version_page = models.OneToOneField(
         on_delete=models.CASCADE, related_name='content',
-        to=DocumentPage, verbose_name=_('Document page')
+        to=DocumentVersionPage, verbose_name=_('Document version page')
     )
     content = models.TextField(
         blank=True, help_text=_(
@@ -30,8 +30,8 @@ class DocumentPageContent(models.Model):
     objects = DocumentPageContentManager()
 
     class Meta:
-        verbose_name = _('Document page content')
+        verbose_name = _('Document version page content')
-        verbose_name_plural = _('Document pages contents')
+        verbose_name_plural = _('Document version pages contents')
 
     def __str__(self):
         return force_text(self.document_page)

mayan/apps/document_parsing/serializers.py

@@ -2,10 +2,10 @@ from __future__ import unicode_literals
 
 from rest_framework import serializers
 
-from .models import DocumentPageContent
+from .models import DocumentVersionPageContent
 
 
 class DocumentPageContentSerializer(serializers.ModelSerializer):
     class Meta:
         fields = ('content',)
-        model = DocumentPageContent
+        model = DocumentVersionPageContent

mayan/apps/document_parsing/urls.py

@@ -7,7 +7,9 @@ from .views import (
     DocumentContentView, DocumentContentDeleteView,
     DocumentContentDownloadView, DocumentPageContentView,
     DocumentParsingErrorsListView, DocumentSubmitView,
-    DocumentTypeSettingsEditView, DocumentTypeSubmitView, ParseErrorListView
+    DocumentTypeSettingsEditView, DocumentTypeSubmitView,
+    DocumentVersionPageContentView,
+    ParseErrorListView
 )
 
 urlpatterns = [
@@ -34,6 +36,11 @@ urlpatterns = [
         regex=r'^documents/pages/(?P<pk>\d+)/content/$',
         view=DocumentPageContentView.as_view(), name='document_page_content'
     ),
+    url(
+        regex=r'^documents/versions/pages/(?P<pk>\d+)/content/$',
+        view=DocumentVersionPageContentView.as_view(),
+        name='document_version_page_content'
+    ),
     url(
         regex=r'^documents/(?P<pk>\d+)/submit/$',
         view=DocumentSubmitView.as_view(), name='document_submit'

mayan/apps/document_parsing/views.py

@@ -13,11 +13,11 @@ from mayan.apps.common.generics import (
 from mayan.apps.common.mixins import ExternalObjectMixin
 from mayan.apps.documents.forms import DocumentTypeFilteredSelectForm
 from mayan.apps.documents.models import (
-    Document, DocumentType, DocumentVersionPage
+    Document, DocumentPage, DocumentType, DocumentVersionPage
 )
 
 from .forms import DocumentContentForm, DocumentPageContentForm
-from .models import DocumentPageContent, DocumentVersionParseError
+from .models import DocumentVersionPageContent, DocumentVersionParseError
 from .permissions import (
     permission_content_view, permission_document_type_parsing_setup,
     permission_parse_document
@@ -48,7 +48,7 @@ class DocumentContentDeleteView(MultipleObjectConfirmActionView):
         return result
 
     def object_action(self, form, instance):
-        DocumentPageContent.objects.delete_content_for(
+        DocumentVersionPageContent.objects.delete_content_for(
             document=instance, user=self.request.user
         )
 
@@ -89,7 +89,7 @@ class DocumentContentDownloadView(SingleObjectDownloadView):
 
 class DocumentPageContentView(SingleObjectDetailView):
     form_class = DocumentPageContentForm
-    model = DocumentVersionPage
+    model = DocumentPage
     object_permission = permission_content_view
 
     def dispatch(self, request, *args, **kwargs):
@@ -109,6 +109,30 @@ class DocumentPageContentView(SingleObjectDetailView):
         }
 
 
+class DocumentVersionPageContentView(SingleObjectDetailView):
+    form_class = DocumentPageContentForm
+    model = DocumentVersionPage
+    object_permission = permission_content_view
+
+    def dispatch(self, request, *args, **kwargs):
+        result = super(DocumentPageContentView, self).dispatch(
+            request, *args, **kwargs
+        )
+        self.get_object().document.add_as_recent_document_for_user(
+            request.user
+        )
+        return result
+
+    def get_extra_context(self):
+        return {
+            'hide_labels': True,
+            'object': self.get_object(),
+            'title': _(
+                'Content for document version page: %s'
+            ) % self.get_object(),
+        }
+
+
 class DocumentParsingErrorsListView(SingleObjectListView):
     view_permission = permission_content_view
 

mayan/apps/documents/apps.py

@@ -127,6 +127,7 @@ class DocumentsApp(MayanAppConfig):
         DocumentType = self.get_model(model_name='DocumentType')
         DocumentTypeFilename = self.get_model(model_name='DocumentTypeFilename')
         DocumentVersion = self.get_model(model_name='DocumentVersion')
+        DocumentVersionPage = self.get_model(model_name='DocumentVersionPage')
         DuplicatedDocument = self.get_model(model_name='DuplicatedDocument')
 
         DynamicSerializerField.add_serializer(
@@ -228,7 +229,7 @@ class DocumentsApp(MayanAppConfig):
             model=Document, manager_name='passthrough'
         )
         ModelPermission.register_inheritance(
-            model=DocumentPage, related='document_version__document',
+            model=DocumentPage, related='document',
         )
         ModelPermission.register_manager(
             model=DocumentPage, manager_name='passthrough'
@@ -245,6 +246,9 @@ class DocumentsApp(MayanAppConfig):
         ModelPermission.register_inheritance(
             model=DocumentVersion, related='document',
         )
+        ModelPermission.register_inheritance(
+            model=DocumentVersionPage, related='document_version',
+        )
 
         # Document and document page thumbnail widget
         document_page_thumbnail_widget = DocumentPageThumbnailWidget()

mayan/apps/documents/migrations/0052_rename_document_page.py

@@ -5,7 +5,6 @@ import django.db.models.deletion
 
 
 class Migration(migrations.Migration):
-
     dependencies = [
         ('documents', '0051_documentpage_enabled'),
     ]
@@ -20,8 +19,20 @@ class Migration(migrations.Migration):
             name='document_version',
             field=models.ForeignKey(
                 on_delete=django.db.models.deletion.CASCADE,
-                related_name='version_pages', to='documents.DocumentVersion',
+                related_name='pages', to='documents.DocumentVersion',
                 verbose_name='Document version'
             ),
         ),
+        migrations.RemoveField(
+            model_name='documentversionpage',
+            name='enabled',
+        ),
+        migrations.AlterModelOptions(
+            name='documentversionpage',
+            options={
+                'ordering': ('page_number',),
+                'verbose_name': 'Document version page',
+                'verbose_name_plural': 'Document version pages'
+            },
+        ),
     ]

mayan/apps/documents/migrations/0053_auto_20191007_1922.py

@@ -1,27 +0,0 @@
-from __future__ import unicode_literals
-
-from django.db import migrations
-
-
-class Migration(migrations.Migration):
-    dependencies = [
-        ('documents', '0052_auto_20191007_1921'),
-        ('ocr', '0008_auto_20180917_0646'),
-        ('document_parsing', '0004_auto_20180917_0645'),
-    ]
-
-    operations = [
-        migrations.CreateModel(
-            name='DocumentPageResult',
-            fields=[
-            ],
-            options={
-                'verbose_name': 'Document version page',
-                'verbose_name_plural': 'Document version pages',
-                'ordering': ('document_version__document', 'page_number'),
-                'proxy': True,
-                'indexes': [],
-            },
-            bases=('documents.documentversionpage',),
-        ),
-    ]

mayan/apps/documents/migrations/0053_create_document_page_and_result_models.py

@@ -1,5 +1,3 @@
-# -*- coding: utf-8 -*-
-# Generated by Django 1.11.22 on 2019-10-08 21:16
 from __future__ import unicode_literals
 
 from django.db import migrations, models
@@ -7,10 +5,9 @@ import django.db.models.deletion
 
 
 class Migration(migrations.Migration):
-
     dependencies = [
         ('contenttypes', '0002_remove_content_type_name'),
-        ('documents', '0054_auto_20191008_1522'),
+        ('documents', '0052_rename_document_page'),
     ]
 
     operations = [
@@ -25,22 +22,36 @@ class Migration(migrations.Migration):
                 ('document', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, related_name='pages', to='documents.Document', verbose_name='Document')),
             ],
             options={
+                'unique_together': set([('document', 'page_number')]),
                 'verbose_name': 'Document page',
                 'verbose_name_plural': 'Document pages',
                 'ordering': ('page_number',),
             },
         ),
-        migrations.AlterModelOptions(
+        migrations.CreateModel(
-            name='documentversionpage',
+            name='DocumentPageResult',
-            options={'ordering': ('page_number',), 'verbose_name': 'Document version page', 'verbose_name_plural': 'Document version pages'},
+            fields=[
+            ],
+            options={
+                'verbose_name': 'Document page result',
+                'verbose_name_plural': 'Document pages result',
+                'ordering': ('document', 'page_number'),
+                'proxy': True,
+                'indexes': [],
+            },
+            bases=('documents.documentpage',),
         ),
-        migrations.AlterField(
+        migrations.CreateModel(
-            model_name='documentversionpage',
+            name='DocumentVersionPageResult',
-            name='document_version',
+            fields=[
-            field=models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, related_name='pages', to='documents.DocumentVersion', verbose_name='Document version'),
+            ],
-        ),
+            options={
-        migrations.AlterUniqueTogether(
+                'verbose_name': 'Document version page',
-            name='documentpage',
+                'verbose_name_plural': 'Document version pages',
-            unique_together=set([('document', 'page_number')]),
+                'ordering': ('document_version__document', 'page_number'),
+                'proxy': True,
+                'indexes': [],
+            },
+            bases=('documents.documentversionpage',),
         ),
     ]

mayan/apps/documents/migrations/0054_auto_20191008_1522.py

@@ -1,31 +0,0 @@
-# -*- coding: utf-8 -*-
-# Generated by Django 1.11.22 on 2019-10-08 15:22
-from __future__ import unicode_literals
-
-from django.db import migrations
-
-
-class Migration(migrations.Migration):
-
-    dependencies = [
-        ('documents', '0053_auto_20191007_1922'),
-    ]
-
-    operations = [
-        migrations.DeleteModel(
-            name='DocumentPageResult',
-        ),
-        migrations.CreateModel(
-            name='DocumentVersionPageResult',
-            fields=[
-            ],
-            options={
-                'verbose_name': 'Document version page',
-                'verbose_name_plural': 'Document version pages',
-                'ordering': ('document_version__document', 'page_number'),
-                'proxy': True,
-                'indexes': [],
-            },
-            bases=('documents.documentversionpage',),
-        ),
-    ]

mayan/apps/documents/migrations/0054_reset_document_pages.py

@@ -0,0 +1,62 @@
+from __future__ import unicode_literals
+
+from django.db import migrations
+
+
+def get_latest_version(document):
+    return document.versions.order_by('timestamp').last()
+
+
+def operation_reset_document_pages(apps, schema_editor):
+    Document = apps.get_model(app_label='documents', model_name='Document')
+    DocumentPage = apps.get_model(
+        app_label='documents', model_name='DocumentPage'
+    )
+
+    # Define inside the function to use the migration's apps instance
+    def pages_reset(document):
+        ContentType = apps.get_model('contenttypes', 'ContentType')
+        DocumentVersionPage = apps.get_model(
+            app_label='documents', model_name='DocumentVersionPage'
+        )
+
+        content_type = ContentType.objects.get_for_model(
+            model=DocumentVersionPage
+        )
+
+        for document_page in document.pages.all():
+            document_page.delete()
+
+        for version_page in get_latest_version(document=document).pages.all():
+            document_page = document.pages.create(
+                content_type=content_type,
+                page_number=version_page.page_number,
+                object_id=version_page.pk,
+            )
+
+    for document in Document.objects.using(schema_editor.connection.alias).all():
+        pages_reset(document=document)
+
+
+def operation_reset_document_pages_reverse(apps, schema_editor):
+    Document = apps.get_model(app_label='documents', model_name='Document')
+    DocumentPage = apps.get_model(
+        app_label='documents', model_name='DocumentPage'
+    )
+
+    for document in Document.objects.using(schema_editor.connection.alias).all():
+        for document_page in document.pages.all():
+            document_page.delete()
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ('documents', '0053_create_document_page_and_result_models'),
+    ]
+
+    operations = [
+        migrations.RunPython(
+            code=operation_reset_document_pages,
+            reverse_code=operation_reset_document_pages_reverse
+        ),
+    ]

mayan/apps/documents/models/document_models.py

@@ -116,6 +116,14 @@ class Document(models.Model):
         )
         return partition
 
+    @property
+    def checksum(self):
+        return self.latest_version.checksum
+
+    @property
+    def date_updated(self):
+        return self.latest_version.timestamp
+
     def delete(self, *args, **kwargs):
         to_trash = kwargs.pop('to_trash', True)
 
@@ -140,6 +148,14 @@ class Document(models.Model):
         else:
             return False
 
+    @property
+    def file_mime_encoding(self):
+        return self.latest_version.encoding
+
+    @property
+    def file_mimetype(self):
+        return self.latest_version.mimetype
+
     def get_absolute_url(self):
         return reverse(
             viewname='documents:document_preview', kwargs={'pk': self.pk}
@@ -154,6 +170,10 @@ class Document(models.Model):
     def is_in_trash(self):
         return self.in_trash
 
+    @property
+    def latest_version(self):
+        return self.versions.order_by('timestamp').last()
+
     def natural_key(self):
         return (self.uuid,)
     natural_key.dependencies = ['documents.DocumentType']
@@ -171,8 +191,6 @@ class Document(models.Model):
 
         logger.info('New document version queued for document: %s', self)
 
-        self.reset_pages(update_page_count=False)
-
         return document_version
 
     def open(self, *args, **kwargs):
@@ -197,7 +215,7 @@ class Document(models.Model):
         )
         return DocumentPage.passthrough.filter(document=self)
 
-    def reset_pages(self, update_page_count=True):
+    def pages_reset(self, update_page_count=True):
         with transaction.atomic():
             for page in self.pages.all():
                 page.delete()
@@ -254,25 +272,3 @@ class Document(models.Model):
     @property
     def size(self):
         return self.latest_version.size
-
-    # Compatibility methods
-
-    @property
-    def checksum(self):
-        return self.latest_version.checksum
-
-    @property
-    def date_updated(self):
-        return self.latest_version.timestamp
-
-    @property
-    def file_mime_encoding(self):
-        return self.latest_version.encoding
-
-    @property
-    def file_mimetype(self):
-        return self.latest_version.mimetype
-
-    @property
-    def latest_version(self):
-        return self.versions.order_by('timestamp').last()

mayan/apps/documents/models/document_page_models.py

@@ -29,9 +29,8 @@ from ..settings import (
 )
 
 from .document_models import Document
-#from .document_version_page_models import DocumentVersionPage
 
-__all__ = ('DocumentPage',)# 'DocumentPageResult')
+__all__ = ('DocumentPage', 'DocumentPageResult')
 logger = logging.getLogger(__name__)
 
 
@@ -301,9 +300,9 @@ class DocumentPage(models.Model):
         return '{}-{}'.format(self.document.uuid, self.pk)
 
 
-#class DocumentVersionPageResult(DocumentVersionPage):
+class DocumentPageResult(DocumentPage):
-#    class Meta:
+    class Meta:
-#        ordering = ('document_version__document', 'page_number')
+        ordering = ('document', 'page_number')
-#        proxy = True
+        proxy = True
-#        verbose_name = _('Document version page')
+        verbose_name = _('Document page result')
-#        verbose_name_plural = _('Document version pages')
+        verbose_name_plural = _('Document pages result')

mayan/apps/documents/models/document_version_models.py

@@ -246,12 +246,12 @@ class DocumentVersion(models.Model):
 
             return result
 
-    @property
+    #@property
-    def page_count(self):
+    #def page_count(self):
-        """
+    #    """
-        The number of pages that the document posses.
+    #    The number of pages that the document posses.
-        """
+    #    """
-        return self.pages.count()
+    #    return self.pages.count()
 
     def revert(self, _user=None):
         """
@@ -326,6 +326,8 @@ class DocumentVersion(models.Model):
                         sender=Document, instance=self.document
                     )
 
+                self.document.pages_reset(update_page_count=False)
+
     def save_to_file(self, file_object):
         """
         Save a copy of the document from the document storage backend

mayan/apps/documents/models/document_version_page_models.py

@@ -41,14 +41,12 @@ class DocumentVersionPage(models.Model):
         on_delete=models.CASCADE, related_name='pages', to=DocumentVersion,
         verbose_name=_('Document version')
     )
-    enabled = models.BooleanField(default=True, verbose_name=_('Enabled'))
     page_number = models.PositiveIntegerField(
         db_index=True, default=1, editable=False,
         verbose_name=_('Page number')
     )
 
     objects = DocumentVersionPageManager()
-    #passthrough = models.Manager()
 
     class Meta:
         ordering = ('page_number',)
@@ -261,11 +259,11 @@ class DocumentVersionPage(models.Model):
         return (self.page_number, self.document_version.natural_key())
     natural_key.dependencies = ['documents.DocumentVersion']
 
-    #@property
+    @property
-    #def siblings(self):
+    def siblings(self):
-    #    return DocumentVersionPage.objects.filter(
+        return DocumentVersionPage.objects.filter(
-    #        document_version=self.document_version
+            document_version=self.document_version
-    #    )
+        )
 
     @property
     def uuid(self):

mayan/apps/documents/queues.py

@@ -71,7 +71,7 @@ queue_tools.add_task_type(
 )
 
 queue_uploads.add_task_type(
-    dotted_path='mayan.apps.documents.tasks.task_document_reset_pages',
+    dotted_path='mayan.apps.documents.tasks.task_document_pages_reset',
     label=_('Reset document pages')
 )
 queue_uploads.add_task_type(

mayan/apps/documents/search.py

@@ -17,12 +17,20 @@ def transformation_format_uuid(term_string):
         return term_string
 
 
-def get_queryset_page_search_queryset():
+def get_queryset_document_page_search_queryset():
     # Ignore documents in trash can
     DocumentPage = apps.get_model(
+        app_label='documents', model_name='DocumentPage'
+    )
+    return DocumentPage.objects.filter(document__in_trash=False)
+
+
+def get_queryset_document_version_page_search_queryset():
+    # Ignore documents in trash can
+    DocumentVersionPage = apps.get_model(
         app_label='documents', model_name='DocumentVersionPage'
     )
-    return DocumentPage.objects.filter(document_version__document__in_trash=False)
+    return DocumentVersionPage.objects.filter(document_version__document__in_trash=False)
 
 
 document_search = SearchModel(
@@ -49,25 +57,50 @@ document_search.add_model_field(
 
 document_page_search = SearchModel(
     app_label='documents', list_mode=LIST_MODE_CHOICE_ITEM,
-    model_name='DocumentVersionPage', permission=permission_document_view,
+    model_name='DocumentPage', permission=permission_document_view,
-    queryset=get_queryset_page_search_queryset,
+    queryset=get_queryset_document_page_search_queryset,
     serializer_path='mayan.apps.documents.serializers.DocumentPageSerializer'
 )
 
+document_version_page_search = SearchModel(
+    app_label='documents', list_mode=LIST_MODE_CHOICE_ITEM,
+    model_name='DocumentVersionPage', permission=permission_document_view,
+    queryset=get_queryset_document_version_page_search_queryset,
+    serializer_path='mayan.apps.documents.serializers.DocumentPageVersionSerializer'
+)
+
 document_page_search.add_model_field(
-    field='document_version__document__document_type__label',
+    field='document__document_type__label',
     label=_('Document type')
 )
 document_page_search.add_model_field(
-    field='document_version__document__versions__mimetype',
+    field='document__versions__mimetype',
     label=_('MIME type')
 )
 document_page_search.add_model_field(
+    field='document__label', label=_('Label')
+)
+document_page_search.add_model_field(
+    field='document__description', label=_('Description')
+)
+document_page_search.add_model_field(
+    field='document__document_version__checksum', label=_('Checksum')
+)
+
+document_version_page_search.add_model_field(
+    field='document_version__document__document_type__label',
+    label=_('Document type')
+)
+document_version_page_search.add_model_field(
+    field='document_version__document__versions__mimetype',
+    label=_('MIME type')
+)
+document_version_page_search.add_model_field(
     field='document_version__document__label', label=_('Label')
 )
-document_page_search.add_model_field(
+document_version_page_search.add_model_field(
     field='document_version__document__description', label=_('Description')
 )
-document_page_search.add_model_field(
+document_version_page_search.add_model_field(
     field='document_version__checksum', label=_('Checksum')
 )

mayan/apps/documents/tasks.py

@@ -66,7 +66,7 @@ def task_delete_stubs():
 
 
 @app.task(bind=True, default_retry_delay=RETRY_DELAY_DOCUMENT_RESET_PAGES, ignore_result=True)
-def task_document_reset_pages(self, document_id):
+def task_document_pages_reset(self, document_id):
     Document = apps.get_model(
         app_label='documents', model_name='Document'
     )
@@ -74,7 +74,7 @@ def task_document_reset_pages(self, document_id):
     document = Document.objects.get(pk=document_id)
 
     try:
-        document.reset_pages()
+        document.pages_reset()
     except OperationalError as exception:
         logger.warning(
             'Operational error during attempt to reset pages for '

mayan/apps/documents/tests/mixins.py

@@ -70,6 +70,7 @@ class DocumentTestMixin(object):
         self.test_document = document
         self.test_documents.append(document)
         self.test_document_version = document.latest_version
+        self.test_document_page = document.pages_all.first()
 
 
 class DocumentTypeViewTestMixin(object):
@@ -264,4 +265,3 @@ class DocumentViewTestMixin(object):
 
     def _request_empty_trash_view(self):
         return self.post(viewname='documents:trash_can_empty')
-

mayan/apps/documents/tests/test_api.py

@@ -530,8 +530,7 @@ class DocumentPageAPIViewTestMixin(object):
         page = self.test_document.pages.first()
         return self.get(
             viewname='rest_api:documentpage-image', kwargs={
-                'pk': page.document.pk, 'version_pk': page.document_version.pk,
+                'pk': page.document.pk, 'page_pk': page.pk
-                'page_pk': page.pk
             }
         )
 
@@ -552,6 +551,33 @@ class DocumentPageAPIViewTestCase(
         self.assertEqual(response.status_code, status.HTTP_200_OK)
 
 
+class DocumentVersionPageAPIViewTestMixin(object):
+    def _request_document_version_page_image(self):
+        page = self.test_document_version.pages.first()
+        return self.get(
+            viewname='rest_api:documentversionpage-image', kwargs={
+                'pk': page.document.pk, 'version_pk': page.document_version.pk,
+                'page_pk': page.pk
+            }
+        )
+
+
+class DocumentVersionPageAPIViewTestCase(
+    DocumentVersionPageAPIViewTestMixin, DocumentTestMixin, BaseAPITestCase
+):
+    def test_document_version_page_api_image_view_no_access(self):
+        response = self._request_document_version_page_image()
+        self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+
+    def test_document_version_page_api_image_view_with_access(self):
+        self.grant_access(
+            obj=self.test_document, permission=permission_document_view
+        )
+
+        response = self._request_document_version_page_image()
+        self.assertEqual(response.status_code, status.HTTP_200_OK)
+
+
 class TrashedDocumentAPIViewTestMixin(object):
     def _request_test_document_api_trash_view(self):
         return self.delete(
@@ -580,7 +606,6 @@ class TrashedDocumentAPIViewTestMixin(object):
         return self.get(
             viewname='rest_api:documentpage-image', kwargs={
                 'pk': latest_version.document.pk,
-                'version_pk': latest_version.pk,
                 'page_pk': latest_version.pages.first().pk
             }
         )

mayan/apps/documents/tests/test_document_page_views.py

@@ -9,10 +9,10 @@ from ..permissions import (
 from .base import GenericDocumentViewTestCase
 
 
-class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
+class DocumentPageDisableViewTestMixin(object):
-    def setUp(self):
+    def _disable_test_document_page(self):
-        super(DocumentPageDisableViewTestCase, self).setUp()
+        self.test_document_page.enabled = False
-        self.test_document_page = self.test_document.pages_all.first()
+        self.test_document_page.save()
 
     def _request_test_document_page_disable_view(self):
         return self.post(
@@ -21,6 +21,31 @@ class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
             }
         )
 
+    def _request_test_document_page_enable_view(self):
+        return self.post(
+            viewname='documents:document_page_enable', kwargs={
+                'pk': self.test_document_page.pk
+            }
+        )
+
+    def _request_test_document_page_multiple_disable_view(self):
+        return self.post(
+            viewname='documents:document_page_multiple_disable', data={
+                'id_list': self.test_document_page.pk
+            }
+        )
+
+    def _request_test_document_page_multiple_enable_view(self):
+        return self.post(
+            viewname='documents:document_page_multiple_enable', data={
+                'id_list': self.test_document_page.pk
+            }
+        )
+
+
+class DocumentPageDisableViewTestCase(
+    DocumentPageDisableViewTestMixin, GenericDocumentViewTestCase
+):
     def test_document_page_disable_view_no_permission(self):
         test_document_page_count = self.test_document.pages.count()
 
@@ -45,13 +70,6 @@ class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
             test_document_page_count, self.test_document.pages.count()
         )
 
-    def _request_test_document_page_multiple_disable_view(self):
-        return self.post(
-            viewname='documents:document_page_multiple_disable', data={
-                'id_list': self.test_document_page.pk
-            }
-        )
-
     def test_document_page_multiple_disable_view_no_permission(self):
         test_document_page_count = self.test_document.pages.count()
 
@@ -76,17 +94,6 @@ class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
             test_document_page_count, self.test_document.pages.count()
         )
 
-    def _disable_test_document_page(self):
-        self.test_document_page.enabled = False
-        self.test_document_page.save()
-
-    def _request_test_document_page_enable_view(self):
-        return self.post(
-            viewname='documents:document_page_enable', kwargs={
-                'pk': self.test_document_page.pk
-            }
-        )
-
     def test_document_page_enable_view_no_permission(self):
         self._disable_test_document_page()
 
@@ -114,13 +121,6 @@ class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
             test_document_page_count, self.test_document.pages.count()
         )
 
-    def _request_test_document_page_multiple_enable_view(self):
-        return self.post(
-            viewname='documents:document_page_multiple_enable', data={
-                'id_list': self.test_document_page.pk
-            }
-        )
-
     def test_document_page_multiple_enable_view_no_permission(self):
         self._disable_test_document_page()
         test_document_page_count = self.test_document.pages.count()
@@ -148,7 +148,7 @@ class DocumentPageDisableViewTestCase(GenericDocumentViewTestCase):
         )
 
 
-class DocumentPageViewTestCase(GenericDocumentViewTestCase):
+class DocumentPageViewTestMixin(object):
     def _request_test_document_page_list_view(self):
         return self.get(
             viewname='documents:document_pages', kwargs={
@@ -156,6 +156,18 @@ class DocumentPageViewTestCase(GenericDocumentViewTestCase):
             }
         )
 
+    def _request_test_document_page_view(self, document_page):
+        return self.get(
+            viewname='documents:document_page_view', kwargs={
+                'pk': document_page.pk,
+            }
+        )
+
+
+class DocumentPageViewTestCase(
+    DocumentPageViewTestMixin, GenericDocumentViewTestCase
+):
+
     def test_document_page_list_view_no_permission(self):
         response = self._request_test_document_page_list_view()
         self.assertEqual(response.status_code, 404)
@@ -170,13 +182,6 @@ class DocumentPageViewTestCase(GenericDocumentViewTestCase):
             response=response, text=self.test_document.label, status_code=200
         )
 
-    def _request_test_document_page_view(self, document_page):
-        return self.get(
-            viewname='documents:document_page_view', kwargs={
-                'pk': document_page.pk,
-            }
-        )
-
     def test_document_page_view_no_permissions(self):
         response = self._request_test_document_page_view(
             document_page=self.test_document.pages.first()

mayan/apps/documents/tests/test_search.py

@@ -1,9 +1,11 @@
 from __future__ import unicode_literals
 
 from mayan.apps.common.tests.base import BaseTestCase
-from mayan.apps.documents.permissions import permission_document_view
+
-from mayan.apps.documents.search import document_search, document_page_search
+from ..permissions import permission_document_view
-from mayan.apps.documents.tests.mixins import DocumentTestMixin
+from ..search import document_search, document_page_search
+
+from .mixins import DocumentTestMixin
 
 
 class DocumentSearchTestCase(DocumentTestMixin, BaseTestCase):

mayan/apps/documents/tests/test_trashed_document_views.py

@@ -9,7 +9,7 @@ from ..permissions import (
 from .base import GenericDocumentViewTestCase
 
 
-class TrashedDocumentTestCase(GenericDocumentViewTestCase):
+class TrashedDocumentTestMixin(object):
     def _request_document_restore_get_view(self):
         return self.get(
             viewname='documents:document_restore', kwargs={
@@ -17,6 +17,46 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
             }
         )
 
+    def _request_document_restore_post_view(self):
+        return self.post(
+            viewname='documents:document_restore', kwargs={
+                'pk': self.test_document.pk
+            }
+        )
+
+    def _request_document_trash_get_view(self):
+        return self.get(
+            viewname='documents:document_trash', kwargs={
+                'pk': self.test_document.pk
+            }
+        )
+
+    def _request_document_trash_post_view(self):
+        return self.post(
+            viewname='documents:document_trash', kwargs={
+                'pk': self.test_document.pk
+            }
+        )
+
+    def _request_trashed_document_delete_get_view(self):
+        return self.get(
+            viewname='documents:document_delete', kwargs={
+                'pk': self.test_document.pk
+            }
+        )
+
+    def _request_trashed_document_delete_post_view(self):
+        return self.post(
+            viewname='documents:document_delete', kwargs={
+                'pk': self.test_document.pk
+            }
+        )
+
+    def _request_trashed_document_list_view(self):
+        return self.get(viewname='documents:document_list_deleted')
+
+
+class TrashedDocumentTestCase(GenericDocumentViewTestCase):
     def test_document_restore_get_view_no_permission(self):
         self.test_document.delete()
         self.assertEqual(Document.objects.count(), 0)
@@ -43,13 +83,6 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
 
         self.assertEqual(Document.objects.count(), document_count)
 
-    def _request_document_restore_post_view(self):
-        return self.post(
-            viewname='documents:document_restore', kwargs={
-                'pk': self.test_document.pk
-            }
-        )
-
     def test_document_restore_post_view_no_permission(self):
         self.test_document.delete()
         self.assertEqual(Document.objects.count(), 0)
@@ -74,13 +107,6 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
         self.assertEqual(DeletedDocument.objects.count(), 0)
         self.assertEqual(Document.objects.count(), 1)
 
-    def _request_document_trash_get_view(self):
-        return self.get(
-            viewname='documents:document_trash', kwargs={
-                'pk': self.test_document.pk
-            }
-        )
-
     def test_document_trash_get_view_no_permissions(self):
         document_count = Document.objects.count()
 
@@ -101,13 +127,6 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
 
         self.assertEqual(Document.objects.count(), document_count)
 
-    def _request_document_trash_post_view(self):
-        return self.post(
-            viewname='documents:document_trash', kwargs={
-                'pk': self.test_document.pk
-            }
-        )
-
     def test_document_trash_post_view_no_permissions(self):
         response = self._request_document_trash_post_view()
         self.assertEqual(response.status_code, 404)
@@ -126,13 +145,6 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
         self.assertEqual(DeletedDocument.objects.count(), 1)
         self.assertEqual(Document.objects.count(), 0)
 
-    def _request_document_delete_get_view(self):
-        return self.get(
-            viewname='documents:document_delete', kwargs={
-                'pk': self.test_document.pk
-            }
-        )
-
     def test_document_delete_get_view_no_permissions(self):
         self.test_document.delete()
         self.assertEqual(Document.objects.count(), 0)
@@ -140,7 +152,7 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
 
         trashed_document_count = DeletedDocument.objects.count()
 
-        response = self._request_document_delete_get_view()
+        response = self._request_trashed_document_delete_get_view()
         self.assertEqual(response.status_code, 404)
 
         self.assertEqual(
@@ -158,26 +170,19 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
 
         trashed_document_count = DeletedDocument.objects.count()
 
-        response = self._request_document_delete_get_view()
+        response = self._request_trashed_document_delete_get_view()
         self.assertEqual(response.status_code, 200)
 
         self.assertEqual(
             DeletedDocument.objects.count(), trashed_document_count
         )
 
-    def _request_document_delete_post_view(self):
-        return self.post(
-            viewname='documents:document_delete', kwargs={
-                'pk': self.test_document.pk
-            }
-        )
-
     def test_document_delete_post_view_no_permissions(self):
         self.test_document.delete()
         self.assertEqual(Document.objects.count(), 0)
         self.assertEqual(DeletedDocument.objects.count(), 1)
 
-        response = self._request_document_delete_post_view()
+        response = self._request_trashed_document_delete_post_view()
         self.assertEqual(response.status_code, 404)
 
         self.assertEqual(Document.objects.count(), 0)
@@ -192,19 +197,16 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
             obj=self.test_document, permission=permission_document_delete
         )
 
-        response = self._request_document_delete_post_view()
+        response = self._request_trashed_document_delete_post_view()
         self.assertEqual(response.status_code, 302)
 
         self.assertEqual(DeletedDocument.objects.count(), 0)
         self.assertEqual(Document.objects.count(), 0)
 
-    def _request_document_list_deleted_view(self):
-        return self.get(viewname='documents:document_list_deleted')
-
     def test_deleted_document_list_view_no_permissions(self):
         self.test_document.delete()
 
-        response = self._request_document_list_deleted_view()
+        response = self._request_trashed_document_list_view()
         self.assertNotContains(
             response=response, text=self.test_document.label, status_code=200
         )
@@ -216,7 +218,7 @@ class TrashedDocumentTestCase(GenericDocumentViewTestCase):
             obj=self.test_document, permission=permission_document_view
         )
 
-        response = self._request_document_list_deleted_view()
+        response = self._request_trashed_document_list_view()
         self.assertContains(
             response=response, text=self.test_document.label, status_code=200
         )

mayan/apps/documents/views/document_views.py

@@ -44,7 +44,7 @@ from ..permissions import (
 from ..settings import (
     setting_print_width, setting_print_height, setting_recent_added_count
 )
-from ..tasks import task_document_reset_pages
+from ..tasks import task_document_pages_reset
 from ..utils import parse_range
 
 __all__ = (
@@ -450,7 +450,7 @@ class DocumentPagesResetView(MultipleObjectConfirmActionView):
     def object_action(self, form, instance):
         latest_version = instance.latest_version
         if latest_version:
-            task_document_reset_pages.apply_async(
+            task_document_pages_reset.apply_async(
                 kwargs={'document_id': instance.pk}
             )
         else:

mayan/apps/ocr/admin.py

@@ -3,13 +3,14 @@ from __future__ import unicode_literals
 from django.contrib import admin
 
 from .models import (
-    DocumentPageOCRContent, DocumentTypeSettings, DocumentVersionOCRError
+    DocumentTypeSettings, DocumentVersionPageOCRContent,
+    DocumentVersionOCRError
 )
 
 
-@admin.register(DocumentPageOCRContent)
+@admin.register(DocumentVersionPageOCRContent)
-class DocumentPageOCRContentAdmin(admin.ModelAdmin):
+class DocumentVersionPageOCRContentAdmin(admin.ModelAdmin):
-    list_display = ('document_page',)
+    list_display = ('document_version_page',)
 
 
 @admin.register(DocumentTypeSettings)

mayan/apps/ocr/api_views.py

@@ -8,7 +8,7 @@ from rest_framework.response import Response
 from mayan.apps.documents.models import Document, DocumentVersion
 from mayan.apps.rest_api.permissions import MayanPermission
 
-from .models import DocumentPageOCRContent
+from .models import DocumentVersionPageOCRContent
 from .permissions import permission_ocr_content_view, permission_ocr_document
 from .serializers import DocumentPageOCRContentSerializer
 
@@ -90,8 +90,8 @@ class APIDocumentPageOCRContentView(generics.RetrieveAPIView):
 
         try:
             ocr_content = instance.ocr_content
-        except DocumentPageOCRContent.DoesNotExist:
+        except DocumentVersionPageOCRContent.DoesNotExist:
-            ocr_content = DocumentPageOCRContent.objects.none()
+            ocr_content = DocumentVersionPageOCRContent.objects.none()
 
         serializer = self.get_serializer(ocr_content)
         return Response(serializer.data)

mayan/apps/ocr/apps.py

@@ -128,7 +128,7 @@ class OCRApp(MayanAppConfig):
         )
 
         document_search.add_model_field(
-            field='versions__version_pages__ocr_content__content', label=_('OCR')
+            field='pages__ocr_content__content', label=_('OCR')
         )
 
         document_page_search.add_model_field(

mayan/apps/ocr/forms.py

@@ -8,7 +8,7 @@ from django.utils.translation import ugettext_lazy as _, ugettext
 
 from mayan.apps.common.widgets import TextAreaDiv
 
-from .models import DocumentPageOCRContent
+from .models import DocumentVersionPageOCRContent
 
 
 class DocumentPageOCRContentForm(forms.Form):
@@ -30,7 +30,7 @@ class DocumentPageOCRContentForm(forms.Form):
 
         try:
             page_content = page.ocr_content.content
-        except DocumentPageOCRContent.DoesNotExist:
+        except DocumentVersionPageOCRContent.DoesNotExist:
             pass
         else:
             content = conditional_escape(force_text(page_content))
@@ -66,7 +66,7 @@ class DocumentOCRContentForm(forms.Form):
         for page in document_pages:
             try:
                 page_content = page.ocr_content.content
-            except DocumentPageOCRContent.DoesNotExist:
+            except DocumentVersionPageOCRContent.DoesNotExist:
                 pass
             else:
                 content.append(conditional_escape(force_text(page_content)))

mayan/apps/ocr/migrations/0009_rename_page_content.py

@@ -0,0 +1,28 @@
+from __future__ import unicode_literals
+
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ('ocr', '0008_auto_20180917_0646'),
+        ('documents', '0052_rename_document_page'),
+    ]
+
+    operations = [
+        migrations.RenameModel(
+            'DocumentPageOCRContent', 'DocumentVersionPageOCRContent'
+        ),
+        migrations.AlterField(
+            model_name='documentversionpageocrcontent',
+            name='document_page',
+            field=models.OneToOneField(
+                on_delete=django.db.models.deletion.CASCADE,
+                name='document_version_page',
+                related_name='ocr_content',
+                to='documents.DocumentVersionPage',
+                verbose_name='Document version page'
+            ),
+        ),
+    ]

mayan/apps/ocr/models.py

@@ -5,7 +5,7 @@ from django.utils.encoding import force_text, python_2_unicode_compatible
 from django.utils.translation import ugettext_lazy as _
 
 from mayan.apps.documents.models import (
-    DocumentPage, DocumentType, DocumentVersion
+    DocumentPage, DocumentType, DocumentVersion, DocumentVersionPage
 )
 
 from .managers import (
@@ -38,13 +38,13 @@ class DocumentTypeSettings(models.Model):
 
 
 @python_2_unicode_compatible
-class DocumentPageOCRContent(models.Model):
+class DocumentVersionPageOCRContent(models.Model):
     """
     This model stores the OCR results for a document page.
     """
-    document_page = models.OneToOneField(
+    document_version_page = models.OneToOneField(
         on_delete=models.CASCADE, related_name='ocr_content',
-        to=DocumentPage, verbose_name=_('Document page')
+        to=DocumentVersionPage, verbose_name=_('Document version page')
     )
     content = models.TextField(
         blank=True, help_text=_(
@@ -55,8 +55,8 @@ class DocumentPageOCRContent(models.Model):
     objects = DocumentPageOCRContentManager()
 
     class Meta:
-        verbose_name = _('Document page OCR content')
+        verbose_name = _('Document version page OCR content')
-        verbose_name_plural = _('Document pages OCR contents')
+        verbose_name_plural = _('Document version pages OCR contents')
 
     def __str__(self):
         return force_text(self.document_page)

mayan/apps/ocr/serializers.py

@@ -2,10 +2,10 @@ from __future__ import unicode_literals
 
 from rest_framework import serializers
 
-from .models import DocumentPageOCRContent
+from .models import DocumentVersionPageOCRContent
 
 
 class DocumentPageOCRContentSerializer(serializers.ModelSerializer):
     class Meta:
         fields = ('content',)
-        model = DocumentPageOCRContent
+        model = DocumentVersionPageOCRContent

mayan/apps/ocr/views.py

@@ -17,7 +17,7 @@ from mayan.apps.documents.models import (
 )
 
 from .forms import DocumentPageOCRContentForm, DocumentOCRContentForm
-from .models import DocumentPageOCRContent, DocumentVersionOCRError
+from .models import DocumentVersionPageOCRContent, DocumentVersionOCRError
 from .permissions import (
     permission_ocr_content_view, permission_ocr_document,
     permission_document_type_ocr_setup
@@ -48,7 +48,7 @@ class DocumentOCRContentDeleteView(MultipleObjectConfirmActionView):
         return result
 
     def object_action(self, form, instance):
-        DocumentPageOCRContent.objects.delete_content_for(
+        DocumentVersionPageOCRContent.objects.delete_content_for(
             document=instance, user=self.request.user
         )