Add PDF orientation detection. Closes GitLab issue #387.

Signed-off-by: Roberto Rosario <roberto.rosario.gonzalez@gmail.com>

HISTORY.rst

@@ -3,6 +3,7 @@ XX (2017-XX-XX)
 - Add Django-mathfilters.
 - Improve render of documents with no pages.
 - Add SANE scanner document source.
+- Added PDF orientation detection.
 
 2.3 (2017-06-08)
 ================

mayan/apps/converter/backends/python.py

@@ -10,6 +10,7 @@ except ImportError:
     from StringIO import StringIO
 
 from PIL import Image
+import PyPDF2
 from pdfminer.pdfpage import PDFPage
 import sh
 
@@ -67,6 +68,19 @@ class Python(ConverterBase):
             finally:
                 fs_cleanup(input_filepath)
 
+    def detect_orientation(self, page_number):
+        # Use different ways depending on the file type
+        if self.mime_type == 'application/pdf':
+            pdf = PyPDF2.PdfFileReader(self.file_object)
+            result = pdf.getPage(page_number - 1).get('/Rotate')
+
+            self.file_object.seek(0)
+
+            return result
+
+        # Default rotation: 0 degrees
+        return 0
+
     def get_page_count(self):
         super(Python, self).get_page_count()
 

mayan/apps/converter/classes.py

@@ -214,6 +214,10 @@ class ConverterBase(object):
         except InvalidOfficeFormat as exception:
             logger.debug('Is not an office format document; %s', exception)
 
+    def detect_orientation(self, page_number):
+        # Must be overrided by subclass
+        pass
+
 
 class BaseTransformation(object):
     """

mayan/apps/converter/managers.py

@@ -102,3 +102,11 @@ class TransformationManager(models.Manager):
             return result
         else:
             return transformations
+
+    def add_for_model(self, obj, transformation, arguments=None):
+        content_type = ContentType.objects.get_for_model(obj)
+
+        self.create(
+            content_type=content_type, object_id=obj.pk,
+            name=transformation.name, arguments=arguments
+        )

mayan/apps/documents/models.py

@@ -420,6 +420,7 @@ class DocumentVersion(models.Model):
                     self.update_mimetype(save=False)
                     self.save()
                     self.update_page_count(save=False)
+                    self.fix_orientation()
 
                     logger.info(
                         'New document version "%s" created for document: %s',
@@ -466,6 +467,15 @@ class DocumentVersion(models.Model):
         """
         return self.file.storage.exists(self.file.name)
 
+    def fix_orientation(self):
+        for page in self.pages.all():
+            degrees = page.detect_orientation()
+            if degrees:
+                Transformation.objects.add_for_model(
+                    obj=page, transformation=TransformationRotate,
+                    arguments='{{"degrees": {}}}'.format(360-degrees)
+                )
+
     def get_intermidiate_file(self):
         cache_filename = self.cache_filename
         logger.debug('Intermidiate filename: %s', cache_filename)
@@ -693,6 +703,16 @@ class DocumentPage(models.Model):
     def document(self):
         return self.document_version.document
 
+    def detect_orientation(self):
+        with self.document_version.open() as file_object:
+            converter = converter_class(
+                file_object=file_object,
+                mime_type=self.document_version.mimetype
+            )
+            return converter.detect_orientation(
+                page_number=self.page_number
+            )
+
     def generate_image(self, *args, **kwargs):
         # Convert arguments into transformations
         transformations = kwargs.get('transformations', [])

requirements/base.txt

@@ -28,6 +28,7 @@ fusepy==2.0.4
 
 pdfminer==20140328
 pycountry==1.20
+PyPDF2==1.26.0
 pyocr==0.4.5
 python-dateutil==2.5.3
 python-gnupg==0.3.9

setup.py

@@ -82,6 +82,7 @@ djangorestframework==3.3.2
 djangorestframework-recursive==0.1.1
 fusepy==2.0.4
 pdfminer==20140328
+PyPDF2==1.26.0
 pycountry==1.20
 pyocr==0.4.5
 python-dateutil==2.5.3