[Fixes #10138] Render pdf thumbnails from top margin (#10139) (#10141)

* render pdf thumbnails from top margin

* fix flake

Co-authored-by: Giovanni Allegri <giohappy@gmail.com>

geonode/base/models.py

@@ -1721,7 +1721,7 @@ def has_thumbnail(self):
 
     # Note - you should probably broadcast layer#post_save() events to ensure
     # that indexing (or other listeners) are notified
-    def save_thumbnail(self, filename, image):
+    def save_thumbnail(self, filename, image, **kwargs):
         upload_path = get_unique_upload_path(filename)
         # force convertion to JPEG output file
         upload_path = f'{os.path.splitext(upload_path)[0]}.jpg'
@@ -1745,7 +1745,8 @@ def save_thumbnail(self, filename, image):
                     # Optimize the Thumbnail size and resolution
                     _default_thumb_size = settings.THUMBNAIL_SIZE
                     im = Image.open(storage_manager.open(actual_name))
-                    cover = ImageOps.fit(im, (_default_thumb_size['width'], _default_thumb_size['height'])).convert("RGB")
+                    centering = kwargs.get("centering", (0.5, 0.5))
+                    cover = ImageOps.fit(im, (_default_thumb_size['width'], _default_thumb_size['height']), centering=centering).convert("RGB")
 
                     # Saving the thumb into a temporary directory on file system
                     tmp_location = os.path.abspath(f"{settings.MEDIA_ROOT}/{upload_path}")

geonode/documents/management/commands/sync_geonode_documents.py

@@ -35,13 +35,21 @@ def add_arguments(self, parser):
             action='store_true',
             dest="updatethumbnails",
             default=False,
-            help="Update the document thumbnails.")
+            help="Generate thumbnails for documents. Only documents without a thumbnail will be considered, unless --force is used.")
+
+        parser.add_argument(
+            '--force',
+            action='store_true',
+            dest="force",
+            default=False,
+            help="Force the update of thumbnails for all documents.")
 
     def handle(self, *args, **options):
         updatethumbnails = options.get('updatethumbnails')
+        force = options.get('force')
         for doc in Document.objects.all():
             if updatethumbnails:
-                if doc.thumbnail_url is None or doc.thumbnail_url == '':
+                if (doc.thumbnail_url is None or doc.thumbnail_url == '') or force:
                     try:
                         create_document_thumbnail(doc.id)
                     except Exception:

geonode/documents/tasks.py

@@ -35,6 +35,10 @@
 
 class DocumentRenderer():
     FILETYPES = ['pdf']
+    # See https://pillow.readthedocs.io/en/stable/reference/ImageOps.html#PIL.ImageOps.fit
+    CROP_CENTERING = {
+        'pdf': (0.0, 0.0)
+    }
 
     def __init__(self) -> None:
         pass
@@ -59,6 +63,9 @@ def render_pdf(self, filename):
             logger.warning(f'Cound not generate thumbnail for {filename}: {e}')
             return None
 
+    def preferred_crop_centering(self, filename):
+        return self.CROP_CENTERING.get(self._get_filetype(filename))
+
     def _get_filetype(self, filname):
         return os.path.splitext(filname)[1][1:]
 
@@ -92,6 +99,7 @@ def create_document_thumbnail(self, object_id):
 
     image_file = None
     thumbnail_content = None
+    centering = (0.5, 0.5)
 
     if document.is_image:
         dname = storage_manager.path(document.files[0])
@@ -113,14 +121,17 @@ def create_document_thumbnail(self, object_id):
     elif doc_renderer.supports(document.files[0]):
         try:
             thumbnail_content = doc_renderer.render(document.files[0])
+            preferred_centering = doc_renderer.preferred_crop_centering(document.files[0])
+            if preferred_centering is not None:
+                centering = preferred_centering
         except Exception as e:
             print(e)
     if not thumbnail_content:
         logger.warning(f"Thumbnail for document #{object_id} empty.")
         ResourceBase.objects.filter(id=document.id).update(thumbnail_url=None)
     else:
         filename = f'document-{document.uuid}-thumb.jpg'
-        document.save_thumbnail(filename, thumbnail_content)
+        document.save_thumbnail(filename, thumbnail_content, centering=centering)
         logger.debug(f"Thumbnail for document #{object_id} created.")