Add image aspect ratio for Bluesky

README.md

@@ -317,6 +317,7 @@ Standardize function and method names in all modules to `to_as1`, `from_as`, etc
     * When a `flag` has multiple objects, use the first one that's an ATProto record.
     * Handle URLs more carefully, don't add link facets with invalid `uri`s.
     * Populate `blobs` into external embed `thumb`s.
+    * Parse image blobs and add `aspectRatio` to image record.
     * Bug fix: handle HTML links with `title` in `content` correctly.
     * Bug fix: handle attachments with no `id` or `url`.
   * `to_as1`:

granary/bluesky.py

@@ -22,6 +22,8 @@
 from oauth_dropins.webutil import util
 from oauth_dropins.webutil.util import trim_nulls
 import requests
+from PIL import Image
+from io import BytesIO
 
 from . import as1
 from .as2 import QUOTE_RE_SUFFIX
@@ -46,6 +48,8 @@
 HANDLE_PATTERN = re.compile(r'^' + HANDLE_REGEX)
 DID_WEB_PATTERN = re.compile(r'^did:web:' + HANDLE_REGEX)
 
+MAX_MEDIA_SIZE_BYTES = 5_000_000
+
 # at:// URI regexp
 # https://atproto.com/specs/at-uri-scheme#full-at-uri-syntax
 # https://atproto.com/specs/record-key#record-key-syntax
@@ -422,7 +426,7 @@ def base_object(obj):
   return {}
 
 
-def from_as1(obj, out_type=None, blobs=None, client=None,
+def from_as1(obj, out_type=None, blobs=None, aspects=None, client=None,
              original_fields_prefix=None, as_embed=False):
   """Converts an AS1 object to a Bluesky object.
 
@@ -443,6 +447,8 @@ def from_as1(obj, out_type=None, blobs=None, client=None,
       ``ref``, ``mimeType``, and ``size``) to use in the returned object. If not
       provided, or if this doesn't have an ``image`` or similar URL in the input
       object, its output blob will be omitted.
+    aspects (dict): optional mapping from str URL to int (width,height) tuple.
+        Used to provide aspect ratio in image embeds.
     client (Bluesky or lexrpc.Client): optional; if provided, this will be used
       to make API calls to PDSes to fetch and populate CIDs for records
       referenced by replies, likes, reposts, etc.
@@ -474,6 +480,8 @@ def from_as1(obj, out_type=None, blobs=None, client=None,
   actor = as1.get_object(activity, 'actor')
   if blobs is None:
     blobs = {}
+  if aspects is None:
+    aspects = {}
 
   # validate out_type
   if out_type:
@@ -652,11 +660,17 @@ def from_as1(obj, out_type=None, blobs=None, client=None,
               '$type': 'app.bsky.embed.images',
               'images': [],
             }
-          images_record_embed['images'].append({
+          image_record = {
             '$type': 'app.bsky.embed.images#image',
             'image': blob,
             'alt': alt,
-          })
+          }
+          if aspect := aspects.get(url):
+            image_record['aspectRatio'] = {
+              'width': aspect[0],
+              'height': aspect[1]
+            }
+          images_record_embed['images'].append(image_record)
 
     # first video => embed
     attachments = util.get_list(obj, 'attachments')
@@ -2172,8 +2186,8 @@ def _create(self, obj, preview=None, include_link=OMIT_LINK, ignore_formatting=F
                                description=preview_description)
 
       else:
-        blobs = self.upload_media(images)
-        post_atp = from_as1(obj, blobs=blobs, client=self)
+        blobs, aspects = self.upload_media(images)
+        post_atp = from_as1(obj, blobs=blobs, aspects=aspects, client=self)
         post_atp['text'] = content
 
         # facet for link to original post, if any
@@ -2249,6 +2263,7 @@ def base_object(self, obj):
 
   def upload_media(self, media):
     blobs = {}
+    aspects = {}
 
     for obj in media:
       url = util.get_url(obj, key='stream') or util.get_url(obj)
@@ -2257,14 +2272,20 @@ def upload_media(self, media):
 
       with util.requests_get(url, stream=True) as fetch:
         fetch.raise_for_status()
+        data = BytesIO(util.FileLimiter(fetch.raw, MAX_MEDIA_SIZE_BYTES).read())
+        content_type = fetch.headers.get('Content-Type', '')
+        if content_type.startswith("image/"):
+          with Image.open(data) as image:
+            aspects[url] = image.size
+          data.seek(0)
         upload = self.client.com.atproto.repo.uploadBlob(
-          input=fetch.raw,
-          headers={'Content-Type': fetch.headers['Content-Type']}
+          input=data,
+          headers={'Content-Type': content_type}
         )
 
       blobs[url] = upload['blob']
 
-    return blobs
+    return blobs, aspects
 
   def truncate(self, *args, type=None, **kwargs):
     """Thin wrapper around :meth:`Source.truncate` that sets default kwargs."""

granary/tests/test_bluesky.py

@@ -3,6 +3,7 @@
 Most tests are via files in testdata/.
 """
 import copy
+import os
 from io import BytesIO
 from unittest import skip
 from unittest.mock import ANY, patch
@@ -3481,8 +3482,55 @@ def test_preview_with_media(self):
   @patch('requests.post')
   @patch('requests.get')
   def test_create_with_media(self, mock_get, mock_post):
+    image_path = os.path.join(os.path.dirname(__file__), 'testdata/image.jpg')
+    with open(image_path, 'rb') as f:
+        mock_get.return_value = requests_response(
+          f.read(), headers={'Content-Type': 'image/jpeg'})
+
+    at_uri = 'at://did:plc:me/app.bsky.feed.post/abc123'
+    mock_post.side_effect = [
+      requests_response({'blob': NEW_BLOB}),
+      requests_response({'uri': at_uri, 'cid': 'sydddddd'}),
+    ]
+
+    self.assert_equals({
+      'id': at_uri,
+      'url': 'https://bsky.app/profile/handull/post/abc123',
+    }, self.bs.create(POST_AS_IMAGES['object']).content)
+
+    mock_get.assert_called_with(NEW_BLOB_URL, stream=True, timeout=HTTP_TIMEOUT,
+                                headers={'User-Agent': util.user_agent})
+    mock_post.assert_any_call(
+      'https://bsky.social/xrpc/com.atproto.repo.uploadBlob',
+      json=None,
+      data=ANY,
+      headers={
+        'Authorization': 'Bearer towkin',
+        'Content-Type': 'image/jpeg',
+        'User-Agent': util.user_agent,
+      })
+    # lexrpc.Client passes a BytesIO as data. sadly requests reads from that
+    # buffer and then closes it, so we can't check its contents.
+    # self.assertEqual(b'pic data', repr(mock_post.call_args_list[0][1]['data']))
+
+    expected = copy.deepcopy(POST_BSKY_IMAGES)
+    del expected['fooOriginalText']
+    del expected['fooOriginalUrl']
+    expected['embed']['images'][0]['aspectRatio'] = {
+      'width': 480,
+      'height': 640
+    }
+    self.assert_call(mock_post, 'com.atproto.repo.createRecord', json={
+      'repo': self.bs.did,
+      'collection': 'app.bsky.feed.post',
+      'record': expected,
+    })
+
+  @patch('requests.post')
+  @patch('requests.get')
+  def test_create_with_non_image_media(self, mock_get, mock_post):
     mock_get.return_value = requests_response(
-      'pic data', headers={'Content-Type': 'my/pic'})
+      b'something', headers={'Content-Type': 'video/mpeg'})
 
     at_uri = 'at://did:plc:me/app.bsky.feed.post/abc123'
     mock_post.side_effect = [
@@ -3503,7 +3551,7 @@ def test_create_with_media(self, mock_get, mock_post):
       data=ANY,
       headers={
         'Authorization': 'Bearer towkin',
-        'Content-Type': 'my/pic',
+        'Content-Type': 'video/mpeg',
         'User-Agent': util.user_agent,
       })
     # lexrpc.Client passes a BytesIO as data. sadly requests reads from that

requirements.txt

@@ -70,6 +70,7 @@ multiformats==0.3.1.post4
 multiformats-config==0.3.1
 oauthlib==3.2.2
 packaging==24.2
+Pillow==11.0.0
 pkce==1.0.3
 praw==7.8.1
 prawcore==2.4.0

setup.py

@@ -48,6 +48,7 @@
           'mf2util>=0.5.0',
           'multiformats>=0.3.1',
           'oauth-dropins>=6.4',
+          'pillow',
           'praw>=7.3.0',
           'python-dateutil>=2.8',
           'requests>=2.22',