bluesky truncation bug fix: use the appropriate lexicon, post vs profile

for snarfed/bridgy-fed#1128

granary/bluesky.py

@@ -679,7 +679,7 @@ def from_as1(obj, out_type=None, blobs=None, client=None, original_fields_prefix
     # maybe [Video] label, link that with a facet
     text = Bluesky('unused').truncate(
       full_text, original_post_text_suffix, include_link=include_link,
-      punctuation=('', ''))
+      punctuation=('', ''), type=type)
     truncated = text != full_text
     if truncated:
       text_byte_end = len(text.split(ELLIPSIS, maxsplit=1)[0].encode())
@@ -1436,7 +1436,7 @@ class Bluesky(Source):
   DOMAIN = 'bsky.app'
   BASE_URL = 'https://bsky.app'
   NAME = 'Bluesky'
-  TRUNCATE_TEXT_LENGTH = LEXRPC_BASE.defs['app.bsky.actor.profile']['record']['properties']['description']['maxGraphemes']
+  TRUNCATE_TEXT_LENGTH = None  # different for post text vs profile description
   POST_ID_RE = AT_URI_PATTERN
   TYPE_LABELS = {
     'post': 'post',
@@ -1827,10 +1827,6 @@ def _create(self, obj, preview=None, include_link=OMIT_LINK, ignore_formatting=F
 
     # truncate and ellipsize content if necessary
     url = obj.get('url')
-    content = self.truncate(content, url, include_link=include_link, type=type)
-
-    # TODO linkify mentions and hashtags
-    preview_content = util.linkify(content, pretty=True, skip_bare_cc_tlds=True)
 
     post_label = f"{self.NAME} {self.TYPE_LABELS['post']}"
 
@@ -1878,10 +1874,13 @@ def _create(self, obj, preview=None, include_link=OMIT_LINK, ignore_formatting=F
           'url': at_uri_to_web_url(repost_atp['subject']['uri']) + '/reposted-by'
         })
 
-    elif (type in ('note', 'article') or is_reply or
+    elif (type in as1.POST_TYPES or is_reply or
           (type == 'activity' and verb == 'post')):  # probably a bookmark
       # TODO: add bookmarked URL and facet
       # tricky because we only want to do that if it's not truncated away
+      content = self.truncate(content, url, include_link=include_link, type='note')
+      # TODO linkify mentions and hashtags
+      preview_content = util.linkify(content, pretty=True, skip_bare_cc_tlds=True)
       data = {'status': content}
       if is_reply and base_url:
         preview_description += f"<span class=\"verb\">{self.TYPE_LABELS['comment']}</span> to <a href=\"{base_url}\">this {self.TYPE_LABELS['post']}</a>:"
@@ -2004,11 +2003,18 @@ def upload_media(self, media):
 
     return blobs
 
-  def truncate(self, *args, **kwargs):
+  def truncate(self, *args, type=None, **kwargs):
     """Thin wrapper around :meth:`Source.truncate` that sets default kwargs."""
+    if type in as1.ACTOR_TYPES:
+      length = LEXRPC_BASE.defs['app.bsky.actor.profile']['record']['properties']['description']['maxGraphemes']
+    elif type in POST_TYPES:
+      length = LEXRPC_BASE.defs['app.bsky.feed.post']['record']['properties']['text']['maxGraphemes']
+    else:
+      assert False, f'unexpected type {type}'
+
     kwargs = {
       'include_link': INCLUDE_LINK,
-      'target_length': self.TRUNCATE_TEXT_LENGTH,
+      'target_length': length,
       'link_length': None,
       'ellipsis': ' […]',
       **kwargs,

granary/source.py

@@ -810,7 +810,6 @@ def _content_for_create(self, obj, ignore_formatting=False, prefer_name=False,
       ignore_formatting (bool): whether to use content text as is, instead of
         converting its HTML to plain text styling (newlines, etc.)
       prefer_name (bool): whether to prefer ``displayName`` to ``content``
-
       strip_first_video_tag (bool): if true, removes the first ``<video>`` tag.
         useful when it will be uploaded and attached to the post natively in the
         silo.

granary/tests/test_bluesky.py

@@ -19,6 +19,7 @@
   did_web_to_url,
   from_as1,
   from_as1_to_strong_ref,
+  LEXRPC_BASE,
   MAX_IMAGES,
   NO_AUTHENTICATED_LABEL,
   to_as1,
@@ -778,7 +779,8 @@ def test_from_as1_post_without_tag_indices(self):
     # no facet
     self.assert_equals(POST_BSKY, self.from_as1(post_as))
 
-  @patch.object(Bluesky, 'TRUNCATE_TEXT_LENGTH', 15)
+  @patch.dict(LEXRPC_BASE.defs['app.bsky.feed.post']['record']['properties']['text'],
+              maxGraphemes=15)
   def test_from_as1_post_truncate_adds_link_embed(self):
     self.assert_equals({
       '$type': 'app.bsky.feed.post',
@@ -801,7 +803,18 @@ def test_from_as1_post_truncate_adds_link_embed(self):
       'content': 'more than ten chars long',
     }))
 
-  @patch.object(Bluesky, 'TRUNCATE_TEXT_LENGTH', 45)
+  def test_from_as1_post_truncate_full_length(self):
+    # check that we use the app.bsky.feed.post limit, not app.bsky.actor.profile's
+    # https://github.com/snarfed/bridgy-fed/issues/1128
+    content = 'Das #BSW spricht den ganzen Tag von Frieden und Diplomatie. Beide Seiten anhören angeblich. Aber wenn das Opfer des illegalen Angriffskriegs was zu sagen hat, dann verlasssen sie aus Protest den Saal? Deutlicher kann man nicht machen, dass man möchte, dass der Angreifer gewinnt.'
+    self.assert_equals(content, self.from_as1({
+      'objectType': 'note',
+      'url': 'http://my.inst/post',
+      'content': content,
+    })['text'])
+
+  @patch.dict(LEXRPC_BASE.defs['app.bsky.feed.post']['record']['properties']['text'],
+              maxGraphemes=45)
   def test_from_as1_post_with_images_truncated_puts_original_post_link_in_text(self):
     content = 'hello hello hello hello hello hello hello hello hello'
     self.assert_equals({
@@ -826,7 +839,8 @@ def test_from_as1_post_with_images_truncated_puts_original_post_link_in_text(sel
       'url': 'http://my.inst/post',
     }, blobs={NEW_BLOB_URL: NEW_BLOB}))
 
-  @patch.object(Bluesky, 'TRUNCATE_TEXT_LENGTH', 51)
+  @patch.dict(LEXRPC_BASE.defs['app.bsky.feed.post']['record']['properties']['text'],
+              maxGraphemes=51)
   def test_from_as1_post_with_images_video_truncated_original_post_link_in_text(self):
     self.assert_equals({
       **POST_BSKY_IMAGES,
@@ -852,7 +866,8 @@ def test_from_as1_post_with_images_video_truncated_original_post_link_in_text(se
       }],
     }, blobs={NEW_BLOB_URL: NEW_BLOB}))
 
-  @patch.object(Bluesky, 'TRUNCATE_TEXT_LENGTH', 40)
+  @patch.dict(LEXRPC_BASE.defs['app.bsky.feed.post']['record']['properties']['text'],
+              maxGraphemes=40)
   def test_from_as1_post_with_images_removes_facets_beyond_truncation(self):
     content = 'hello <a href="http://foo">link</a> goodbye goodbye goodbye goodbye'
     self.assert_equals({
@@ -877,7 +892,8 @@ def test_from_as1_post_with_images_removes_facets_beyond_truncation(self):
       'url': 'http://my.inst/post',
     }, blobs={NEW_BLOB_URL: NEW_BLOB}))
 
-  @patch.object(Bluesky, 'TRUNCATE_TEXT_LENGTH', 40)
+  @patch.dict(LEXRPC_BASE.defs['app.bsky.feed.post']['record']['properties']['text'],
+              maxGraphemes=40)
   def test_from_as1_post_with_images_truncates_facet_that_overlaps_truncation(self):
     content = '<a href="http://foo">hello link text</a> goodbye goodbye goodbye goodbye'
     self.assert_equals({
@@ -1194,7 +1210,8 @@ def test_from_as1_html_link(self):
       'content': content,
     }))
 
-  @patch.object(Bluesky, 'TRUNCATE_TEXT_LENGTH', 12)
+  @patch.dict(LEXRPC_BASE.defs['app.bsky.feed.post']['record']['properties']['text'],
+              maxGraphemes=12)
   def test_from_as1_html_omit_link_facet_after_truncation(self):
     content = 'foo bar <a href="http://post">baaaaaaaz</a>'
     self.assert_equals({
@@ -2382,9 +2399,9 @@ def test_post_id(self):
       with self.subTest(input=input):
         self.assertEqual(expected, self.bs.post_id(input))
 
+  @patch.dict(LEXRPC_BASE.defs['app.bsky.feed.post']['record']['properties']['text'],
+              maxGraphemes=20)
   def test_preview_post(self):
-    self.bs.TRUNCATE_TEXT_LENGTH = 20
-
     for content, expected in (
         ('foo ☕ bar', 'foo ☕ bar'),
         ('too long, will be ellipsized', 'too long, will […]'),