bluesky: convert HTML summary and content to plain text

snarfed · Oct 5, 2023 · a1d1592 · a1d1592
1 parent 1259e63
commit a1d1592
Show file tree

Hide file tree

Showing 3 changed files with 32 additions and 7 deletions.
diff --git a/README.md b/README.md
@@ -290,6 +290,7 @@ Changelog
   * Add new `blob_to_url` function.
   * `from_as1`: add `out_type` kwarg to specify desired output type, eg `app.bsky.actor.profile` vs `app.bsky.actor.defs#profileViewBasic` vs `app.bsky.actor.defs#profileView`.
   * Delete `as1_to_profile`, switch `from_as1` to return `$type: app.bsky.actor.profile`.
+  * Convert HTML `summary` and `content` to plain text.
 * `facebook`:
   * Remove `Facebook.fql_stream_to_post`. [Facebook turned down FQL in 2016.](https://en.wikipedia.org/wiki/Facebook_Query_Language#History)
 

diff --git a/granary/bluesky.py b/granary/bluesky.py
@@ -16,7 +16,7 @@
 from oauth_dropins.webutil.util import trim_nulls
 
 from . import as1
-from .source import FRIENDS, Source, OMIT_LINK
+from .source import FRIENDS, html_to_text, Source, OMIT_LINK
 
 logger = logging.getLogger(__name__)
 
@@ -250,7 +250,7 @@ def from_as1(obj, out_type=None):
   if type in as1.ACTOR_TYPES:
     ret = {
       'displayName': obj.get('displayName'),
-      'description': obj.get('summary'),
+      'description': html_to_text(obj.get('summary')),
     }
     if not out_type or out_type == 'app.bsky.actor.profile':
       return trim_nulls({**ret, '$type': 'app.bsky.actor.profile'})
@@ -351,7 +351,7 @@ def from_as1(obj, out_type=None):
     src = Bluesky('unused')
     content = obj.get('content')
     text = obj.get('summary') or content or obj.get('name') or ''
-    text = src.truncate(text, None, OMIT_LINK)
+    text = src.truncate(html_to_text(text), None, OMIT_LINK)
 
     facets = []
     if text == content:

diff --git a/granary/tests/test_bluesky.py b/granary/tests/test_bluesky.py
@@ -368,16 +368,21 @@ def test_from_as1_post_with_author(self):
     got = from_as1(POST_AUTHOR_AS, out_type='app.bsky.feed.defs#postView')
     self.assert_equals(expected, got)
 
-  def test_from_as1_post_html_with_tag_indices_not_implemented(self):
+  def test_from_as1_post_html_skips_tag_indices(self):
     post_as = copy.deepcopy(POST_AS)
     post_as['object'].update({
       'content': '<em>some html</em>',
       'content_is_html': True,
-      'tags': [FACET_TAG],
+      # not set because content is HTML
+      # 'tags': [FACET_TAG],
     })
 
-    with self.assertRaises(NotImplementedError):
-      from_as1(post_as)
+    # with self.assertRaises(NotImplementedError):
+    self.assert_equals({
+      '$type': 'app.bsky.feed.post',
+      'text': '_some html_',
+      'createdAt': '2007-07-07T03:04:05',
+    },from_as1(post_as))
 
   def test_from_as1_post_without_tag_indices(self):
     post_as = copy.deepcopy(POST_AS)
@@ -404,6 +409,16 @@ def test_from_as1_post_view_with_image(self):
     got = from_as1(POST_AS_IMAGES, out_type='app.bsky.feed.defs#postView')
     self.assert_equals(expected, got)
 
+  def test_from_as1_post_content_html(self):
+    self.assertEqual({
+      '$type': 'app.bsky.feed.post',
+      'text': 'Some\n_HTML_',
+      'createdAt': '',
+    }, from_as1({
+      'objectType': 'note',
+      'content': '<p>Some <br> <em>HTML</em></p>',
+    }))
+
   def test_from_as1_object_vs_activity(self):
     obj = {
       'objectType': 'note',
@@ -454,6 +469,15 @@ def test_from_as1_actor_id_not_url(self):
       'id': 'tag:foo.com,2001:bar',
     }, out_type='app.bsky.actor.defs#profileView')['did'])
 
+  def test_from_as1_actor_description_html(self):
+    self.assertEqual({
+      '$type': 'app.bsky.actor.profile',
+      'description': 'Some\n_HTML_',
+    }, from_as1({
+      'objectType': 'person',
+      'summary': '<p>Some <br> <em>HTML</em></p>',
+    }))
+
   def test_from_as1_composite_url(self):
     self.assertEqual({
       '$type': 'app.bsky.actor.defs#profileView',