as2: add new link_tags function

for snarfed/bridgy-fed#959, snarfed/bridgy-fed#1013

README.md

@@ -289,6 +289,7 @@ Changelog
 * `as2`:
   * Add support for the `Application`, `Block`, `Flag`, and `Link` types.
   * Generalize actor logic in `to/from_as1` across all actor types, not just `Person`.
+  * Add new `link_tags` function.
 * `atom`:
   * `activities_to_atom`: handle image attachments without `url` field.
 * `bluesky`:

granary/as2.py

@@ -567,3 +567,43 @@ def address(actor):
       match = re.match(r'^https?://(.+)/(users/|profile/|@)(.+)$', url)
       if match:
         return match.expand(r'@\3@\1')
+
+
+def link_tags(obj):
+  """Adds HTML links to ``content`` for tags with ``startIndex`` and ``length``.
+
+  ``content`` is modified in place. If ``content_is_html`` is ``true``, does
+  nothing. Otherwise, sets it to ``true`` if at least one link is added. Tags
+  without ``startIndex``/``length`` are ignored.
+
+  TODO: duplicated in :func:`microformats2.render_content`. unify?
+
+  Args:
+    obj (dict): AS2 JSON object
+  """
+  content = obj.get('content')
+  if not content or obj.get('content_is_html'):
+    return
+
+  # extract indexed tags, preserving order
+  tags = [tag for tag in obj.get('tag', [])
+          if 'startIndex' in tag and 'length' in tag and 'href' in tag]
+  tags.sort(key=lambda t: t['startIndex'])
+
+  # linkify embedded mention tags inside content.
+  last_end = 0
+  orig = content
+  linked = ''
+  for tag in tags:
+    start = tag['startIndex']
+    if start < last_end:
+      logger.warning(f'tag indices overlap! skipping {tag["href"]}')
+      continue
+    end = start + tag['length']
+    linked = f"{linked}{orig[last_end:start]}<a href=\"{tag['href']}\">{orig[start:end]}</a>"
+    last_end = end
+    obj['content_is_html'] = True
+    del tag['startIndex']
+    del tag['length']
+
+  obj['content'] = linked + orig[last_end:]

granary/microformats2.py

@@ -998,6 +998,7 @@ def render_content(obj, include_location=True, synthesize_content=True,
       tags.setdefault(as1.object_type(t), []).append(t)
 
   # linkify embedded mention tags inside content.
+  # TODO: duplicated in :func:`as2.link_tags`. unify?
   if mentions:
     mentions.sort(key=lambda t: t['startIndex'])
     last_end = 0

granary/tests/test_as2.py

@@ -472,3 +472,35 @@ def test_to_as1_flag(self):
       # note that this is the user being reported
       'to': 'http://bob',
     }))
+
+  def test_link_tags(self):
+    # no indices, shouuld be a noop
+    obj = {
+      'content': 'foo\nbar\nbaz',
+      'tag': [
+        {'href': 'http://bar'},
+        {'href': 'http://baz'},
+      ],
+    }
+    as2.link_tags(obj)
+    self.assert_equals('foo\nbar\nbaz', obj['content'])
+
+    # with indices, should link and then remove indices
+    obj['tag'] = [
+      {'href': 'http://bar', 'startIndex': 4, 'length': 3},
+      {'href': 'http://baz', 'startIndex': 8, 'length': 3},
+    ]
+
+    as2.link_tags(obj)
+    self.assert_equals({
+      'content': """\
+foo
+<a href="http://bar">bar</a>
+<a href="http://baz">baz</a>
+""",
+      'content_is_html': True,
+      'tag': [
+        {'href': 'http://bar'},
+        {'href': 'http://baz'},
+      ],
+    }, obj)