Add `process_text_links` text utility to linkify URL's in text

Moved from Socialhome.

CHANGELOG.md

@@ -8,6 +8,8 @@
 
   If Django is configured, a profile will be retrieved using the configured profile
   getter function and the profile name or username will be used for the link.
+  
+* Add `process_text_links` text utility to linkify URL's in text.
 
 ### Changed
 

federation/tests/utils/test_text.py

@@ -1,4 +1,4 @@
-from federation.utils.text import decode_if_bytes, encode_if_text, validate_handle
+from federation.utils.text import decode_if_bytes, encode_if_text, validate_handle, process_text_links
 
 
 def test_decode_if_bytes():
@@ -11,6 +11,29 @@ def test_encode_if_text():
     assert encode_if_text("foobar") == b"foobar"
 
 
+class TestProcessTextLinks:
+    def test_link_at_start_or_end(self):
+        assert process_text_links('https://example.org example.org\nhttp://example.org') == \
+               '<a href="https://example.org" rel="nofollow" target="_blank">https://example.org</a> ' \
+               '<a href="http://example.org" rel="nofollow" target="_blank">example.org</a>\n' \
+               '<a href="http://example.org" rel="nofollow" target="_blank">http://example.org</a>'
+
+    def test_existing_links_get_attrs_added(self):
+        assert process_text_links('<a href="https://example.org">https://example.org</a>') == \
+               '<a href="https://example.org" rel="nofollow" target="_blank">https://example.org</a>'
+
+    def test_code_sections_are_skipped(self):
+        assert process_text_links('<code>https://example.org</code><code>\nhttps://example.org\n</code>') == \
+               '<code>https://example.org</code><code>\nhttps://example.org\n</code>'
+
+    def test_emails_are_skipped(self):
+        assert process_text_links('foo@example.org') == 'foo@example.org'
+
+    def test_does_not_add_target_blank_if_link_is_internal(self):
+        assert process_text_links('<a href="/streams/tag/foobar">#foobar</a>') == \
+               '<a href="/streams/tag/foobar">#foobar</a>'
+
+
 def test_validate_handle():
     assert validate_handle("foo@bar.com")
     assert validate_handle("Foo@baR.com")

federation/utils/text.py

@@ -1,6 +1,9 @@
 import re
 from urllib.parse import urlparse
 
+import bleach
+from bleach import callbacks
+
 
 def decode_if_bytes(text):
     try:
@@ -24,6 +27,29 @@ def get_path_from_url(url: str) -> str:
     return parsed.path
 
 
+def process_text_links(text):
+    """Process links in text, adding some attributes and linkifying textual links."""
+    link_callbacks = [callbacks.nofollow, callbacks.target_blank]
+
+    def link_attributes(attrs, new=False):
+        """Run standard callbacks except for internal links."""
+        href_key = (None, "href")
+        if attrs.get(href_key).startswith("/"):
+            return attrs
+
+        # Run the standard callbacks
+        for callback in link_callbacks:
+            attrs = callback(attrs, new)
+        return attrs
+
+    return bleach.linkify(
+        text,
+        callbacks=[link_attributes],
+        parse_email=False,
+        skip_tags=["code"],
+    )
+
+
 def validate_handle(handle):
     """
     Very basic handle validation as per

setup.py

@@ -28,6 +28,7 @@ setup(
     license="BSD 3-clause",
     install_requires=[
         "attrs",
+        "bleach>3.0",
         "commonmark",
         "cryptography",
         "cssselect>=0.9.2",