Scope HTML link text capture to each anchor

humanfriendly/terminal/html.py

@@ -204,7 +204,7 @@ def handle_data(self, data):
         if self.link_url:
             # Link text is captured literally so that we can reliably check
             # whether the text and the URL of the link are the same string.
-            self.link_text = data
+            self.link_text += data
         elif self.callback and self.preformatted_text_level == 0:
             # Text that is not part of a link and not preformatted text is
             # passed to the user defined callback to allow for arbitrary
@@ -227,7 +227,7 @@ def handle_endtag(self, tag):
                 self.stack.pop(-1)
             new_style = self.current_style
             if tag == 'a':
-                if self.urls_match(self.link_text, self.link_url):
+                if not self.link_url or self.urls_match(self.link_text, self.link_url):
                     # Don't render the URL when it's part of the link text.
                     self.emit_style(new_style)
                 else:
@@ -237,6 +237,8 @@ def handle_endtag(self, tag):
                     self.output.write(self.render_url(self.link_url))
                     self.emit_style(new_style)
                     self.output.write(')')
+                self.link_url = None
+                self.link_text = None
             else:
                 self.emit_style(new_style)
             if tag in ('code', 'pre'):
@@ -270,6 +272,7 @@ def handle_starttag(self, tag, attrs):
             # this is the most intuitive way to present a link in a plain text
             # interface).
             self.link_url = next((v for n, v in attrs if n == 'href'), '')
+            self.link_text = ''
         elif tag == 'b' or tag == 'strong':
             self.push_styles(bold=True)
         elif tag == 'br':

humanfriendly/tests.py

@@ -1078,6 +1078,24 @@ def callback(text):
                 title="VPS hosting" rel="external"&gt;True</a>
         ''')
 
+    def test_html_callback_after_link(self):
+        """Text callbacks resume after each hyperlink closes."""
+        result = html_to_ansi('before <a href="url">link</a> after', callback=lambda text: text.upper())
+        self.assertEqual(ansi_strip(result), 'BEFORE link (url) AFTER')
+
+    def test_html_link_text_in_multiple_fragments(self):
+        """Nested markup contributes all link text to URL comparison."""
+        result = html_to_ansi('<a href="https://example.com">https://<b>example</b>.com</a>')
+        self.assertEqual(ansi_strip(result), 'https://example.com')
+
+    def test_html_empty_links(self):
+        """Empty anchors do not reuse previous link text or raise errors."""
+        result = html_to_ansi('<a href="first">first</a><a href="second"></a>')
+        self.assertEqual(ansi_strip(result), 'first (second)')
+        self.assertEqual(ansi_strip(html_to_ansi('<a href="url"></a>')), ' (url)')
+        self.assertEqual(ansi_strip(html_to_ansi('<a>text</a>')), 'text')
+        self.assertEqual(ansi_strip(html_to_ansi('text</a>')), 'text')
+
     def test_generate_output(self):
         """Test the :func:`humanfriendly.terminal.output()` function."""
         text = "Standard output generated by output()"