Catch ValueError when parsing URLs, show a better error/warning

Author: oprypin

properdocs/config/config_options.py

@@ -17,7 +17,6 @@
 from urllib.parse import urlsplit, urlunsplit
 
 import markdown
-import pathspec
 import pathspec.gitignore
 
 from properdocs import plugins, theme, utils
@@ -497,8 +496,8 @@ def run_validation(self, value: object) -> str:
             return value
         try:
             parsed_url = urlsplit(value)
-        except (AttributeError, TypeError):
-            raise ValidationError("Unable to parse the URL.")
+        except Exception:
+            raise ValidationError("The URL is invalid")
 
         if parsed_url.scheme and parsed_url.netloc:
             if self.is_dir and not parsed_url.path.endswith('/'):

properdocs/structure/nav.py

@@ -164,9 +164,16 @@ def get_navigation(files: Files, config: ProperDocsConfig) -> Navigation:
 
     links = _get_by_type(items, Link)
     for link in links:
-        scheme, netloc, path, query, fragment = urlsplit(link.url)
-        if scheme or netloc:
-            log.debug(f"An external link to '{link.url}' is included in the 'nav' configuration.")
+        invalid: str | None = None
+        try:
+            scheme, netloc, path, query, fragment = urlsplit(link.url)
+        except Exception as e:
+            invalid = 'invalid'
+        else:
+            if scheme or netloc:
+                invalid = 'external'
+        if invalid:
+            log.debug(f"An {invalid} link to '{link.url}' is included in the 'nav' configuration.")
         elif (
             link.url.startswith('/')
             and config.validation.nav.absolute_links

properdocs/structure/pages.py

@@ -417,7 +417,15 @@ def _possible_target_uris(
                     tried.add(guess)
 
     def path_to_url(self, url: str) -> str:
-        scheme, netloc, path, query, anchor = urlsplit(url)
+        try:
+            scheme, netloc, path, query, anchor = urlsplit(url)
+        except ValueError:  # Invalid URL, e.g. invalid IPv6.
+            log.log(
+                self.config.validation.links.unrecognized_links,
+                f"Doc file '{self.file.src_uri}' contains an invalid link '{url}', "
+                f"it was left as is.",
+            )
+            return url
 
         absolute_link = None
         warning_level, warning = 0, ''

properdocs/tests/config/config_options_tests.py

@@ -424,14 +424,24 @@ class Schema(Config):
         conf = self.get_config(Schema, {'option': None})
         self.assertEqual(conf.option, None)
 
+    @unittest.skipUnless(sys.version_info >= (3, 11), "new error kind in Python 3.11")
     def test_invalid_url(self) -> None:
         class Schema(Config):
             option = c.URL()
 
         with self.expect_error(option="Required configuration not provided."):
             self.get_config(Schema, {'option': None})
 
-        for url in "properdocs.org", "//properdocs.org/test", "http:/properdocs.org/", "/hello/":
+        for url in ["http://[a]/"]:
+            with self.subTest(url=url):
+                with self.expect_error(option="The URL is invalid"):
+                    self.get_config(Schema, {'option': url})
+
+    def test_invalid_url_no_http(self) -> None:
+        class Schema(Config):
+            option = c.URL()
+
+        for url in ["properdocs.org", "//properdocs.org/test", "http:/properdocs.org/", "/hello/"]:
             with self.subTest(url=url):
                 with self.expect_error(
                     option="The URL isn't valid, it should include the http:// (scheme)"

properdocs/tests/structure/page_tests.py

@@ -1221,6 +1221,19 @@ def test_absolute_win_local_path(self):
                 '<a href="\\image.png">absolute local path</a>',
             )
 
+    @unittest.skipUnless(sys.version_info >= (3, 11), "new error kind in Python 3.11")
+    def test_invalid_link_ipv6(self):
+        for use_directory_urls in True, False:
+            self.assertEqual(
+                self.get_rendered_result(
+                    use_directory_urls=use_directory_urls,
+                    content='[invalid link](http://[a]/)',
+                    files=['index.md'],
+                    logs="INFO:Doc file 'index.md' contains an invalid link 'http://[a]/', it was left as is.",
+                ),
+                '<a href="http://[a]">absolute local path</a>',
+            )
+
     def test_email_link(self):
         self.assertEqual(
             self.get_rendered_result(content='<mail@example.com>', files=['index.md']),

properdocs/utils/__init__.py

@@ -221,8 +221,11 @@ def _get_norm_url(path: str) -> tuple[str, int]:
             f"That will be unsupported in a future release. Please change it to '/'."
         )
         path = path.replace('\\', '/')
+    try:
+        parsed = urlsplit(path)
+    except ValueError:
+        return path, -1
     # Allow links to be fully qualified URLs
-    parsed = urlsplit(path)
     if parsed.scheme or parsed.netloc or path.startswith(('/', '#')):
         return path, -1