Check QUrl objects for validity

doc/HACKING.asciidoc

@@ -276,7 +276,7 @@ following guidelines:
 * Convert a string to a QUrl object as early as possible, i.e. directly after
 the user did enter it.
     - Use `utils.url.fuzzy_url` if the URL is entered by the user somewhere.
-    - Be sure you handle `utils.url.SearchEngineError` and display an error
+    - Be sure you handle `utils.url.FuzzyError` and display an error
       message to the user.
 * Convert a `QUrl` object to a string as late as possible, e.g. before
 displaying it to the user.
@@ -288,6 +288,7 @@ displaying it to the user.
 * Name a string URL something like `urlstr`, and a `QUrl` something like `url`.
 * Mention in the docstring whether your function needs a URL string or a
 `QUrl`.
+* Call QUrl.isValid() and take appropriate action if not.
 
 
 Style conventions

qutebrowser/app.py

@@ -294,7 +294,7 @@ class Application(QApplication):
                 self._opened_urls.append(cmd)
                 try:
                     url = urlutils.fuzzy_url(cmd)
-                except urlutils.SearchEngineError as e:
+                except urlutils.FuzzyUrlError as e:
                     message.error("Error in startup argument '{}': {}".format(
                         cmd, e))
                 else:
@@ -305,7 +305,7 @@ class Application(QApplication):
             for urlstr in self.config.get('general', 'startpage'):
                 try:
                     url = urlutils.fuzzy_url(urlstr)
-                except urlutils.SearchEngineError as e:
+                except urlutils.FuzzyUrlError as e:
                     message.error("Error when opening startpage: {}".format(e))
                 else:
                     self.mainwindow.tabs.tabopen(url)

qutebrowser/browser/commands.py

@@ -162,7 +162,7 @@ class CommandDispatcher:
         tab = self._tabs.cntwidget(count)
         try:
             url = urlutils.fuzzy_url(urlstr)
-        except urlutils.SearchEngineError as e:
+        except urlutils.FuzzyUrlError as e:
             raise CommandError(e)
         if tab is None:
             if count is None:
@@ -434,7 +434,7 @@ class CommandDispatcher:
         """Open a new tab with a given url."""
         try:
             url = urlutils.fuzzy_url(urlstr)
-        except urlutils.SearchEngineError as e:
+        except urlutils.FuzzyUrlError as e:
             raise CommandError(e)
         self._tabs.tabopen(url, background=False)
 
@@ -443,7 +443,7 @@ class CommandDispatcher:
         """Open a new tab in background."""
         try:
             url = urlutils.fuzzy_url(urlstr)
-        except urlutils.SearchEngineError as e:
+        except urlutils.FuzzyUrlError as e:
             raise CommandError(e)
         self._tabs.tabopen(url, background=True)
 
@@ -518,7 +518,7 @@ class CommandDispatcher:
         log.misc.debug("Clipboard contained: '{}'".format(text))
         try:
             url = urlutils.fuzzy_url(text)
-        except urlutils.SearchEngineError as e:
+        except urlutils.FuzzyUrlError as e:
             raise CommandError(e)
         if tab:
             self._tabs.tabopen(url)
@@ -640,7 +640,10 @@ class CommandDispatcher:
     def quickmark_load(self, name):
         """Load a quickmark."""
         urlstr = quickmarks.get(name)
-        self._tabs.currentWidget().openurl(QUrl(urlstr))
+        url = QUrl(urlstr)
+        if not url.isValid():
+            raise CommandError("Invalid URL {}".format(urlstr))
+        self._tabs.currentWidget().openurl(url)
 
     @cmdutils.register(instance='mainwindow.tabs.cmd')
     def quickmark_load_tab(self, name):

qutebrowser/browser/hints.py

@@ -400,7 +400,7 @@ class HintManager(QObject):
             raise CommandError("No {} links found!".format(
                 "prev" if prev else "forward"))
         url = self._resolve_url(elem, baseurl)
-        if url is None:
+        if url is None or not url.isValid():
             raise CommandError("No {} links found!".format(
                 "prev" if prev else "forward"))
         self.openurl.emit(url, newtab)

qutebrowser/browser/quickmarks.py

@@ -97,4 +97,8 @@ def get(name):
     if name not in marks:
         raise CommandError("Quickmark '{}' does not exist!".format(name))
     urlstr = marks[name]
-    return QUrl(urlstr)
+    url = QUrl(urlstr)
+    if not url.isValid():
+        raise CommandError("Invalid URL for quickmark {}: {}".format(
+            name, urlstr))
+    return url

qutebrowser/test/utils/test_url.py

@@ -129,7 +129,7 @@ class SearchUrlTests(TestCase):
 
     def test_engine_wrong(self):
         """Test with wrong search engine."""
-        with self.assertRaises(urlutils.SearchEngineError):
+        with self.assertRaises(urlutils.FuzzyUrlError):
             _ = urlutils._get_search_url('!blub testfoo')
 
     def tearDown(self):

qutebrowser/utils/url.py

@@ -43,7 +43,7 @@ def _get_search_url(txt):
         The search URL as a QUrl.
 
     Raise:
-        SearchEngineError if there is no template or no search term was found.
+        FuzzyUrlError if there is no template or no search term was found.
     """
     logger.debug("Finding search engine for '{}'".format(txt))
     r = re.compile(r'(^|\s+)!(\w+)($|\s+)')
@@ -53,7 +53,7 @@ def _get_search_url(txt):
         try:
             template = config.get('searchengines', engine)
         except config.NoOptionError:
-            raise SearchEngineError("Search engine {} not found!".format(
+            raise FuzzyUrlError("Search engine {} not found!".format(
                 engine))
         term = r.sub('', txt)
         logger.debug("engine {}, term '{}'".format(engine, term))
@@ -62,7 +62,7 @@ def _get_search_url(txt):
         term = txt
         logger.debug("engine: default, term '{}'".format(txt))
     if not term:
-        raise SearchEngineError("No search term given")
+        raise FuzzyUrlError("No search term given")
     return QUrl.fromUserInput(template.format(urllib.parse.quote(term)))
 
 
@@ -79,7 +79,9 @@ def _is_url_naive(urlstr):
     url = QUrl.fromUserInput(urlstr)
     # We don't use url here because fromUserInput appends http://
     # automatically.
-    if QUrl(urlstr).scheme() in schemes:
+    if not url.isValid():
+        return False
+    elif QUrl(urlstr).scheme() in schemes:
         return True
     elif '.' in url.host():
         return True
@@ -98,6 +100,8 @@ def _is_url_dns(url):
     Return:
         True if the URL really is a URL, False otherwise.
     """
+    if not url.isValid():
+        return False
     host = url.host()
     logger.debug("DNS request for {}".format(host))
     if not host:
@@ -136,6 +140,8 @@ def is_special_url(url):
     Args:
         url: The URL as QUrl.
     """
+    if not url.isValid():
+        return False
     special_schemes = ('about', 'qute', 'file')
     return url.scheme() in special_schemes
 
@@ -184,8 +190,8 @@ def is_url(urlstr):
         raise ValueError("Invalid autosearch value")
 
 
-class SearchEngineError(Exception):
+class FuzzyUrlError(Exception):
 
-    """Exception raised when a search engine wasn't found."""
+    """Exception raised by fuzzy_url on problems."""
 
     pass