Store QUTE_TEXT/QUTE_HTML in files for userscripts.

Fixes #644.

CHANGELOG.asciidoc

@@ -14,6 +14,14 @@ This project adheres to http://semver.org/[Semantic Versioning].
 // `Fixed` for any bug fixes.
 // `Security` to invite users to upgrade in case of vulnerabilities.
 
+v0.3.0 (unreleased)
+-------------------
+
+Changed
+~~~~~~~
+
+- `QUTE_HTML` and `QUTE_TEXT` for userscripts now don't store the contents directly, and instead contain a filename.
+
 https://github.com/The-Compiler/qutebrowser/releases/tag/v0.2.1[v0.2.1]
 -----------------------------------------------------------------------
 

doc/userscripts.asciidoc

@@ -24,8 +24,8 @@ The following environment variables will be set when an userscript is launched:
   command or key binding).
 - `QUTE_USER_AGENT`: The currently set user agent.
 - `QUTE_FIFO`: The FIFO or file to write commands to.
-- `QUTE_HTML`: The HTML source of the current page.
-- `QUTE_TEXT`: The plaintext of the current page.
+- `QUTE_HTML`: Path of a file containing the HTML source of the current page.
+- `QUTE_TEXT`: Path of a file containing the plaintext of the current page.
 
 In `command` mode:
 

qutebrowser/browser/commands.py

@@ -876,14 +876,13 @@ class CommandDispatcher:
             env['QUTE_TITLE'] = tabbed_browser.page_title(idx)
 
         webview = tabbed_browser.currentWidget()
-        if webview is not None:
+        if webview is None:
+            mainframe = None
+        else:
             if webview.hasSelection():
                 env['QUTE_SELECTED_TEXT'] = webview.selectedText()
                 env['QUTE_SELECTED_HTML'] = webview.selectedHtml()
             mainframe = webview.page().mainFrame()
-            if mainframe is not None:
-                env['QUTE_HTML'] = mainframe.toHtml()
-                env['QUTE_TEXT'] = mainframe.toPlainText()
 
         try:
             url = tabbed_browser.current_url()
@@ -892,6 +891,7 @@ class CommandDispatcher:
         else:
             env['QUTE_URL'] = url.toString(QUrl.FullyEncoded)
 
+        env.update(userscripts.store_source(mainframe))
         userscripts.run(cmd, *args, win_id=self._win_id, env=env)
 
     @cmdutils.register(instance='command-dispatcher', scope='window')

qutebrowser/browser/hints.py

@@ -523,12 +523,11 @@ class HintManager(QObject):
             'QUTE_MODE': 'hints',
             'QUTE_SELECTED_TEXT': str(elem),
             'QUTE_SELECTED_HTML': elem.toOuterXml(),
-            'QUTE_HTML': frame.toHtml(),
-            'QUTE_TEXT': frame.toPlainText(),
         }
         url = self._resolve_url(elem, context.baseurl)
         if url is not None:
             env['QUTE_URL'] = url.toString(QUrl.FullyEncoded)
+        env.update(userscripts.store_source(frame))
         userscripts.run(cmd, *args, win_id=self._win_id, env=env)
 
     def _spawn(self, url, context):

qutebrowser/commands/userscripts.py

@@ -101,6 +101,7 @@ class _BaseUserscriptRunner(QObject):
         self._win_id = win_id
         self._filepath = None
         self._proc = None
+        self._env = None
 
     def _run_process(self, cmd, *args, env):
         """Start the given command via QProcess.
@@ -110,6 +111,7 @@ class _BaseUserscriptRunner(QObject):
             *args: The arguments to hand to the command
             env: A dictionary of environment variables to add.
         """
+        self._env = env
         self._proc = QProcess(self)
         procenv = QProcessEnvironment.systemEnvironment()
         procenv.insert('QUTE_FIFO', self._filepath)
@@ -122,17 +124,26 @@ class _BaseUserscriptRunner(QObject):
         self._proc.start(cmd, args)
 
     def _cleanup(self):
-        """Clean up the temporary file."""
-        log.procs.debug("Deleting temporary file {}.".format(self._filepath))
-        try:
-            os.remove(self._filepath)
-        except OSError as e:
-            # NOTE: Do not replace this with "raise CommandError" as it's
-            # executed async.
-            message.error(self._win_id,
-                          "Failed to delete tempfile... ({})".format(e))
+        """Clean up temporary files."""
+        tempfiles = [self._filepath]
+        if self._env is not None:
+            if 'QUTE_HTML' in self._env:
+                tempfiles.append(self._env['QUTE_HTML'])
+            if 'QUTE_TEXT' in self._env:
+                tempfiles.append(self._env['QUTE_TEXT'])
+        for fn in tempfiles:
+            log.procs.debug("Deleting temporary file {}.".format(fn))
+            try:
+                os.remove(fn)
+            except OSError as e:
+                # NOTE: Do not replace this with "raise CommandError" as it's
+                # executed async.
+                message.error(
+                    self._win_id, "Failed to delete tempfile {} ({})!".format(
+                        fn, e))
         self._filepath = None
         self._proc = None
+        self._env = None
 
     def run(self, cmd, *args, env=None):
         """Run the userscript given.
@@ -305,6 +316,37 @@ else:
     UserscriptRunner = _DummyUserscriptRunner
 
 
+def store_source(frame):
+    """Store HTML/plaintext in files.
+
+    This writes files containing the HTML/plaintext source of the page, and
+    returns a dict with the paths as QUTE_HTML/QUTE_TEXT.
+
+    Args:
+        frame: The QWebFrame to get the info from, or None to do nothing.
+
+    Return:
+        A dictionary with the needed environment variables.
+
+    Warning:
+        The caller is responsible to delete the files after using them!
+    """
+    if frame is None:
+        return {}
+    env = {}
+    with tempfile.NamedTemporaryFile(mode='w', encoding='utf-8',
+                                     suffix='.html',
+                                     delete=False) as html_file:
+        html_file.write(frame.toHtml())
+        env['QUTE_HTML'] = html_file.name
+    with tempfile.NamedTemporaryFile(mode='w', encoding='utf-8',
+                                     suffix='.txt',
+                                     delete=False) as txt_file:
+        txt_file.write(frame.toPlainText())
+        env['QUTE_TEXT'] = txt_file.name
+    return env
+
+
 def run(cmd, *args, win_id, env):
     """Convenience method to run an userscript.