Change actions.txt to actions.xml

tools/metrics/actions/extract_actions.py

 #!/usr/bin/env python
 #
 # Copyright (c) 2012 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 """Extract UserMetrics "actions" strings from the Chrome source.
 
 This program generates the list of known actions we expect to see in the
 user behavior logs.  It walks the Chrome source, looking for calls to
 UserMetrics functions, extracting actions and warning on improper calls,
 as well as generating the lists of possible actions in situations where
 there are many possible actions.
 
 See also:
   base/metrics/user_metrics.h
   http://wiki.corp.google.com/twiki/bin/view/Main/ChromeUserExperienceMetrics
 
-If run with a "--hash" argument, chromeactions.txt will be updated.
+After extracting all actions, the content will go through a pretty print
+function to make sure it's well formatted. If the file content needs to be
+changed, a window will be prompted asking for user's consent. The old version
+will also be saved in a backup file.
 """
 
 __author__ = 'evanm (Evan Martin)'
 
-import hashlib
 from HTMLParser import HTMLParser
+import logging
 import os
 import re
+import shutil
 import sys
+from xml.dom import minidom
+
+import print_style
 
 sys.path.insert(1, os.path.join(sys.path[0], '..', '..', 'python'))
 from google import path_utils
 
+# Import the metrics/common module for pretty print xml.
+sys.path.append(os.path.join(os.path.dirname(__file__), '..', 'common'))
+import diff_util
+import pretty_print_xml
+
 # Files that are known to use content::RecordComputedAction(), which means
 # they require special handling code in this script.
 # To add a new file, add it to this list and add the appropriate logic to
 # generate the known actions to AddComputedActions() below.
 KNOWN_COMPUTED_USERS = (
   'back_forward_menu_model.cc',
   'options_page_view.cc',
   'render_view_host.cc',  # called using webkit identifiers
   'user_metrics.cc',  # method definition
   'new_tab_ui.cc',  # most visited clicks 1-9
@@ skipping 65 matching lines @@
   'xkb:sk::slo', 'xkb:tr::tur', 'xkb:ua::ukr', 'xkb:us::eng',
   'xkb:us:altgr-intl:eng', 'xkb:us:colemak:eng', 'xkb:us:dvorak:eng',
   'xkb:us:intl:eng',
 )
 
 # The path to the root of the repository.
 REPOSITORY_ROOT = os.path.join(path_utils.ScriptDir(), '..', '..', '..')
 
 number_of_files_total = 0
 
+# Tags that need to be inserted to each 'action' tag and their default content.
+TAGS = {'description': 'Please enter the description of the metric.',
+        'owner': ('Please list the metric\'s owners. Add more owner tags as '
+                  'needed.')}
+
 
 def AddComputedActions(actions):
   """Add computed actions to the actions list.
 
   Arguments:
     actions: set of actions to add to.
   """
 
   # Actions for back_forward_menu_model.cc.
   for dir in ('BackMenu_', 'ForwardMenu_'):
@@ skipping 444 matching lines @@
   actions.add('AutomaticReset_WebUIBanner_BannerShown')
   actions.add('AutomaticReset_WebUIBanner_ManuallyClosed')
   actions.add('AutomaticReset_WebUIBanner_ResetClicked')
 
   # These actions relate to the the automatic settings reset banner shown as
   # a result of settings hardening.
   actions.add('AutomaticSettingsReset_WebUIBanner_BannerShown')
   actions.add('AutomaticSettingsReset_WebUIBanner_ManuallyClosed')
   actions.add('AutomaticSettingsReset_WebUIBanner_LearnMoreClicked')
 
+
+class Error(Exception):
+  pass
+
+
+def _ExtractText(parent_dom, tag_name):
+  """Extract the text enclosed by |tag_name| under |parent_dom|
+
+  Args:
+    parent_dom: The parent Element under which text node is searched for.
+    tag_name: The name of the tag which contains a text node.
+
+  Returns:
+    A (list of) string enclosed by |tag_name| under |parent_dom|.
+  """
+  texts = []
+  for child_dom in parent_dom.getElementsByTagName(tag_name):
+    text_dom = child_dom.childNodes
+    if text_dom.length != 1:
+      raise Error('More than 1 child node exists under %s' % tag_name)
+    if text_dom[0].nodeType != minidom.Node.TEXT_NODE:
+      raise Error('%s\'s child node is not a text node.' % tag_name)
+    texts.append(text_dom[0].data)
+  return texts
+
+
+class Action(object):
+  def __init__(self, name, description, owners, obsolete=None):
+    self.name = name
+    self.description = description
+    self.owners = owners
+    self.obsolete = obsolete
+
+
+def ParseActionFile(file_path):
+  """Parse the XML data currently stored in the file.
+
+  Args:
+    file_path: the path to the action XML file.
+
+  Returns:
+    (actions, actions_dict) actions is a set with all user actions' names.
+    actions_dict is a dict from user action name to Action object.
+  """
+  actions = set()
+  actions_dict = {}
+
+  dom = minidom.parse(file_path)
+  for action_dom in dom.getElementsByTagName('action'):
+    action_name = action_dom.getAttribute('name')
+    actions.add(action_name)
+
+    owners = _ExtractText(action_dom, 'owner')
+    # There is only one description for each user action. Get the first element
+    # of the returned list.
+    description_list = _ExtractText(action_dom, 'description')
+    if len(description_list) > 1:
+      logging.error('user actions "%s" has more than one descriptions. Exactly '
+                    'one description is needed for each user action. Please '
+                    'fix.', action_name)
+      sys.exit(1)
+    description = description_list[0] if description_list else None
+    # There is at most one obsolete tag for each user action.
+    obsolete_list = _ExtractText(action_dom, 'obsolete')
+    if len(obsolete_list) > 1:
+      logging.error('user actions "%s" has more than one obsolete tag. At most '
+                    'one obsolete tag can be added for each user action. Please'
+                    ' fix.', action_name)
+      sys.exit(1)
+    obsolete = obsolete_list[0] if obsolete_list else None
+    actions_dict[action_name] = Action(action_name, description, owners,
+                                       obsolete)
+  return actions, actions_dict
+
+
+def _CreateActionTag(top_dom, action_name, action_object):
+  """Create a new action tag.
+
+  Format of an action tag:
+  <action name="name">
+    <owner>Owner</owner>
+    <description>Description.</description>
+    <obsolete>Deprecated.</obsolete>
+  </action>
+
+  <obsolete> is an optional tag. It's added to user actions that are no longer
+  used any more.
+
+  If action_name is in actions_dict, the values to be inserted are based on the
+  corresponding Action object. If action_name is not in actions_dict, the
+  default value from TAGS is used.
+
+  Args:
+    top_dom: The parent node under which the new action tag is created.
+    action_name: The name of an action.
+    action_object: An action object representing the data to be inserted.
+
+  Returns:
+    An action tag Element with proper children elements.
+  """
+  action_dom = top_dom.createElement('action')
+  action_dom.setAttribute('name', action_name)
+
+  # Create owner tag.
+  if action_object and action_object.owners:
+    # If owners for this action is not None, use the stored value. Otherwise,
+    # use the default value.
+    for owner in action_object.owners:
+      owner_dom = top_dom.createElement('owner')
+      owner_dom.appendChild(top_dom.createTextNode(owner))
+      action_dom.appendChild(owner_dom)
+  else:
+    # Use default value.
+    owner_dom = top_dom.createElement('owner')
+    owner_dom.appendChild(top_dom.createTextNode(TAGS.get('owner', '')))
+    action_dom.appendChild(owner_dom)
+
+  # Create description tag.
+  description_dom = top_dom.createElement('description')
+  action_dom.appendChild(description_dom)
+  if action_object and action_object.description:
+    # If description for this action is not None, use the store value.
+    # Otherwise, use the default value.
+    description_dom.appendChild(top_dom.createTextNode(
+        action_object.description))
+  else:
+    description_dom.appendChild(top_dom.createTextNode(
+        TAGS.get('description', '')))
+
+  # Create obsolete tag.
+  if action_object and action_object.obsolete:
+    obsolete_dom = top_dom.createElement('obsolete')
+    action_dom.appendChild(obsolete_dom)
+    obsolete_dom.appendChild(top_dom.createTextNode(
+        action_object.obsolete))
+
+  return action_dom
+
+
+def PrettyPrint(actions, actions_dict):
+  """Given a list of action data, create a well-printed minidom document.
+
+  Args:
+    actions: A list of action names.
+    actions_dict: A mappting from action name to Action object.
+
+  Returns:
+    A well-printed minidom document that represents the input action data.
+  """
+  # Form new minidom nodes based on updated |actions|.
+  new_dom = minidom.getDOMImplementation().createDocument(None, 'actions', None)
+  top_element = new_dom.documentElement
+
+  # Print out the actions as a sorted list.
+  for action in sorted(actions):
+    top_element.appendChild(_CreateActionTag(new_dom, action,
+                                             actions_dict.get(action, None)))
+
+  return print_style.GetPrintStyle().PrettyPrintNode(new_dom)
+
+
 def main(argv):
-  if '--hash' in argv:
-    hash_output = True
-  else:
-    hash_output = False
-    print >>sys.stderr, "WARNING: If you added new UMA tags, you must" + \
-           " use the --hash option to update chromeactions.txt."
-  # if we do a hash output, we want to only append NEW actions, and we know
-  # the file we want to work on
-  actions = set()
+  presubmit = ('--presubmit' in argv)
+  actions_xml_path = os.path.join(path_utils.ScriptDir(), 'actions.xml')
 
-  chromeactions_path = os.path.join(path_utils.ScriptDir(), "chromeactions.txt")
+  # Save the original file content.
+  with open(actions_xml_path, 'rb') as f:
+    original_xml = f.read()
 
-  if hash_output:
-    f = open(chromeactions_path)
-    for line in f:
-      part = line.rpartition("\t")
-      part = part[2].strip()
-      actions.add(part)
-    f.close()
-
+  actions, actions_dict = ParseActionFile(actions_xml_path)
 
   AddComputedActions(actions)
   # TODO(fmantek): bring back webkit editor actions.
   # AddWebKitEditorActions(actions)
   AddAboutFlagsActions(actions)
   AddWebUIActions(actions)
 
   AddLiteralActions(actions)
 
   # print "Scanned {0} number of files".format(number_of_files_total)
   # print "Found {0} entries".format(len(actions))
 
   AddAndroidActions(actions)
   AddAutomaticResetBannerActions(actions)
   AddBookmarkManagerActions(actions)
   AddChromeOSActions(actions)
   AddClosedSourceActions(actions)
   AddExtensionActions(actions)
   AddHistoryPageActions(actions)
   AddKeySystemSupportActions(actions)
 
-  if hash_output:
-    f = open(chromeactions_path, "wb")
+  pretty = PrettyPrint(actions, actions_dict)
+  if original_xml == pretty:
+    print 'actions.xml is correctly pretty-printed.'
+    sys.exit(0)
+  if presubmit:
+    logging.info('actions.xml is not formatted correctly; run '
+                 'extract_actions.py to fix.')
+    sys.exit(1)
 
+  # Prompt user to consent on the change.
+  if not diff_util.PromptUserToAcceptDiff(
+      original_xml, pretty, 'Is the new version acceptable?'):
+    logging.error('Aborting')
+    sys.exit(1)
 
-  # Print out the actions as a sorted list.
-  for action in sorted(actions):
-    if hash_output:
-      hash = hashlib.md5()
-      hash.update(action)
-      print >>f, '0x%s\t%s' % (hash.hexdigest()[:16], action)
-    else:
-      print action
+  print 'Creating backup file: actions.old.xml.'
+  shutil.move(actions_xml_path, 'actions.old.xml')
 
-  if hash_output:
-    print "Done. Do not forget to add chromeactions.txt to your changelist"
+  with open(actions_xml_path, 'wb') as f:
+    f.write(pretty)
+  print ('Updated %s. Don\'t forget to add it to your changelist' %
+         actions_xml_path)
   return 0
 
 
 if '__main__' == __name__:
   sys.exit(main(sys.argv))