Convert gcl.GetSVNFileInfo() to xml format.

gcl.py

 #!/usr/bin/python
 # Copyright (c) 2006-2009 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.
 #
 # Wrapper script around Rietveld's upload.py that groups files into
 # changelists.
 
 import getpass
 import os
 import random
 import re
 import string
 import subprocess
 import sys
 import tempfile
 import upload
 import urllib2
+import xml.dom.minidom
 
 
 __version__ = '1.0'
 
 
 CODEREVIEW_SETTINGS = {
   # Default values.
   "CODE_REVIEW_SERVER": "codereview.chromium.org",
   "CC_LIST": "chromium-reviews@googlegroups.com",
   "VIEW_VC": "http://src.chromium.org/viewvc/chrome?view=rev&revision=",
@@ skipping 11 matching lines @@
 # Filename where we store repository specific information for gcl.
 CODEREVIEW_SETTINGS_FILE = "codereview.settings"
 
 # Warning message when the change appears to be missing tests.
 MISSING_TEST_MSG = "Change contains new or modified methods, but no new tests!"
 
 # Caches whether we read the codereview.settings file yet or not.
 read_gcl_info = False
 
 
+### Simplified XML processing functions.
+
+def ParseXML(output):
+  try:
+    return xml.dom.minidom.parseString(output)
+  except xml.parsers.expat.ExpatError:
+    return None
+
+def GetNamedNodeText(node, node_name):
+  child_nodes = node.getElementsByTagName(node_name)
+  if not child_nodes:
+    return None
+  assert len(child_nodes) == 1 and child_nodes[0].childNodes.length == 1
+  return child_nodes[0].firstChild.nodeValue
+
+
+def GetNodeNamedAttributeText(node, node_name, attribute_name):
+  child_nodes = node.getElementsByTagName(node_name)
+  if not child_nodes:
+    return None
+  assert len(child_nodes) == 1
+  return child_nodes[0].getAttribute(attribute_name)
+
+
+### SVN Functions
+
 def IsSVNMoved(filename):
   """Determine if a file has been added through svn mv"""
   info = GetSVNFileInfo(filename)
   return (info.get('Copied From URL') and
           info.get('Copied From Rev') and
           info.get('Schedule') == 'add')
 
 
 def GetSVNFileInfo(file):
   """Returns a dictionary from the svn info output for the given file."""
-  output = RunShell(["svn", "info", file])
+  dom = ParseXML(RunShell(["svn", "info", "--xml", file]))
   result = {}
-  re_key_value_pair = re.compile('^(.*)\: (.*)$')
-  for line in output.splitlines():
-    key_value_pair = re_key_value_pair.match(line)
-    if key_value_pair:
-      result[key_value_pair.group(1)] = key_value_pair.group(2)
+  if dom:
+    # /info/entry/
+    #   url
+    #   reposityory/(root|uuid)
+    #   wc-info/(schedule|depth)
+    #   commit/(author|date)
+    result['Node Kind'] = GetNodeNamedAttributeText(dom, 'entry', 'kind')
+    result['Repository Root'] = GetNamedNodeText(dom, 'root')
+    result['Schedule'] = GetNamedNodeText(dom, 'schedule')
+    result['URL'] = GetNamedNodeText(dom, 'url')
+    result['Path'] = GetNodeNamedAttributeText(dom, 'entry', 'path')
+    result['Copied From URL'] = GetNamedNodeText(dom, 'copy-from-url')
+    result['Copied From Rev'] = GetNamedNodeText(dom, 'copy-from-rev')
   return result
 
 
 def GetSVNFileProperty(file, property_name):
   """Returns the value of an SVN property for the given file.
 
   Args:
     file: The file to check
     property_name: The name of the SVN property, e.g. "svn:mime-type"
 
   Returns:
     The value of the property, which will be the empty string if the property
     is not set on the file.  If the file is not under version control, the
     empty string is also returned.
   """
   output = RunShell(["svn", "propget", property_name, file])
   if (output.startswith("svn: ") and
       output.endswith("is not under version control")):
     return ""
   else:
     return output
 
 
+def GetSVNStatus(file):
+  """Returns the svn 1.5 svn status emulated output.
+  
+  @file can be a string (one file) or a list of files."""
+  command = ["svn", "status", "--xml"]
+  if file is None:
+    pass
+  elif isinstance(file, basestring):
+    command.append(file)
+  else:
+    command.extend(file)
+
+  status_letter = {
+    '': ' ',
+    'unversioned': '?',
+    'modified': 'M',
+    'added': 'A',
+    'conflicted': 'C',
+    'deleted': 'D',
+    'ignored': 'I',
+    'replaced': 'R',
+    # TODO(maruel): Find the corresponding strings for X, !, ~
+  }
+  dom = ParseXML(RunShell(command))
+  results = []
+  if dom:
+    # /status/target/entry/(wc-status|commit|author|date)
+    for target in dom.getElementsByTagName('target'):
+      base_path = target.getAttribute('path')
+      for entry in target.getElementsByTagName('entry'):
+        file = entry.getAttribute('path')
+        wc_status = entry.getElementsByTagName('wc-status')
+        assert len(wc_status) == 1
+        # Emulate svn 1.5 status ouput...
+        statuses = [' ' for i in range(7)]
+        # Col 0
+        xml_item_status = wc_status[0].getAttribute('item')
+        if xml_item_status in status_letter:
+          statuses[0] = status_letter[xml_item_status]
+        else:
+          raise Exception('Unknown item status "%s"; please implement me!' %
+                          xml_item_status)
+        # Col 1
+        xml_props_status = wc_status[0].getAttribute('props')
+        if xml_props_status == 'modified':
+          statuses[1] = 'M'
+        elif xml_props_status == 'conflicted':
+          statuses[1] = 'C'
+        elif (not xml_props_status or xml_props_status == 'none' or
+              xml_props_status == 'normal'):
+          pass
+        else:
+          raise Exception('Unknown props status "%s"; please implement me!' %
+                          xml_props_status)
+        # Col 3
+        if wc_status[0].getAttribute('copied') == 'true':
+          statuses[3] = '+'
+        item = (''.join(statuses), file)
+        results.append(item)
+  return results
+
+
+def UnknownFiles(extra_args):
+  """Runs svn status and prints unknown files.
+
+  Any args in |extra_args| are passed to the tool to support giving alternate
+  code locations.
+  """
+  return [item[1] for item in GetSVNStatus(extra_args) if item[0][0] == '?']
+
+
 def GetRepositoryRoot():
   """Returns the top level directory of the current repository.
 
   The directory is returned as an absolute path.
   """
   global repository_root
   if not repository_root:
     cur_dir_repo_root = GetSVNFileInfo(os.getcwd()).get("Repository Root")
     if not cur_dir_repo_root:
       raise Exception("gcl run outside of repository")
@@ skipping 311 matching lines @@
   files = []
   for line in split_data[1].splitlines():
     status = line[:7]
     file = line[7:]
     files.append((status, file))
   description = split_data[2]
   save = False
   if update_status:
     for file in files:
       filename = os.path.join(GetRepositoryRoot(), file[1])
-      status = RunShell(["svn", "status", filename])[:7]
-      if not status:  # File has been reverted.
+      status_result = GetSVNStatus(filename)
+      if not status_result or not status_result[0][0]:
+        # File has been reverted.
         save = True
         files.remove(file)
       elif status != file[0]:
         save = True
         files[files.index(file)] = (status, file[1])
   change_info = ChangeInfo(changename, issue, description, files)
   if save:
     change_info.Save()
   return change_info
 
@@ skipping 34 matching lines @@
   dir_prefix = os.getcwd()[len(GetRepositoryRoot()):].strip(os.sep)
 
   # Get a list of all files in changelists.
   files_in_cl = {}
   for cl in GetCLs():
     change_info = LoadChangelistInfo(cl)
     for status, filename in change_info.files:
       files_in_cl[filename] = change_info.name
 
   # Get all the modified files.
-  status = RunShell(["svn", "status"])
-  for line in status.splitlines():
-    if not len(line) or line[0] == "?":
+  status_result = GetSVNStatus(None)
+  for line in status_result:
+    status = line[0]
+    filename = line[1]
+    if status[0] == "?":
       continue
-    status = line[:7]
-    filename = line[7:].strip()
     if dir_prefix:
       filename = os.path.join(dir_prefix, filename)
     change_list_name = ""
     if filename in files_in_cl:
       change_list_name = files_in_cl[filename]
     files.setdefault(change_list_name, []).append((status, filename))
 
   return files
 
 
@@ skipping 30 matching lines @@
       ErrorExit("Error accessing url %s" % request_path)
     else:
       return None
 
 
 def GetIssueDescription(issue):
   """Returns the issue description from Rietveld."""
   return SendToRietveld("/" + issue + "/description")
 
 
-def UnknownFiles(extra_args):
-  """Runs svn status and prints unknown files.
-
-  Any args in |extra_args| are passed to the tool to support giving alternate
-  code locations.
-  """
-  args = ["svn", "status"]
-  args += extra_args
-  p = subprocess.Popen(args, stdout = subprocess.PIPE,
-                       stderr = subprocess.STDOUT, shell = use_shell)
-  while 1:
-    line = p.stdout.readline()
-    if not line:
-      break
-    if line[0] != '?':
-      continue  # Not an unknown file to svn.
-    # The lines look like this:
-    # "?      foo.txt"
-    # and we want just "foo.txt"
-    print line[7:].strip()
-  p.wait()
-  p.stdout.close()
-
-
 def Opened():
   """Prints a list of modified files in the current directory down."""
   files = GetModifiedFiles()
   cl_keys = files.keys()
   cl_keys.sort()
   for cl_name in cl_keys:
     if cl_name:
       note = ""
       if len(LoadChangelistInfo(cl_name).files) != len(files[cl_name]):
         note = " (Note: this changelist contains files outside this directory)"
@@ skipping 530 matching lines @@
     # the files. This allows commands such as 'gcl diff xxx' to work.
     args =["svn", command]
     root = GetRepositoryRoot()
     args.extend([os.path.join(root, x) for x in change_info.FileList()])
     RunShell(args, True)
   return 0
 
 
 if __name__ == "__main__":
   sys.exit(main())