diff --git a/tools/README b/tools/README
new file mode 100644
index 00000000..712e4efb
--- /dev/null
+++ b/tools/README
@@ -0,0 +1,15 @@
+pretty_vcproj:
+  Usage: pretty_vcproj.py "c:\path\to\vcproj.vcproj" [key1=value1] [key2=value2]
+
+  They key/value pair are used to resolve vsprops name.
+
+  For example, if I want to diff the base.vcproj project:
+
+  pretty_vcproj.py z:\dev\src-chrome\src\base\build\base.vcproj "$(SolutionDir)=z:\dev\src-chrome\src\chrome\\" "$(CHROMIUM_BUILD)=" "$(CHROME_BUILD_TYPE)=" > orignal.txt
+  pretty_vcproj.py z:\dev\src-chrome\src\base\base_gyp.vcproj "$(SolutionDir)=z:\dev\src-chrome\src\chrome\\" "$(CHROMIUM_BUILD)=" "$(CHROME_BUILD_TYPE)=" > gyp.txt
+
+  And you can use your favorite diff tool to see the changes.
+
+  Note: In the case of base.vcproj, the original vcproj is one level up the generated one.
+        I suggest you do a search and replace for '"..\' and replace it with '"' in original.txt
+        before you perform the diff.
\ No newline at end of file
diff --git a/tools/pretty_vcproj.py b/tools/pretty_vcproj.py
new file mode 100644
index 00000000..e0e8c597
--- /dev/null
+++ b/tools/pretty_vcproj.py
@@ -0,0 +1,271 @@
+#!/usr/bin/python2.5
+# Copyright 2009 Google Inc.
+# All Rights Reserved.
+
+"""Make the format of a vcproj really pretty.
+
+   This script normalize and sort an xml. It also fetches all the properties
+   inside linked vsprops and include them explicitly in the vcproj.
+   
+   It outputs the resulting xml to stdout.
+"""
+
+__author__ = 'nsylvain (Nicolas Sylvain)'
+
+import os
+import sys
+
+from xml.dom.minidom import parse
+from xml.dom.minidom import Node
+
+REPLACEMENTS = dict()
+
+class CmpTuple:
+  """Compare function between 2 tuple."""
+  def __call__(self, x, y):
+    (key1, value1) = x
+    (key2, value2) = y
+    return cmp(key1, key2)
+
+class CmpNode:
+  """Compare function between 2 xml nodes."""
+
+  def get_string(self, node):
+    node_string = "node"
+    node_string += node.nodeName
+    if node.nodeValue:
+      node_string += node.nodeValue
+
+    if node.attributes:
+      # We first sort by name, if present.
+      node_string += node.getAttribute("Name")
+
+      all_nodes = []
+      for (name, value) in node.attributes.items():
+        all_nodes.append((name, value))
+
+      all_nodes.sort(CmpTuple())
+      for (name, value) in all_nodes:
+        node_string += name
+        node_string += value
+
+    return node_string
+
+  def __call__(self, x, y):
+    return cmp(self.get_string(x), self.get_string(y))
+
+def PrettyPrintNode(node, indent=0):
+  if node.nodeType == Node.TEXT_NODE:
+    if node.data.strip():
+      print '%s%s' % (' '*indent, node.data.strip())
+    return
+
+  if node.childNodes:
+    node.normalize()
+  # Get the number of attributes
+  attr_count = 0
+  if node.attributes:
+    attr_count = node.attributes.length
+
+  # Print the main tag
+  if attr_count == 0:
+    print '%s<%s>' % (' '*indent, node.nodeName)
+  else:
+    print '%s<%s' % (' '*indent, node.nodeName)
+
+    all_attributes = []
+    for (name, value) in node.attributes.items():
+      all_attributes.append((name, value))
+      all_attributes.sort(CmpTuple())
+    for (name, value) in all_attributes:
+      print '%s  %s="%s"' % (' '*indent, name, value)
+    print '%s>' % (' '*indent)
+  if node.nodeValue:
+    print '%s  %s' % (' '*indent, node.nodeValue)
+
+  for sub_node in node.childNodes:
+    PrettyPrintNode(sub_node, indent=indent+2)
+  print '%s</%s>' % (' '*indent, node.nodeName)
+  
+def CleanupVcproj(node):
+  # For each sub node, we call recursively this function.
+  for sub_node in node.childNodes:
+    CleanupVcproj(sub_node)
+
+  # Normalize the node, and remove all extranous whitespaces.
+  for sub_node in node.childNodes:
+    if sub_node.nodeType == Node.TEXT_NODE:
+        sub_node.data = sub_node.data.replace("\r", "")
+        sub_node.data = sub_node.data.replace("\n", "")
+        sub_node.data = sub_node.data.rstrip()
+        
+  # Fix all the semicolon separated attributes to be sorted.
+  if node.attributes:
+    for (name, value) in node.attributes.items():
+      node.setAttribute(name, ';'.join(sorted(value.split(';'))))
+      if not value:
+        node.removeAttribute(name)
+
+  if node.childNodes:
+    node.normalize()
+    
+  # For each node, take a copy, and remove it from the list.
+  node_array = []
+  while node.childNodes and node.childNodes[0]:
+    node_array.append(node.childNodes[0])
+    node.removeChild(node.childNodes[0])
+    
+  # Sort the list.
+  node_array.sort(CmpNode())
+
+  # Insert the nodes in the correct order.
+  for new_node in node_array:
+    # But don't append empty tool node.
+    if new_node.nodeName == 'Tool':
+      if new_node.attributes and new_node.attributes.length == 1:
+        # This one was empty.
+        continue
+    # We also don't add the "_excluded_files" filter.
+    if new_node.nodeName == 'Filter':
+      if (new_node.attributes and
+          new_node.getAttribute('Name') == '_excluded_files'):
+        continue
+    node.appendChild(new_node)
+
+def GetConfiguationNodes(vcproj):
+  #TODO(nsylvain): Find a better way to navigate the xml.
+  nodes = []
+  for node in vcproj.childNodes:
+    if node.nodeName == "Configurations":
+      for sub_node in node.childNodes:
+        if sub_node.nodeName == "Configuration":
+          nodes.append(sub_node)
+
+  return nodes
+
+def FixFilenames(filenames, current_directory):
+  replacements = REPLACEMENTS
+  #{
+  #                 '$(SolutionDir)': 'z:\\dev\\src-chrome\\src\\chrome\\',
+  #                 '$(CHROMIUM_BUILD)' :'',
+  #                 '$(CHROME_BUILD_TYPE)' :'',
+  #               }
+
+  new_list = []
+  for filename in filenames:
+    if filename:
+      for key in replacements:
+        filename = filename.replace(key, replacements[key])
+      os.chdir(current_directory)
+      new_list.append(os.path.abspath(filename))
+  return new_list
+
+def GetChildrenVsprops(filename):
+  dom = parse(filename)
+  if dom.documentElement.attributes:
+    vsprops = dom.documentElement.getAttribute('InheritedPropertySheets')
+    return FixFilenames(vsprops.split(';'), os.path.dirname(filename))
+  return []
+
+def SeekToNode(node1, child2):
+  # A text node does not have properties.
+  if child2.nodeType == Node.TEXT_NODE:
+    return None
+
+  # Get the name of the current node.
+  current_name = child2.getAttribute("Name")
+  if not current_name:
+    # There is no name. We don't know how to merge.
+    return None
+
+  # Look through all the nodes to find a match.
+  for sub_node in node1.childNodes:
+    if sub_node.nodeName == child2.nodeName:
+      name = sub_node.getAttribute("Name")
+      if name == current_name:
+        return sub_node
+
+  # No match. We give up.
+  return None
+
+def MergeAttributes(node1, node2):
+  # No attributes to merge?
+  if not node2.attributes:
+    return
+
+  for (name, value2) in node2.attributes.items():
+    # Don't merge the 'Name' attribute.
+    if name == 'Name':
+      continue
+    value1 = node1.getAttribute(name)
+    if value1:
+      # The attribute exist in the main node. If it's equal, we leave it
+      # untouched, otherwise we concatenate it.
+      if value1 != value2:
+        node1.setAttribute(name, ';'.join([value1, value2]))
+    else:
+      # The attribute does nto exist in the main node. We append this one.
+      node1.setAttribute(name, value2)
+
+    # If the attribute was a property sheet attributes, we remove it, since
+    # they are useless.
+    if name == 'InheritedPropertySheets':
+      node1.removeAttribute(name)
+
+def MergeProperties(node1, node2):
+  MergeAttributes(node1, node2)
+  for child2 in node2.childNodes:
+    child1 = SeekToNode(node1, child2)
+    if child1:
+      MergeProperties(child1, child2)
+    else:
+      node1.appendChild(child2.cloneNode(True))
+
+def main():
+  """Main function of this vcproj prettifier."""
+
+  # check if we have exactly 1 parameter.
+  if len(sys.argv) < 2:
+    print ('Usage: %s "c:\\path\\to\\vcproj.vcproj" [key1=value1] '
+           '[key2=value2]' % sys.argv[0])
+    return
+
+  # Parse the keys
+  for i in range(2, len(sys.argv)):
+    print sys.argv[i].split('=')
+    (key, value) = sys.argv[i].split('=')
+    REPLACEMENTS[key] = value
+
+  # Open the vcproj and parse the xml.
+  dom = parse(sys.argv[1])
+
+  # First thing we need to do is find the Configuration Node and merge them
+  # with the vsprops they include.
+  for configuration_node in GetConfiguationNodes(dom.documentElement):
+    # Get the property sheets associated with this configuration.
+    vsprops = configuration_node.getAttribute('InheritedPropertySheets')
+
+    # Fix the filenames to be absolute.
+    vsprops_list = FixFilenames(vsprops.strip().split(';'),
+                                os.path.dirname(sys.argv[1]))
+
+    # Extend the list of vsprops with all vsprops contained in the current
+    # vsprops.
+    for current_vsprops in vsprops_list:
+     vsprops_list.extend(GetChildrenVsprops(current_vsprops))
+
+    # Now that we have all the vsprops, we need to merge them.
+    for current_vsprops in vsprops_list:
+      MergeProperties(configuration_node,
+                      parse(current_vsprops).documentElement)
+
+  # Now that everything is merged, we need to cleanup the xml.
+  CleanupVcproj(dom.documentElement)
+
+  # Finally, we use the prett xml function to print the vcproj back to the
+  # user.
+  #print dom.toprettyxml(newl="\n")
+  PrettyPrintNode(dom.documentElement)
+  
+if __name__ == '__main__':
+  main()
\ No newline at end of file