Auto-updating script for Windows toolchain

tools/win/toolchain/get_toolchain_if_necessary.py

Index: tools/win/toolchain/get_toolchain_if_necessary.py
diff --git a/tools/win/toolchain/get_toolchain_if_necessary.py b/tools/win/toolchain/get_toolchain_if_necessary.py
new file mode 100644
index 0000000000000000000000000000000000000000..8d9e25fec83c6f9dad2d3e2595333b1710d1d9c7
--- /dev/null
+++ b/tools/win/toolchain/get_toolchain_if_necessary.py
@@ -0,0 +1,69 @@
+# Copyright 2013 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.
+
+import hashlib
+import os
+import subprocess
+import sys
+
+
+def RunOrDie(command):
+  rc = subprocess.call(command, shell=True)

[M-A Ruel, 2013/11/26 02:07:43]

that's called subprocess.check_call()

[scottmg, 2013/11/26 05:07:19]

Done.

+  if rc != 0:
+    raise SystemExit('%s failed.' % command)
+
+
+def CalculateHash(root):
+  """Calculates the sha1 of the paths to all files in the given |root| and the
+  contents of those files, and returns as a hex string."""
+  digest = hashlib.sha1()
+  count = 0
+  for root, dirs, files in os.walk(root):
+    for name in files:

[M-A Ruel, 2013/11/26 02:07:43]

I'm not sure the files are listed in deterministic orders or if the directories
are traversed in deterministic order.

You can fix this with:
dirs.sort()
for name in sorted(files):
  ...

Then there's issue with path casing. You could .lower() everything.

And you are including 'root' in sha-1 calculation. You should base it at the
relative directory instead.

Finally, the code will generate different hash if the input is given '/' or
'\\'.

[scottmg, 2013/11/26 05:07:19]

On 2013/11/26 02:07:43, M-A Ruel wrote:

Good points, thanks.
Done.
 
Done, and before sorted() too, in case the file case is inconsistent on disk.
The root is the path that's passed to os.walk, so
'src/third_party/win_toolchain/...' I think that's right in case we want to move
it.
Done.

+      path = os.path.join(root, name)
+      with open(path, 'rb') as f:
+        digest.update(path)
+        digest.update(f.read())

[M-A Ruel, 2013/11/26 02:07:43]

Make this a function that reads chunks. It is necessary to support large files
anyway, then you can reuse the function for line 42.

[scottmg, 2013/11/26 05:07:19]

The largest file is 42M, so it seems unnecessary.

+  return digest.hexdigest()
+
+
+def main():
+  if sys.platform not in ('win32', 'cygwin'):
+    return 0
+

[M-A Ruel, 2013/11/26 02:07:43]

if len(sys.argv) != 1:
  print >> sys.stderr, 'Unsupported arguments'
  return 1

[scottmg, 2013/11/26 05:07:19]

Done.

+  # cwd is assumed to be where .gclient is.
+  toolchain_dir = 'src/third_party/win_toolchain'

[M-A Ruel, 2013/11/26 02:07:43]

os.path.join()

[scottmg, 2013/11/26 05:07:19]

I'm not sure what you mean. This script is a DEPS helper, so it only makes sense
to run with cwd as the location .gclient.

+  target_dir = os.path.join(toolchain_dir, 'files')
+
+  sha1path = os.path.join(toolchain_dir, 'toolchain.sha1')
+  desired_hash = ''
+  if os.path.isfile(sha1path):
+    with open(sha1path, 'rb') as f:
+      desired_hash = f.read()
+
+  # If the current hash doesn't match what we want in the file, nuke and pave.
+  # Note that this script is only run when a .sha1 file is updated (per DEPS)
+  # so this relatively expensive step of hashing everything only happens when
+  # the toolchain is updated.
+  current_hash = CalculateHash(target_dir)
+  if current_hash != desired_hash:
+    print 'Windows toolchain out of date or doesn\'t exist, updating...'
+    if os.path.isdir(target_dir):
+      RunOrDie('rmdir /s/q "%s"' % target_dir)
+    RunOrDie([
+        sys.executable,

[M-A Ruel, 2013/11/26 02:07:43]

This shouldn't be run under shell=True.

[scottmg, 2013/11/26 05:07:19]

Done.

+        'src/tools/win/toolchain/toolchain2013.py',

[M-A Ruel, 2013/11/26 02:07:43]

os.path.join()

also, you assume about the current working directory.

+        '--targetdir=' + target_dir])
+
+  current_hash = CalculateHash(target_dir)
+  if current_hash != desired_hash:
+    raise SystemExit(

[M-A Ruel, 2013/11/26 02:07:43]

print >> sys.stderr, 'Got ...'
return 1

[scottmg, 2013/11/26 05:07:19]

Done.

+        'Got wrong hash after pulling a new toolchain. '
+        'Wanted \'%s\', got \'%s\'.' % (
+            desired_hash, current_hash))
+  return 0
+
+
+if __name__ == '__main__':
+  sys.exit(main())