tools/hook-scripts/verify-po.py

   1 #!/usr/bin/env python
   2 """This is a pre-commit hook that checks whether the contents of PO files
   3 committed to the repository are encoded in UTF-8.
   4 """
   5
   6 import codecs
   7 import string
   8 import sys
   9 from svn import core, fs, delta, repos
  10
  11 # Set to the path of the 'msgfmt' executable to use msgfmt to check
  12 # the syntax of the po file
  13
  14 USE_MSGFMT = None
  15
  16 if USE_MSGFMT is not None:
  17   import popen2
  18   class MsgFmtChecker:
  19     def __init__(self):
  20       self.pipe = popen2.Popen3("%s -c -o /dev/null -" % (USE_MSGFMT))
  21       self.pipe.fromchild.close()
  22       self.io_error = 0
  23
  24     def write(self, data):
  25       if self.io_error:
  26         return
  27       try:
  28         self.pipe.tochild.write(data)
  29       except IOError:
  30         self.io_error = 1
  31
  32     def close(self):
  33       try:
  34         self.pipe.tochild.close()
  35       except IOError:
  36         self.io_error = 1
  37       return self.pipe.wait() == 0 and not self.io_error
  38 else:
  39   class MsgFmtChecker:
  40     def write(self, data):
  41       pass
  42     def close(self):
  43       return 1
  44
  45
  46 class ChangeReceiver(delta.Editor):
  47   def __init__(self, txn_root, base_root, pool):
  48     self.txn_root = txn_root
  49     self.base_root = base_root
  50     self.pool = pool
  51
  52   def add_file(self, path, parent_baton,
  53                copyfrom_path, copyfrom_revision, file_pool):
  54     return [0, path]
  55
  56   def open_file(self, path, parent_baton, base_revision, file_pool):
  57     return [0, path]
  58
  59   def apply_textdelta(self, file_baton, base_checksum):
  60     file_baton[0] = 1
  61     # no handler
  62     return None
  63
  64   def close_file(self, file_baton, text_checksum):
  65     changed, path = file_baton
  66     if len(path) < 3 or path[-3:] != '.po' or not changed:
  67       # This is not a .po file, or it hasn't changed
  68       return
  69
  70     try:
  71       # Read the file contents through a validating UTF-8 decoder
  72       subpool = core.svn_pool_create(self.pool)
  73       checker = MsgFmtChecker()
  74       try:
  75         stream = core.Stream(fs.file_contents(self.txn_root, path, subpool))
  76         reader = codecs.getreader('UTF-8')(stream, 'strict')
  77         writer = codecs.getwriter('UTF-8')(checker, 'strict')
  78         while 1:
  79           data = reader.read(core.SVN_STREAM_CHUNK_SIZE)
  80           if not data:
  81             break
  82           writer.write(data)
  83         if not checker.close():
  84           sys.exit("PO format check failed for '" + path + "'")
  85       except UnicodeError:
  86         sys.exit("PO file is not in UTF-8: '" + path + "'")
  87     finally:
  88       core.svn_pool_destroy(subpool)
  89
  90
  91 def check_po(pool, repos_path, txn):
  92   def authz_cb(root, path, pool):
  93     return 1
  94
  95   fs_ptr = repos.fs(repos.open(repos_path, pool))
  96   txn_ptr = fs.open_txn(fs_ptr, txn, pool)
  97   txn_root = fs.txn_root(txn_ptr, pool)
  98   base_root = fs.revision_root(fs_ptr, fs.txn_base_revision(txn_ptr), pool)
  99   editor = ChangeReceiver(txn_root, base_root, pool)
 100   e_ptr, e_baton = delta.make_editor(editor, pool)
 101   repos.dir_delta(base_root, '', '', txn_root, '',
 102                   e_ptr, e_baton, authz_cb, 0, 1, 0, 0, pool)
 103
 104
 105 if __name__ == '__main__':
 106   assert len(sys.argv) == 3
 107   core.run_app(check_po, sys.argv[1], sys.argv[2])