subversion/tests/cmdline/utf8_tests.py

   1 #!/usr/bin/env python
   2 #  -*- coding: utf-8 -*-
   3 #
   4 #  utf8_tests.py:  testing the svn client's utf8 (i18n) handling
   5 #
   6 #  Subversion is a tool for revision control.
   7 #  See http://subversion.tigris.org for more information.
   8 #
   9 # ====================================================================
  10 # Copyright (c) 2000-2004 CollabNet.  All rights reserved.
  11 #
  12 # This software is licensed as described in the file COPYING, which
  13 # you should have received as part of this distribution.  The terms
  14 # are also available at http://subversion.tigris.org/license-1.html.
  15 # If newer versions of this license are posted there, you may use a
  16 # newer version instead, at your option.
  17 #
  18 ######################################################################
  19
  20 # General modules
  21 import sys, re, os, locale
  22
  23 # Our testing module
  24 import svntest
  25 from svntest import wc
  26
  27 # (abbreviation)
  28 Item = wc.StateItem
  29 Skip = svntest.testcase.Skip
  30
  31 #--------------------------------------------------------------------
  32 # Data
  33
  34 # Here's a filename and a log message which contain some high-ascii
  35 # data.  In theory this data has different interpretations when
  36 # converting from 2 different charsets into UTF-8.
  37
  38 ### "bÔçÅ" in ISO-8859-1 encoding:
  39 i18n_filename = 'b\xd4\xe7\xc5'
  40
  41 ### "drieÃ«ntwintig keer was Ã©Ã©n keer teveel" in ISO-8859-1 encoding:
  42 i18n_logmsg = 'drie\xc3\xabntwintig keer was \xc3\xa9\xc3\xa9n keer teveel'
  43
  44
  45 ######################################################################
  46 # Tests
  47 #
  48 #   Each test must return on success or raise on failure.
  49
  50
  51 def basic_utf8_conversion(sbox):
  52   "conversion of paths and logs to/from utf8"
  53
  54   sbox.build()
  55   wc_dir = sbox.wc_dir
  56
  57   # Create the new i18n file and schedule it for addition
  58   svntest.main.file_append(os.path.join(wc_dir, i18n_filename), "hi")
  59   svntest.actions.run_and_verify_svn(
  60     "Failed to schedule i18n filename for addition", None, [],
  61     'add', os.path.join(wc_dir, i18n_filename))
  62
  63   svntest.actions.run_and_verify_svn(
  64     "Failed to commit i18n filename", None, [],
  65     'commit', '-m', i18n_logmsg, wc_dir)
  66
  67 # Here's how the test should really work:
  68
  69 # 1. sh LC_ALL=ISO-8859-1 svn commit <filename> -m "<logmsg>"
  70
  71 # 2. sh LC_ALL=UTF-8 svn log -rHEAD > output
  72
  73 # 3. verify that output is the exact UTF-8 data that we expect.
  74
  75 # 4. repeat the process using some other locale other than ISO8859-1,
  76 #    preferably some locale which will convert the high-ascii data to
  77 #    *different* UTF-8.
  78
  79
  80
  81 #----------------------------------------------------------------------
  82
  83 ########################################################################
  84 # Run the tests
  85
  86 try:
  87   # Generic setlocale so that getlocale returns something sensible
  88   locale.setlocale(locale.LC_ALL, '')
  89
  90   # Try to make these test run in an ISO-8859-1 environment, otherwise
  91   # they would run in whatever random locale the testing platform
  92   # happens to have, and then we couldn't predict the exact results.
  93   if svntest.main.windows:
  94     # In this case, it would probably be "english_usa.1252", but you should
  95     # be able to set just the encoding by using ".1252" (that's codepage
  96     # 1252, which is almost but not quite entirely unlike tea; um, I mean
  97     # it's very similar to ISO-8859-1).
  98     #                                     -- Branko Čibej <brane@xbc.nu>
  99     locale.setlocale(locale.LC_ALL, '.1252')
 100   else:
 101     locale.setlocale(locale.LC_ALL, 'en_US.ISO8859-1')
 102
 103     if os.putenv:
 104       # propagate to the svn* executables, so they do the correct translation
 105       # the line below works for Linux systems if they have the particular
 106       # locale installed
 107       os.environ['LC_ALL'] = "en_US.ISO8859-1"
 108 except:
 109   pass
 110
 111 # Check to see if the locale uses ISO-8859-1 encoding.  The regex is necessary
 112 # because some systems ommit the first hyphen or use lowercase letters for ISO.
 113 if sys.platform == 'win32':
 114   localematch = 1
 115 else:
 116   localeenc = locale.getlocale()[1]
 117   if localeenc:
 118     localeregex = re.compile('^ISO-?8859-1$', re.I)
 119     localematch = localeregex.search(localeenc)
 120     try:
 121       svntest.actions.run_and_verify_svn(None, svntest.SVNAnyOutput, [],"help")
 122     except:
 123       # We won't be able to run the client; this might be because the
 124       # system does not support the iso-8859-1 locale. Anyhow, it makes
 125       # no sense to run the test.
 126       localematch = None
 127   else:
 128     localematch = None
 129
 130 # Also check that the environment contains the expected locale settings
 131 # either by default, or because we set them above.
 132 if localematch:
 133   localeregex = re.compile('^en_US\.ISO-?8859-1$', re.I)
 134   for env in [ 'LC_ALL', 'LC_CTYPE', 'LANG' ]:
 135     env_value = os.getenv(env)
 136     if env_value:
 137       if localeregex.search(env_value):
 138         break
 139       else:
 140         localematch = None
 141         break
 142
 143
 144 ########################################################################
 145 # Run the tests
 146
 147 # list all tests here, starting with None:
 148 test_list = [ None,
 149               Skip(basic_utf8_conversion)
 150              ]
 151
 152 if __name__ == '__main__':
 153   svntest.main.run_tests(test_list)
 154   # NOTREACHED
 155
 156
 157 ### End of file.