tools/test_pogrep.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 from translate.storage import po
   5 from translate.storage import xliff
   6 from translate.tools import pogrep
   7 from translate.misc import wStringIO
   8
   9 class TestPOGrep:
  10     def poparse(self, posource):
  11         """helper that parses po source without requiring files"""
  12         dummyfile = wStringIO.StringIO(posource)
  13         pofile = po.pofile(dummyfile)
  14         return pofile
  15
  16     def pogrep(self, posource, searchstring, cmdlineoptions=None):
  17         """helper that parses po source and passes it through a filter"""
  18         if cmdlineoptions is None:
  19             cmdlineoptions = []
  20         options, args = pogrep.cmdlineparser().parse_args(["xxx.po"] + cmdlineoptions)
  21         grepfilter = pogrep.GrepFilter(searchstring, options.searchparts, options.ignorecase, options.useregexp, options.invertmatch, options.accelchar)
  22         tofile = grepfilter.filterfile(self.poparse(posource))
  23         print str(tofile)
  24         return str(tofile)
  25
  26     def test_simplegrep_msgid(self):
  27         """grep for a string in the source"""
  28         posource = '#: test.c\nmsgid "test"\nmsgstr "rest"\n'
  29         poresult = self.pogrep(posource, "test", ["--search=msgid"])
  30         assert poresult == posource
  31         poresult = self.pogrep(posource, "rest", ["--search=msgid"])
  32         assert poresult == ""
  33
  34     def test_simplegrep_msgstr(self):
  35         """grep for a string in the target"""
  36         posource = '#: test.c\nmsgid "test"\nmsgstr "rest"\n'
  37         poresult = self.pogrep(posource, "rest", ["--search=msgstr"])
  38         assert poresult == posource
  39         poresult = self.pogrep(posource, "test", ["--search=msgstr"])
  40         assert poresult == ""
  41
  42     def test_simplegrep_locations(self):
  43         """grep for a string in the location comments"""
  44         posource = '#: test.c\nmsgid "test"\nmsgstr "rest"\n'
  45         poresult = self.pogrep(posource, "test.c", ["--search=locations"])
  46         assert poresult == posource
  47         poresult = self.pogrep(posource, "rest.c", ["--search=locations"])
  48         assert poresult == ""
  49
  50     def test_simplegrep_comments(self):
  51         """grep for a string in the comments"""
  52         posource = '# (review) comment\n#: test.c\nmsgid "test"\nmsgstr "rest"\n'
  53         poresult = self.pogrep(posource, "review", ["--search=comment"])
  54         assert poresult == posource
  55         poresult = self.pogrep(posource, "test", ["--search=comment"])
  56         assert poresult == ""
  57
  58     def test_unicode_message_searchstring(self):
  59         """check that we can grep unicode messages and use unicode search strings"""
  60         poascii = '# comment\n#: test.c\nmsgid "test"\nmsgstr "rest"\n'
  61         pounicode = '# comment\n#: test.c\nmsgid "test"\nmsgstr "rešṱ"\n'
  62         queryascii = 'rest'
  63         queryunicode = 'rešṱ'
  64         for source, search, expected in [(poascii, queryascii, poascii),
  65                                          (poascii, queryunicode, ''),
  66                                          (pounicode, queryascii, ''),
  67                                          (pounicode, queryunicode, pounicode)]:
  68             print "Source:\n%s\nSearch: %s\n" % (source, search)
  69             poresult = self.pogrep(source, search)
  70             assert poresult == expected
  71
  72     def test_unicode_message_regex_searchstring(self):
  73         """check that we can grep unicode messages and use unicode regex search strings"""
  74         poascii = '# comment\n#: test.c\nmsgid "test"\nmsgstr "rest"\n'
  75         pounicode = '# comment\n#: test.c\nmsgid "test"\nmsgstr "rešṱ"\n'
  76         queryascii = 'rest'
  77         queryunicode = 'rešṱ'
  78         for source, search, expected in [(poascii, queryascii, poascii),
  79                                          (poascii, queryunicode, ''),
  80                                          (pounicode, queryascii, ''),
  81                                          (pounicode, queryunicode, pounicode)]:
  82             print "Source:\n%s\nSearch: %s\n" % (source, search)
  83             poresult = self.pogrep(source, search, ["--regexp"])
  84             assert poresult == expected
  85
  86     def test_unicode_normalise(self):
  87         """check that we normlise unicode strings before comparing"""
  88         source_template = u'# comment\n#: test.c\nmsgid "test"\nmsgstr "t%sst"\n'
  89         # é, e + '
  90         # Ḽ, L + ^
  91         # Ṏ
  92         groups = [(u"\u00e9", u"\u0065\u0301"), \
  93                   (u"\u1e3c", u"\u004c\u032d"), \
  94                   (u"\u1e4e", u"\u004f\u0303\u0308", u"\u00d5\u0308")]
  95         for letters in groups:
  96             for source_letter in letters:
  97                 source = source_template % source_letter
  98                 for search_letter in letters:
  99                     print search_letter.encode('utf-8')
 100                     poresult = self.pogrep(source, search_letter)
 101                     assert poresult == source.encode('utf-8')
 102
 103 class TestXLiffGrep:
 104     xliff_skeleton = '''<?xml version="1.0" ?>
 105 <xliff version="1.1" xmlns="urn:oasis:names:tc:xliff:document:1.1">
 106   <file original="filename.po" source-language="en-US" datatype="po">
 107     <body>
 108         %s
 109     </body>
 110   </file>
 111 </xliff>'''
 112
 113     xliff_text = xliff_skeleton % '''<trans-unit>
 114   <source>red</source>
 115   <target>rooi</target>
 116 </trans-unit>'''
 117
 118     def xliff_parse(self, xliff_text):
 119         """helper that parses po source without requiring files"""
 120         dummyfile = wStringIO.StringIO(xliff_text)
 121         xliff_file = xliff.xlifffile(dummyfile)
 122         return xliff_file
 123
 124     def xliff_grep(self, xliff_text, searchstring, cmdlineoptions=None):
 125         """helper that parses xliff text and passes it through a filter"""
 126         if cmdlineoptions is None:
 127             cmdlineoptions = []
 128         options, args = pogrep.cmdlineparser().parse_args(["xxx.xliff"] + cmdlineoptions)
 129         grepfilter = pogrep.GrepFilter(searchstring, options.searchparts, options.ignorecase, options.useregexp, options.invertmatch, options.accelchar)
 130         tofile = grepfilter.filterfile(self.xliff_parse(xliff_text))
 131         return str(tofile)
 132
 133     def test_simplegrep(self):
 134         """grep for a simple string."""
 135         xliff_text = self.xliff_text
 136         xliff_file = self.xliff_parse(xliff_text)
 137         xliff_result = self.xliff_parse(self.xliff_grep(xliff_text, "red"))
 138         assert xliff_result.units[0].getsource() == u"red"
 139         assert xliff_result.units[0].gettarget() == u"rooi"
 140
 141         xliff_result = self.xliff_parse(self.xliff_grep(xliff_text, "unavailable string"))
 142         assert xliff_result.isempty()
 143
 144