tools/test_pocount.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 from translate.storage import po
   5 from translate.storage import statsdb
   6
   7 class TestPOCount:
   8     def count(self, source, expectedsource, target=None, expectedtarget=None):
   9         """simple helper to check the respective word counts"""
  10         poelement = po.pounit(source)
  11         if target is not None:
  12             poelement.target = target
  13         wordssource, wordstarget = statsdb.wordsinunit(poelement)
  14         print 'Source (expected=%d; actual=%d): "%s"' % (expectedsource, wordssource, source)
  15         assert wordssource == expectedsource
  16         if target is not None:
  17             print 'Target (expected=%d; actual=%d): "%s"' % (expectedtarget, wordstarget, target)
  18             assert wordstarget == expectedtarget
  19
  20     def test_simple_count_zero(self):
  21         """no content"""
  22         self.count("", 0)
  23
  24     def test_simple_count_one(self):
  25         """simplest one word count"""
  26         self.count("One", 1)
  27
  28     def test_simple_count_two(self):
  29         """simplest one word count"""
  30         self.count("One two", 2)
  31
  32     def test_punctuation_divides_words(self):
  33         """test that we break words when there is punctuation"""
  34         self.count("One. Two", 2)
  35         self.count("One.Two", 2)
  36
  37     def test_xml_tags(self):
  38         """test that we do not count XML tags as words"""
  39         # <br> is a word break
  40         self.count("A word<br>Another word", 4)
  41         self.count("A word<br/>Another word", 4)
  42         self.count("A word<br />Another word", 4)
  43         # \n is a word break
  44         self.count("<p>A word</p>\n<p>Another word</p>", 4)
  45
  46     def test_newlines(self):
  47         """test to see that newlines divide words"""
  48         # newlines break words
  49         self.count("A word.\nAnother word", 4)
  50         self.count(r"A word.\\n\nAnother word", 4)
  51
  52     def test_variables_are_words(self):
  53         """test that we count variables as words"""
  54         self.count("%PROGRAMNAME %PROGRAM% %s $file $1", 5)
  55
  56     def test_plurals(self):
  57         """test that we can handle plural PO elements"""
  58         # #: gdk-pixbuf/gdk-pixdata.c:430
  59         # #, c-format
  60         # msgid "failed to allocate image buffer of %u byte"
  61         # msgid_plural "failed to allocate image buffer of %u bytes"
  62         # msgstr[0] "e paletšwe go hwetša sešireletši sa seswantšho sa paete ya %u"
  63         # msgstr[1] "e paletšwe go hwetša sešireletši sa seswantšho sa dipaete tša %u"
  64
  65     def test_plurals_kde(self):
  66         """test that we correcly count old style KDE plurals"""
  67         self.count("_n: Singular\\n\nPlural", 2, "Een\\n\ntwee\\n\ndrie", 3)
  68
  69     def test_msgid_blank(self):
  70         """counts a message id"""
  71         self.count("   ", 0)
  72
  73     # Counting strings
  74     #  We need to check how we count strings also and if we call it translated or untranslated
  75     # ie an all spaces msgid should be translated if there are spaces in the msgstr
  76
  77     # Make sure we don't count obsolete messages
  78
  79     # Do we correctly identify a translated yet blank message?
  80
  81     # Need to test that we can differentiate between fuzzy, translated and untranslated
  82