예제 #1
0
 def test_clean_up_word_list_in_another_language(self):
     translated_wordlists = TranslatedWordlist("test")
     caps = JobAd("Fancy some sourdough, sake&glendronach or a sun-downer?")
     self.assertEqual(
         caps.clean_up_word_list(translated_wordlists),
         [
             "fancy",
             "some",
             "sourdough",
             "sake",
             "glendronach",
             "or",
             "a",
             "sun-downer",
         ],
     )
예제 #2
0
 def test_clean_up_word_list(self):
     caps = JobAd("Sharing is as important as ambition")
     self.assertEqual(caps.clean_up_word_list(),
         ['sharing', 'is', 'as', 'important', 'as', 'ambition'])
     tab = JobAd("Qualities: sharing\tambition")
     self.assertEqual(tab.clean_up_word_list(),
         ['qualities', 'sharing', 'ambition'])
     semicolon = JobAd("Sharing;ambitious")
     self.assertEqual(semicolon.clean_up_word_list(),
         ['sharing', 'ambitious'])
     slash = JobAd(u"Sharing/ambitious")
     self.assertEqual(slash.clean_up_word_list(), ['sharing', 'ambitious'])
     hyphen = JobAd(u"Sharing, co-operative, 'servant-leader'")
     self.assertEqual(hyphen.clean_up_word_list(),
         ['sharing', 'co-operative', 'servant', 'leader'])
     mdash = JobAd(u"Sharing—ambitious")
     self.assertEqual(mdash.clean_up_word_list(), ['sharing', 'ambitious'])
     bracket = JobAd(u"Sharing(ambitious) and (leader)")
     self.assertEqual(bracket.clean_up_word_list(), ['sharing', 'ambitious',
         'and', 'leader'])
     sqbracket = JobAd(u"Sharing[ambitious] and [leader]")
     self.assertEqual(sqbracket.clean_up_word_list(), ['sharing',
         'ambitious', 'and', 'leader'])
     abracket = JobAd(u"Sharing<ambitious> and <leader>")
     self.assertEqual(abracket.clean_up_word_list(), ['sharing',
         'ambitious', 'and', 'leader'])
     space = JobAd(u"Sharing ambitious ")
     self.assertEqual(space.clean_up_word_list(), ['sharing', 'ambitious'])
     amp = JobAd(u"Sharing&ambitious, empathy&kindness,")
     self.assertEqual(amp.clean_up_word_list(),
         ['sharing', 'ambitious', 'empathy', 'kindness'])
     asterisk = JobAd(u"Sharing&ambitious*, empathy*kindness,")
     self.assertEqual(asterisk.clean_up_word_list(),
         ['sharing', 'ambitious', 'empathy', 'kindness'])
     atandquestion = JobAd(u"Lead \"Developer\" Who is Connect@HBS? We ")
     self.assertEqual(atandquestion.clean_up_word_list(),
         ['lead', 'developer', 'who', 'is', 'connect', 'hbs', 'we'])
     exclaim = JobAd(u"Lead Developer v good!")
     self.assertEqual(exclaim.clean_up_word_list(),
         ['lead', 'developer', 'v', 'good'])
     curls = JobAd(u"“Lead” ‘Developer’ v good!")
     self.assertEqual(exclaim.clean_up_word_list(),
         ['lead', 'developer', 'v', 'good'])
예제 #3
0
 def test_clean_up_word_list(self):
     caps = JobAd("Sharing is as important as ambition")
     self.assertEqual(caps.clean_up_word_list(),
         ['sharing', 'is', 'as', 'important', 'as', 'ambition'])
     tab = JobAd("Qualities: sharing\tambition")
     self.assertEqual(tab.clean_up_word_list(),
         ['qualities', 'sharing', 'ambition'])
     semicolon = JobAd("Sharing;ambitious")
     self.assertEqual(semicolon.clean_up_word_list(),
         ['sharing', 'ambitious'])
     slash = JobAd(u"Sharing/ambitious")
     self.assertEqual(slash.clean_up_word_list(), ['sharing', 'ambitious'])
     hyphen = JobAd(u"Sharing, co-operative, 'servant-leader'")
     self.assertEqual(hyphen.clean_up_word_list(),
         ['sharing', 'co-operative', 'servant', 'leader'])
     mdash = JobAd(u"Sharing—ambitious")
     self.assertEqual(mdash.clean_up_word_list(), ['sharing', 'ambitious'])
     bracket = JobAd(u"Sharing(ambitious) and (leader)")
     self.assertEqual(bracket.clean_up_word_list(), ['sharing', 'ambitious',
         'and', 'leader'])
     sqbracket = JobAd(u"Sharing[ambitious] and [leader]")
     self.assertEqual(sqbracket.clean_up_word_list(), ['sharing',
         'ambitious', 'and', 'leader'])
     abracket = JobAd(u"Sharing<ambitious> and <leader>")
     self.assertEqual(abracket.clean_up_word_list(), ['sharing',
         'ambitious', 'and', 'leader'])
     space = JobAd(u"Sharing ambitious ")
     self.assertEqual(space.clean_up_word_list(), ['sharing', 'ambitious'])
     amp = JobAd(u"Sharing&ambitious, empathy&kindness,")
     self.assertEqual(amp.clean_up_word_list(),
         ['sharing', 'ambitious', 'empathy', 'kindness'])
     asterisk = JobAd(u"Sharing&ambitious*, empathy*kindness,")
     self.assertEqual(asterisk.clean_up_word_list(),
         ['sharing', 'ambitious', 'empathy', 'kindness'])
     atandquestion = JobAd(u"Lead \"Developer\" Who is Connect@HBS? We ")
     self.assertEqual(atandquestion.clean_up_word_list(),
         ['lead', 'developer', 'who', 'is', 'connect', 'hbs', 'we'])
     exclaim = JobAd(u"Lead Developer v good!")
     self.assertEqual(exclaim.clean_up_word_list(),
         ['lead', 'developer', 'v', 'good'])
     curls = JobAd(u"“Lead” ‘Developer’ v good!")
     self.assertEqual(exclaim.clean_up_word_list(),
         ['lead', 'developer', 'v', 'good'])
예제 #4
0
 def test_clean_up_word_list(self):
     translated_wordlists = TranslatedWordlist("en")
     caps = JobAd("Sharing is as important as ambition")
     self.assertEqual(
         caps.clean_up_word_list(translated_wordlists),
         ["sharing", "is", "as", "important", "as", "ambition"],
     )
     tab = JobAd("Qualities: sharing\tambition")
     self.assertEqual(
         tab.clean_up_word_list(translated_wordlists),
         ["qualities", "sharing", "ambition"],
     )
     semicolon = JobAd("Sharing;ambitious")
     self.assertEqual(semicolon.clean_up_word_list(translated_wordlists),
                      ["sharing", "ambitious"])
     slash = JobAd(u"Sharing/ambitious")
     self.assertEqual(slash.clean_up_word_list(translated_wordlists),
                      ["sharing", "ambitious"])
     hyphen = JobAd(u"Sharing, co-operative, 'servant-leader'")
     self.assertEqual(
         hyphen.clean_up_word_list(translated_wordlists),
         ["sharing", "co-operative", "servant", "leader"],
     )
     mdash = JobAd(u"Sharing—ambitious")
     self.assertEqual(mdash.clean_up_word_list(translated_wordlists),
                      ["sharing", "ambitious"])
     bracket = JobAd(u"Sharing(ambitious) and (leader)")
     self.assertEqual(
         bracket.clean_up_word_list(translated_wordlists),
         ["sharing", "ambitious", "and", "leader"],
     )
     sqbracket = JobAd(u"Sharing[ambitious] and [leader]")
     self.assertEqual(
         sqbracket.clean_up_word_list(translated_wordlists),
         ["sharing", "ambitious", "and", "leader"],
     )
     abracket = JobAd(u"Sharing<ambitious> and <leader>")
     self.assertEqual(
         abracket.clean_up_word_list(translated_wordlists),
         ["sharing", "ambitious", "and", "leader"],
     )
     space = JobAd(u"Sharing ambitious ")
     self.assertEqual(space.clean_up_word_list(translated_wordlists),
                      ["sharing", "ambitious"])
     amp = JobAd(u"Sharing&ambitious, empathy&kindness,")
     self.assertEqual(
         amp.clean_up_word_list(translated_wordlists),
         ["sharing", "ambitious", "empathy", "kindness"],
     )
     asterisk = JobAd(u"Sharing&ambitious*, empathy*kindness,")
     self.assertEqual(
         asterisk.clean_up_word_list(translated_wordlists),
         ["sharing", "ambitious", "empathy", "kindness"],
     )
     atandquestion = JobAd(u'Lead "Developer" Who is Connect@HBS? We ')
     self.assertEqual(
         atandquestion.clean_up_word_list(translated_wordlists),
         ["lead", "developer", "who", "is", "connect", "hbs", "we"],
     )
     exclaim = JobAd(u"Lead Developer v good!")
     self.assertEqual(
         exclaim.clean_up_word_list(translated_wordlists),
         ["lead", "developer", "v", "good"],
     )
     curls = JobAd(u"“Lead” ‘Developer’ v good")
     self.assertEqual(
         curls.clean_up_word_list(translated_wordlists),
         ["lead", "developer", "v", "good"],
     )
     accents = JobAd(u"cariñoso colaboración pingüino")
     self.assertEqual(
         accents.clean_up_word_list(translated_wordlists),
         ["cariñoso", "colaboración", "pingüino"],
     )