def test_replace_invalid_chars(self):
        """solrutils - replacement of invalid characters"""
        self.assertEqual(u'abc\nde', replace_invalid_solr_characters(u'abc\uD800\nde'))
        self.assertEqual(u'abc\nde', replace_invalid_solr_characters(u'abc\uDF12\nde'))
        self.assertEqual(u'abc\nde', replace_invalid_solr_characters(u'abc\uDFFF\nde'))

        self.assertEqual(u'abc\nde', replace_invalid_solr_characters(u'abc\uFFFE\nde'))
        self.assertEqual(u'abc\nde', replace_invalid_solr_characters(u'abc\uFFFF\nde'))
Example #2
0
    def test_replace_invalid_chars(self):
        """solrutils - replacement of invalid characters"""
        self.assertEqual(u'abc\nde',
                         replace_invalid_solr_characters(u'abc\uD800\nde'))
        self.assertEqual(u'abc\nde',
                         replace_invalid_solr_characters(u'abc\uDF12\nde'))
        self.assertEqual(u'abc\nde',
                         replace_invalid_solr_characters(u'abc\uDFFF\nde'))

        self.assertEqual(u'abc\nde',
                         replace_invalid_solr_characters(u'abc\uFFFE\nde'))
        self.assertEqual(u'abc\nde',
                         replace_invalid_solr_characters(u'abc\uFFFF\nde'))
def solr_add(recid, abstract, author, fulltext, keyword, title):
    """
    Helper function that adds word similarity ranking relevant indexes to Solr.
    """
    try:
        SOLR_CONNECTION.add(id=recid,
                            abstract=replace_invalid_solr_characters(abstract),
                            author=replace_invalid_solr_characters(author),
                            fulltext=replace_invalid_solr_characters(fulltext),
                            keyword=replace_invalid_solr_characters(keyword),
                            title=replace_invalid_solr_characters(title))
    except:
        register_exception(alert_admin=True)
Example #4
0
def solr_add(recid, abstract, author, fulltext, keyword, title):
    """
    Helper function that adds word similarity ranking relevant indexes to Solr.
    """
    try:
        SOLR_CONNECTION.add(id=recid,
                            abstract=replace_invalid_solr_characters(abstract),
                            author=replace_invalid_solr_characters(author),
                            fulltext=replace_invalid_solr_characters(fulltext),
                            keyword=replace_invalid_solr_characters(keyword),
                            title=replace_invalid_solr_characters(title))
    except:
        register_exception(alert_admin=True)
    def test_replace_control_characters(self):
        """solrutils - replacement of control characters"""
        self.assertEqual(u'abc \nde', replace_invalid_solr_characters(u'abc\u0000\nde'))
        self.assertEqual(u'abc \nde', replace_invalid_solr_characters(u'abc\u0003\nde'))
        self.assertEqual(u'abc \nde', replace_invalid_solr_characters(u'abc\u0008\nde'))

        self.assertEqual(u'abc \nde', replace_invalid_solr_characters(u'abc\u000B\nde'))
        self.assertEqual(u'abc \nde', replace_invalid_solr_characters(u'abc\u000C\nde'))

        self.assertEqual(u'abc \nde', replace_invalid_solr_characters(u'abc\u000E\nde'))
        self.assertEqual(u'abc \nde', replace_invalid_solr_characters(u'abc\u0012\nde'))
        self.assertEqual(u'abc \nde', replace_invalid_solr_characters(u'abc\u001F\nde'))
Example #6
0
    def test_replace_control_characters(self):
        """solrutils - replacement of control characters"""
        self.assertEqual(u'abc \nde',
                         replace_invalid_solr_characters(u'abc\u0000\nde'))
        self.assertEqual(u'abc \nde',
                         replace_invalid_solr_characters(u'abc\u0003\nde'))
        self.assertEqual(u'abc \nde',
                         replace_invalid_solr_characters(u'abc\u0008\nde'))

        self.assertEqual(u'abc \nde',
                         replace_invalid_solr_characters(u'abc\u000B\nde'))
        self.assertEqual(u'abc \nde',
                         replace_invalid_solr_characters(u'abc\u000C\nde'))

        self.assertEqual(u'abc \nde',
                         replace_invalid_solr_characters(u'abc\u000E\nde'))
        self.assertEqual(u'abc \nde',
                         replace_invalid_solr_characters(u'abc\u0012\nde'))
        self.assertEqual(u'abc \nde',
                         replace_invalid_solr_characters(u'abc\u001F\nde'))
 def test_no_replacement(self):
     """solrutils - no characters to replace"""
     utext_in = unicode('foo\nbar\tfab\n\r', 'utf-8')
     utext_out = unicode('foo\nbar\tfab\n\r', 'utf-8')
     self.assertEqual(utext_out, replace_invalid_solr_characters(utext_in))
Example #8
0
 def test_no_replacement(self):
     """solrutils - no characters to replace"""
     utext_in = unicode('foo\nbar\tfab\n\r', 'utf-8')
     utext_out = unicode('foo\nbar\tfab\n\r', 'utf-8')
     self.assertEqual(utext_out, replace_invalid_solr_characters(utext_in))