| 1 | import unittest |
|---|
| 2 | |
|---|
| 3 | import whoosh.analysis as analysis |
|---|
| 4 | import whoosh.highlight as highlight |
|---|
| 5 | |
|---|
| 6 | |
|---|
| 7 | class TestHighlighting(unittest.TestCase): |
|---|
| 8 | _doc = u"alfa bravo charlie delta echo foxtrot golf hotel india juliet kilo lima" |
|---|
| 9 | |
|---|
| 10 | def test_null_fragment(self): |
|---|
| 11 | terms = frozenset(("bravo", "india")) |
|---|
| 12 | sa = analysis.StandardAnalyzer() |
|---|
| 13 | nf = highlight.NullFragmenter |
|---|
| 14 | uc = highlight.UppercaseFormatter() |
|---|
| 15 | htext = highlight.highlight(self._doc, terms, sa, nf, uc) |
|---|
| 16 | self.assertEqual(htext, "alfa BRAVO charlie delta echo foxtrot golf hotel INDIA juliet kilo lima") |
|---|
| 17 | |
|---|
| 18 | def test_simple_fragment(self): |
|---|
| 19 | terms = frozenset(("bravo", "india")) |
|---|
| 20 | sa = analysis.StandardAnalyzer() |
|---|
| 21 | sf = highlight.SimpleFragmenter(size=20) |
|---|
| 22 | uc = highlight.UppercaseFormatter() |
|---|
| 23 | htext = highlight.highlight(self._doc, terms, sa, sf, uc) |
|---|
| 24 | self.assertEqual(htext, "alfa BRAVO charlie...hotel INDIA juliet kilo") |
|---|
| 25 | |
|---|
| 26 | def test_sentence_fragment(self): |
|---|
| 27 | text = u"This is the first sentence. This one doesn't have the word. This sentence is the second. Third sentence here." |
|---|
| 28 | terms = ("sentence", ) |
|---|
| 29 | sa = analysis.StandardAnalyzer(stoplist=None) |
|---|
| 30 | sf = highlight.SentenceFragmenter() |
|---|
| 31 | uc = highlight.UppercaseFormatter() |
|---|
| 32 | htext = highlight.highlight(text, terms, sa, sf, uc) |
|---|
| 33 | self.assertEqual(htext, "This is the first SENTENCE...This SENTENCE is the second...Third SENTENCE here") |
|---|
| 34 | |
|---|
| 35 | def test_context_fragment(self): |
|---|
| 36 | terms = frozenset(("bravo", "india")) |
|---|
| 37 | sa = analysis.StandardAnalyzer() |
|---|
| 38 | cf = highlight.ContextFragmenter(terms, surround=6) |
|---|
| 39 | uc = highlight.UppercaseFormatter() |
|---|
| 40 | htext = highlight.highlight(self._doc, terms, sa, cf, uc) |
|---|
| 41 | self.assertEqual(htext, "alfa BRAVO charlie...hotel INDIA juliet") |
|---|
| 42 | |
|---|
| 43 | def test_html_format(self): |
|---|
| 44 | terms = frozenset(("bravo", "india")) |
|---|
| 45 | sa = analysis.StandardAnalyzer() |
|---|
| 46 | cf = highlight.ContextFragmenter(terms, surround=6) |
|---|
| 47 | hf = highlight.HtmlFormatter() |
|---|
| 48 | htext = highlight.highlight(self._doc, terms, sa, cf, hf) |
|---|
| 49 | self.assertEqual(htext, 'alfa <strong class="match term0">bravo</strong> charlie...hotel <strong class="match term1">india</strong> juliet') |
|---|
| 50 | |
|---|
| 51 | def test_maxclasses(self): |
|---|
| 52 | terms = frozenset(("alfa", "bravo", "charlie", "delta", "echo")) |
|---|
| 53 | sa = analysis.StandardAnalyzer() |
|---|
| 54 | cf = highlight.ContextFragmenter(terms, surround=6) |
|---|
| 55 | hf = highlight.HtmlFormatter(tagname="b", termclass="t", maxclasses=2) |
|---|
| 56 | htext = highlight.highlight(self._doc, terms, sa, cf, hf) |
|---|
| 57 | self.assertEqual(htext, '<b class="match t0">alfa</b> <b class="match t1">bravo</b> <b class="match t0">charlie</b>...<b class="match t1">delta</b> <b class="match t0">echo</b> foxtrot') |
|---|
| 58 | |
|---|
| 59 | |
|---|
| 60 | |
|---|
| 61 | if __name__ == '__main__': |
|---|
| 62 | unittest.main() |
|---|
