Returns word set used in page content >>> sorted(getPageWordSet(u' foobar test ')) == [u'foobar', u'test'] True
(page)
| 2243 | return retVal |
| 2244 | |
| 2245 | def getPageWordSet(page): |
| 2246 | """ |
| 2247 | Returns word set used in page content |
| 2248 | |
| 2249 | >>> sorted(getPageWordSet(u'<html><title>foobar</title><body>test</body></html>')) == [u'foobar', u'test'] |
| 2250 | True |
| 2251 | """ |
| 2252 | |
| 2253 | retVal = set() |
| 2254 | |
| 2255 | # only if the page's charset has been successfully identified |
| 2256 | if isinstance(page, six.string_types): |
| 2257 | retVal = set(_.group(0) for _ in re.finditer(r"\w+", getFilteredPageContent(page))) |
| 2258 | |
| 2259 | return retVal |
| 2260 | |
| 2261 | def showStaticWords(firstPage, secondPage, minLength=3): |
| 2262 | """ |
no test coverage detected
searching dependent graphs…