def parseTranslation(self, languageHeader, text):
matcher = re.search(ENTranslationHandler.TRANSLATION, text)
if matcher is None:
return None
prefix = matcher.group("prefix")
content = matcher.group("content")
postfix = matcher.group("postfix")
if content.startswith("{{"):
translation = self.parseTemplate(content)
else:
translation = WiktionaryTranslation(
languageHeader,
StringUtils.cleanText(WikiString.removeWikiLinks(content)))
if translation is not None:
additionalInformation = ""
if prefix is not None:
additionalInformation += prefix.strip()
if translation.getGender() is not None:
additionalInformation += " {{" + translation.getGender(
) + "}} "
additionalInformation += postfix
translation.setAdditionalInformation(
StringUtils.cleanText(additionalInformation.strip()))
if self.currentSense is not None and len(
self.currentSense.strip()) > 0:
translation.setRawSense(self.currentSense.strip())
return translation
else:
return None
def processBody(self, text, context):
text = text.strip()
if text.startswith("{{trans-mid}}") or text.startswith("{{mid}}"):
return True
if text.startswith("{{trans-top|") and ("}}" in text):
template = TemplateParser.parseTemplate(text[2:text.find("}}")])
if template is not None and template.getNumberedParamsCount() >= 1:
self.currentSense = template.getNumberedParam(0)
return True
if text.startswith("{{top}}"):
self.currentSense = ""
return True
if text.startswith("{{trans-bottom}}") or text.startswith(
"{{bottom}}"
): # This template indicates the end of the translation block
return False
if text.startswith("{{") or text.startswith(
"=="): # Indicates that a new block has just started.
return False
matcher = re.search(ENTranslationHandler.LANGUAGE, text)
if matcher is None:
return False
languageText = WikiString.removeWikiLinks(matcher.group(1).strip())
language = Language.findByName(languageText)
endOffSet = matcher.end()
if endOffSet > len(text) - 1:
return False
remainingText = text[endOffSet:]
for part in self.splitTranslationParts(remainingText):
translation = self.parseTranslation(language, part)
if translation is not None:
# Save the translation
if self.currentSense not in self.sensNum2trans:
self.sensNum2trans[self.currentSense] = list()
translations = self.sensNum2trans[self.currentSense]
translations.append(translation)
return True
def findMatchingSense(cls, entry, marker):
""" @return the word sense whose sense definition
corresponds to the specified comment (sense marker). The matching
of the corresponding word sense is achieved by word similarity
metrics. Returns <code>None</code> if no matching word sense
could be found. """
if entry.getSenseCount() == 1:
return entry.getSense(1)
# Empty sense marker.
if marker is None or not len(marker):
return None
best1Gram = None
best3Gram = None
best1GramScore = -1
best3GramScore = -1
for sense in entry.senses:
if sense.getIndex() <= 0:
continue # Skip unassigned sense.
gloss = WikiString.removeWikiLinks(
sense.getGloss().getText()).lower()
similarity = SimilarityUtils.wordSim(marker, gloss)
if similarity > best1GramScore:
best1GramScore = similarity
best1Gram = sense
similarity = SimilarityUtils.textSim(marker, gloss)
if similarity > best3GramScore:
best3GramScore = similarity
best3Gram = sense
if best1Gram is None and best3Gram is None:
return None
if best1GramScore <= 0 and best3GramScore <= 0:
return None
if best1GramScore > best3GramScore:
return best1Gram
else:
return best3Gram
def parseTemplate(self, templateString):
template = TemplateParser.parseTemplate(templateString[2:-2])
if template is None or template.getNumberedParamsCount() <= 1:
return None
translationText = StringUtils.cleanText(
WikiString.removeWikiLinks(template.getNumberedParam(1)))
if not translationText:
return None
languageCode = template.getNumberedParam(0)
transliteration = template.getNamedParam("tr")
translation = WiktionaryTranslation(Language.findByCode(languageCode),
translationText)
if template.getNumberedParamsCount(
) > 2 and "=" not in template.getNumberedParam(2):
translation.setGender(template.getNumberedParam(2))
translation.setCheckNeeded("check" in template.getName())
if transliteration is not None:
translation.setTransliteration(
StringUtils.cleanText(transliteration))
return translation
def testremoveWikiLinks(self): # throws Exception
self.assertEqual("Leader",
WikiString.removeWikiLinks("[[leader|Leader]]"))
self.assertEqual("Leader", WikiString.removeWikiLinks("[[Leader]]"))