def _getActualExpansions(self, articleID, article):
extractor = AcronymExtractor_v1()
acronymExpansions = extractor.get_acronyms(article)
expander = Expander_fromText()
actualExpansions, ignore_boolean = expander.try_to_expand_acronyms(
article, acronymExpansions)
result = [item for item in actualExpansions.items() if len(item)>0]
return dict(result)
def _removeInTextExpansions(self, article, acronyms):
expander = Expander_fromText()
for acronym in acronyms:
for pattern in expander.definition_patterns(acronym):
results = pattern.findall(article)
for result in results:
article = article.replace(result, " ")
self.__verifyRemovalOfInTextExpansions(article, acronyms)
return article
def __verifyRemovalOfInTextExpansions(self, article_for_testing, acronyms):
expander = Expander_fromText()
for acronym in acronyms:
for pattern in expander.definition_patterns(acronym):
if(len(pattern.findall(article_for_testing)) > 0):
errorString = "Verification of text expansion removal failed:\nAcronym: " +\
acronym + "\nText:\n" + article_for_testing
common_logger.error(errorString)
raise RuntimeError(errorString)
if acronym not in article_for_testing:
errorMessage = "Acronym removed from text while cleaning:\nAcronym: " +\
acronym + "\nText:\n" + article_for_testing
common_logger.error(errorMessage)
raise RuntimeError(errorMessage)
