def getReferences(self, json=False):
references = self.soup.find_all('ref')
referenceList = []
for reference in references:
extractedReference = Reference()
try:
extractedReference.id = reference.attrs['id']
except AttributeError:
print('id not present')
continue
try:
authors = reference.find_all('string-name')
extractedReference.authors[:] = []
for author in authors:
surname = author.find('surname').get_text().encode("utf-8")
givenName = author.find('given-names').get_text().encode(
"utf-8")
authorObject = Author()
authorObject.surname = surname
authorObject.givenName = givenName
extractedReference.authors.append(authorObject)
except AttributeError:
print('author not present')
try:
extractedReference.year = reference.find(
'year').get_text().encode("utf-8")
except AttributeError:
print('year not present')
try:
extractedReference.articleTitle = reference.find_all(
'article-title')[-1].get_text().encode("utf-8")
print(extractedReference.articleTitle)
except AttributeError:
print('article title not present')
try:
extractedReference.source = reference.find(
'source').get_text().encode("utf-8")
except AttributeError:
print('source not present')
try:
extractedReference.volume = reference.find(
'volume').get_text().encode("utf-8")
except AttributeError:
print('volume not present')
try:
extractedReference.fromPage = reference.find(
'fpage').get_text().encode("utf-8")
except AttributeError:
print('fromPage not present')
try:
extractedReference.toPage = reference.find(
'lpage').get_text().encode("utf-8")
except AttributeError:
print('toPage not present')
referenceList.append(extractedReference)
if json is True:
referenceListJsons = []
for reference in referenceList:
referenceListJsons.append(self.convertToJson(reference))
return self.convertToJson(referenceList)
return referenceList
