def get_references(self, element):
'''
Get references using the handy comments added by Scielo. But then remove all
of the HTML fluff to get just the text we want.
'''
refs = re.findall(r"<!-- ref -->(.*?)<!-- end-ref -->",
self.wd.page_source)
refs = self._clean_refs(refs)
return mining.TagList(refs, "reference")
def get_keywords(self, element):
"""Gather article keywords from centaurminer.Element object.
Args:
element(:obj: `centaurminer.Element`): Page element to gather keywords from.
Returns:
String comprising keywords separated by HTML-like tags.
"""
keys = self.__parse_keywords(self.get(element))
if keys:
return mining.TagList(keys, "keyword")
return None
def get_organization(self, element):
return mining.TagList(self.get(element, several=True), tag='organization')
def get_authors(self, element):
return mining.TagList(self.get(element, several=True), tag='author')
def get_organization_affiliated(self, element):
"""
Gets several `organizations_affiliated` fields and wraps them
inside `html` like tags.
"""
return mining.TagList(self.get(element, several=True), tag='org')
def get_authors(self, element):
""" Gets several `author` fields and wraps them inside
`html` like tags. """
return mining.TagList(self.get(element, several=True),
tag='author')
def get_references(self, element):
return mining.TagList(self.get(element, several=True), tag='ref')
def get_authors(self, element):
"""Returns a string with article authors from search engine, separated by HTML-like elements"""
authors = map(self.__format_author, self.get(element,
several=True))
return mining.TagList(list(dict.fromkeys(authors)), "author")
def get_organization_affiliated(self, element):
"""Returns a string with article authors organizations, separated by HTML-like elements"""
orgs = [
o.split('</sup>')[-1] for o in self.get(element, several=True)
]
return mining.TagList(orgs, "orgs")
def get_organization_affiliated(self, element):
orgs = self.get(element, several=True)
return mining.TagList(orgs, "org")
def get_title_translated(self, element):
titles = self.get(element, several=True)
return mining.TagList(titles, "title")
def get_organization_affiliated(self, element):
orgs = self.get(element, several=True)
# All orgs are doubled in this format
orgs = orgs[::2]
return mining.TagList(orgs, "org")
def get_references(self, element):
references = self.get(element, several=True)
references = self._clean_refs(references)
return mining.TagList(references, "reference")
