def getArticleCited():
'''Give all articles cited
:returns: a list of object artCitedBib
'''
artCitedBib = ArtCitedBib.all()
data = []
for artiCited in artCitedBib:
data.append(artiCited)
return data
def getArtCitedBiblio(wordNumber = 10):
'''Give name all articles cited
:param wordNumber: the max number of results.
:return name of all the article cited, order by number of occurrence
'''
artCitedBib = ArtCitedBib.all()
dataDict = []
results = artCitedBib.fetch(limit=wordNumber)
for artCited in results:
dataDict.append(artCited.nameArticle)
return dataDict
def getArtMostFreqCited(wordNumber = 10):
'''Give article most frequently cited
:param wordNumber: the max number of results.
:returns: a list of object artCitedBib
'''
artCitedBib = ArtCitedBib.all()
artCitedBib.order('-count')
dataDict = []
results = artCitedBib.fetch(limit=wordNumber)
for artCited in results:
dataDict.append(artCited)
return dataDict
def getArtCitedFromArt(nameArt):
''' Give articles cited in an article
:param nameArt: name of the article
:returns: list of names of articles
'''
article = Article.all()
article.filter('name =', nameArt)
result = article.get()
artCitedBib = ArtCitedBib.all()
artCitedBib.filter('keyArticle = ', result)
data = []
for artiCited in artCitedBib:
data.append(artiCited.nameArticle)
return data
def getReferences(fic, articlePrinc):
"""
Get the references of the article
"""
fic = fic.replace("\n", " ")
objFic = re.match(r"(.*)REFERENCES(.*)", fic)
if objFic:
references = re.split("(\[[0-9]+\])", objFic.group(2))
for oneRef in references:
sepAuthName = oneRef.split(",")
authorsArt = []
for l in sepAuthName:
l = l.strip()
matchAuth = re.match(r"([A-Z]\..*)", l)
if matchAuth:
author = re.sub(r"[^a-zA-Z\s\.]", "", matchAuth.group(1))
authorsArt.append(author)
else:
derMatchAuth = re.match(r" and (.*)", l)
if derMatchAuth:
author = re.sub(r"[^a-zA-Z\s\.]", "", derMatchAuth.group(1))
authorsArt.append(author)
else:
nameMatch = re.match(r"(.*)\..*", l)
if nameMatch:
nameArt = nameMatch.group(1)
else:
nameArt = l
nameArt = re.sub(r"[^a-zA-Z\s]", "", nameArt)
if nameArt == "" or len(authorsArt) == 0:
break
articleCitedBib = ArtCitedBib.all()
articleCitedBib.filter("nameArticle =", nameArt)
if articleCitedBib.count() > 0:
artCitedBib = articleCitedBib.get()
artCitedBib.count = artCitedBib.count + 1
else:
artCitedBib = ArtCitedBib(
keyArticle=articlePrinc, nameArticle=nameArt, authors=authorsArt, count=1
)
artCitedBib.put()
break
else:
print "pas de references"
def deleteData():
"""
Delete all the data
"""
masters = Master.all()
for master in masters:
Master.delete(master)
mapReduces = MapReduce.all()
for mapReduce in mapReduces:
MapReduce.delete(mapReduce)
articles = Article.all()
for article in articles:
Article.delete(article)
authors = Author.all()
for author in authors:
Author.delete(author)
artCitedBibs = ArtCitedBib.all()
for artCitedBib in artCitedBibs:
ArtCitedBib.delete(artCitedBib)