def saveMapReduce(namefic):
"""
Converted the pdf file in text
Do the mapper and reduce in the text
Get the references cited in the article
Save data Author, Article, ArtCitedBib, MapReduce, Master
:param namefic : the name file
"""
# save_pdf(namefic)
fic = convert_pdf_to_txt(namefic)
dataDict = mapper(fic)
dataDict = reducer(dataDict)
lines = re.split(r"\n", fic)
authorStr = re.sub(r"[^a-zA-Z\s]", " ", lines[4])
author = Author(name=authorStr)
author.put()
titre = re.sub(r"[^a-zA-Z\s]", " ", lines[0] + lines[1])
titre = titre.strip()
article = Article(name=titre, fileName=namefic)
article.put()
getReferences(fic, article)
artiAuth = ArtiAuth(keyAuthor=author, keyArticle=article)
artiAuth.put()
for cle in dataDict.keys():
mapReduce = MapReduce(keyWord=cle, keyArticle=article, count=dataDict[cle])
mapReduce.put()
checkMaster = Master.all()
checkMaster.filter("keyWord =", cle)
if checkMaster.count() > 0:
master = checkMaster.get()
master.count = master.count + dataDict[cle]
else:
master = Master(keyWord=cle, count=dataDict[cle])
master.put()
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)
