if temp is not None: for pat in temp: pat = CleanPatent(pat) if pat not in ListeBrevetAug and pat != '': if pat['label'] in DejaVu: temporar = [patent for patent in temp if patent['label'] == pat['label']][0] #hum should be unique temporar=UnNest(temporar) for cle in temporar.keys(): temporar[cle] = UnNest(temporar[cle]) temporar = CleanPatent(Update(temporar, pat)) temporar = CleanPatent(temporar) ListeBrevetAug.append(temporar) #temp.append(temporar) else: pat = CleanPatent(pat) for cle in pat.keys(): pat[cle] = UnNest(pat[cle]) ListeBrevetAug.append(CleanPatent(pat)) DejaVu.append(pat['label']) elif pat['label'] in ListeBrevetAug and pat != '': temporar = [patent for patent in ListeBrevetAug if patent['label'] == pat['label']][0] #hum should be unique ListeBrevetAug.remove(temporar) temporar = CleanPatent(Update(temporar, pat)) temporar = CleanPatent(temporar) for cle in temporar.keys(): temporar[cle] = UnNest(temporar[cle]) ListeBrevetAug.append(temporar) # time.sleep(7) Done.append(Brev) with open(ResultPathFamilies+'//Families'+ ndf, 'w') as ndfLstBrev:
if pat not in ListeBrevetAug and pat != '': if pat['label'] in DejaVu: temporar = [ patent for patent in temp if patent['label'] == pat['label'] ][0] #hum should be unique temporar = UnNest(temporar) for cle in temporar.keys(): temporar[cle] = UnNest(temporar[cle]) temporar = CleanPatent(Update(temporar, pat)) temporar = CleanPatent(temporar) ListeBrevetAug.append(temporar) #temp.append(temporar) else: pat = CleanPatent(pat) for cle in pat.keys(): pat[cle] = UnNest(pat[cle]) ListeBrevetAug.append(CleanPatent(pat)) DejaVu.append(pat['label']) elif pat['label'] in ListeBrevetAug and pat != '': temporar = [ patent for patent in ListeBrevetAug if patent['label'] == pat['label'] ][0] #hum should be unique ListeBrevetAug.remove(temporar) temporar = CleanPatent(Update(temporar, pat)) temporar = CleanPatent(temporar) for cle in temporar.keys(): temporar[cle] = UnNest(temporar[cle]) ListeBrevetAug.append(temporar)
brev[cle] = UnNest(brev[cle]) brev= CleanPatent(brev) brev= CleanPatentOthers2(brev) ## tempo = brev # this one for DataTable tempo2 = dict() #the one for pitable PaysInv= [] #new field PaysApp = [] #tempo = CleanPatent(tempo) tempo2 = copy.deepcopy(tempo) #ugly tempo3 = dict() #what the problem LstExp.append(tempo) for ket in brev.keys(): tempo3[ket] = tempo[ket] if isinstance(tempo2[ket], list): tempo2[ket] = UnNest(tempo2[ket]) else: tempo2[ket] = tempo[ket] clesRef2 = ['label', 'date', 'citations','family lenght', 'priority-active-indicator', 'IPCR4', 'IPCR7', 'portee', 'applicant', 'pays', 'inventeur', 'representative', 'prior', "Inventor-Country", "Applicant-Country"] tempoBrev = Decoupe(tempo2) for nb in tempoBrev: brev2 = tempoBrev[nb] brev2 = CleanPatent(brev2)
brev[cle] = UnNest(brev[cle]) brev = CleanPatent(brev) brev = CleanPatentOthers2(brev) ## tempo = brev # this one for DataTable tempo2 = dict() #the one for pitable PaysInv = [] #new field PaysApp = [] #tempo = CleanPatent(tempo) tempo2 = copy.deepcopy(tempo) #ugly tempo3 = dict() #what the problem LstExp.append(tempo) for ket in brev.keys(): tempo3[ket] = tempo[ket] if isinstance(tempo2[ket], list): tempo2[ket] = UnNest(tempo2[ket]) else: tempo2[ket] = tempo[ket] clesRef2 = [ 'label', 'date', 'citations', 'family lenght', 'priority-active-indicator', 'IPCR4', 'IPCR7', 'portee', 'applicant', 'pays', 'inventeur', 'representative', 'prior', "Inventor-Country", "Applicant-Country" ] tempoBrev = Decoupe(tempo2) for nb in tempoBrev:
print "Found ", data["number"], " patents! Formating to HMTL tables" LstExp = [] LstExp2 = [] for brev in LstBrevet: brev = CleanPatent(brev) tempo = dict() # this one for DataTable tempo2 = dict() #the one for pitable PaysInv= [] #new field PaysApp = [] # tempo = CleanPatent(brev) # brevet= SeparateCountryField(tempo) #cleaning classification cles = [key for key in brev.keys() if brev[key]==None or brev[key] == [u'None', None] or brev[key] == [None]] for cle in cles: if cle=='date': brev[cle] = unicode(datetime.date.today().year) elif cle=="dateDate": brev[cle] = datetime.date.today() else: brev[cle] = u'empty' for key in clesRef: if key =='inventeur' or key =='applicant': if isinstance(brev[key], list): tempo[key] = ' '.join(brev[key]).title().strip() else: tempo[key] = brev[key].title().strip() elif key =='titre': if isinstance(brev[key], list):