dateMini = datetime.date(3000,1,1)
dateMaxi = datetime.date(1000,1,1)
NeededInfo .extend(mixNet) # list of needed field for building the net
G1 = nx.DiGraph() # dynamic network for Gephi
attr_dict = dict() # attributes for the net
G2 = nx.DiGraph() # flat net for gexf.js may be it is possible to use previous instead of this one...
print network, ": loading data with ", " and ".join(mixNet), " fields."
with open (BiblioPath+'//'+ndf, 'r') as fic:
DataBrevet = pickle.load(fic)
for brev in DataBrevet["brevets"]:
#tempo = pickle.load(fic) # we only memorize needed nfo
pat = OrderedDict ()
for key in NeededInfo:
if isinstance(brev[key], list):
brev[key]= flatten(brev[key])
brev[key]= [cont for cont in brev[key] if (cont !='empty' or cont != 'Empty' or cont !='')]
elif isinstance (brev[key], unicode) or isinstance (brev[key], str):
brev[key].replace('empty', '')
pat[key] = brev[key]
for flatPat in DecoupeOnTheFly(pat, []):
if flatPat not in ListeBrevet:
ListeBrevet.append(flatPat)
if pat['label'] not in Patents:
Patents.add(pat['label'])
for lab in Patents:
temp = []
for bre in [brev for brev in ListeBrevet if brev['label']==lab]:
for cat in mixNet:
tempFiltered = []
LabList = [pat['label'] for pat in temp]
for pat in LabList:
tempoPat = [
patent for patent in temp if patent['label'] == pat
] # fusionning several patents wwith same label
# OPS model seems to save one entry for several status documents...
# in P2N model, label is unique key... so properties are lists.. this is the jobs of update function hereafter
tempoRar = dict()
for pate in tempoPat:
tempoRar = Update(pate, tempoRar)
for clef in tempoRar.keys():
if isinstance(tempoRar[clef], list):
tempoRar[clef] = flatten(tempoRar[clef])
tempo = []
for contenu in tempoRar[clef]:
if contenu is not None:
tempo.append(contenu)
else:
if '' not in tempo and len(
tempo) == 0:
tempo.append('')
tempoRar[clef] = tempo
else:
pass #should be good here
if pat not in YetIn:
tempFiltered.append(dictCleaner(tempoRar))
YetIn.append(pat)
else:
print "please use Comptatibilizer"
print "Nice, ", len(
DataBrevet["brevets"]
), " patents found. Pre-formating ", sys.argv[1], " net."
for brev in DataBrevet["brevets"]:
#tempo = pickle.load(fic) # we only memorize needed nfo
pat = OrderedDict()
if "date" not in brev.keys():
brev['date'] = '1-1-1'
if isinstance(brev['label'], list):
brev['label'] = brev['label'][0]
for key in NeededInfo:
if key.count('nice') > 0 and isinstance(brev[key], list):
brev[key] = flatten(brev[key])
machin = []
for truc in brev[key]:
Tt = Cleaning(truc)
if Tt is not None and len(Tt) > 0:
machin.append(Tt)
if len(machin) > 0:
pat[key] = machin #can you do more ugly ?
else:
pat[key] = ''
elif isinstance(brev[key], list):
pat[key] = flatten(brev[key])
if key.count('Date') == 0:
pat[key] = [
cont for cont in brev[key]
dateMaxi = datetime.date(1000, 1, 1)
NeededInfo.extend(mixNet) # list of needed field for building the net
G1 = nx.DiGraph() # dynamic network for Gephi
attr_dict = dict() # attributes for the net
G2 = nx.DiGraph(
) # flat net for gexf.js may be it is possible to use previous instead of this one...
print network, ": loading data with ", " and ".join(mixNet), " fields."
with open(BiblioPath + '//' + ndf, 'r') as fic:
DataBrevet = pickle.load(fic)
for brev in DataBrevet["brevets"]:
#tempo = pickle.load(fic) # we only memorize needed nfo
pat = OrderedDict()
for key in NeededInfo:
if isinstance(brev[key], list):
brev[key] = flatten(brev[key])
brev[key] = [
cont for cont in brev[key]
if (cont != 'empty' or cont != 'Empty' or cont != '')
]
elif isinstance(brev[key], unicode) or isinstance(
brev[key], str):
brev[key].replace('empty', '')
pat[key] = brev[key]
for flatPat in DecoupeOnTheFly(pat, []):
if flatPat not in ListeBrevet:
ListeBrevet.append(flatPat)
if pat['label'] not in Patents:
Patents.add(pat['label'])