Mercurial > hg > drupalISMI
comparison importFromOpenMind/converter/addDRI.py @ 0:124ef8f3b22d
initial
| author | Dirk Wintergruen <dwinter@mpiwg-berlin.mpg.de> |
|---|---|
| date | Fri, 27 Mar 2015 19:21:42 +0100 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:124ef8f3b22d |
|---|---|
| 1 from urllib import request | |
| 2 | |
| 3 class converter: | |
| 4 | |
| 5 out ="/var/tmp/out.csv" | |
| 6 url = "http://md.mpiwg-berlin.mpg.de/purls/search?q=" # /permanent/library/W028ATN3 | |
| 7 def read(self,fn): | |
| 8 | |
| 9 of = open(self.out,"w") | |
| 10 | |
| 11 fl = open(fn,"r") | |
| 12 | |
| 13 for line in fl.readlines(): | |
| 14 line = line.replace("\n","") | |
| 15 line = line.replace('"','') | |
| 16 splitted = line.split(",") | |
| 17 | |
| 18 path = splitted[0].replace('"','') | |
| 19 path = path.replace("/mpiwg/online","") | |
| 20 | |
| 21 qr = self.url + path | |
| 22 | |
| 23 g = request.urlopen(qr, data=bytearray(path,"utf-8")) | |
| 24 | |
| 25 p = g.read().decode('utf-8') | |
| 26 | |
| 27 spl = p.split("\t") | |
| 28 | |
| 29 if len(spl) > 1: | |
| 30 dri = spl[1] | |
| 31 | |
| 32 splitted.append(dri) | |
| 33 | |
| 34 print (splitted) | |
| 35 | |
| 36 of.write(",".join(splitted)+"\n") | |
| 37 | |
| 38 | |
| 39 of.close() | |
| 40 | |
| 41 | |
| 42 | |
| 43 | |
| 44 cv =converter() | |
| 45 | |
| 46 cv.read("/var/tmp/archive.csv") | |
| 47 | |
| 48 | |
| 49 |