5
|
1 '''
|
|
2 Created on 01.11.2012
|
|
3
|
|
4 @author: dwinter
|
|
5 '''
|
|
6 import os
|
|
7 import manageIndexMetaPURLs
|
|
8 import re
|
|
9 from lxml import etree
|
|
10
|
|
11 from os.path import join, getsize
|
|
12
|
6
|
13 errorFile = file("/tmp/addDRIErrors.txt","w")
|
5
|
14
|
|
15 def addPURL(fl,purl,test=False):
|
|
16 tree = etree.parse(fl)
|
|
17
|
|
18 dris = tree.xpath("/resource/meta/dri[@type='mpiwg']")
|
|
19
|
|
20 if len(dris)==0: # erzeuge neu
|
|
21 newDri = etree.Element("dri",type="mpiwg")
|
|
22 newDri.text=purl
|
|
23 metas=tree.xpath("/resource/meta")
|
|
24 if len(metas)==0:
|
|
25 print "no resource/meta in %s"%fl
|
|
26 return False
|
|
27 else:
|
|
28 metas[0].append(newDri)
|
|
29 else:
|
|
30 dris[0].text=purl
|
|
31
|
|
32 print etree.tostring(tree, pretty_print=True)
|
|
33
|
|
34 if not test:
|
6
|
35 try:
|
|
36 os.rename(fl, fl+"_mpiwg_dri")
|
|
37 out = etree.tostring(tree, encoding="UTF-8",xml_declaration=False)
|
|
38 fo = file(fl,"w")
|
|
39 fo.write(out)
|
|
40 fo.close
|
|
41 except:
|
|
42 errorFile.write(fl+"\n")
|
5
|
43 return True
|
|
44
|
6
|
45 def addDriToIndexMeta(path,delpath="",replacepath="",test=False):
|
5
|
46
|
|
47
|
|
48
|
|
49 for root, dirs, files in os.walk(path):
|
|
50
|
|
51
|
|
52 for name in files:
|
|
53 if name.endswith(".meta"):
|
|
54 fl=join(root, name)
|
|
55 shortPath=re.sub("^"+delpath,replacepath,fl)
|
|
56 purl=manageIndexMetaPURLs.IndexMetaPURLManager().getPurl(shortPath)
|
|
57
|
|
58 addPURL(fl,purl,test)
|
|
59
|
|
60 if 'pageimg' in dirs:
|
|
61 dirs.remove('pageimg') # don't visit pageimf
|
|
62 for dir in dirs:
|
|
63 if dir== "pageimg":
|
|
64 dirs.remove('pageimg')
|
|
65 if dir.startswith("."):
|
|
66 dirs.remove(dir)
|
|
67
|
|
68 if __name__ == '__main__':
|
|
69 addDriToIndexMeta("/mpiwg/online/permanent/annalen",delpath="/mpiwg/online") |