5
|
1 '''
|
|
2 Created on 01.11.2012
|
|
3
|
|
4 @author: dwinter
|
|
5 '''
|
|
6 import os
|
10
|
7 import managePurls.manageIndexMetaPURLs as manageIndexMetaPURLs
|
5
|
8 import re
|
|
9 from lxml import etree
|
30
|
10 import sys
|
5
|
11
|
|
12 from os.path import join, getsize
|
|
13
|
6
|
14 errorFile = file("/tmp/addDRIErrors.txt","w")
|
7
|
15 parseErrorFile = file("/tmp/addDRIParseErrors.txt","w")
|
22
|
16 alreadyExistsFile = file("/tmp/addDRIalreadyExists.txt","w")
|
5
|
17
|
30
|
18
|
|
19 def correctAuthor(tree):
|
|
20 """ersetzt in den autor felder "\r" durch ;"""
|
|
21
|
|
22
|
|
23 authors = tree.xpath("/resource/meta/bib/author")
|
|
24 for author in authors:
|
|
25
|
|
26 if author.text is not None:
|
|
27 splitted =author.text.split("\n")
|
|
28 txt = "; ".join(splitted)
|
|
29
|
|
30 author.text=txt
|
|
31
|
|
32
|
|
33
|
5
|
34 def addPURL(fl,purl,test=False):
|
7
|
35 try:
|
|
36 tree = etree.parse(fl)
|
|
37 except:
|
|
38 parseErrorFile.write("PARSE ERROR:"+fl+"\n")
|
|
39 return False
|
30
|
40
|
5
|
41 dris = tree.xpath("/resource/meta/dri[@type='mpiwg']")
|
|
42
|
30
|
43
|
|
44 correctAuthor(tree)
|
|
45
|
|
46
|
|
47
|
5
|
48 if len(dris)==0: # erzeuge neu
|
|
49 newDri = etree.Element("dri",type="mpiwg")
|
|
50 newDri.text=purl
|
|
51 metas=tree.xpath("/resource/meta")
|
|
52 if len(metas)==0:
|
7
|
53 parseErrorFile.write("no resource/meta: %s \n"%fl)
|
5
|
54 return False
|
|
55 else:
|
|
56 metas[0].append(newDri)
|
|
57 else:
|
|
58 dris[0].text=purl
|
22
|
59 alreadyExistsFile.write("%s \n"%fl)
|
30
|
60 #return True
|
22
|
61
|
5
|
62 print etree.tostring(tree, pretty_print=True)
|
|
63
|
30
|
64
|
5
|
65 if not test:
|
6
|
66 try:
|
30
|
67
|
6
|
68 os.rename(fl, fl+"_mpiwg_dri")
|
|
69 out = etree.tostring(tree, encoding="UTF-8",xml_declaration=False)
|
|
70 fo = file(fl,"w")
|
|
71 fo.write(out)
|
|
72 fo.close
|
|
73 except:
|
30
|
74
|
|
75 print sys.exc_info()[0]
|
|
76 print sys.exc_info()[1]
|
6
|
77 errorFile.write(fl+"\n")
|
5
|
78 return True
|
|
79
|
6
|
80 def addDriToIndexMeta(path,delpath="",replacepath="",test=False):
|
5
|
81
|
8
|
82 md=manageIndexMetaPURLs.IndexMetaPURLManager()
|
5
|
83
|
|
84 for root, dirs, files in os.walk(path):
|
|
85
|
|
86
|
|
87 for name in files:
|
11
|
88 if name=="index.meta":
|
5
|
89 fl=join(root, name)
|
|
90 shortPath=re.sub("^"+delpath,replacepath,fl)
|
8
|
91 purl=md.getPurl(shortPath)
|
5
|
92
|
|
93 addPURL(fl,purl,test)
|
|
94
|
|
95 if 'pageimg' in dirs:
|
|
96 dirs.remove('pageimg') # don't visit pageimf
|
|
97 for dir in dirs:
|
|
98 if dir== "pageimg":
|
|
99 dirs.remove('pageimg')
|
|
100 if dir.startswith("."):
|
|
101 dirs.remove(dir)
|
|
102
|
|
103 if __name__ == '__main__':
|
30
|
104 addDriToIndexMeta("/mpiwg/online/permanent/vlp",delpath="/mpiwg/online",test=False)
|