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")
|
31
|
24
|
|
25 changed = False
|
30
|
26 for author in authors:
|
|
27
|
|
28 if author.text is not None:
|
|
29 splitted =author.text.split("\n")
|
|
30 txt = "; ".join(splitted)
|
|
31
|
31
|
32 if txt!=author.text:
|
|
33 author.text=txt
|
|
34 changed=True
|
30
|
35
|
31
|
36 return changed
|
36
|
37
|
|
38
|
|
39 def getDRIfromIndexMeta(fl):
|
|
40 try:
|
|
41 tree = etree.parse(fl)
|
|
42 except:
|
|
43 parseErrorFile.write("PARSE ERROR:"+fl+"\n")
|
|
44 return False
|
30
|
45
|
36
|
46 dris = tree.xpath("/resource/meta/dri[@type='mpiwg']")
|
|
47
|
|
48 if len(dris)==0:
|
|
49 return None
|
|
50 else:
|
|
51 return dris[0].text
|
|
52
|
|
53
|
5
|
54 def addPURL(fl,purl,test=False):
|
7
|
55 try:
|
|
56 tree = etree.parse(fl)
|
|
57 except:
|
|
58 parseErrorFile.write("PARSE ERROR:"+fl+"\n")
|
|
59 return False
|
30
|
60
|
5
|
61 dris = tree.xpath("/resource/meta/dri[@type='mpiwg']")
|
|
62
|
30
|
63
|
31
|
64 changed = correctAuthor(tree)
|
30
|
65
|
|
66
|
|
67
|
5
|
68 if len(dris)==0: # erzeuge neu
|
|
69 newDri = etree.Element("dri",type="mpiwg")
|
|
70 newDri.text=purl
|
|
71 metas=tree.xpath("/resource/meta")
|
|
72 if len(metas)==0:
|
7
|
73 parseErrorFile.write("no resource/meta: %s \n"%fl)
|
5
|
74 return False
|
|
75 else:
|
|
76 metas[0].append(newDri)
|
|
77 else:
|
|
78 dris[0].text=purl
|
22
|
79 alreadyExistsFile.write("%s \n"%fl)
|
31
|
80 if not changed: #nothing has to be done
|
|
81 return True
|
22
|
82
|
5
|
83 print etree.tostring(tree, pretty_print=True)
|
|
84
|
30
|
85
|
5
|
86 if not test:
|
6
|
87 try:
|
30
|
88
|
6
|
89 os.rename(fl, fl+"_mpiwg_dri")
|
|
90 out = etree.tostring(tree, encoding="UTF-8",xml_declaration=False)
|
|
91 fo = file(fl,"w")
|
|
92 fo.write(out)
|
|
93 fo.close
|
|
94 except:
|
30
|
95
|
|
96 print sys.exc_info()[0]
|
|
97 print sys.exc_info()[1]
|
6
|
98 errorFile.write(fl+"\n")
|
5
|
99 return True
|
|
100
|
6
|
101 def addDriToIndexMeta(path,delpath="",replacepath="",test=False):
|
5
|
102
|
8
|
103 md=manageIndexMetaPURLs.IndexMetaPURLManager()
|
5
|
104
|
|
105 for root, dirs, files in os.walk(path):
|
|
106
|
|
107
|
|
108 for name in files:
|
11
|
109 if name=="index.meta":
|
5
|
110 fl=join(root, name)
|
|
111 shortPath=re.sub("^"+delpath,replacepath,fl)
|
8
|
112 purl=md.getPurl(shortPath)
|
5
|
113
|
|
114 addPURL(fl,purl,test)
|
|
115
|
|
116 if 'pageimg' in dirs:
|
|
117 dirs.remove('pageimg') # don't visit pageimf
|
|
118 for dir in dirs:
|
|
119 if dir== "pageimg":
|
|
120 dirs.remove('pageimg')
|
|
121 if dir.startswith("."):
|
|
122 dirs.remove(dir)
|
|
123
|
|
124 if __name__ == '__main__':
|
30
|
125 addDriToIndexMeta("/mpiwg/online/permanent/vlp",delpath="/mpiwg/online",test=False)
|