annotate addDriToIndexMeta.py @ 30:bcd8076ff7ec

random selection of entries bug fixes
author dwinter
date Wed, 05 Jun 2013 17:37:09 +0200
parents f748e2b684c9
children 0190f49bce88
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
1 '''
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
2 Created on 01.11.2012
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
3
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
4 @author: dwinter
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
5 '''
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
6 import os
10
1b2d74f94ca8 repackaging
dwinter
parents: 8
diff changeset
7 import managePurls.manageIndexMetaPURLs as manageIndexMetaPURLs
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
8 import re
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
9 from lxml import etree
30
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
10 import sys
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
11
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
12 from os.path import join, getsize
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
13
6
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
14 errorFile = file("/tmp/addDRIErrors.txt","w")
7
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
15 parseErrorFile = file("/tmp/addDRIParseErrors.txt","w")
22
f748e2b684c9 bug in xml output fixed
dwinter
parents: 11
diff changeset
16 alreadyExistsFile = file("/tmp/addDRIalreadyExists.txt","w")
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
17
30
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
18
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
19 def correctAuthor(tree):
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
20 """ersetzt in den autor felder "\r" durch ;"""
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
21
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
22
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
23 authors = tree.xpath("/resource/meta/bib/author")
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
24 for author in authors:
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
25
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
26 if author.text is not None:
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
27 splitted =author.text.split("\n")
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
28 txt = "; ".join(splitted)
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
29
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
30 author.text=txt
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
31
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
32
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
33
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
34 def addPURL(fl,purl,test=False):
7
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
35 try:
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
36 tree = etree.parse(fl)
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
37 except:
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
38 parseErrorFile.write("PARSE ERROR:"+fl+"\n")
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
39 return False
30
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
40
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
41 dris = tree.xpath("/resource/meta/dri[@type='mpiwg']")
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
42
30
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
43
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
44 correctAuthor(tree)
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
45
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
46
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
47
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
48 if len(dris)==0: # erzeuge neu
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
49 newDri = etree.Element("dri",type="mpiwg")
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
50 newDri.text=purl
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
51 metas=tree.xpath("/resource/meta")
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
52 if len(metas)==0:
7
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
53 parseErrorFile.write("no resource/meta: %s \n"%fl)
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
54 return False
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
55 else:
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
56 metas[0].append(newDri)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
57 else:
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
58 dris[0].text=purl
22
f748e2b684c9 bug in xml output fixed
dwinter
parents: 11
diff changeset
59 alreadyExistsFile.write("%s \n"%fl)
30
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
60 #return True
22
f748e2b684c9 bug in xml output fixed
dwinter
parents: 11
diff changeset
61
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
62 print etree.tostring(tree, pretty_print=True)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
63
30
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
64
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
65 if not test:
6
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
66 try:
30
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
67
6
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
68 os.rename(fl, fl+"_mpiwg_dri")
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
69 out = etree.tostring(tree, encoding="UTF-8",xml_declaration=False)
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
70 fo = file(fl,"w")
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
71 fo.write(out)
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
72 fo.close
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
73 except:
30
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
74
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
75 print sys.exc_info()[0]
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
76 print sys.exc_info()[1]
6
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
77 errorFile.write(fl+"\n")
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
78 return True
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
79
6
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
80 def addDriToIndexMeta(path,delpath="",replacepath="",test=False):
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
81
8
733d43b30a82 connection handling changed
dwinter
parents: 7
diff changeset
82 md=manageIndexMetaPURLs.IndexMetaPURLManager()
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
83
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
84 for root, dirs, files in os.walk(path):
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
85
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
86
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
87 for name in files:
11
fad73212354b wsgi addedx
dwinter
parents: 10
diff changeset
88 if name=="index.meta":
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
89 fl=join(root, name)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
90 shortPath=re.sub("^"+delpath,replacepath,fl)
8
733d43b30a82 connection handling changed
dwinter
parents: 7
diff changeset
91 purl=md.getPurl(shortPath)
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
92
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
93 addPURL(fl,purl,test)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
94
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
95 if 'pageimg' in dirs:
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
96 dirs.remove('pageimg') # don't visit pageimf
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
97 for dir in dirs:
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
98 if dir== "pageimg":
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
99 dirs.remove('pageimg')
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
100 if dir.startswith("."):
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
101 dirs.remove(dir)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
102
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
103 if __name__ == '__main__':
30
bcd8076ff7ec random selection of entries
dwinter
parents: 22
diff changeset
104 addDriToIndexMeta("/mpiwg/online/permanent/vlp",delpath="/mpiwg/online",test=False)