annotate addDriToIndexMeta.py @ 7:78dd28ade713

error handling
author dwinter
date Fri, 02 Nov 2012 09:25:11 +0100
parents f0417a01690a
children 733d43b30a82
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
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
7 import manageIndexMetaPURLs
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
8 import re
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
9 from lxml import etree
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
10
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
11 from os.path import join, getsize
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
12
6
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
13 errorFile = file("/tmp/addDRIErrors.txt","w")
7
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
14 parseErrorFile = file("/tmp/addDRIParseErrors.txt","w")
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
15
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
16 def addPURL(fl,purl,test=False):
7
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
17 try:
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
18 tree = etree.parse(fl)
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
19 except:
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
20 parseErrorFile.write("PARSE ERROR:"+fl+"\n")
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
21 return False
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
22 dris = tree.xpath("/resource/meta/dri[@type='mpiwg']")
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
23
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
24 if len(dris)==0: # erzeuge neu
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
25 newDri = etree.Element("dri",type="mpiwg")
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
26 newDri.text=purl
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
27 metas=tree.xpath("/resource/meta")
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
28 if len(metas)==0:
7
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
29 parseErrorFile.write("no resource/meta: %s \n"%fl)
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
30 return False
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
31 else:
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
32 metas[0].append(newDri)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
33 else:
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
34 dris[0].text=purl
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
35
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
36 print etree.tostring(tree, pretty_print=True)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
37
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
38 if not test:
6
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
39 try:
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
40 os.rename(fl, fl+"_mpiwg_dri")
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
41 out = etree.tostring(tree, encoding="UTF-8",xml_declaration=False)
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
42 fo = file(fl,"w")
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
43 fo.write(out)
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
44 fo.close
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
45 except:
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
46 errorFile.write(fl+"\n")
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
47 return True
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
48
6
f0417a01690a error handling if file renaming not possible
dwinter
parents: 5
diff changeset
49 def addDriToIndexMeta(path,delpath="",replacepath="",test=False):
5
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
50
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
51
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
52
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
53 for root, dirs, files in os.walk(path):
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
54
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
55
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
56 for name in files:
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
57 if name.endswith(".meta"):
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
58 fl=join(root, name)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
59 shortPath=re.sub("^"+delpath,replacepath,fl)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
60 purl=manageIndexMetaPURLs.IndexMetaPURLManager().getPurl(shortPath)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
61
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
62 addPURL(fl,purl,test)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
63
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
64 if 'pageimg' in dirs:
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
65 dirs.remove('pageimg') # don't visit pageimf
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
66 for dir in dirs:
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
67 if dir== "pageimg":
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
68 dirs.remove('pageimg')
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
69 if dir.startswith("."):
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
70 dirs.remove(dir)
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
71
3ebe37d81071 addDri added
dwinter
parents:
diff changeset
72 if __name__ == '__main__':
7
78dd28ade713 error handling
dwinter
parents: 6
diff changeset
73 addDriToIndexMeta("/mpiwg/online/",delpath="/mpiwg/online",test=False)