Mercurial > hg > cc > cirrus_home

annotate bin/plinks.py @ 9:7a93e190c74d

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
logging tweaks, preparing for timeout on problem pdfs
author Henry S. Thompson <ht@inf.ed.ac.uk>
date Tue, 25 Feb 2020 10:34:41 +0000
parents 25ca3505b4d7
children a33db8e3f51c
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
65a56c0d1c1f bolting the barn door...
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
1 #!/lustre/sw/miniconda3/bin/python3
7
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
2 import sys,pdfx,traceback,os
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
3 from datetime import datetime
1
a4b0359456bc switch to file loop inside python, assume file index integer in pipe as well as filename, check /dev/shm/stopJob
Henry Thompson <ht@markup.co.uk>
parents: 0
diff changeset
4 from os import path
a4b0359456bc switch to file loop inside python, assume file index integer in pipe as well as filename, check /dev/shm/stopJob
Henry Thompson <ht@markup.co.uk>
parents: 0
diff changeset
5
a4b0359456bc switch to file loop inside python, assume file index integer in pipe as well as filename, check /dev/shm/stopJob
Henry Thompson <ht@markup.co.uk>
parents: 0
diff changeset
6 def run(file):
0
65a56c0d1c1f bolting the barn door...
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
7 global pdf
9
7a93e190c74d logging tweaks, preparing for timeout on problem pdfs
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 7
diff changeset
8 pdf=pdfx.PDFx(file,limit=30)
1
a4b0359456bc switch to file loop inside python, assume file index integer in pipe as well as filename, check /dev/shm/stopJob
Henry Thompson <ht@markup.co.uk>
parents: 0
diff changeset
9 return pdf.get_references_as_dict()
0
65a56c0d1c1f bolting the barn door...
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
10
6
0f494c76a887 refactor to address tarred-up pdfs
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
11 tarnum=sys.argv[1]
7
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
12 print(tarnum, sys.argv, os.getcwd(),file=sys.stderr)
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
13 gf=0
6
0f494c76a887 refactor to address tarred-up pdfs
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
14 with open('badpdfs_%s'%tarnum,'w') as bf:
1
a4b0359456bc switch to file loop inside python, assume file index integer in pipe as well as filename, check /dev/shm/stopJob
Henry Thompson <ht@markup.co.uk>
parents: 0
diff changeset
15 for l in sys.stdin:
6
0f494c76a887 refactor to address tarred-up pdfs
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
16 (fno,f)=l.split()
1
a4b0359456bc switch to file loop inside python, assume file index integer in pipe as well as filename, check /dev/shm/stopJob
Henry Thompson <ht@markup.co.uk>
parents: 0
diff changeset
17 try:
2
Henry Thompson <ht@markup.co.uk>
parents: 1
diff changeset
18 links=run(f)
4
462179da7dc2 try harder not to write empty links files
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 2
diff changeset
19 if bool(links) and (links.get('scrape',False) or
462179da7dc2 try harder not to write empty links files
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 2
diff changeset
20 links.get('annot',False)):
7
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
21 gf+=1
6
0f494c76a887 refactor to address tarred-up pdfs
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
22 with open('links_%s_%s'%(tarnum,fno),'w') as of:
3
cbd13beb0922 only create links file if there are some
Henry Thompson <ht@markup.co.uk>
parents: 2
diff changeset
23 for k in links.keys():
cbd13beb0922 only create links file if there are some
Henry Thompson <ht@markup.co.uk>
parents: 2
diff changeset
24 for l in links[k]:
cbd13beb0922 only create links file if there are some
Henry Thompson <ht@markup.co.uk>
parents: 2
diff changeset
25 print("%s\t%s"%(k,l),file=of)
1
a4b0359456bc switch to file loop inside python, assume file index integer in pipe as well as filename, check /dev/shm/stopJob
Henry Thompson <ht@markup.co.uk>
parents: 0
diff changeset
26 except Exception as e:
6
0f494c76a887 refactor to address tarred-up pdfs
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
27 if str(e)=='Unexpected EOF':
7
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
28 print("%s:\t%s\t%s\t%s"%(datetime.now().isoformat(),
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
29 tarnum,fno,e),file=bf)
6
0f494c76a887 refactor to address tarred-up pdfs
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
30 else:
7
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
31 print("%s: "%(datetime.now().isoformat()),end='',file=bf)
6
0f494c76a887 refactor to address tarred-up pdfs
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
32 traceback.print_exc(file=bf)
0f494c76a887 refactor to address tarred-up pdfs
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 5
diff changeset
33
1
a4b0359456bc switch to file loop inside python, assume file index integer in pipe as well as filename, check /dev/shm/stopJob
Henry Thompson <ht@markup.co.uk>
parents: 0
diff changeset
34 if (path.exists('/dev/shm/stopJob')):
7
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
35 print("%s: Quiting early: %s %s"%(datetime.now().isoformat(),tarnum,fno),
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
36 file=sys.stderr)
1
a4b0359456bc switch to file loop inside python, assume file index integer in pipe as well as filename, check /dev/shm/stopJob
Henry Thompson <ht@markup.co.uk>
parents: 0
diff changeset
37 exit(1)
7
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
38 now=datetime.now().isoformat()
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
39 print('%s: exiting from %s having found %s files with links out of %s'%(now,
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
40 tarnum,
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
41 gf,
25ca3505b4d7 more logging
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 6
diff changeset
42 fno))