diff bin/plinks.py @ 6:0f494c76a887

refactor to address tarred-up pdfs
author Henry S. Thompson <ht@inf.ed.ac.uk>
date Sun, 23 Feb 2020 16:48:34 +0000
parents a28d731977da
children 25ca3505b4d7
line wrap: on
line diff
--- a/bin/plinks.py	Wed Feb 19 10:41:59 2020 +0000
+++ b/bin/plinks.py	Sun Feb 23 16:48:34 2020 +0000
@@ -7,20 +7,24 @@
   pdf=pdfx.PDFx(file)
   return pdf.get_references_as_dict()
 
-me=sys.argv[1]
-with open('/dev/shm/x/badpdfs_%s'%me,'w') as bf:
+tarnum=sys.argv[1]
+with open('badpdfs_%s'%tarnum,'w') as bf:
   for l in sys.stdin:
-    (fno,f)=l.rstrip().split()
+    (fno,f)=l.split()
     try:
       links=run(f)
       if bool(links) and (links.get('scrape',False) or
                           links.get('annot',False)):
-        with open('/dev/shm/x/links_%s'%fno,'w') as of:
+        with open('links_%s_%s'%(tarnum,fno),'w') as of:
           for k in links.keys():
             for l in links[k]:
               print("%s\t%s"%(k,l),file=of)
     except Exception as e:
-      print("%s\t%s"%(fno,e),file=bf)
+      if str(e)=='Unexpected EOF':
+        print("%s\t%s\t%s"%(tarnum,fno,e),file=bf)
+      else:
+        traceback.print_exc(file=bf)
+
     if (path.exists('/dev/shm/stopJob')):
-      print("Quiting early: %s %s"%(me,fno),file=sys.stderr)
+      print("Quiting early: %s %s"%(tarnum,fno),file=sys.stderr)
       exit(1)