def main():
fails = open('bulkpippy.fails', 'a')
for line in sys.stdin:
(doc1, doc2) = line.strip().split('\t')
print "[%d] %s, %s" % (os.getpid(), doc1, doc2)
try:
d1 = getDoc(doc1)
except:
print "!!!!PIPPI ERROR: load doc", doc1
sys.stderr.write("%s\t%s\n" % (doc1, doc2))
traceback.print_exc(file=sys.stderr)
fails.write("%s\t%s\n" % (doc1, doc2))
continue
if doc2 in d1.pippiDocs:
print "already done: %s %s" % (doc1, doc2)
continue
try:
d2 = getDoc(doc2)
except:
print "!!!!PIPPI ERROR: load doc", doc2
sys.stderr.write("%s\t%s\n" % (doc1, doc2))
traceback.print_exc(file=sys.stderr)
fails.write("%s\t%s\n" % (doc1, doc2))
continue
try:
lcs.pippi(d1, d2)
except:
print "!!!!PIPPI ERROR: lcs", doc1, doc2
sys.stderr.write("%s\t%s\n" % (doc1, doc2))
traceback.print_exc(file=sys.stderr)
fails.write("%s\t%s\n" % (doc1, doc2))
tfidf.save()
fails.close()
def main():
fails=open('bulkpippy.fails','a')
for line in sys.stdin:
(doc1,doc2)=line.strip().split('\t')
print "[%d] %s, %s" % (os.getpid(),doc1,doc2)
try:
d1=getDoc(doc1)
except:
print "!!!!PIPPI ERROR: load doc",doc1
sys.stderr.write("%s\t%s\n" % (doc1,doc2))
traceback.print_exc(file=sys.stderr)
fails.write("%s\t%s\n" % (doc1,doc2))
continue
if doc2 in d1.pippiDocs:
print "already done: %s %s" % (doc1,doc2)
continue
try:
d2=getDoc(doc2)
except:
print "!!!!PIPPI ERROR: load doc",doc2
sys.stderr.write("%s\t%s\n" % (doc1,doc2))
traceback.print_exc(file=sys.stderr)
fails.write("%s\t%s\n" % (doc1,doc2))
continue
try:
lcs.pippi(d1,d2)
except:
print "!!!!PIPPI ERROR: lcs",doc1,doc2
sys.stderr.write("%s\t%s\n" % (doc1,doc2))
traceback.print_exc(file=sys.stderr)
fails.write("%s\t%s\n" % (doc1,doc2))
tfidf.save()
fails.close()
def job(request):
d1=request.GET.get('d1','')
d2=request.GET.get('d2','')
try:
D1=Doc(docid=d1)
except:
return render_to_response('error.html', {'error': 'wrong document: "%s"!' % d1}, context_instance=RequestContext(request))
try:
D2=Doc(docid=d2)
except:
return render_to_response('error.html', {'error': 'specify document: "%s"!' % d2}, context_instance=RequestContext(request))
lcs.pippi(D1,D2)
return HttpResponseRedirect('/doc/%s' % (d1))
def jobs(request):
rdoc=request.POST.get('doc')
try:
refdoc=Doc(oid=ObjectId(rdoc))
except:
return render_to_response('error.html', {'error': 'wrong document: "%s"!' % rdoc}, context_instance=RequestContext(request))
failed=[]
for doc in request.POST.getlist('ids'):
try:
od=Doc(oid=ObjectId(doc))
except:
failed.append(doc)
lcs.pippi(refdoc,od)
return HttpResponseRedirect('/doc/%s' % (refdoc.docid))
def job(request):
d1 = request.GET.get('d1', '')
d2 = request.GET.get('d2', '')
try:
D1 = Doc(docid=d1, owner=request.user)
except:
return render_to_response('error.html',
{'error': 'wrong document: "%s"!' % d1},
context_instance=RequestContext(request))
try:
D2 = Doc(docid=d2, owner=request.user)
except:
return render_to_response('error.html',
{'error': 'specify document: "%s"!' % d2},
context_instance=RequestContext(request))
lcs.pippi(D1, D2)
return HttpResponseRedirect('/doc/%s' % (d1))
def jobs(request):
rdoc = request.GET.get('doc')
try:
refdoc = Doc(oid=ObjectId(rdoc))
except:
return render_to_response('error.html',
{'error': 'wrong document: "%s"!' % rdoc},
context_instance=RequestContext(request))
failed = []
for doc in request.GET.getlist('ids'):
if doc == rdoc: continue
try:
od = Doc(oid=ObjectId(doc))
except:
failed.append(doc)
continue
lcs.pippi(refdoc, od)
return HttpResponseRedirect('/doc/%s' % (refdoc.docid))