def search(request):
requestDict = request.POST
template = loader.get_template('search.html')
if 'question' not in requestDict:
return HttpResponse(template.render(
RequestContext(request,
{'title': 'No question',
'question': 'Please write the question URL first'
})))
questionURL = requestDict['question']
parsed = tparser.parse(questionURL)
if 'drupal' in questionURL:
posts = choose_answers('ent-sam-dru', parsed)
bestpost = choose_answer('ent-sam-dru', parsed)
elif 'ubuntu' in questionURL:
posts = choose_answers('ent-sam-ubu', parsed)
bestpost = choose_answer('ent-sam-ubu', parsed)
else:
posts = choose_answers('ent-sam-mix', parsed)
bestpost = choose_answer('ent-sam-mix', parsed)
#pdb.set_trace()
#Do some NLP magic here...
#vimpdb.set_trace()
#posts = nlpsort.magic(posts)
#response = ['good response', 'not so good response']
#response = test.post(0)
context = RequestContext(request,
{'title': 'Natural language search engine',
'question': questionURL,
'response': bestpost,
'others': posts,
'link': bestpost.link})
return HttpResponse(template.render(context))
def main(argv):
import wedt.training
import wedt.features
start_time = time.time()
if argv[1] == "gather":
if len(argv) < 4:
print_help()
return
for filename in argv[3:]:
if not os.path.exists(os.path.join(
wedt.training.train_path, argv[2])):
os.makedirs(os.path.join(
wedt.training.train_path, argv[2]))
with open(os.path.join(
'wedt', 'site_lists', filename), 'r') as file:
addresslist = file.read().splitlines()
wedt.training.gather_topics(addresslist, argv[2])
elif argv[1] == "train":
if len(argv) < 6:
print_help()
return
try:
feature = getattr(wedt.features, argv[4])
threshold = argv[6] if len(argv) > 6 else None
wedt.training.train_classifier(
argv[3], argv[5], feature, argv[2], threshold)
except AttributeError:
print "Wrong feature function name - see wedt/features.py"
elif argv[1] == "classify":
if len(argv) < 4:
print_help()
return
from wedt.tparser import parse
from wedt.nlpclassify import choose_answers
answers = choose_answers(argv[2], parse(argv[3]))
for answer in answers:
print answer
elif argv[1] == "best":
if len(argv) < 4:
print_help()
return
from wedt.tparser import parse
from wedt.nlpclassify import choose_answer
answer = choose_answer(argv[2], parse(argv[3]))
print answer
print answer.link
elif argv[1] == "check":
if len(argv) < 4:
print_help()
return
from wedt.nlpclassify import check_classifier
print "Accuracy: " + str(check_classifier(argv[2], argv[3]))
elif argv[1] == "fullcheck":
if len(argv) < 4:
print_help()
return
from wedt.report import Report
report = Report.generate(argv[2], argv[3])
i = 0
print "\\begin{tabular}{|l|c|c|c|}"
print "\\hline"
print "\\multicolumn{{4}}{{|c|}}{{Klasyfikator: {0}}}\\\\"\
.format(argv[2])
print "\\hline"
print "Nr. testu & Dokładność & Poprawna & Najlepsza\\\\"
print "\\hline"
for best_hit, c_accuracy, c_recall in zip(
report.best_hits, report.c_accuracy, report.c_recall):
i = i + 1
print "{0} & {1:.2f} & {2} & {3:.2f} \\\\".format(
i, c_accuracy, c_recall, best_hit)
#print "\nTopic {0}:".format(i)
#print "Classification accuracy: {0}".format(c_accuracy)
#print "Properly accepted: {0}".format(c_recall)
#print "Best answer correct: {0}".format(best_hit)
print "\\hline"
print "Razem & {0:.2f} & {1:.2f} & {2:.2f} \\\\".format(
report.total_accuracy, report.total_recall, report.best_accuracy)
print "\\hline"
print "\\end{tabular}"
else:
print "Unknown command."
print_help()
end_time = time.time()
if show_time:
print "\nTime elapsed: " + str(end_time - start_time)
