def index(request):
corpora = Corpus.objects.all()
langs = Lang.objects.all()
context = {'my_corpora' : Corpus.objects.filter(user=request.user),
'other_corpora' : Corpus.objects.exclude(user=request.user),
'langs' : langs}
if request.method == "POST":
form = CorpusForm(request.POST, request.FILES)
if form.is_valid():
path_type = form.cleaned_data["path_type"]
if (path_type == "file_upload"):
name = form.cleaned_data["name"]
file = form.cleaned_data["upload_path"]
path = files.write_tmp_file(file, name)
else:
location = form.cleaned_data["location"].strip("/")
file_name = form.cleaned_data["file_name"].strip("/")
path = os.path.join(location, file_name)
sample = files.corpus_sample(path, 15)
context = {'form' : form, 'sample' : sample }
return render(request, 'corpora/confirm_upload.html', context)
else:
context["form"] = form
else:
context["form"] = CorpusForm()
return render(request, 'corpora/index.html', context)
def confirm(request):
form = CorpusForm(request.POST, request.FILES)
# so it passes validation
form.fields["upload_path"].initial = True
if form.is_valid():
c = Corpus(user = request.user, date_added = timezone.now())
c.name = form.cleaned_data["name"]
c.description = form.cleaned_data["description"]
c.process_steps = form.cleaned_data["process_steps"]
c.metadata_url = form.cleaned_data["metadata_url"]
lang_id = form.cleaned_data["lang_select"]
if(lang_id == "new"):
new_lang = form.cleaned_data["new_lang"]
c.lang = Lang.objects.create(name=new_lang)
else:
c.lang = Lang.objects.get(id=lang_id)
#To take value of the label field from form to the database label field
c.label = c.name + " " + str(c.lang) + " " + str(c.user) + " " + form.cleaned_data["label"]
path_type = form.cleaned_data["path_type"]
if (path_type == "file_upload"):
(c.location, c.file_name) = files.save_tmp_file(c.name)
else:
c.location = form.cleaned_data["location"].strip("/")
c.file_name = form.cleaned_data["file_name"].strip("/")
# To find out the size of the corpora
#name2 = c.name + ".txt"
#c.size = c.current_size()
#To find out the word count of the corpora
'''count = 0
with open(files.corpus_path(c),'r') as f:
for line in f:
for word in line.split():
count+=1
c.count = count'''
#c.count = c.wc()
c.save()
return HttpResponseRedirect(reverse("corpora:corpus", kwargs={"corpus_id" : c.id}))
# we should ideally never get here, so just redirect to the main index
# in case something unexpected went wrong
else:
return HttpResponseRedirect(reverse("corpora:index"))
def index(request):
corpora = Corpus.objects.all()
langs = Lang.objects.all()
context = {'my_corpora' : Corpus.objects.filter(user=request.user),
'other_corpora' : Corpus.objects.exclude(user=request.user),
'langs' : langs}
if request.method == 'GET':
if 'q' in request.GET :
query = request.GET['q']
context['my_corpora'] = context['my_corpora'].filter(label__contains=query)
context['other_corpora'] = context['other_corpora'].filter(label__contains=query)
if request.method == "POST":
form = CorpusForm(request.POST, request.FILES)
if form.is_valid():
path_type = form.cleaned_data["path_type"]
if (path_type == "file_upload"):
name = form.cleaned_data["name"]
file = form.cleaned_data["upload_path"]
path = files.write_tmp_file(file, name)
else:
location = form.cleaned_data["location"].strip("/")
file_name = form.cleaned_data["file_name"].strip("/")
path = os.path.join(location, file_name)
sample = files.corpus_sample(path, 15)
size = files.file_size(path)
wc = files.wc(path)
context = {'form' : form, 'sample' : sample, 'size' : size, 'wc' : wc}
return render(request, 'corpora/confirm_upload.html', context)
else:
context["form"] = form
else:
context["form"] = CorpusForm()
return render(request, 'corpora/index.html', context)
def confirm(request):
form = CorpusForm(request.POST, request.FILES)
# so it passes validation
form.fields["upload_path"].initial = True
if form.is_valid():
c = Corpus(user = request.user, date_added = timezone.now())
c.name = form.cleaned_data["name"]
c.description = form.cleaned_data["description"]
c.process_steps = form.cleaned_data["process_steps"]
c.metadata_url = form.cleaned_data["metadata_url"]
lang_id = form.cleaned_data["lang_select"]
if(lang_id == "new"):
new_lang = form.cleaned_data["new_lang"]
c.lang = Lang.objects.create(name=new_lang)
else:
c.lang = Lang.objects.get(id=lang_id)
path_type = form.cleaned_data["path_type"]
if (path_type == "file_upload"):
(c.location, c.file_name) = files.save_tmp_file(c.name)
else:
c.location = form.cleaned_data["location"].strip("/")
c.file_name = form.cleaned_data["file_name"].strip("/")
c.save()
return HttpResponseRedirect(reverse("corpora:corpus", kwargs={"corpus_id" : c.id}))
# we should ideally never get here, so just redirect to the main index
# in case something unexpected went wrong
else:
return HttpResponseRedirect(reverse("corpora:index"))
