def piindex(request):
wdc = WDCService('PI')
service_creds = wdc.getCreds()
form = Form_pi()
service_creds['form'] = form
return render(request, 'Watson/piindex.html', service_creds)
def nlcnewx(request):
# This is a call that is not done through REST.
# The request is for a new classifier. The file is sent through a form
# The return is a redirect back to the classifier list, which forces it to refresh.
wdc = WDCService('LC')
service_creds = wdc.getCreds()
if request.POST:
form = UploadClassifierForm(request.FILES)
if request.FILES and 'classInputFile' in request.FILES:
f = request.FILES['classInputFile']
module_dir = os.path.dirname(__file__)
file_path = os.path.join(module_dir, '../static/', 'xx.json')
with open(file_path, 'wb+') as destination:
for chunk in f.chunks():
destination.write(chunk)
destination.close()
with open(file_path) as fj:
data = json.loads(fj.read())
wdc = WDCService('LC')
service_creds = wdc.getCreds()
nlcService = wdc.nlcService()
if nlcService is not None:
nlcResults = nlcService.createClassifier(data)
if nlcResults and 'error' in nlcResults:
service_creds['error'] = nlcResults['description']
return render(request, 'Watson/lcindex.html', service_creds)
return redirect('watson:nlclassifier')
def classifyTranscript(classURL, transcript):
# Runs classification against a transcript
# The classifier url must be passed in, as it contains the classifier id.
classifiedData = {}
classifiedData["message"] = transcript
theData = {}
wdc = WDCService('LC')
nlcService = wdc.nlcService()
if nlcService is not None:
nlcResult = nlcService.getNLClassification({"text":transcript,}, classURL)
if "error" in nlcResult:
theData = {"error": nlcResult["error"]}
else:
if "top_class" in nlcResult:
classifiedData["top_class"] = nlcResult["top_class"]
if "classes" in nlcResult:
classes = nlcResult["classes"]
for c in classes:
if classifiedData["top_class"] == c["class_name"]:
classifiedData["confidence"] = c["confidence"]
else:
theData = {"error": "Natural Language Classifier service not found"}
if "error" not in theData:
theData = {"classification" : classifiedData}
return theData
def classifyTweets(classURL, twitTimeLine):
# This is an internal helper method that runs a classification on each tweet.
tweetDataArray = []
theData = {}
wdc = WDCService('LC')
nlcService = wdc.nlcService()
if nlcService is not None:
tweets = twitTimeLine["tweets"]
for t in tweets:
tweetData = {}
tweetData["message"] = t
nlcResult = nlcService.getNLClassification({"text":t,}, classURL)
if "error" in nlcResult:
theData = {"error": nlcResult["error"]}
break
if "top_class" in nlcResult:
tweetData["top_class"] = nlcResult["top_class"]
if "classes" in nlcResult:
classes = nlcResult["classes"]
for c in classes:
if tweetData["top_class"] == c["class_name"]:
tweetData["confidence"] = c["confidence"]
tweetDataArray.append(tweetData)
else:
theData = {"error": "Natural Language Classifier service not found"}
if "error" not in theData:
theData = {"classification" : tweetDataArray}
return theData
def lcindex(request):
# These are the views for the Natural Language Classifier portion of the application.
# It makes use of both the NLC and a Twitter service, and all calls to the services
# are done from the client via AJAX calls to REST APIs
wdc = WDCService('LC')
service_creds = wdc.getCreds()
nlcService = wdc.nlcService()
return render(request, 'Watson/lcindex.html', service_creds)
def staudio(request):
# This request receives an Audio BLOB file which is passed to the
# Speech to text service. The response is then forwarded to the
# classifier service.
results = {}
theData = {"error": "Error detected in REST API"}
module_dir = os.path.dirname(__file__)
if request.POST and request.FILES:
form = UploadAudioForm(request.POST, request.FILES)
# Don't bother checking the form, as it is always invalid
#if form.is_valid():
# print("Valid Form")
#else:
# print("Invalid Form")
filename = ""
classURL = ""
if "fname" in request.POST:
filename = request.POST["fname"]
if "classifierurl" in request.POST:
classURL = request.POST["classifierurl"]
# Saving the file and reading it again, as this ensures that all the data has
# been received. This gives a better result from the service.
f = request.FILES['data']
if f:
file_path = os.path.join(module_dir, '../static/', filename)
with open(file_path, 'wb+') as destination:
for chunk in f.chunks():
destination.write(chunk)
destination.close()
# Remember to switch
yy_file_path = os.path.join(module_dir, '../static/', filename)
#yy_file_path = os.path.join(module_dir, '../static/', 'yy.wav')
with open(yy_file_path, 'rb') as fj:
audiodata = fj.read()
if audiodata:
wdc = WDCService('ST')
service_creds = wdc.getCreds()
stService = wdc.stService()
if stService is None:
theData = {"error": "No Speech to Text service found"}
else:
theData = stService.processAudio(audiodata)
if "results" in theData:
if list is type(theData["results"]):
res = theData["results"][0]
if "alternatives" in res:
alt = res["alternatives"][0]
if "transcript" in alt:
theData = classifyTranscript(classURL, alt["transcript"])
fj.close()
results["results"] = theData
return HttpResponse(json.dumps(results), content_type="application/json")
def taindex(request):
"""
This is the initial (or refresh) page request, not much to do other than send the form
"""
wdc = WDCService('TA')
service_creds = wdc.getCreds()
form = Form_ta()
service_creds['form'] = form
return render(request, 'Watson/taindex.html', service_creds)
def lclist(request):
# This is a request for a list of the classifiers available. We expect no input
# so all the classifiers found are returned.
results = {}
theData = {"error":"No Response as list request has failed"}
wdc = WDCService('LC')
nlcService = wdc.nlcService()
if nlcService is not None:
nlcResults = nlcService.listClassifiers()
theData = {"classifiers": nlcResults}
results["results"] = theData
return HttpResponse(json.dumps(results), content_type="application/json")
def taapi(request):
"""
Acting as a proxy between the Tradeoff Analytics widget and service.
"""
theData= {}
theResponse = {"Error":"Request could not be submitted"}
theRequest = json.loads(request.body.decode("utf-8"))
#with open('dumpedrequestjson.txt', 'w') as outfile:
# json.dump(theRequest, outfile)
if theRequest:
wdc = WDCService('TA')
theResponse = wdc.performOperation(theRequest)
if 'error' in theResponse:
print('Error Detected', theResponse['error'])
return HttpResponse(json.dumps(theResponse), content_type="application/json")
def nlcnew(request):
# This is a call that is not done through REST. Though it should also change to REST.
# This is the original new classifier request, but has been superseded by nlcnewx method.
# Will be removed in a future iteration, but is here for reference.
#
# The request is for a new classifier. Currently the file chosen is hardcoded and sits in the
# static directory. This should be modified to allow a file from the client to be submitted.
# The return is a redirect back to the classifier list, which forces it to refresh.
wdc = WDCService('LC')
data = ""
module_dir = os.path.dirname(__file__)
file_path = os.path.join(module_dir, '../static/', 'nlcsample.json')
with open(file_path) as f:
data = json.loads(f.read())
nlcService = wdc.nlcService()
if nlcService is not None:
nlcResults = nlcService.createClassifier(data)
return redirect('watson:nlclassifier')
def drop(request):
# Has we are only allowed 2 classifiers, we added in an option of dropping them to free up
# resources and to allow us to modify the training.
# In the input we are expecting a link to the classifier, which will be used to drop it.
results = {}
theData = {"error":"No Response as drop request has failed"}
if request.POST and "data" in request.POST:
d = request.POST["data"]
theRequestData = json.loads(d)
if theRequestData:
if "url" in theRequestData:
wdc = WDCService('LC')
nlcService = wdc.nlcService()
if nlcService is not None:
classURL = theRequestData["url"]
nlcResult = nlcService.dropClassifier(classURL)
if "error" in nlcResult:
theData = {"error": nlcResult["error"]}
else:
theData = {"ok": classURL}
results["results"] = theData
return HttpResponse(json.dumps(results), content_type="application/json")
def processPIRequest(request, theData):
results = {}
data = {}
errorFound = False
wdc = WDCService('PI')
service_creds = wdc.getCreds()
personality = theData["personality"]
performTwitterScan = theData["twitterScan"]
if performTwitterScan and personality and 0 < len(personality):
# Data needs to come from twitter
pulledData = pullTwitterTimeline(personality)
if "error" in pulledData:
errorFound = True
elif "data" in pulledData:
data = pulledData["data"]
else:
data = theData["data"].encode('utf8', 'replace')
if not errorFound:
results = wdc.performOperation(data)
service_creds['results'] = results
pi = Pidata(results)
keys = pi.big5keys()
traits = {}
for trait in keys:
traits[trait] = pi.getTraitValue(trait, False)
service_creds["traits"] = traits
savePersonality(personality, pi)
results = service_creds
else:
results["results"] = data
return results
def staudio_with_nlc(request):
# This request receives an Audio BLOB file which is passed to the
# Speech to text service. The response is then forwarded to the
# classifier service.
results = {}
tts_input = ''
theData = {"error": "Error detected in REST API"}
module_dir = os.path.dirname(__file__)
if request.POST and request.FILES:
form = UploadAudioForm(request.POST, request.FILES)
# Don't bother checking the form, as it is always invalid
#if form.is_valid():
# print("Valid Form")
#else:
# print("Invalid Form")
filename = ""
classURL = ""
textscript = ""
if "fname" in request.POST:
filename = request.POST["fname"]
if "textscript" in request.POST:
textscript = request.POST["textscript"]
if "classifierurl" in request.POST:
classURL = request.POST["classifierurl"]
# Saving the file and reading it again, as this ensures that all the data has
# been received. This gives a better result from the service.
f = request.FILES['data']
if f:
file_path = os.path.join(module_dir, '../static/', filename)
with open(file_path, 'wb+') as destination:
for chunk in f.chunks():
destination.write(chunk)
destination.close()
# Remember to switch
yy_file_path = os.path.join(module_dir, '../static/', filename)
#yy_file_path = os.path.join(module_dir, '../static/', 'yy.wav')
with open(yy_file_path, 'rb') as fj:
audiodata = fj.read()
if audiodata:
wdc = WDCService('ST')
service_creds = wdc.getCreds()
stService = wdc.stService()
if stService is None:
theData = {"error": "No Speech to Text service found"}
else:
theData = stService.processAudio(audiodata)
if "results" in theData:
if list is type(theData["results"]):
res = theData["results"][0]
if "alternatives" in res:
alt = res["alternatives"][0]
if "transcript" in alt:
question = alt["transcript"]
theData = classifyTranscript(classURL, alt["transcript"])
data = {}
if textscript != "":
data = {"txt": textscript,
"conversation_id":"",
"client_id":""
}
else:
data = {"txt": alt["transcript"],
"conversation_id":"",
"client_id":""
}
if not 'classification' in theData:
raise Exception('Classificatio failed: {}'.format(theData))
category = theData["classification"]["top_class"]
res = sendDialogAPI(category, {'data': json.dumps(data)})
theData = res['results']
theData['category'] = category
theData['question'] = question
tts_input = theData['conversationData']['response']
fj.close()
results["results"] = theData
textToSpeech(tts_input)
return HttpResponse(json.dumps(results), content_type="application/json")