def home():
if(request.method == 'GET'):
return render_template('index.html')
else:
data = request.form
print(request.form)
return (predict_sentiment(request.form['text']) and 'negative' or 'positive')
def search_tweets():
query = request.json['query']
tweets = getTweets(query)
predictions = predict_sentiment(tweets)
predictions = [
sentiments[prediction.index(max(prediction))]
for prediction in predictions
]
sentiment_counts = {'negative': 0, 'neutral': 0, 'positive': 0}
for sentiment in predictions:
sentiment_counts[sentiment] += 1
return jsonify({'results': sentiment_counts}), 200
def predict():
"""
Return json response with sentiment
for a client sentence from PyTorch
Transformer + GRU model trained on
Yelp Reviews dataset
"""
if request.method == "POST":
sentence = request.get_data()
sentence = str(sentence)
sentiment = predict_sentiment(model, tokenizer, sentence)
return jsonify({'sentiment': sentiment})
def home():
if request.method == "POST":
user_id = request.form.get("userid")
user_id=user_id.lower().strip()
if len(user_id)==0:
return render_template('base.html') + 'PLEASE ENTER USER ID'
if user_id not in recc_df.index:
return render_template('base.html') + 'THE USER ID IS NOT AVILABLE IN DATASET PLEASE USE VALID USER ID'
else:
result_df=predict_sentiment(user_id,recc_df)
return render_template('home.html',predict=result_df.head(5),user=user_id)
else:
return render_template('base.html')
def chat():
"""API point handeling chatbot requests
Args:
Json input from via POST request
Request should contain 'input' field
containing chatbot qery
Returns:
JSON response:
{
"message": message,
"source": source-to-scrap-data-from,
"data": data-relavent-to-source,
"company": company-name,
}
"""
req_data = request.get_json()
input_query = req_data["input"]
# Get analysis from diagnolflow
try:
flow_output = dflow.collect_intent(input_query)
except Exception as error:
pprint(error)
return "Error"
# Extract message from DiagnolFlow Response
message = flow_output.query_result.fulfillment_messages[0].text.text[0]
# Extract company from DiagnolFlow Response
try:
company_field = flow_output.query_result.parameters.fields["Company"]
company = company_field.list_value.values[0].string_value
except:
company = None
# Extract source from DiagnolFlow Response
try:
source_field = flow_output.query_result.parameters.fields["source"]
source = source_field.list_value.values[0].string_value
except:
source = None
# Scrape data from corresponding source and do sentiment analysis on it
raw_data = None
if company is not None:
if source == "reddit":
raw_data = scrap.scrape_reddit(company)
for key in raw_data.keys():
predictions = predict_sentiment(raw_data[key])
raw_data[key] = [
(sentence, int(prediction))
for sentence, prediction in zip(raw_data[key], predictions)
]
elif source == "news":
raw_data = scrap.scrape_news(company)
sentences = [news_url["title"] for news_url in raw_data]
predictions = predict_sentiment(sentences)
for data, prediction in zip(raw_data, predictions):
data["sentiment"] = int(prediction)
elif source == "twitter":
raw_data = scrap.scrape_twitter(company)
sentences = [tweet_object["tweet"] for tweet_object in raw_data]
predictions = predict_sentiment(sentences)
for data, prediction in zip(raw_data, predictions):
data["sentiment"] = int(prediction)
# Return response back to frontend
response = {
"message": message,
"source": source,
"data": raw_data,
"company": company,
}
return response
def predict():
text_to_predict = request.json["text"]
logger.debug("Received: %s" % text_to_predict)
prediction = ml_predict.predict_sentiment(text_to_predict)
return jsonify(prediction)
def predict_tweet():
tweet = request.json['tweet']
prediction = predict_sentiment([tweet])
sentiment = sentiments[prediction[0].index(max(prediction[0]))]
return jsonify({'sentiment': sentiment}), 200
from predict import predict_sentiment text = input() print(predict_sentiment(text))