turicreate flash

gistfile1.txt

from flask_api import FlaskAPI
from flask import Flask, request, json, jsonify
import datetime
from datetime import datetime
import turicreate as tc
import pandas as pd
import requests
import os
from flask import Flask, request, redirect, url_for
from werkzeug.utils import secure_filename
import string
import random
from random import randint,choice
import flask_uploads
from flask_uploads import UploadSet, configure_uploads, IMAGES,UploadNotAllowed

UPLOAD_FOLDER = '/models'
ALLOWED_EXTENSIONS = set(['txt', 'csv'])

app = FlaskAPI(__name__)
app.config['UPLOAD_FOLDER'] = UPLOAD_FOLDER
app.config['MAX_CONTENT_LENGTH'] = 0.0016 * 1024 * 1024
	
files = UploadSet('files', extensions="csv")

@app.route('/',methods=['GET'])
def hello_world:
	return "Hello World"

@app.route('/upload',methods=['POST'])
def upload_files():
	allchar = string.ascii_letters + string.digits
	model = "".join(choice(allchar) for x in range(randint(50, 50)))
	
	if request.method == 'POST':
	    f = request.files['file']
	    f.save(secure_filename(model+".csv"))
	    return model
	else:
		pass    
#Creates or gets a user based of an ID
@app.route('/identify',methods=['POST'])
def identify():
	#Loading of train data
	model_id = request.headers['model']
	sf_train   = tc.SFrame.read_csv(model_id+".csv",error_bad_lines=True)

	#input_post_json
	json_data = request.get_json(force = True)

	#converts the JSON to a dataframe then a S Frame
	df = pd.DataFrame(data = json_data,index=[0])
	sf_new = tc.SFrame(data=df)

	#Gets the maximum number of options we would like to get
	#You will need to assign a category colum
	a = len((sf_train["category"].unique()))

	#Starts the prediction part
	m = tc.nearest_neighbor_classifier.create(sf_train, target='category')
	ystar = m.predict_topk(sf_new, max_neighbors=10, k=a)

	empty_json={}
	result = False
	while result is False:
	    try:
	        for i in range(a):
	            empty_json[ystar["class"][i]]=ystar["probability"][i]
	            result = True
	    except:
	        pass
	return json.dumps(empty_json)

if __name__ == '__main__':
    app.run(host='0.0.0.0', port=5000, debug=True)