Exemplo n.º 1
0
from xml.etree.ElementTree import iterparse
import urllib
import socket
from sklearn.feature_extraction.text import TfidfTransformer

import tornado.httpserver
import tornado.ioloop
import tornado.web
from tornado.httpclient import AsyncHTTPClient
from tornado import gen
from tornado.httpserver import HTTPServer
from tornado.process import fork_processes
from tornado.netutil import bind_sockets

from constants import color
bcolors = color.bcolors()


numIdx = 1
numDoc = 0

# deciding ports
from constants import inventory
allserver = inventory.Inventory()
Baseport = allserver.callBasePort(13000)
allserver.findPorts( socket.gethostname(), numIdx, numDoc, Baseport)
Idxservers = allserver.getIndexers()
Docservers = allserver.getDocuments()

# print basic port information
print bcolors.HEADER + "====== BASIC INFO ======" + bcolors.ENDC
Exemplo n.º 2
0
import json, operator
import pickle
import urllib
from nltk.tokenize import RegexpTokenizer

from tornado.httpclient import AsyncHTTPClient
from tornado import gen
from tornado.options import define, options

from operator import mul

invertedIndex = {}
tokenizer = None

from constants import color
bcolors = color.bcolors()


def top10String(sorted_List):
    outputStr = "<h1>Top10 related industry to your stock</h1>"
    for (industry, count) in sorted_List[1:10]:
        outputStr += industry + " " + str(count) + "<br>"

    return outputStr


def Tosnippet(text, keywords, extend):
    returnText = '...'

    for keyword in keywords:
        loc = 0