import json import sys from cleaning import convertTime eventTimes = [] for input in sys.stdin.readlines(): input = input.strip() if input.endswith("_"): input = input[7:] parts = input.split() start = parts[0] end = parts[1][2:] + "-0100" elif input.startswith("bar"): input = input[7:] start, end = input.split() end = start[:11] + end + start[16:] else: start, end = input.split() start = start[2:] + "-0100" end = end[2:] + "-0100" eventTimes.append({"end": convertTime(start), "start": convertTime(end)}) print json.dumps(eventTimes)
import pymongo import sys from cleaning import convertTime, removeLineBreaks, removeHttp connection = pymongo.Connection() db = connection.olympics_database posts = db.posts keyWord = sys.argv[1] tweets = [] for tweet in posts.find( { "user.lang" : "en" }, { "text": 1, "created_at" : 1} ): ts = convertTime(tweet['created_at']) if keyWord in tweet['text']: tweets.append((ts, tweet['created_at'], tweet['text'])) tweets.sort() print "TimeStamp Tweet" for time, created_at, text in tweets: print time, removeHttp(removeLineBreaks(text)).encode('utf-8')
import pymongo import sys from cleaning import convertTime, removeLineBreaks, removeHttp, removeOddities connection = pymongo.Connection() db = connection.olympics_database posts = db.posts nTweets = int(sys.argv[1]) tweets = [] for tweet in posts.find({ "user.lang": "en" }, { "text": 1, "created_at": 1 }).limit(nTweets): ts = convertTime(tweet['created_at']) tweets.append((ts, tweet['created_at'], tweet['text'])) tweets.sort() print "TimeStamp Tweet" for time, created_at, text in tweets: print time, removeOddities(removeHttp( removeLineBreaks(text))).encode('utf-8')
import json import sys from cleaning import convertTime eventTimes = [] for input in sys.stdin.readlines(): input = input.strip() if input.endswith("_"): input = input[7:] parts = input.split() start = parts[0] end = parts[1][2:] + "-0100" elif input.startswith("bar"): input = input[7:] start, end = input.split() end = start[:11] + end + start[16:] else: start, end = input.split() start = start[2:] +"-0100" end = end[2:] + "-0100" eventTimes.append( { "end": convertTime(start), "start": convertTime(end)} ) print json.dumps(eventTimes)