from pattern.web import Twitter from pattern.en import Sentence, parse, pprint, modality, sentiment, parsetree, ngrams, mood import db import MySQLdb as mdb import sys import logging import pattern_engine as p logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO) t = Twitter() trending = t.trends(cached=False) tweet_db = [] for trend in trending: for tweet in t.search(trend, count=10): if tweet.language == u'en': tweet_db.append((tweet, trend)) tweet_db = [[tweet.id, tweet.author, tweet.text, tweet.date, tweet.profile, tweet.url, tweet.language, trend] for (tweet, trend) in tweet_db] tweet_db = p.add_sentiment(tweet_db) tweet_db = p.add_modality(tweet_db) con = db.con cur = con.cursor(mdb.cursors.DictCursor) query = 'INSERT into predictive.dev_pattern \ (tweet_id, author, text, date, profile, url, language, trend, polarity, subjectivity, mood, modality) \ VALUES %s, %s, %s, %s, %s, %s, %s, %s, %f, %f, %s, %f);' for tweet in tweet_db: cur.execute(query, tuple(tweet))
def poli_twitter_analysis():
"""This function parses Twitter to determine the average sentiment towards a user-defined political figure"""
print 'This program measures the average sentiment of the populous towards a political candidate through the analysis of recent tweets' #introduce program to user
print
print 'Enter the name of a candidate:'
x = raw_input('> ') #receives name of candidate to search for
print 'Enter number of tweets to search (max = 100)'
twtNumstr = raw_input('> ') #recieve number of tweets to search for
twtNum = int(twtNumstr) #convert to int to use in search
if twtNum <= 1: #check if an invalid number was entered, and if so, correct it to either the minimum or maximum allowed
twtNum = 2
print 'Invalid number entered. The minimum of 2 tweets will be used.'
elif twtNum > 100:
twtNum = 100
print 'Invalid number entered. The maximum of 100 tweets will be used.'
t = Twitter() #search for tweets containing user-defined key word
i = None
twts = []
for j in range(1):
for tweet in t.search(x, start=i, count=twtNum):
twts.append(tweet.text)
senti_count = 0 #predefine sentiment variable
for entry in twts: #loop through all tweets found
[senti,objec] = sentiment(entry) #sentiment analysis of tweets
senti_count = senti_count + senti
res = senti_count/twtNum #get average sentiment of all tweets parsed
print 'Result:'
if res >= -0.01 and res <= 0.01:
print 'Overwhelmingly neutral!'
elif res > 0.01 and res <= 0.1:
print 'Somewhat positive'
print("%s out of 1.0" % res)
elif res > 0.1 and res <= 0.5:
print 'Pretty positive'
print("%s out of 1.0" % res)
elif res > 0.5:
print 'Very positive!'
print("%s out of 1.0" % res)
elif res < -0.01 and res >= -0.1:
print 'Somewhat negative'
print("%s out of -1.0" % res)
elif res < -0.1 and res >= -0.5:
print 'Pretty negative'
print("%s out of -1.0" % res)
elif res < 0.5:
print 'Very negative...'
print("%s out of -1.0" % res)
twtTrends = t.trends(cached=False) #import current twitter trends (worldwide)
twtTrendsLower = [runthrough.lower() for runthrough in twtTrends]
xLower = x.lower()
if any(xLower in s for s in twtTrendsLower): #check to see if the user's search matches any of the current trends
print 'Your search is trending!'
matching = [s for s in twtTrendsLower if xLower in s]
for entry in matching:
if entry[0] == '#':
print entry
else:
print '#' + entry
print 'Do you want to read the tweets? (y/n)' #user can read the tweets used
yay_or_nay = raw_input('> ')
if yay_or_nay == 'y': #only complete if user requests
for entry in twts: #loop through all tweets found
print entry #print each tweet with a blank line separating each
print
