class Sentiment(object):
'''
Makes a Sentiment analysis of the posts
'''
def __init__(self):
self.dbc = DBConnection()
self.postBlockVersionAltered = True
self.postHistoryAltered = True
self.postsAltered = True
self.commentsAltered = True
# calculates sentiment score of all text blocks
#self.postblockversion_sentiment()
# calculates sentiment score of text blocks from history entries
#self.posthistory_sentiment()
# calculates sentiment score of most recent post versions
#self.posts_sentiment()
# calculates sentiment score of the post comments
#self.comment_sentiment()
def postblockversion_sentiment(self):
# alter posts table: add metric columns
if (not self.postBlockVersionAltered):
self.dbc.add_sentiment_columns("postblockversion")
""" For every entry in the database """
entries = self.dbc.get_id_content_from_postblockversion()
counter = 0
for entry in entries:
# id from table entry
id_ = entry[0]
# Calculates the metrics of the text
results = self.calc_sentiment(entry[1])
if (results != 0):
# Print the metrics in stdout
#self.print_sentiment(results)
# Stores the metrics in the database
self.dbc.store_sentiment(id_, "postblockversion", results)
counter += 1
if counter % 10000 == 0:
print str(counter) + ' statements executed!'
def posthistory_sentiment(self):
# adds the columns for the sentiment
# analysis to the posthistory table
if (not self.postHistoryAltered):
self.dbc.add_sentiment_columns('posthistory')
count = 0
phIds = self.dbc.get_ids_from_posthistory()
for id_ in phIds:
# get all textblocks from the posthistory entry
try:
textblocks = self.dbc.get_content_from_posthistory(id_[0])
# get all the text blocks from the history version of the post
text = ""
for block in textblocks:
text = text + "\n" + block[0]
# Calculates the metrics of the text
results = self.calc_sentiment(text)
if (results != 0):
# Print the metrics in stdout
# self.print_sentiment(results)
# Stores the metrics in the database
self.dbc.store_sentiment(id_[0], "posthistory", results)
except UnboundLocalError:
print "No post block verion"
continue
count += 1
if count % 10000 == 0:
print str(count) + " statements executed!"
def posts_sentiment(self):
# adds the columns for the sentiment analysis to the posts table
if (not self.postsAltered):
self.dbc.add_sentiment_columns('posts')
results = self.dbc.get_most_recent_sentiment()
count = 0
for result in results:
count += 1
if count % 10000 == 0:
print "10.000 querys executed!"
# post id
id_ = result[0]
# sentiment metrics
metrics = {
'neg': result[1],
'neu': result[2],
'pos': result[3],
'compound': result[4]
}
# store in posts table
#print id_
#print metrics
self.dbc.store_sentiment(id_, "posts", metrics)
def comment_sentiment(self):
# adds the columns for the sentiment analysis to the comments table
if (not self.commentsAltered):
self.dbc.add_sentiment_columns('comments')
comments = self.dbc.get_comment_text()
count = 0
for comment in comments:
id_ = comment[0]
result = self.calc_sentiment(comment[1])
#print id_
#print result
self.dbc.store_sentiment(id_, 'comments', result)
count += 1
if count % 10000 == 0:
print str(count) + ' Comments done!'
def calc_sentiment(self, text):
# in the csv files are these unconverted special chars
text = text.replace('
', '\n')
sid = SentimentIntensityAnalyzer()
return sid.polarity_scores(text)
def print_sentiment(self, result):
for k in result:
print('{0}: {1}, '.format(k, result[k]))
