def prepare_hist_and_plot(self, n_tweets, users, n_bins, campaign_id):
import numpy
import matplotlib.pyplot as plot
plot_graphs = False
hist = {
'user_creation': {
'data': None,
'bins': None,
},
'user_n_tweets': {
'data': None,
'bins': None,
},
'user_n_tweets_overall': {
'data': None,
'bins': None,
},
'n_tweets': None,
'n_unique_users': None,
'n_default_profile_image': None,
'n_lower_than_threshold': None,
}
self.logger.debug("How many tweets? %d" % n_tweets)
hist['n_tweets'] = n_tweets
# TODO: abort if there are more than 200000 tweets.
if n_tweets > 200000:
return
#
# How many unique users?
#
n_unique_users = len(users)
self.logger.debug("How many unique users? %d" % n_unique_users)
hist['n_unique_users'] = n_unique_users
######
sec_title = "Histogram of user creation dates?"
#
tmp_dates = []
for x in users:
tmp_date = x['user']['created_at']
if type(tmp_date) != float:
tmp_date = py_utc_time2drnj_time(tmp_date)
tmp_dates.append(tmp_date)
# tmp_dates = [py_utc_time2drnj_time(x['user']['created_at']) for x in users]
(hist['user_creation']['data'], hist['user_creation']['bins']) = numpy.histogram(tmp_dates, bins=n_bins)
if plot_graphs:
bins = hist['user_creation']['bins'][:-1]
width = (hist['user_creation']['bins'][1] - hist['user_creation']['bins'][0])/2
plot.bar(bins, hist['user_creation']['data'], width=width, align='center')
xticklabels = [time.strftime('%d %b %Y', time.gmtime(drnj_time2py_time(x))) for x in bins]
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('1.pdf', dpi=600)
#####
sec_title = "Histogram of number of tweets of each user in this campaign"
tmp_counts = [int(x['n_user_tweets']) for x in users]
#
(hist['user_n_tweets']['data'], hist['user_n_tweets']['bins']) = numpy.histogram(tmp_counts, bins=n_bins)
if plot_graphs:
bins = hist['user_n_tweets']['bins'][:-1]
data = hist['user_n_tweets']['data']
width = (hist['user_n_tweets']['bins'][1] - hist['user_n_tweets']['bins'][0])/2
plot.bar(bins, data, width=width, align='center')
xticklabels = bins
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('2.pdf', dpi=600)
#####
sec_title = "What percentage of them used the default profile image?"
#
n_default_profile_image = 0
for u in users:
if u['user']['default_profile_image']:
n_default_profile_image += 1
hist['n_default_profile_image'] = n_default_profile_image
self.logger.debug("%s: %0.2f%%" % (sec_title, 100*(float(n_default_profile_image)/n_unique_users)))
#####
sec_title = "Histogram of tweet counts of unique users"
tmp_counts = [int(x['user']['statuses_count']) for x in users]
(hist['user_n_tweets_overall']['data'],
hist['user_n_tweets_overall']['bins']) = numpy.histogram(tmp_counts, bins=n_bins)
if plot_graphs:
bins = hist['user_n_tweets_overall']['bins'][:-1]
data = hist['user_n_tweets_overall']['data']
width = (hist['user_n_tweets_overall']['bins'][1] - hist['user_n_tweets_overall']['bins'][0])/2
plot.bar(bins, data, width=width, align='center')
xticklabels = bins
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('3.pdf', dpi=600)
#
sec_title = "What percentage of them have lower than 5 tweets?"
n_lower_than_threshold = 0
for u in users:
if u['user']['statuses_count'] < 5:
n_lower_than_threshold += 1
hist['n_lower_than_threshold'] = n_lower_than_threshold
self.logger.debug("%s: %0.2f%%" % (sec_title, 100*(float(n_lower_than_threshold)/n_unique_users)))
self.logger.debug(hist)
# converting numpy.array's to normal python lists.
for k in hist.keys():
if type(hist[k]) == dict:
for k2 in hist[k].keys():
if type(hist[k][k2]) == type(numpy.array([])):
hist[k][k2] = list(hist[k][k2])
hist = {'campaign_id': campaign_id,
'histogram': hist,
'created_at': now_in_drnj_time()}
return hist
def insert_tweet(self, tweet_obj_array):
# actual tweet insertion
yield self.motor_column.tweets.insert(tweet_obj_array)
for tweet_obj in tweet_obj_array:
# build the analytics
freq = {}
campaign_id = tweet_obj['campaign_id']
# strip unnecessary fields
tweet_obj = tweet_obj['tweet']
freq['campaigns'] = {campaign_id: 1}
# freq['tokens'] = {'ali': 1, 'veli': 1}
freq['hashtags'] = {}
if 'entities' in tweet_obj and 'hashtags' in tweet_obj['entities']:
for hashtag in tweet_obj['entities']['hashtags']:
if 'text' in hashtag:
item_key = hashtag['text']
if item_key in freq['hashtags']:
freq['hashtags'][item_key] += 1
else:
freq['hashtags'][item_key] = 1
else:
# log this missing attribute.
pass
freq['mentions'] = {}
if 'entities' in tweet_obj and 'user_mentions' in tweet_obj['entities']:
for mention in tweet_obj['entities']['user_mentions']:
if 'id_str' in mention:
item_key = "|".join([mention['id_str'], mention['screen_name']])
if item_key in freq['mentions']:
freq['mentions'][item_key] += 1
else:
freq['mentions'][item_key] = 1
else:
# log this missing attribute.
pass
freq['urls'] = {}
if 'entities' in tweet_obj and 'urls' in tweet_obj['entities']:
for url in tweet_obj['entities']['urls']:
if 'expanded_url' in url:
item_key = url['expanded_url']
if item_key in freq['urls']:
freq['urls'][item_key] += 1
else:
freq['urls'][item_key] = 1
else:
# log this missing attribute.
pass
if 'created_at' in tweet_obj:
# turns out that we've already transformed into drnj_time
t = drnj_time2py_time(tweet_obj['created_at'])
else:
t = time.time()
gm_t = time.gmtime(t)
today_str = time.strftime('%Y-%m-%d', gm_t)
hour = time.strftime('%H', gm_t)
minute = "%04d" % (int(hour)*60 + int(time.strftime('%M', gm_t)))
for key in freq:
for item in freq[key].keys():
count = freq[key][item]
yield self.colls[key].update({'campaign_id': campaign_id, 'date': today_str, 'key': item},
{'$inc': {('hour.%s' % hour): count, ('minute.%s' % minute): count, ('day_total'): count},
'$set': {'last_updated_minute': minute}}, upsert=True)
sec_title = "Histogram of user creation dates?"
#
tmp_dates = [py_utc_time2drnj_time(x['user']['created_at']) for x in users]
(hist['user_creation']['data'],
hist['user_creation']['bins']) = numpy.histogram(tmp_dates, bins=100)
if plotGraphs:
bins = hist['user_creation']['bins'][:-1]
width = (hist['user_creation']['bins'][1] -
hist['user_creation']['bins'][0]) / 2
plot.bar(bins, hist['user_creation']['data'], width=width, align='center')
xticklabels = [
time.strftime('%d %b %Y', time.gmtime(drnj_time2py_time(x)))
for x in bins
]
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('1.pdf', dpi=600)
#####
sec_title = "Histogram of number of tweets of each user in this campaign"
tmp_counts = [int(x['n_user_tweets']) for x in users]
#
(hist['user_n_tweets']['data'],
hist['user_n_tweets']['bins']) = numpy.histogram(tmp_counts, bins=100)
def prepare_hist_and_plot(self, n_tweets, users, n_bins, campaign_id):
import numpy
import matplotlib.pyplot as plot
plot_graphs = False
hist = {
'user_creation': {
'data': None,
'bins': None,
},
'user_n_tweets': {
'data': None,
'bins': None,
},
'user_n_tweets_overall': {
'data': None,
'bins': None,
},
'n_tweets': None,
'n_unique_users': None,
'n_default_profile_image': None,
'n_lower_than_threshold': None,
}
self.logger.debug("How many tweets? %d" % n_tweets)
hist['n_tweets'] = n_tweets
# TODO: abort if there are more than 200000 tweets.
if n_tweets > 200000:
return
#
# How many unique users?
#
n_unique_users = len(users)
self.logger.debug("How many unique users? %d" % n_unique_users)
hist['n_unique_users'] = n_unique_users
######
sec_title = "Histogram of user creation dates?"
#
tmp_dates = []
for x in users:
tmp_date = x['user']['created_at']
if type(tmp_date) != float:
tmp_date = py_utc_time2drnj_time(tmp_date)
tmp_dates.append(tmp_date)
# tmp_dates = [py_utc_time2drnj_time(x['user']['created_at']) for x in users]
(hist['user_creation']['data'],
hist['user_creation']['bins']) = numpy.histogram(tmp_dates,
bins=n_bins)
if plot_graphs:
bins = hist['user_creation']['bins'][:-1]
width = (hist['user_creation']['bins'][1] -
hist['user_creation']['bins'][0]) / 2
plot.bar(bins,
hist['user_creation']['data'],
width=width,
align='center')
xticklabels = [
time.strftime('%d %b %Y', time.gmtime(drnj_time2py_time(x)))
for x in bins
]
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('1.pdf', dpi=600)
#####
sec_title = "Histogram of number of tweets of each user in this campaign"
tmp_counts = [int(x['n_user_tweets']) for x in users]
#
(hist['user_n_tweets']['data'],
hist['user_n_tweets']['bins']) = numpy.histogram(tmp_counts,
bins=n_bins)
if plot_graphs:
bins = hist['user_n_tweets']['bins'][:-1]
data = hist['user_n_tweets']['data']
width = (hist['user_n_tweets']['bins'][1] -
hist['user_n_tweets']['bins'][0]) / 2
plot.bar(bins, data, width=width, align='center')
xticklabels = bins
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('2.pdf', dpi=600)
#####
sec_title = "What percentage of them used the default profile image?"
#
n_default_profile_image = 0
for u in users:
if u['user']['default_profile_image']:
n_default_profile_image += 1
hist['n_default_profile_image'] = n_default_profile_image
self.logger.debug("%s: %0.2f%%" %
(sec_title, 100 *
(float(n_default_profile_image) / n_unique_users)))
#####
sec_title = "Histogram of tweet counts of unique users"
tmp_counts = [int(x['user']['statuses_count']) for x in users]
(hist['user_n_tweets_overall']['data'],
hist['user_n_tweets_overall']['bins']) = numpy.histogram(tmp_counts,
bins=n_bins)
if plot_graphs:
bins = hist['user_n_tweets_overall']['bins'][:-1]
data = hist['user_n_tweets_overall']['data']
width = (hist['user_n_tweets_overall']['bins'][1] -
hist['user_n_tweets_overall']['bins'][0]) / 2
plot.bar(bins, data, width=width, align='center')
xticklabels = bins
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('3.pdf', dpi=600)
#
sec_title = "What percentage of them have lower than 5 tweets?"
n_lower_than_threshold = 0
for u in users:
if u['user']['statuses_count'] < 5:
n_lower_than_threshold += 1
hist['n_lower_than_threshold'] = n_lower_than_threshold
self.logger.debug("%s: %0.2f%%" %
(sec_title, 100 *
(float(n_lower_than_threshold) / n_unique_users)))
self.logger.debug(hist)
# converting numpy.array's to normal python lists.
for k in hist.keys():
if type(hist[k]) == dict:
for k2 in hist[k].keys():
if type(hist[k][k2]) == type(numpy.array([])):
hist[k][k2] = list(hist[k][k2])
hist = {
'campaign_id': campaign_id,
'histogram': hist,
'created_at': now_in_drnj_time()
}
return hist
def insert_tweet(self, tweet_obj_array):
# actual tweet insertion
yield self.motor_column.tweets.insert(tweet_obj_array)
for tweet_obj in tweet_obj_array:
# build the analytics
freq = {}
campaign_id = tweet_obj['campaign_id']
# strip unnecessary fields
tweet_obj = tweet_obj['tweet']
freq['campaigns'] = {campaign_id: 1}
# freq['tokens'] = {'ali': 1, 'veli': 1}
freq['hashtags'] = {}
if 'entities' in tweet_obj and 'hashtags' in tweet_obj['entities']:
for hashtag in tweet_obj['entities']['hashtags']:
if 'text' in hashtag:
item_key = hashtag['text']
if item_key in freq['hashtags']:
freq['hashtags'][item_key] += 1
else:
freq['hashtags'][item_key] = 1
else:
# log this missing attribute.
pass
freq['mentions'] = {}
if 'entities' in tweet_obj and 'user_mentions' in tweet_obj[
'entities']:
for mention in tweet_obj['entities']['user_mentions']:
if 'id_str' in mention:
item_key = "|".join(
[mention['id_str'], mention['screen_name']])
if item_key in freq['mentions']:
freq['mentions'][item_key] += 1
else:
freq['mentions'][item_key] = 1
else:
# log this missing attribute.
pass
freq['urls'] = {}
if 'entities' in tweet_obj and 'urls' in tweet_obj['entities']:
for url in tweet_obj['entities']['urls']:
if 'expanded_url' in url:
item_key = url['expanded_url']
if item_key in freq['urls']:
freq['urls'][item_key] += 1
else:
freq['urls'][item_key] = 1
else:
# log this missing attribute.
pass
if 'created_at' in tweet_obj:
# turns out that we've already transformed into drnj_time
t = drnj_time2py_time(tweet_obj['created_at'])
else:
t = time.time()
gm_t = time.gmtime(t)
today_str = time.strftime('%Y-%m-%d', gm_t)
hour = time.strftime('%H', gm_t)
minute = "%04d" % (int(hour) * 60 + int(time.strftime('%M', gm_t)))
for key in freq:
for item in freq[key].keys():
count = freq[key][item]
yield self.colls[key].update(
{
'campaign_id': campaign_id,
'date': today_str,
'key': item
}, {
'$inc': {
('hour.%s' % hour): count,
('minute.%s' % minute): count,
('day_total'): count
},
'$set': {
'last_updated_minute': minute
}
},
upsert=True)
hist['n_unique_users'] = n_unique_users
######
sec_title = "Histogram of user creation dates?"
#
tmp_dates = [py_utc_time2drnj_time(x['user']['created_at']) for x in users]
(hist['user_creation']['data'], hist['user_creation']['bins']) = numpy.histogram(tmp_dates, bins=100)
if plotGraphs:
bins = hist['user_creation']['bins'][:-1]
width = (hist['user_creation']['bins'][1] - hist['user_creation']['bins'][0])/2
plot.bar(bins, hist['user_creation']['data'], width=width, align='center')
xticklabels = [time.strftime('%d %b %Y', time.gmtime(drnj_time2py_time(x))) for x in bins]
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('1.pdf', dpi=600)
#####
sec_title = "Histogram of number of tweets of each user in this campaign"
tmp_counts = [int(x['n_user_tweets']) for x in users]
#
(hist['user_n_tweets']['data'], hist['user_n_tweets']['bins']) = numpy.histogram(tmp_counts, bins=100)
if plotGraphs:
bins = hist['user_n_tweets']['bins'][:-1]
data = hist['user_n_tweets']['data']
