Untitled

from html.parser import HTMLParser
import re
import plotly as py
import plotly.graph_objs as go


class LinksParser(HTMLParser):
    def __init__(self):
        HTMLParser.__init__(self)
        self.recordingData = False
        self.recordUser = False
        self.message_data = [()]
        self.user_data = []
        self.filter_tag = 'p'
        self.user_tag = 'span'
        self.user = ''
        self.persons = []

    def handle_starttag(self, tag, attributes):
        # print('TAG : ' + tag)
        if tag == self.filter_tag:
            self.recordingData = True
        if tag == self.user_tag:
            self.recordUser = True

    def handle_endtag(self, tag):
        return

    def handle_data(self, data):
        # print('USEDATA : ' + data)

        if self.recordUser and '@facebook' in data:
            result = ''
            for i in data:
                if i.isdigit():
                    result += i
            # print('USERID : ' + result)
            self.recordUser = False
            self.user_id = result
            if result not in self.persons:
                self.persons.append(result)
        if self.recordingData:
            # print('Message: ' + data)
            self.message_data.append((data, self.user_id))
            self.recordingData = False


parser = LinksParser()
"""
parser.feed('<html><head><title>Test</title></head>'
            '<body><h1>Parse me!</h1></body></html>'
            '<p>PRINT ME</p>'
            '<div>NOT ME</div>')
# print(parser.data)"""
with open('messages.htm', encoding="utf8") as content_file:
    content = content_file.read()
parser.feed(content)

from pprint import pprint

# pprint(parser.message_data)
# pprint(parser.persons)
smiley_count_dict = dict.fromkeys(parser.persons, 0)
message_count_dict = dict.fromkeys(parser.persons, 0)
# pprint(count_dict[parser.message_data[5][1]])

for message in parser.message_data:
    # pprint(message)

    if len(message) == 2:
        key = message[1]
        smileys = 0
        smileys += message[0].count(':)')
        smileys += message[0].count(':-')
        smileys += message[0].count(':(')
        smileys += message[0].count(':p')
        smileys += message[0].count(':s')
        smileys += message[0].count(':P')
        smileys += message[0].count(':')
        smileys += message[0].count('^^')
        smileys += len(re.findall('[\U0001f600-\U0001f650]', message[0]))

        smiley_count_dict[key] += smileys
        message_count_dict[key] += 1

data = []
for key in smiley_count_dict.keys():
    message = message_count_dict[key]
    smileys = smiley_count_dict[key]
    if message > 0:
        rel = smileys / message
    # print("User: " +  str(key) + " messages: " + str(message)+ " smileys: " + str(smileys) + " Rel = "+str(rel))
    data.append((smileys, message, rel, key))
data.sort(key=lambda tup: tup[2])
topTen = []
for data_instance in data:
    if data_instance[1] > 50:
        topTen.append(data_instance)
#pprint(topTen)
#topTen=topTen[-10:]
"""
data = [go.Bar(
            x=['giraffes', 'orangutans', 'monkeys'],
            y=[20, 14, 23]
    )]
"""
keys = [x[3] for x in topTen]
messages = [x[1] for x in topTen]
smileys = [x[1] for x in topTen]
print(topTen[0])
print(keys)
#print(messages)
# trace1 = go.Bar(key,messages,'Messages')
trace1 = go.Bar(
    # x=keys,
    y=messages,
    name='Messages'
)
trace2 = go.Bar(
    # x=keys,
    y=smileys,
    name='Smileys'
)
g_rel = [x[2] for x in topTen]
g_data = [go.Bar(
    y=g_rel
)]
py.offline.plot(g_data, filename='basic-bar')

g_data = [trace1, trace2]
layout = go.Layout(barmode='stack')
fig = go.Figure(data=g_data, layout=layout)
# py.offline.plot(fig, filename='stacked-bar')