Merge branch 'dev'

README.md

@@ -8,13 +8,19 @@ Don't expect too fast development tho'.
 
 More info soon...
 
+## Visualizer CLI
+A command-line interface for some visualization capabilities. The feature is not even alpha yet.
+
+For running Visualizer CLI:
+```shell script
+export PYTHONPATH="$PWD"
+python miner/Visualizer.py --help
+```
+
 ## Contribution
 Help is more than welcome. If somebody feel the urge to contribute, I would share my plans with them.
 
 Ideas are welcome too. Feel free to open a new issue.
 
 
-For running VIsualizer CLI:
-```shell script
-export PYTHONPATH="$PWD"
-```
+

miner/Visualizer.py

@@ -1,40 +1,126 @@
 import matplotlib.pyplot as plt
-import seaborn as sns
 import pandas as pd
+import numpy as np
+import argparse
+import os
+
 from miner.People import People
-from miner.ConversationAnalyzer import ConversationAnalyzer
+from miner.Analyzer import Analyzer
+from miner import utils
+
+TEST_DATA_PATH = f'{os.getcwd()}/tests/test_data'
 
-# plt.rcParams.update({'figure.figsize': (10, 7), 'figure.dpi': 120})
 
-TEST_DATA_PATH = '/home/levente/projects/facebook-data-miner/tests/test_data'
+# TEST_DATA_PATH = f'{os.getcwd()}/data'
 
 
 class Visualizer:
     def __init__(self):
         pass
 
-    def plot_convos(self, names):
-        people = People(path=TEST_DATA_PATH)
-        for name in names:
-            data = self.set_up_data(people, name, period='d')
-            df = pd.DataFrame(data.items(), columns=['date', 'value'])
-            v.plot_time_series(x=df.date, y=df.value, title=name)
+    @staticmethod
+    def set_up_data(people, period, stat='msg_count', **kwargs):
+        analyzer = Analyzer(people)
+        interval_stats = analyzer.get_time_series_data(period, **kwargs)
+        return analyzer.get_stat_count(interval_stats, statistic=stat)
+
+    def setup_data_for_all_subjects(self, people, period, stat=None, **kwargs):
+        all_data = self.set_up_data(people, period, stat=stat, subject='all', **kwargs)
+        me_data = self.set_up_data(people, period, stat=stat, subject='me', **kwargs)
+        partner_data = self.set_up_data(people, period, stat=stat, subject='partner', **kwargs)
+        return all_data, me_data, partner_data
+
+    def plot_time_series_data_of_messages(self, period, name=None, stat='msg_count', **kwargs):
+        people = People(path=TEST_DATA_PATH, name=name)
+        all_data, me_data, partner_data = self.setup_data_for_all_subjects(people, period, stat, **kwargs)
+        d = {
+            'date': list(all_data.keys()),
+            'all': list(all_data.values()),
+            'me': list(me_data.values()),
+            'partner': list(partner_data.values())
+        }
+        df = pd.DataFrame(d).set_index('date')
+        self.plot_time_series(df, title=name, stat=stat)
+
+    @staticmethod
+    def plot_time_series(df, title="Time series analysis", xlabel='Date', stat='msg_count'):
+        ylabel = f'Stat for {stat}'
+        df.plot(kind='line', linestyle='dashdot', figsize=(16, 5))
+        plt.gca().set(xlabel=xlabel, ylabel=ylabel)
+        plt.title(title)  # does not work
+        plt.legend()
+        plt.show()
+
+    def bar_plot_stat_per_time_period(self, period, name=None, stat='msg_count', **kwargs):
+        people = People(path=TEST_DATA_PATH, name=name)
+        analyzer = Analyzer(people)
+        me_stat = analyzer.stat_per_period(period, statistic=stat, subject='me', **kwargs)
+        partner_stat = analyzer.stat_per_period(period, statistic=stat, subject='partner', **kwargs)
+        d = {
+            'date': list(me_stat.keys()),
+            'me': list(me_stat.values()),
+            'partner': list(partner_stat.values())
+        }
+        df = pd.DataFrame(d).set_index('date')
+        self.plot_stat_per_time(df, stat=stat)
 
     @staticmethod
-    def set_up_data(people, name, period='y'):
-        analyzer = ConversationAnalyzer(name, people.data.get(name).messages)
-        interval_stats = analyzer.get_time_series_data(subject='all', start=None, end=None, period=period)
-        return analyzer.get_plottable_time_series_data(interval_stats, statistic='msg_count')
+    def plot_stat_per_time(df, stat='msg_count', title="Stat per timeperiod", xlabel='Timeperiod', dpi=100):
+        # Stacked bar chart
+        width = 0.35
+        ylabel = f'Stat for {stat}'
+        plt.figure(figsize=(12, 12), dpi=dpi)
+        plt.bar(list(df.index), list(df.me), width, label='me', color='r')
+        plt.bar(list(df.index), list(df.partner), width, label='partner', bottom=list(df.me), color='g')
+        plt.gca().set(xlabel=xlabel, ylabel=ylabel)
+        plt.title(title)  # does not work
+        plt.legend()
+        plt.show()
+
+    def plot_ranking_of_friends_by_message_stats(self, stat='msg_count'):
+        p = People(path=TEST_DATA_PATH, name='Foo Bar')
+        analyzer = Analyzer(p)
+        ranks_dict = analyzer.get_ranking_of_partners_by_messages(attribute=stat)
+        # NOTE maybe this could be done by pandas, but maybe we will use these functions elsewhere
+        sorted_dict = utils.sort_dict(ranks_dict, func=lambda item: item[1], reverse=True)
+        sliced_dict = utils.slice_dict(sorted_dict, 20) if len(sorted_dict) > 20 else sorted_dict
+        cleared_dict = utils.remove_items_where_value_is_falsible(sliced_dict)
+
+        df = pd.DataFrame(cleared_dict, index=[0])
+        self.horizontal_bar_chart(df, stat=stat)
 
     @staticmethod
-    def plot_time_series(x, y, title="Time series", xlabel='Date', ylabel='Value', dpi=100):
-        plt.figure(figsize=(16, 5), dpi=dpi)
-        plt.plot(x, y, color='tab:red')
-        plt.gca().set(title=title, xlabel=xlabel, ylabel=ylabel)
+    def horizontal_bar_chart(df, stat='msg_count', title='Ranking of friends {}', dpi=100):
+        people = list(df.columns)
+        y_pos = np.arange(len(people))
+        performance = df.iloc[0]
+
+        plt.figure(figsize=(12, 12), dpi=dpi)
+        plt.barh(y_pos, performance, align='center', tick_label=people)
+        plt.gca().invert_yaxis()  # labels read top-to-bottom
+        plt.title(title.format(stat))
         plt.show()
 
 
 if __name__ == "__main__":
-    names = ['Teflon Musk', 'Tőke Hal']
+    parser = argparse.ArgumentParser(description='Visualize chat statistics')
+    parser.add_argument(
+        '-p', '--period', metavar='period', type=str, default='y',
+        help='One of {y|m|d|h}, standing for yearly, monthly, daily and hourly breakdown of statisctics.')
+    parser.add_argument(
+        '-n', '--name', metavar='name', type=str, default=None,
+        help="A person's name if you only want data filtered only for them.")
+    parser.add_argument(
+        '-s', '--stat', metavar='stat', type=str, default='msg',
+        help="One of {msg|word|char}, indicating which statistics do you want to get.")
+
+    # TODO add possibility for adding dates from teh command line
+    # https://docs.python.org/3/library/argparse.html#the-add-argument-method
+    args = parser.parse_args()
+    period = args.period
+    name = args.name
+    stat = args.stat
     v = Visualizer()
-    v.plot_convos(names)
+    v.plot_time_series_data_of_messages(period, name=None, stat=f'{stat}_count')
+    v.bar_plot_stat_per_time_period(period, name=None, stat=f'{stat}_count')
+    v.plot_ranking_of_friends_by_message_stats()