Merge branch 'refactor' into 'master'

Refactor

See merge request !1

1 files changed, 70 insertions, 31 deletions

diff --git a/main.py b/main.py
index 0d12a1e..da7c6ae 100644
--- a/main.py
+++ b/main.py
@@ -1,62 +1,101 @@
-from banapedia.Ban import *
-import bandict
 from collections import Counter
+import webbrowser
+
 import pygal
 import numpy as np
+import pygeoip
+
+from wikibania.ban.BanDB import BanDB
+from wikibania.ban.BanDBWrapper import BanDBWrapper
+import sysproxy
+
+
+# PARAMS
+
+GEOIP_DB = "/usr/share/GeoIP/GeoIP.dat"
 
 BAN_MAP_FILE = "output/ban-map.svg"
 BAN_DURATION_MAP_FILE = "output/ban-duration-map.svg"
 HIST_FILE = "output/histogram.svg"
+STATS_FILE = "output/stats.txt"
+
+BAN_DB_FILE = "resources/ban_list.json"
+
+FETCH_SAMPLES = 2000
+FETCH_DB = False
+DUMP_DB = False
+LOAD_DB = False
+
+OPEN_FILES = False
+
+
+# SETUP
 
-SAMPLES = 1000
+sysproxy.configure_system_proxy()
+geoip_looker = pygeoip.GeoIP(GEOIP_DB)
 
-BAN_FILE = "resources/ban_list.json"
+ban_db = BanDB(geoip_looker)
 
-ban_dict_list = bandict.BanList(BAN_FILE, samples=SAMPLES, from_internet=True)
+if FETCH_DB:
+    ban_db.fetch(FETCH_SAMPLES)
+if LOAD_DB:
+    ban_db.load_file(BAN_DB_FILE)
+if DUMP_DB:
+    ban_db.dump_file(BAN_DB_FILE)
 
-# ======== HISTOGRAM ======= #
+ban_db_wrapper = BanDBWrapper(ban_db)
 
-ban_durations = ban_dict_list.get_durations()
-(ban_durations_bars, bins) = np.histogram(ban_durations,
-                                          bins=[round(365/12*x)
-                                                for x in range(1, 50+2)
-                                                ]
-                                          )
 
-print("[INFO]", "Generating histogram")
+# HISTOGRAM
+
+ban_durations = ban_db_wrapper.get_all_durations()
+duration_bins = [round(365 / 12 * x) for x in range(1, 50 + 2)]
+(ban_durations_bars, bins) = np.histogram(ban_durations, bins=duration_bins)
+
 bar_chart = pygal.Bar(legend_at_bottom=True)
-bar_chart.title = "Active Wikipedia bans by duration (%d samples)" % SAMPLES
-bar_chart.x_labels = map(str, range(1, len(ban_durations_bars)+1))
+bar_chart.title = "Active Wikipedia bans by duration (%d samples)" % len(ban_db.list())
+bar_chart.x_labels = map(str, range(1, len(ban_durations_bars) + 1))
 bar_chart.add("Number of active bans", ban_durations_bars)
 bar_chart.render_to_file(HIST_FILE)
-print("[INFO]", "Histogram generation complete")
 
-# ======= NB BAN MAP ======= #
+if OPEN_FILES:
+    webbrowser.open(HIST_FILE, 2)
+
+
+# BAN DURATION STATS
+
+with open(STATS_FILE, "w") as stats_file:
+    stats_file.write("Mean: %.2f days\n" % np.mean(ban_durations))
+    stats_file.write("Median: %.2f days\n" % np.median(ban_durations))
+    stats_file.write("Deviation: %.2f\n" % np.std(ban_durations))
+    stats_file.write("Variance: %.2f\n" % np.var(ban_durations))
 
-country_ban_list = ban_dict_list.get_countries()
+if OPEN_FILES:
+    webbrowser.open(STATS_FILE, 2)
+
+
+# NB BAN MAP
+
+country_ban_list = ban_db_wrapper.get_all_countries()
 nb_bans_by_country = Counter(country_ban_list)
 
-print("[INFO]", "Generating ban map")
 worldmap_chart = pygal.Worldmap(legend_at_bottom=True)
-worldmap_chart.title = "World active Wikipedia bans by country (%d samples)" % SAMPLES
+worldmap_chart.title = "World active Wikipedia bans by country (%d samples)" % len(ban_db.list())
 worldmap_chart.add("Active bans", nb_bans_by_country)
 worldmap_chart.render_to_file(BAN_MAP_FILE)
-print("[INFO]", "Ban map generation complete")
+
+if OPEN_FILES:
+    webbrowser.open(BAN_MAP_FILE, 2)
 
 
-# ======= BAN DURATION MAP ======= #
+# BAN DURATION MAP
 
-average_ban_duration_by_country = ban_dict_list.average_ban_by_country()
+average_ban_duration_by_country = ban_db_wrapper.calc_average_duration_by_country()
 
-print("[INFO]", "Generating ban duration map")
 worldmap_chart = pygal.Worldmap(legend_at_bottom=True)
-worldmap_chart.title = "Average Wikipedia ban duration by country (%d samples)" % SAMPLES
+worldmap_chart.title = "Average Wikipedia ban duration by country (%d samples)" % len(ban_db.list())
 worldmap_chart.add("Average ban duration (months)", average_ban_duration_by_country)
 worldmap_chart.render_to_file(BAN_DURATION_MAP_FILE)
-print("[INFO]", "Ban duration map generation complete")
 
-print("Some additional stats about ban durations:")
-print("  Mean: %.2f days" % np.mean(ban_durations))
-print("  Median: %.2f days" % np.median(ban_durations))
-print("  Deviation: %.2f" % np.std(ban_durations))
-print("  Variance: %.2f" % np.var(ban_durations))
+if OPEN_FILES:
+    webbrowser.open(BAN_DURATION_MAP_FILE, 2)