Initial commit

6 files changed, 146 insertions, 0 deletions

diff --git a/banapedia/Ban.py b/banapedia/Ban.py
new file mode 100644
index 0000000..d8666b4
--- /dev/null
+++ b/banapedia/Ban.py
@@ -0,0 +1,76 @@
+from banapedia.wapi.WikipediaQuery import BlockQuery
+from datetime import datetime
+import pygeoip
+
+__author__ = 'pacien'
+
+
+GEOIP_FILE = "/usr/share/GeoIP/GeoIP.dat"
+geoip = pygeoip.GeoIP(GEOIP_FILE)
+
+ISO_TIMESTAMP = "%Y-%m-%dT%H:%M:%SZ"
+
+
+class Ban:
+    def __init__(self, ip, start, end):
+        self.ip = ip
+        self.start = start
+        self.end = end
+        self.country_code = None
+
+    def get_duration(self):
+        return (self.end - self.start).days
+
+    def get_country_code(self):
+        if self.country_code is not None:
+            return self.country_code
+
+        country_code = ""
+
+        try:
+            country_code = geoip.country_code_by_addr(self.ip).lower()
+        except pygeoip.GeoIPError:
+            print("[ERROR]", "Could not determine country for ip", self.ip)
+
+        self.country_code = country_code
+        return country_code
+
+
+def map_ban(ban_dict):
+    return Ban(
+        ban_dict["user"],
+        datetime.strptime(ban_dict["timestamp"], ISO_TIMESTAMP),
+        datetime.strptime(ban_dict["expiry"], ISO_TIMESTAMP),
+    )
+
+
+def map_bans(ban_dict_list):
+    ban_list = []
+    for ban_dict in ban_dict_list:
+        ban_list.append(map_ban(ban_dict))
+
+    return ban_list
+
+
+def fetch_multipart_ban_dict(n, query_limit):
+    ban_dict_list = []
+    n_fetched = 0
+    continue_token = None
+
+    print("[INFO]", "Fetching %d bans" % n)
+    while n_fetched < n:
+        to_fetch = min(query_limit, n - n_fetched)
+        query = BlockQuery(
+            bkprop=["user", "timestamp", "expiry"],
+            bkshow=["temp", "ip"],
+            limit=to_fetch,
+            continue_token=continue_token,
+        )
+        results = query.fetch_result()
+        ban_dict_list.extend(results["query"]["blocks"])
+        continue_token = results["query-continue"]["blocks"]["bkcontinue"]
+        n_fetched += to_fetch
+        print("[INFO]", "Fetched %d over %d bans" % (n_fetched, n))
+
+    print("[INFO]", "Bans fetching complete")
+    return ban_dict_list
diff --git a/banapedia/__init__.py b/banapedia/__init__.py
new file mode 100644
index 0000000..93efc48
--- /dev/null
+++ b/banapedia/__init__.py
@@ -0,0 +1,3 @@
+__author__ = 'pacien'
+
+
diff --git a/banapedia/api/Query.py b/banapedia/api/Query.py
new file mode 100644
index 0000000..7453df9
--- /dev/null
+++ b/banapedia/api/Query.py
@@ -0,0 +1,23 @@
+import urllib.parse
+import urllib.request
+import json
+
+__author__ = 'pacien'
+
+
+class Query:
+    def __init__(self, base_url="", params={}, encoding="utf8"):
+        self.base_url = base_url
+        self.params = params
+        self.encoding = encoding
+
+    def fetch_raw_result(self):
+        post_query = urllib.parse.urlencode(self.params)
+        post_query = post_query.encode(self.encoding)
+        document = urllib.request.urlopen(self.base_url, post_query)
+        return document.read().decode(self.encoding)
+
+
+class JSONQuery(Query):
+    def fetch_result(self):
+        return json.loads(self.fetch_raw_result())
diff --git a/banapedia/api/__init__.py b/banapedia/api/__init__.py
new file mode 100644
index 0000000..a0f842f
--- /dev/null
+++ b/banapedia/api/__init__.py
@@ -0,0 +1 @@
+__author__ = 'pacien'
diff --git a/banapedia/wapi/WikipediaQuery.py b/banapedia/wapi/WikipediaQuery.py
new file mode 100644
index 0000000..d3d2f94
--- /dev/null
+++ b/banapedia/wapi/WikipediaQuery.py
@@ -0,0 +1,42 @@
+from ..api.Query import JSONQuery
+
+__author__ = 'pacien'
+
+WIKIPEDIA_QUERY_BASE_URL = "https://en.wikipedia.org/w/api.php"
+LIST_SEPARATOR = "|"
+DEFAULT_BKPROP = ["id", "user", "userid", "by", "byid", "timestamp", "expiry", "reason", "range", "flags"]
+DEFAULT_BKSHOW = ["account", "temp", "ip", "range"]
+
+
+class WikipediaQuery(JSONQuery):
+    def __init__(self, params={}):
+        params.update({
+            "action": "query",
+            "format": "json",
+        })
+        JSONQuery.__init__(self, base_url=WIKIPEDIA_QUERY_BASE_URL, params=params)
+
+
+class ListQuery(WikipediaQuery):
+    def __init__(self, list_name, params={}):
+        params.update({
+            "list": list_name,
+        })
+        WikipediaQuery.__init__(self, params)
+
+
+class BlockQuery(ListQuery):
+    def __init__(self, bkprop=DEFAULT_BKPROP, bkshow=DEFAULT_BKSHOW, bkdir="newer", limit=500, continue_token=None):
+        params = {
+            "bkprop": LIST_SEPARATOR.join(bkprop),
+            "bkshow": LIST_SEPARATOR.join(bkshow),
+            "bkdir": bkdir,
+            "bklimit": limit,
+        }
+
+        if continue_token is not None:
+            params.update({"bkcontinue": continue_token})
+
+        ListQuery.__init__(self, "blocks", params=params)
+
+
diff --git a/banapedia/wapi/__init__.py b/banapedia/wapi/__init__.py
new file mode 100644
index 0000000..a0f842f
--- /dev/null
+++ b/banapedia/wapi/__init__.py
@@ -0,0 +1 @@
+__author__ = 'pacien'