Merge pull request #1 from flannelJesus/top20teams

Add a top 20 teams function (grabs a lot more data as well), a function to parse ints, and refactored webpage parsing

main.py

@@ -2,14 +2,17 @@ import requests
 import json
 from bs4 import BeautifulSoup
 
-home = requests.get("http://hltv.org/").text
-home = BeautifulSoup(home, "lxml")
 
-matches = requests.get("http://www.hltv.org/matches/").text
-matches = BeautifulSoup(matches, "lxml")
+def get_parsed_page(url):
+    return BeautifulSoup(requests.get(url).text, "lxml")
+
+
+def parseint(string):
+    return int(''.join([x for x in string if x.isdigit()]))
 
 
 def top5teams():
+    home = get_parsed_page("http://hltv.org/")
     count = 0
     teams = []
     for team in home.find_all("div", {"class": "vsbox", })[:5]:
@@ -18,8 +21,30 @@ def top5teams():
         teams.append(teamname)
     return teams
 
+
+def top20teams():
+    page = get_parsed_page("http://www.hltv.org/ranking/teams/")
+    teams = page.select("div.ranking-box")
+    teamlist = []
+    for team in teams:
+        newteam = {'name': team.select('.ranking-teamName > a')[0].text.strip(),
+                   'rank': parseint(team.select('.ranking-number')[0].text.strip()),
+                   'rank-points': parseint(team.select('.ranking-teamName > span')[0].text),
+                   'team-id': parseint(team.select('.ranking-delta')[0].get('id')),
+                   'team-players': []}
+        for player_div in team.select('.ranking-lineup > div'):
+            player = {}
+            player_anchor = player_div.select('.ranking-playerNick > a')[0]
+            player['name'] = player_anchor.text.strip()
+            player['player-id'] = parseint(player_anchor.get('href'))
+            newteam['team-players'].append(player)
+        teamlist.append(newteam)
+    return teamlist
+
+
+
 def getmatches():
-    match_data = []
+    matches = get_parsed_page("http://www.hltv.org/matches/")
     matchlist = matches.find_all("div", {"class": ["matchListBox", "matchListDateBox"]})
     for match in matchlist:
         if match['class'][0] == "matchListDateBox":
@@ -34,4 +59,4 @@ def getmatches():
                 print(match.text[:7].strip(), match.text[7:-7].strip())
 
 if __name__ == "__main__":
-    getmatches()
+    print(top20teams())