import urllib2
import re
import csv
from bs4 import BeautifulSoup

teams = {'ana', 'bos', 'buf', 'cal', 'car', 'chi', 'col', 'cbj', 'dal', 'det', 'edm', 'flo', 'lak', 'min', 'mtl', 'nsh', 'njd', 'nyi', 'nyr', 'ott', 'phi', 'phx', 'pit', 'sjs', 'stl', 'tbl', 'tor', 'van', 'wpg', 'wsh'}
gameidset = set()

print("Collecting game IDs. This may take a while.")

for i in teams:
	response = urllib2.urlopen('http://www.nhl.com/ice/schedulebyseason.htm?team=' + i)
	html = response.read()
	soup = BeautifulSoup(html)
	for link in soup.find_all('a'):
		linkCandidate = str(link.get('href'))
		if linkCandidate.startswith('http://www.nhl.com/gamecenter/en/recap?id='):
			gameidset.add(re.sub(re.escape('http://www.nhl.com/gamecenter/en/recap?id='), '', linkCandidate))

print("Finished collecting game IDs.")

datafile = open('data.csv', 'wb')
wr = csv.writer(datafile, quoting=csv.QUOTE_NONE)
wr.writerow(['GameID', 'HomeTeam', 'AwayTeam', 'HomePIM', 'AwayPIM', 'Referee1', 'Referee2'])

for i in gameidset:
	response = urllib2.urlopen('http://www.nhl.com/gamecenter/en/boxscore?id=' + i)
	html = response.read()
	soup = BeautifulSoup(html)
	gamePack = [i]
	gamePack.append(re.sub('<[^<]+?>', '', str(soup.find("th", "ht"))))
	gamePack.append(re.sub('<[^<]+?>', '', str(soup.find("th", "at"))))
	gamePack.append(re.sub('<[^<]+?>', '', str(soup.find("td", "hPIM"))))
	gamePack.append(re.sub('<[^<]+?>', '', str(soup.find("td", "aPIM"))))
	extraInfo = str(re.sub('<[^<]+?>', '', str(soup.find_all("li"))))
	refs = str(re.findall('Referees:.*Linesmen', extraInfo))
	refs = refs[12:]
	refs = refs[:-12]
	gamePack.extend([x.strip() for x in refs.split(',')])
	print(gamePack)
	wr.writerow(gamePack)

datafile.close()

refList = set()
teams = {'ana', 'bos', 'buf', 'cal', 'car', 'chi', 'col', 'cbj', 'dal', 'det', 'edm', 'flo', 'lak', 'min', 'mtl', 'nsh', 'njd', 'nyi', 'nyr', 'ott', 'phi', 'phx', 'pit', 'sjs', 'stl', 'tbl', 'tor', 'van', 'wpg', 'wsh'}

datafile = open('data.csv', 'rb')
reader = csv.reader(datafile, delimiter=',')
next(reader, None)
for row in reader:
	refList.add(row[5])
	refList.add(row[6])
print(refList)
datafile.close()

datafile = open('data.csv', 'rb')
datafile2 = open('longdata.csv', 'wb')
wr = csv.writer(datafile2, quoting=csv.QUOTE_NONE)
wr.writerow(['Team', 'PIM', 'Referee1', 'Referee2'])
reader = csv.reader(datafile, delimiter=',')
next(reader, None)
for row in reader:
	print([row[1], row[3], row[5], row[6]])
	print([row[2], row[4], row[5], row[6]])
	wr.writerow([row[1], row[3], row[5], row[6]])
	wr.writerow([row[2], row[4], row[5], row[6]])
datafile.close()
datafile2.close()