gistfile1.py

from bs4 import BeautifulSoup
import requests
import re

def getYearIDs(year):
	pass

def writeLine(fileName, listOfLists):
	pass

def parseSheet(gameid):
	url = "http://ontariohockeyleague.com/schedule/show/game/%s" % (gameid)
	data = requests.get(url).text
	soup =  BeautifulSoup(data, "html5lib")
	# game summary meta information
	date = soup.findAll("td", "content")[0].findAll('b')[1].text
	print gameid
	print date

	# need to get home / away teams here for meta - can get from next section
	# scoring by period
	topTwo = soup.findAll("table", "margin10 floatLeft")
	scoring = topTwo[0].findAll('tr', 'light')
	cells = scoring[0].findAll('td')
	awayScoring = [gameid, 'Away']+[datum.text for datum in cells]
	cells = scoring[1].findAll('td')
	homeScoring = [gameid, 'Home']+[datum.text for datum in cells]
	# need to check for length
	for teamScoring in [awayScoring, homeScoring]:
		if len(teamScoring) < 9: teamScoring[len(teamScoring)-1:len(teamScoring)-1] \
														= [u'0']*(9-len(teamScoring))
	print awayScoring
	print homeScoring

	# shots by period
	shots = topTwo[1].findAll('tr','light')
	cells = shots[0].findAll('td')
	awayShots = [gameid, 'Away']+[datum.text for datum in cells]
	cells = shots[1].findAll('td')
	homeShots = [gameid, 'Home']+[datum.text for datum in cells]
	# need to check for length
	for teamShots in [awayShots, homeShots]:
		if len(teamShots) < 9: teamShots[len(teamShots)-1:len(teamShots)-1] \
														= [u'0']*(9-len(teamShots))
	print awayShots
	print homeShots

	# rosters of both team
	# goalies of both teams
	middleTables = soup.findAll('td', valign="top") # away is 0, home is 1
	
	awayRows = middleTables[0].findAll('tr', 'light')
	awayRoster = []
	awayGoalies = []
	for row in awayRows[:-2]:
		cells = row.findAll('td')
		if cells[1].text != u'\xa0': awayRoster.append([str(gameid), 'Away']+[datum.text for datum in cells])
	for row in awayRows[-2:]:
		cells = row.findAll('td')
		awayGoalies.append([str(gameid), 'Away']+[datum.text for datum in cells])

	homeRows = middleTables[1].findAll('tr', 'light')
	homeRoster = []
	homeGoalies = []
	for row in homeRows[:-2]:
		cells = row.findAll('td')
		if cells[1].text != u'\xa0': homeRoster.append([str(gameid), 'Away']+[datum.text for datum in cells])
	for row in homeRows[-2:]:
		cells = row.findAll('td')
		homeGoalies.append([str(gameid), 'Away']+[datum.text for datum in cells])	

	print awayRoster
	print homeRoster
	print awayGoalies
	print homeGoalies
	# scoring data
	scoringData = middleTables[2].findAll('tr', 'light')
	scoringOutput = []
	onIceGoals = []
	for scoring in scoringData:
		cells = scoring.findAll('td')
		scoringOutput.append([str(gameid)]+[d.text for d in cells[0].findAll('a')])

		onice = scoring.findAll('a', href="#")
		newHTML = onice[0]['onmouseover']
		start = newHTML.find("'")
		end = newHTML.rfind("'")
		newHTML = BeautifulSoup(newHTML[start+1:end])
		teams = newHTML.findAll("td", "dark")
		plus = []
		minus = []
		onIceData = newHTML.findAll("td", "light")
		for data in onIceData[0].findAll("tr"): plus += data.text.split(" ", 1)
		for data in onIceData[1].findAll("tr"): minus += data.text.split(" ", 1)
		onIceGoals.append([str(gameid), teams[0].text]+plus+[teams[1].text]+minus)

	print scoringOutput
	print onIceGoals

	# penalties data
	penalties = middleTables[3].findAll('tr', 'light')
	penaltiesOutput = []
	for pen in penalties:
		per = pen.findAll('i')[0].text
		person = pen.findAll("a")[0].text
		time = re.findall(r'\b\d?\d:\d\d\b', pen.text)[0]
		penLen = re.findall(r'\b\d min\b', pen.text)[0]
		penTeam = re.findall(r'\b\w\w\w\b', pen.text)[0]
		thePenalty =  pen.text[pen.text.rfind("-")+2:pen.text.rfind(",")]
		penaltiesOutput.append([per, penTeam, person, time, penLen, thePenalty])

	print penaltiesOutput

	# shootout data
	if "shootout".lower() in middleTables[4].text.lower():
		pass

	# PP / PK data
	# arena
	# stars
	# on ice officials
	# off ice officials

parseSheet(63565)

# 66528