def get_live_games(seasonID, gamedate, home_team, away_team):
gameid = ""
gameVector = []
#Read in season schedule
scheduleUrl = "http://stats.swehockey.se/ScheduleAndResults/Live/" + str(
seasonID)
response = urllib.urlopen(scheduleUrl)
page_source = str(response.read())
#If vectors dont exist then get vectors
page_source = page_source.replace("\\xc3\\xa5", "å")
page_source = page_source.replace("\\xc3\\xa4", "ä")
page_source = page_source.replace("\\xc2\\xa0", " ")
page_source = page_source.replace("\\xc3\\xa9", "é")
page_source = page_source.replace("\\xc3\\xb6", "ö")
page_source = page_source.replace("\\xc3\\x84", "Ä")
page_source = page_source.replace("\\xc3\\x85", "Å")
page_source = page_source.replace("\\xc3\\x96", "Ö")
page_source = page_source.replace("\\r", " ")
page_source = page_source.replace("\\n", " ")
for i in range(1, len(page_source) - 10):
if isnumber(page_source[i:i +
4]) and page_source[i + 4] == '-' and isnumber(
page_source[i + 5:i + 7]) and page_source[
i + 7] and isnumber(
page_source[i + 8:i + 10]):
currdate = page_source[i:i + 10]
if page_source[i:i + 8] == "/Events/":
gameID = 0
for j in range(1, 10):
if isnumber(page_source[i + 8 + j]) == False:
if gameID == 0:
gameID = page_source[i + 8:i + 8 + j]
stats = get_stats(gameID, gamedate)
if home_team == stats[2] and away_team == stats[3]:
gameid = gameID
return gameid
def get_official_roster(seasonID, season, serie):
gameUrl = "http://stats.swehockey.se/Teams/Info/TeamRoster/" + str(seasonID) + ""
response = urllib.urlopen(gameUrl)
page_source = str(response.read())
output = []
page_source = page_source.replace("\\xc3\\xa5", "å")
page_source = page_source.replace("\\xc3\\xa4", "ä")
page_source = page_source.replace("\\xc2\\xa0", " ")
page_source = page_source.replace("\\xc3\\xa9", "é")
page_source = page_source.replace("\\xc3\\xb6", "ö")
page_source = page_source.replace("\\xc3\\x84", "Ä")
page_source = page_source.replace("\\xc3\\x85", "Å")
page_source = page_source.replace("\\xc3\\x96", "Ö")
page_source = page_source.replace("\\xc3\\xa8", "é")
page_source = page_source.replace("\\r", " ")
page_source = page_source.replace("\\n", " ")
content = get_td_content(page_source)
for i in range(3,len(content)-3):
player = []
if "Team Roster" in content[i]:
team = content[i - 2]
if "Youth club" in content[i] or (isnumber(content[i+1]) and "," in content[i+2] and "-" in content[i+3]):
player.append(season)
player.append(team)
player.append(serie)
player.append(content[i + 1])
n = content[i + 2].find(",")
name = content[i + 2][0:n]
surname = content[i+2][n+1:len(content[i+2])]
if surname[0] == " ":
surname = surname[1:len(surname)]
player.append(name)
player.append(surname)
player.append(content[i + 3])
player.append(content[i + 4])
player.append(content[i + 5])
player.append(content[i + 6])
player.append(content[i + 7])
player.append(content[i + 8][0:3])
#print(player)
output.append(player)
#print(output)
return output
def test_coords(x_diff):
im1 = ImageGrab.grab(bbox=(cV[0] + xV[0] + x_diff, cV[1] + yV[0],
cV[0] + xV[1] + x_diff, cV[1] + yV[1]))
bigimage1 = im1.resize((int(
(xV[0] - xV[1]) * 2), int((yV[0] - yV[1]) * 2)), Image.NEAREST)
bigimage2 = im1.resize((int(
(xV[0] - xV[1]) * 3), int((yV[0] - yV[1]) * 3)), Image.NEAREST)
bigimage3 = im1.resize((int(
(xV[0] - xV[1]) * 5), int((yV[0] - yV[1]) * 5)), Image.NEAREST)
image_string1 = pytesseract.image_to_string(
bigimage1, config='-psm 7 -c tessedit_char_whitelist=0123456789')
image_string2 = pytesseract.image_to_string(
bigimage2, config='-psm 7 -c tessedit_char_whitelist=0123456789')
image_string3 = pytesseract.image_to_string(
bigimage3, config='-psm 7 -c tessedit_char_whitelist=0123456789')
nm = -1
if isnumber(image_string3):
nm = int(image_string3)
elif isnumber(image_string2):
nm = int(image_string2)
elif isnumber(image_string1):
nm = int(image_string1)
if pc in [2, 3, 4, 5, 6, 7]:
category = "Attacks"
elif pc in [8, 9, 10, 11]:
category = "Shots"
elif pc in [16, 17]:
category = "Corners"
elif pc in [12, 13, 14, 15]:
category = "Shots"
if x_diff != 0:
savestring = "./pics/" + category + "/" + str(matchID) + "-" + str(
pc) + "-" + str(x_diff) + ".jpg"
bigimage1 = bigimage1.convert("RGB")
bigimage1.save(savestring, "JPEG")
nm = test_number(nm, pc, min, c, matchID)
return nm
def OCRscore(cV):
h1 = (cV[2] - cV[0]) * 0.715
h2 = (cV[2] - cV[0]) * 0.74
h3 = (cV[2] - cV[0]) * 0.75
h4 = (cV[2] - cV[0]) * 0.775
v1 = (cV[3] - cV[1]) * 0.21
v2 = (cV[3] - cV[1]) * 0.295
im1 = ImageGrab.grab(bbox=(cV[0] + h1, cV[1] + v1, cV[0] + h2, cV[1] + v2))
im2 = ImageGrab.grab(bbox=(cV[0] + h3, cV[1] + v1, cV[0] + h4, cV[1] + v2))
bigimage1 = im1.resize((int((h2 - h1) * 10), int((v2 - v1) * 10)),
Image.NEAREST)
bigimage2 = im2.resize((int((h4 - h3) * 10), int((v2 - v1) * 10)),
Image.NEAREST)
bigimage1 = bigimage1.convert("RGB")
bigimage1.save("pic17n.jpg", "JPEG")
bigimage2 = bigimage2.convert("RGB")
bigimage2.save("pic16n.jpg", "JPEG")
image_string1 = pytesseract.image_to_string(
bigimage1, config='-psm 10 -c tessedit_char_whitelist=0123456789')
image_string2 = pytesseract.image_to_string(
bigimage2, config='-psm 10 -c tessedit_char_whitelist=0123456789')
score1 = -1
score2 = -1
if isnumber(image_string1) == True:
score1 = int(image_string1)
if isnumber(image_string2) == True:
score2 = int(image_string2)
scoreResult = [score1, score2]
return scoreResult
def check_time(pre_val, pre_sec, start_time, start_min, c, matchID):
if isnumber(pre_val):
mins = []
c.execute("SELECT time as dayTime FROM tradedata where ID = ?", [str(matchID)])
dayTime = c.fetchall()
if len(dayTime) > 0:
currTime = max(dayTime)[0]
c.execute("SELECT timeMin as times FROM tradedata where ID = ? and time = ?", [str(matchID),currTime])
mins = c.fetchall()
if len(mins) > 0:
start_time = str(max(dayTime)[0])
start_min = max(mins)[0]
check_number = 1
# Check if time numbers are reasonable
if pre_val < 0 or pre_val > 120:
check_number = 0
d1 = datetime.datetime.strptime((start_time)[0:19], '%Y-%m-%d %H:%M:%S')
d2 = datetime.datetime.strptime(str(datetime.datetime.now())[0:19], '%Y-%m-%d %H:%M:%S')
diff = (d2 - d1).total_seconds() / 60
exp_min = start_min + diff
if len(mins) == 0 and pre_val in [0,1]:
exp_min = 0
if pre_val == 45 and pre_sec == 0:
if start_min > 45:
pass
else:
if abs(exp_min - pre_val) > 2:
pre_val = -1
else:
print("Expected time " + str(round(exp_min, 1)) + " actual " + str(pre_val) + " time accepted")
return pre_val
def add_team_games(seasonID, seasonYear, serie):
scheduleUrl = "http://stats.swehockey.se/ScheduleAndResults/Schedule/" + str(
seasonID)
gameVector = []
venueVector = []
audVector = []
dateVector = []
response = urllib.urlopen(scheduleUrl)
page_source = str(response.read())
import sqlite3
conn = sqlite3.connect('hockeystats.db')
c = conn.cursor()
# Check if vectors exist
c.execute("SELECT * FROM schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
sc = c.fetchall()
if len(sc) == 0:
page_source = page_source.replace("\\xc3\\xa5", "å")
page_source = page_source.replace("\\xc3\\xa4", "ä")
page_source = page_source.replace("\\xc2\\xa0", " ")
page_source = page_source.replace("\\xc3\\xa9", "é")
page_source = page_source.replace("\\xc3\\xb6", "ö")
page_source = page_source.replace("\\xc3\\x84", "Ä")
page_source = page_source.replace("\\xc3\\x85", "Å")
page_source = page_source.replace("\\xc3\\x96", "Ö")
page_source = page_source.replace("\\r", " ")
page_source = page_source.replace("\\n", " ")
for i in range(1, len(page_source) - 10):
if isnumber(page_source[i:i + 4]) and page_source[i + 4] == '-' and isnumber(page_source[i + 5:i + 7]) and \
page_source[i + 7] and isnumber(page_source[i + 8:i + 10]):
dateVector.append(page_source[i:i + 11])
if page_source[i:i + 8] == "/Events/":
gameID = 0
for j in range(1, 10):
if isnumber(page_source[i + 8 + j]) == False:
if gameID == 0:
gameID = page_source[i + 8:i + 8 + j]
gameVector.append(gameID)
audience = ""
tds = get_td_content(
page_source[i:max(len(page_source) - 10, i + 200)])
inserted = 0
for j in range(0, 10):
if isnumber(tds[j]) and inserted == 0:
inserted = 1
audVector.append(int(tds[j]))
venueVector.append(tds[j + 1])
for j in range(0, len(gameVector)):
c.execute(
"INSERT INTO schedule (SEASONID, SERIE, GAMEID, GAMEDATE, AUD, VENUE) VALUES (?,?,?,?,?,?)",
[
seasonYear, serie, gameVector[j], dateVector[j + 1],
audVector[j], venueVector[j]
])
c.execute(
"SELECT GAMEID from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
gameVector = c.fetchall()
c.execute(
"SELECT GAMEDATE from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
dateVector = c.fetchall()
c.execute("SELECT AUD from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
audVector = c.fetchall()
c.execute(
"SELECT VENUE from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
venueVector = c.fetchall()
else:
c.execute(
"SELECT GAMEID from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
gameVector = c.fetchall()
c.execute(
"SELECT GAMEDATE from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
dateVector = c.fetchall()
c.execute("SELECT AUD from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
audVector = c.fetchall()
c.execute(
"SELECT VENUE from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
venueVector = c.fetchall()
for j in range(0, len(gameVector)):
stats = get_stats(gameVector[j][0], dateVector[j][0])
c.execute("SELECT GAMEID as GAMEID FROM stats where GAMEID = ?",
[stats[0]])
hits = c.fetchall()
if len(hits) == 0:
c.execute(
"""INSERT INTO
stats (
SEASONID,SERIE,GAMEID,GAMEDATE,HOMETEAM,AWAYTEAM,HOMESCORE,AWAYSCORE,HOMESHOTS,AWAYSHOTS,HOMESAVES,AWAYSAVES,HOMEPENALTY,AWAYPENALTY,HSCORE1,HSCORE2,HSCORE3,HSCORE4,ASCORE1,ASCORE2,ASCORE3,ASCORE4,
HSHOTS1,HSHOTS2,HSHOTS3,HSHOTS4,ASHOTS1,ASHOTS2,ASHOTS3,ASHOTS4,HSAVES1,HSAVES2,HSAVES3,HSAVES4,ASAVES1,ASAVES2,ASAVES3,ASAVES4,HPENALTY1,HPENALTY2,HPENALTY3,HPENALTY4,APENALTY1,APENALTY2,APENALTY3,
APENALTY4)
VALUES
(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)""",
(seasonYear, serie, stats[0], stats[1], stats[2], stats[3],
stats[4], stats[5], stats[6], stats[7], stats[8], stats[9],
stats[10], stats[11], stats[12], stats[13], stats[14],
stats[15], stats[16], stats[17], stats[18], stats[19],
stats[20], stats[21], stats[22], stats[23], stats[24],
stats[25], stats[26], stats[27], stats[28], stats[29],
stats[30], stats[31], stats[32], stats[33], stats[34],
stats[35], stats[36], stats[37], stats[38], stats[39],
stats[40], stats[41], stats[42], stats[43]))
else:
pass
conn.commit()
create_teamgames(seasonYear, serie)
def import_base_data(pV, input_text, t):
from functions import contains
from functions import get_isolated_number
from functions import isnumber
import re
# Variables to get
# Split up text in lines and delete empty lines
text_list = input_text[t].splitlines()
#print(text_list)
if len(text_list) > 0:
#print(text_list)
for k in range(0, len(text_list)):
if len(text_list[k]) < 4:
text_list[k] = ""
for k in range(0, len(text_list)):
text_list[k] = text_list[k].replace("O", "0")
text_list[k] = text_list[k].replace("'I", "1")
text_list[k] = text_list[k].upper()
#Get time:
time_row = len(text_list)
for i in range(1, len(text_list)):
if ":" in text_list[i]:
if len(text_list[i]) == 5:
pV[0].append(text_list[i])
time_row = i
elif len(text_list[i]) > 5:
p = text_list[i].find(":")
if isnumber(text_list[i][p - 2:p]):
pV[0].append(text_list[i][p - 2:p + 3])
time_row = i
#Get attacks, dangerous attacks, possession
for i in range(0, len(text_list)):
if len(text_list[i]) > 6 and contains(text_list[i].upper(),
"TARGET", 4) == False:
text_list[i] = text_list[i].replace(">", "")
text_list[i] = text_list[i].replace(")", "")
text_list[i] = text_list[i].replace("(", "")
text_list[i] = text_list[i].replace("C", "")
#print(text_list[i])
if len(get_isolated_number(text_list[i])) > 2:
list = get_isolated_number(text_list[i])
#print(list)
for j in range(0, len(list)):
list[j] = re.sub("[^0-9]", "", list[j])
if len(list) == 4:
pV[1].append(list[0])
pV[2].append(list[1])
pV[3].append(list[2])
pV[4].append(list[3])
pos1 = float(int(list[0])) * 1.5
pos2 = float(int(list[1])) * 1.5
pos3 = float(int(list[2]))
pos4 = float(int(list[3]))
ep1 = 100 * (pos1 + pos3) / (pos1 + pos2 + pos3 + pos4)
ep2 = 100 - ep1
est_pos1 = str(round(ep1))
est_pos2 = str(round(ep2))
pV[5].append(est_pos1)
pV[6].append(est_pos2)
elif len(list) == 6:
pV[1].append(list[0])
pV[2].append(list[1])
pV[3].append(list[2])
pV[4].append(list[3])
pV[5].append(list[4])
pV[6].append(list[5])
elif len(list) > 6:
pass
#Get Shots on target and shot outside
if len(text_list[i]) > 10 and contains(text_list[i], "OFF TARGET",
4) == True:
for k in range(0, len(text_list[i])):
text_list[i] = text_list[i].replace("l", "1")
text_list[i] = text_list[i].replace("'", "")
text_list[i] = text_list[i].replace("´", "")
text_list[i] = text_list[i].replace("‘", "")
text_list[i] = text_list[i].replace(">", "")
text_list[i] = text_list[i].replace(")", "")
text_list[i] = text_list[i].replace("(", "")
text_list[i] = text_list[i].replace("C", "")
list = get_isolated_number(text_list[-1])
#print(list)
if len(list) == 8:
pV[9].append(list[3])
pV[10].append(list[4])
pV[11].append(list[2])
pV[12].append(list[5])
pV[13].append(list[1])
pV[14].append(list[6])
pV[15].append(list[0])
pV[16].append(list[7])
for i in range(0, len(text_list)):
if len(text_list[i]) > 8 and contains(
text_list[i], "ON TARGET",
4) == True and "OF" not in text_list[
i] and "FF" not in text_list[i]:
#print(text_list[i])
list = get_isolated_number(text_list[i])
if len(list) == 2:
pV[7].append(list[0])
pV[8].append(list[1])
if len(list) == 3:
if list[1] == "0N":
pV[7].append(list[0])
pV[8].append(list[2])
#Get score
for i in range(0, time_row):
#print(text_list[i])
if len(get_isolated_number(text_list[i])) == 2:
pV[17].append(get_isolated_number(text_list[i])[0])
pV[18].append(get_isolated_number(text_list[i])[1])
elif len(get_isolated_number(text_list[i])) > 2:
nums = get_isolated_number(text_list[i])
for j in range(0, len(nums)):
if len(nums) > j:
if isnumber(nums[j]):
if int(nums[j]) > 9:
del nums[j]
if len(nums) == 2:
pV[17].append(nums[0])
pV[18].append(nums[1])
return pV
#If vectors dont exist then get vectors
page_source = page_source.replace("\\xc3\\xa5", "å")
page_source = page_source.replace("\\xc3\\xa4", "ä")
page_source = page_source.replace("\\xc2\\xa0", " ")
page_source = page_source.replace("\\xc3\\xa9", "é")
page_source = page_source.replace("\\xc3\\xb6", "ö")
page_source = page_source.replace("\\xc3\\x84", "Ä")
page_source = page_source.replace("\\xc3\\x85", "Å")
page_source = page_source.replace("\\xc3\\x96", "Ö")
page_source = page_source.replace("\\r", " ")
page_source = page_source.replace("\\n", " ")
for i in range(1, len(page_source) - 10):
if isnumber(
page_source[i:i + 4]) and page_source[i + 4] == '-' and isnumber(
page_source[i + 5:i + 7]) and page_source[i + 7] and isnumber(
page_source[i + 8:i + 10]):
currdate = page_source[i:i + 10]
if page_source[i:i + 8] == "/Events/":
gameID = 0
for j in range(1, 10):
if isnumber(page_source[i + 8 + j]) == False:
if gameID == 0:
gameID = page_source[i + 8:i + 8 + j]
gameVector.append(gameID)
dateVector.append(currdate)
def get_year_statistics(id,seasonYear,serie):
gameUrl = "http://stats.swehockey.se/Teams/Info/PlayersByTeam/" + str(id)
response = urllib.urlopen(gameUrl)
page_source = str(response.read())
page_source = page_source.replace("\\xc3\\xa5", "å")
page_source = page_source.replace("\\xc3\\xa4", "ä")
page_source = page_source.replace("\\xc2\\xa0", " ")
page_source = page_source.replace("\\xc3\\xa9", "é")
page_source = page_source.replace("\\xc3\\xb6", "ö")
page_source = page_source.replace("\\xc3\\x84", "Ä")
page_source = page_source.replace("\\xc3\\x85", "Å")
page_source = page_source.replace("\\xc3\\x96", "Ö")
page_source = page_source.replace("\\r", " ")
page_source = page_source.replace("\\n", " ")
tds = get_td_content(page_source)
import sqlite3
conn = sqlite3.connect('hockeystats.db')
c = conn.cursor()
c.execute("SELECT DISTINCT TEAM FROM rosters where seasonid = ? and serie = ?", [seasonYear, serie])
teams = c.fetchall()
for j in range(0,len(teams)):
last_n = 0
search_keepers = 0
team_found = 0
for i in range(0,len(tds)):
if teams[j][0] in tds[i]:
team_found += 1
if team_found == 2:
team = tds[i]
if "," in tds[i] and tds[i+1] in ["GK","LD","RD","LW","RW","CE"]:
forname = tds[i][tds[i].find(",")+2:len(tds[i])]
forname = forname.replace("*","")
surname = tds[i][0:tds[i].find(",")]
games = tds[i+2]
goals = tds[i+3]
assist = tds[i+4]
penalty = tds[i+6]
plus = tds[i+7]
minus = tds[i+8]
shots = 0
saves = 0
c.execute("UPDATE rosters SET GAMES = ?, GOALS = ?, ASSIST = ?, PENALTY = ?, PLUS = ?, MINUS = ?, SHOTS = ?, SAVES = ? WHERE SEASONID = ? and SERIE = ? and TEAM = ? and FORNAME = ? and SURNAME = ?",[games, goals, assist, penalty, plus, minus, shots, saves, seasonYear, serie, team, forname, surname])
conn.commit()
if isnumber(tds[i]) and isnumber(tds[i + 1]) == True and "," in tds[i + 2]:
if last_n > int(tds[i]):
if search_keepers == 0:
search_keepers = 1
else:
search_keepers = 0
last_n = int(tds[i])
if search_keepers == 1:
if ":" in tds[i + 6]:
forname = tds[i+2][tds[i+2].find(",") + 2:len(tds[i+2])]
forname = forname.replace("*", "")
surname = tds[i+2][0:tds[i+2].find(",")]
games = tds[i+5]
goals = 0
assist = 0
penalty = 0
plus = 0
minus = 0
shots = tds[i+9]
saves = tds[i+8]
c.execute("UPDATE rosters SET GAMES = ?, SHOTS = ?, SAVES = ? WHERE SEASONID = ? and SERIE = ? and TEAM = ? and FORNAME = ? and SURNAME = ?",
[games, shots, saves, seasonYear, serie, team, forname,
surname])
conn.commit()
pV = [[], [], [], [], [], [], [], [], [], [], [], [], [], [], [], [], [],
[], []]
fV = [
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-1, -1
]
for t in range(0, n_iter):
pV = import_base_data(pV, input_text, t)
#print(input_text[t])
### Create final suggested vector based on the five versions ###
if pV[0]:
if isnumber(most_common(pV[0][:])[0:2]) and isnumber(
most_common(pV[0][:])[3:5]):
fV[0] = int(most_common(pV[0][:])[0:2])
fV[1] = int(most_common(pV[0][:])[3:5])
for i in range(1, 19):
if len(pV[i][:]) > 1:
if isnumber(most_common(pV[i][:])):
fV[i + 1] = int(most_common(pV[i][:]))
else:
fV[i + 1] = -1
elif len(pV[i][:]) == 1:
if isnumber((pV[i][0])):
fV[i + 1] = int(pV[i][0])
else:
def get_actions(id, audience, venue, season, team1, team2, c):
gameUrl = "http://stats.swehockey.se/Game/Events/" + str(id)
response = urllib.urlopen(gameUrl)
page_source = str(response.read())
page_source = page_source.replace("\\xc3\\xa5", "å")
page_source = page_source.replace("\\xc3\\xa4", "ä")
page_source = page_source.replace("\\xc2\\xa0", " ")
page_source = page_source.replace("\\xc3\\xa9", "é")
page_source = page_source.replace("\\xc3\\xb6", "ö")
page_source = page_source.replace("\\xc3\\x84", "Ä")
page_source = page_source.replace("\\xc3\\x85", "Å")
page_source = page_source.replace("\\xc3\\x96", "Ö")
page_source = page_source.replace("\\r", " ")
page_source = page_source.replace("\\n", " ")
content = get_td_content(page_source)
period = 0
events = []
for i in range(0, len(content)):
stn = content[i]
if "Overtime" in stn:
period = 4
elif "3rd" in stn:
period = 3
elif "2nd" in stn:
period = 2
elif "1st" in stn:
period = 1
if "%" in content[i] and "(" in content[i + 1] and ")" in content[
i + 1] and "/" in content[i + 1]:
#print(content[i+1])
[saves, shots] = get_all_numbers(content[i + 1])
event = create_goalie_event(id, period, content[i - 2:i + 1],
shots, saves, audience, venue, season)
events.append(event)
if isnumber(content[i][0:2]) and isnumber(
content[i][3:5]) and content[i][2] == ":":
event = create_event(id, period, content[i - 1:i + 6], audience,
venue, season)
events.append(event)
if event[3] == "Goal":
extra = event[8]
for j in range(5, 9):
if isnumber(content[i + j][0]) and content[i + j].find(
".") > 0 and content[i + j].find(".") < 5:
event = create_assist_event(content[i + j], event,
audience, venue, season,
extra)
events.append(event)
if "Neg." in content[i + j]:
numbers = (get_all_numbers(content[i + j]))
for k in range(0, len(numbers)):
event = create_plus_minus_event(
event, -1, numbers[k], audience, venue, season,
extra)
events.append(event)
if "Pos." in content[i + j]:
numbers = (get_all_numbers(content[i + j]))
for k in range(0, len(numbers)):
event = create_plus_minus_event(
event, 1, numbers[k], audience, venue, season,
extra)
events.append(event)
home_team_short = events[-1][4]
for i in range(0, len(events)):
if events[i][3] != -1:
if events[i][4] == home_team_short:
events[i][4] = team1
else:
events[i][4] = team2
else:
if events[i][4] == home_team_short:
events[i][4] = team2
else:
events[i][4] = team1
if events[i][3] in [-1, 1]:
c.execute(
"SELECT FORNAME, SURNAME FROM lineups where GAMEID = ? and TEAM = ? and NUMBER = ?",
[events[i][0], events[i][4], events[i][5]])
player_name = c.fetchall()
events[i][6] = ""
events[i][7] = ""
try:
events[i][6] = player_name[0][1]
except IndexError:
pass
try:
events[i][7] = player_name[0][0]
except IndexError:
pass
#print(events)
return events
def scrape_sh(seasonID, seasonYear, serie, score_update):
#Vectors to scrape in first step
gameVector = []
venueVector = []
audVector = []
dateVector = []
lineVector = []
#Read in season schedule
scheduleUrl = "http://stats.swehockey.se/ScheduleAndResults/Schedule/" + str(
seasonID)
response = urllib.urlopen(scheduleUrl)
page_source = str(response.read())
#Establish connection to database
import sqlite3
conn = sqlite3.connect('hockeystats.db')
c = conn.cursor()
#Check if vectors exist
c.execute("SELECT * FROM schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
sc = c.fetchall()
if len(sc) > 0:
c.execute("DELETE FROM SCHEDULE")
#If vectors dont exist then get vectors
page_source = page_source.replace("\\xc3\\xa5", "å")
page_source = page_source.replace("\\xc3\\xa4", "ä")
page_source = page_source.replace("\\xc2\\xa0", " ")
page_source = page_source.replace("\\xc3\\xa9", "é")
page_source = page_source.replace("\\xc3\\xb6", "ö")
page_source = page_source.replace("\\xc3\\x84", "Ä")
page_source = page_source.replace("\\xc3\\x85", "Å")
page_source = page_source.replace("\\xc3\\x96", "Ö")
page_source = page_source.replace("\\r", " ")
page_source = page_source.replace("\\n", " ")
#Update rosters
update_rosters(seasonID, seasonYear, serie, c, conn)
for i in range(1, len(page_source) - 10):
if isnumber(page_source[i:i +
4]) and page_source[i + 4] == '-' and isnumber(
page_source[i + 5:i + 7]) and page_source[
i + 7] and isnumber(
page_source[i + 8:i + 10]):
currdate = page_source[i:i + 10]
if page_source[i:i + 8] == "/Events/":
gameID = 0
for j in range(1, 10):
if isnumber(page_source[i + 8 + j]) == False:
if gameID == 0:
gameID = page_source[i + 8:i + 8 + j]
gameVector.append(gameID)
dateVector.append(currdate)
audience = ""
tds = get_td_content(
page_source[i:max(len(page_source) - 10, i + 200)])
inserted = 0
for j in range(0, 10):
if isnumber(tds[j]) and inserted == 0:
inserted = 1
audVector.append(int(tds[j]))
venueVector.append(tds[j + 1])
for j in range(0, len(gameVector)):
c.execute(
"INSERT INTO schedule (SEASONID, SERIE, GAMEID, GAMEDATE, AUD, VENUE) VALUES (?,?,?,?,?,?)",
[
seasonYear, serie, gameVector[j], dateVector[j], audVector[j],
venueVector[j]
])
c.execute("SELECT GAMEID from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
gameVector = c.fetchall()
c.execute("SELECT GAMEDATE from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
dateVector = c.fetchall()
c.execute("SELECT AUD from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
audVector = c.fetchall()
c.execute("SELECT VENUE from schedule where SEASONID = ? and SERIE = ?",
[seasonYear, serie])
venueVector = c.fetchall()
conn.commit()
########################################################################################################################
################################ Get game specific statistics (Lineups) ##########################################
########################################################################################################################
#Loop through games
for j in range(0, len(gameVector)):
#Check if game already has been updated, then skip update
c.execute("SELECT * FROM lineups where GAMEID = ?", [gameVector[j][0]])
check = c.fetchall()
if len(check) == 0:
stats = get_stats(gameVector[j][0], dateVector[j][0])
lineups = get_lineups(gameVector[j][0], audVector[j][0],
venueVector[j][0], seasonYear, stats[2],
stats[3])
[refs, lines] = get_refs(gameVector[j][0], audVector[j][0],
venueVector[j][0], seasonYear)
# Create stats table
c.execute("SELECT GAMEID as GAMEID FROM stats where GAMEID = ?",
[stats[0]])
hits = c.fetchall()
if len(hits) == 0:
c.execute(
"""INSERT INTO
stats (
SEASONID,SERIE,GAMEID,GAMEDATE,HOMETEAM,AWAYTEAM,HOMESCORE,AWAYSCORE,HOMESHOTS,AWAYSHOTS,HOMESAVES,AWAYSAVES,HOMEPENALTY,AWAYPENALTY,HSCORE1,HSCORE2,HSCORE3,HSCORE4,ASCORE1,ASCORE2,ASCORE3,ASCORE4,
HSHOTS1,HSHOTS2,HSHOTS3,HSHOTS4,ASHOTS1,ASHOTS2,ASHOTS3,ASHOTS4,HSAVES1,HSAVES2,HSAVES3,HSAVES4,ASAVES1,ASAVES2,ASAVES3,ASAVES4,HPENALTY1,HPENALTY2,HPENALTY3,HPENALTY4,APENALTY1,APENALTY2,APENALTY3,
APENALTY4, HOMEPP, AWAYPP)
VALUES
(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)""",
(seasonYear, serie, stats[0], stats[1][0:10], stats[2],
stats[3], stats[4], stats[5], stats[6], stats[7],
stats[8], stats[9], stats[10], stats[11], stats[12],
stats[13], stats[14], stats[15], stats[16], stats[17],
stats[18], stats[19], stats[20], stats[21], stats[22],
stats[23], stats[24], stats[25], stats[26], stats[27],
stats[28], stats[29], stats[30], stats[31], stats[32],
stats[33], stats[34], stats[35], stats[36], stats[37],
stats[38], stats[39], stats[40], stats[41], stats[42],
stats[43], stats[44], stats[45]))
else:
pass
conn.commit()
#Create lineup table
for i in range(0, len(lineups)):
c.execute(
"SELECT ID as ID FROM lineups where GAMEID = ? and TEAM = ? and NUMBER = ? and FORNAME = ? and SURNAME = ?",
[
lineups[i][0], lineups[i][1], lineups[i][2],
lineups[i][3], lineups[i][4]
])
hits = c.fetchall()
c.execute("SELECT ID as ID FROM lineups")
ids = c.fetchall()
if len(ids) > 0:
id = max(ids)[0] + 1
else:
id = 1
if len(hits) == 0:
c.execute(
"""INSERT INTO
lineups (
ID,GAMEID,SEASONID,SERIE,AUDIENCE,VENUE,HOMETEAM,AWAYTEAM,TEAM,GAMEDATE,NUMBER,FORNAME,SURNAME,POSITION,START_PLAYER,
GOALS, PPGOALS, SHGOALS, ASSISTS, PLUS, MINUS, PENALTY, INPOWERPLAY, INBOXPLAY, SHOTSAT, SAVES, SCORE, FINALSCORE, SCORE5, GOALS5, ASSIST5, GAMES5, SCORE_CURRENT, GOALS_CURRENT, ASSIST_CURRENT, GAMES_CURRENT)
VALUES
(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)""",
(id, lineups[i][0], lineups[i][9], serie,
lineups[i][7], lineups[i][8], stats[2], stats[3],
lineups[i][1], stats[1][0:10], lineups[i][2],
lineups[i][3], lineups[i][4], lineups[i][5],
lineups[i][6]))
else:
pass
conn.commit()
c.execute(
"SELECT PERSONNR from rosters where SEASONID = ? and TEAM = ? and NUMBER = ?",
[seasonYear, lineups[i][1], lineups[i][2]])
personnr = c.fetchall()
persnr = ''
if len(personnr) > 0:
persnr = personnr[0][0]
c.execute(
"UPDATE lineups SET PERSONNR = ? WHERE SEASONID = ? and TEAM = ? and NUMBER = ? and FORNAME = ? and SURNAME = ?",
[
persnr, seasonYear, lineups[i][1], lineups[i][2],
lineups[i][3], lineups[i][4]
])
conn.commit()
# Get events data from each game
events = get_actions(gameVector[j][0], audVector[j][0],
venueVector[j][0], seasonYear, stats[2],
stats[3], c)
# Create event table
for i in range(0, len(events)):
c.execute(
"SELECT ID FROM events where GAMEID = ? and TIME = ? and EVENT = ? and TEAM = ? and NUMBER = ? and FORNAME = ? and SURNAME = ?",
[
events[i][0], events[i][2], events[i][3], events[i][4],
events[i][5], events[i][7], events[i][6]
])
hits = c.fetchall()
c.execute("SELECT ID as ID FROM events")
ids = c.fetchall()
if len(ids) > 0:
id = max(ids)[0] + 1
else:
id = 1
if len(hits) == 0:
c.execute(
"""SELECT PERSONNR FROM rosters WHERE SEASONID = ? and TEAM = ? and NUMBER = ? and FORNAME = ? AND SURNAME = ? """,
(seasonYear, events[i][4], events[i][5], events[i][7],
events[i][6]))
personnr = c.fetchall()
if personnr == []:
pnr = ""
else:
pnr = personnr[0][0]
c.execute(
"""INSERT INTO
events (
ID,GAMEID,SEASONID,AUDIENCE,VENUE,PERIOD,TIME,EVENT,TEAM,NUMBER,PERSONNR,FORNAME,SURNAME,EXTRA1,EXTRA2)
VALUES
(?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)""",
(id, events[i][0], events[i][12], events[i][10],
events[i][11], events[i][1], events[i][2],
events[i][3], events[i][4], events[i][5], pnr,
events[i][7], events[i][6], events[i][8],
events[i][9]))
else:
pass
conn.commit()
# Create ref table
if serie == "SHL":
for i in range(0, len(events)):
c.execute(
"SELECT GAMEID as GAMEID FROM refs where GAMEID = ?",
[stats[0]])
hits = c.fetchall()
if len(hits) == 0:
c.execute(
"""INSERT INTO
refs (
GAMEID,SEASONID,HOMETEAM,AWAYTEAM,REF1,REF2,LINE1,LINE2)
VALUES
(?,?,?,?,?,?,?,?)""",
(stats[0], seasonYear, refs[0], refs[1], stats[2],
stats[3], lines[0], lines[1]))
else:
pass
conn.commit()
print(str(stats[0]) + " lineups loaded")
#Update lineups with stats
c.execute(
"SELECT TEAM, NUMBER, FORNAME, SURNAME, GAMEDATE FROM lineups where GAMEID = ?",
[gameVector[j][0]])
lineups = c.fetchall()
for i in range(0, len(lineups)):
c.execute(
"SELECT SUM(CASE WHEN EVENT = ? then 1 else 0 end) as X FROM events where GAMEID = ? and TEAM = ? and NUMBER = ?",
['Goal', gameVector[j][0], lineups[i][0], lineups[i][1]])
goals = c.fetchall()[0][0]
if goals == None:
goals = 0
c.execute(
"SELECT SUM(CASE WHEN EVENT = ? and EXTRA1 = ? then 1 else 0 end) as X FROM events where GAMEID = ? and TEAM = ? and NUMBER = ?",
[
'Goal', 'PP', gameVector[j][0], lineups[i][0],
lineups[i][1]
])
PP = c.fetchall()[0][0]
if PP == None:
PP = 0
c.execute(
"SELECT SUM(CASE WHEN EVENT = ? and EXTRA1 = ? then 1 else 0 end) as X FROM events where GAMEID = ? and TEAM = ? and NUMBER = ?",
[
'Goal', 'SH', gameVector[j][0], lineups[i][0],
lineups[i][1]
])
SH = c.fetchall()[0][0]
if SH == None:
SH = 0
c.execute(
"SELECT SUM(CASE WHEN EVENT = ? then 1 else 0 end) as X FROM events where GAMEID = ? and TEAM = ? and NUMBER = ?",
['Assist', gameVector[j][0], lineups[i][0], lineups[i][1]])
assist = c.fetchall()[0][0]
if assist == None:
assist = 0
c.execute(
"SELECT SUM(CASE WHEN EVENT = ? then 1 else 0 end) as X FROM events where GAMEID = ? and TEAM = ? and NUMBER = ?",
['1', gameVector[j][0], lineups[i][0], lineups[i][1]])
plus = c.fetchall()[0][0]
if plus == None:
plus = 0
c.execute(
"SELECT SUM(CASE WHEN EVENT = ? then 1 else 0 end) as X FROM events where GAMEID = ? and TEAM = ? and NUMBER = ? and (extra1 = ? or extra1 = ?)",
[
'-1', gameVector[j][0], lineups[i][0], lineups[i][1],
'', 'PP'
])
minus = c.fetchall()[0][0]
if minus == None:
minus = 0
c.execute(
"SELECT SUM(CASE WHEN EVENT = ? then CAST(EXTRA2 as INT) else 0 end) as X FROM events where GAMEID = ? and TEAM = ? and NUMBER = ?",
[
'Penalty', gameVector[j][0], lineups[i][0],
lineups[i][1]
])
penalty = c.fetchall()[0][0]
if penalty == None:
penalty = 0
c.execute(
"SELECT SUM(CASE WHEN EXTRA1 = ? then 1 else 0 end) as X FROM events where GAMEID = ? and TEAM = ? and NUMBER = ?",
['PP', gameVector[j][0], lineups[i][0], lineups[i][1]])
activePP = c.fetchall()[0][0]
if activePP == None:
activePP = 0
elif activePP > 1:
activePP = 1
c.execute(
"SELECT SUM(CASE WHEN EXTRA1 = ? then 1 else 0 end) as X FROM events where GAMEID = ? and TEAM = ? and NUMBER = ?",
['SH', gameVector[j][0], lineups[i][0], lineups[i][1]])
activeBP = c.fetchall()[0][0]
if activeBP == None:
activeBP = 0
elif activeBP > 1:
activeBP = 1
#Addera kod för shots/saves
c.execute(
"SELECT EXTRA1, EXTRA2 from events where EVENT = ? and GAMEID = ? and TEAM = ? and NUMBER = ?",
[
'Keeper stat', gameVector[j][0], lineups[i][0],
lineups[i][1]
])
golieStats = c.fetchall()
if golieStats == []:
shotsAt = 0
saves = 0
else:
shotsAt = golieStats[0][0]
saves = golieStats[0][1]
c.execute(
"UPDATE lineups SET GOALS = ?, PPGOALS = ?, SHGOALS = ?, ASSISTS = ?, PLUS = ?, MINUS = ?, PENALTY = ?, INPOWERPLAY = ?, INBOXPLAY = ?, SHOTSAT = ?, SAVES = ? WHERE GAMEID = ? and TEAM = ? and NUMBER = ?",
[
goals, PP, SH, assist, plus, minus, penalty, activePP,
activeBP, shotsAt, saves, gameVector[j][0],
lineups[i][0], lineups[i][1]
])
conn.commit()
# Update lineups with old stats
# Last five games
c.execute(
"SELECT SCORE, GOALS, ASSISTS, GAMEDATE, TEAM, FORNAME, SURNAME FROM lineups WHERE GAMEDATE < ? and GAMEDATE > ? and FORNAME = ? and SURNAME = ? and PERSONNR = ? ORDER BY GAMEDATE DESC",
[
lineups[i][4],
transform_date(lineups[i][4], 20), lineups[i][2],
lineups[i][3], persnr
])
output = np.array(c.fetchall())
games5 = min(len(output), 5)
score5 = 0
goals5 = 0
assist5 = 0
for k in range(0, games5):
score5 += float(output[k][0]) / games5
goals5 += float(output[k][1]) / games5
assist5 += float(output[k][2]) / games5
# Current season
c.execute(
"SELECT SCORE, GOALS, ASSISTS, GAMEDATE, TEAM, FORNAME, SURNAME FROM lineups WHERE SEASONID = ? and FORNAME = ? and SURNAME = ? and PERSONNR = ? ORDER BY GAMEDATE DESC",
[seasonYear, lineups[i][2], lineups[i][3], persnr])
output = np.array(c.fetchall())
gamescurr = len(output)
scorecurr = 0
goalscurr = 0
assistcurr = 0
for k in range(0, gamescurr):
scorecurr += float(output[k][0]) / gamescurr
goalscurr += float(output[k][1]) / gamescurr
assistcurr += float(output[k][2]) / gamescurr
c.execute(
"""UPDATE lineups SET SCORE5 = ?, GOALS5 = ?, ASSIST5 = ?, SCORE_CURRENT = ?, GOALS_CURRENT = ?, ASSIST_CURRENT = ?, GAMES5 = ?, GAMES_CURRENT = ?
WHERE GAMEID = ? and PERSONNR = ? and FORNAME = ? and SURNAME = ?""",
[
score5, goals5, assist5, scorecurr, goalscurr,
assistcurr, games5, gamescurr, gameVector[j][0],
persnr, lineups[i][2], lineups[i][3]
])
conn.commit()
#Create teamgames table
create_teamgames(seasonYear, serie, c)
print(str(stats[0]) + " stats, events loaded")
else:
print("Game already loaded")
if t_count == 1:
t += 1
print(str(t) + "/" + str(len(gameVector)) + " done")
#Update score for games
for j in range(0, len(gameVector)):
c.execute("SELECT GAMEID FROM TEAMSCORE WHERE GAMEID = ?",
[gameVector[j][0]])
check = c.fetchall()
if score_update == "Full" or len(check) == 0:
# Add score to lineups
c.execute(
"SELECT TEAM, NUMBER, FORNAME, SURNAME, GAMEDATE, HOMETEAM, AWAYTEAM FROM lineups where GAMEID = ?",
[gameVector[j][0]])
lineups = c.fetchall()
for i in range(0, len(lineups)):
c.execute(
"SELECT * from lineups where GAMEID = ? and TEAM = ? and NUMBER = ?",
[gameVector[j][0], lineups[i][0], lineups[i][1]])
lineup = c.fetchall()
score = create_game_rating(lineup, lineups[i][0], c, conn)
if len(score) < 4:
score = ['0', '0', '0', '0']
c.execute(
"UPDATE lineups SET SCORE = ?, FINALSCORE = ?, OFFSCORE = ?, DEFSCORE = ? WHERE GAMEID = ? and TEAM = ? and NUMBER = ?",
[
score[0], score[1], score[2], score[3],
gameVector[j][0], lineups[i][0], lineups[i][1]
])
#Calculate team strenght
if len(lineups) > 0:
# Check score home team
[team_strenght, form_score, last_seasons_score, player_score
] = calculate_team_strength(lineups[0][5], lineups[0][4], "",
c)
c.execute(
"SELECT * FROM TEAMSCORE WHERE GAMEDATE = ? AND TEAM = ?",
[lineups[0][4], lineups[0][5]])
chk = c.fetchall()
if len(chk) == 0:
c.execute(
"INSERT INTO TEAMSCORE (SEASONID, SERIE, GAMEID, GAMEDATE, TEAM, SCORE, FORM_SCORE, LAST_SEASONS_SCORE, PLAYER_SCORE) VALUES (?,?,?,?,?,?,?,?,?)",
[
seasonYear, serie, gameVector[j][0], lineups[0][4],
lineups[0][5], team_strenght, form_score,
last_seasons_score, player_score
])
else:
c.execute(
"UPDATE TEAMSCORE SET SCORE = ?, FORM_SCORE = ?, LAST_SEASONS_SCORE = ?, PLAYER_SCORE = ? WHERE SEASONID = ? AND SERIE = ? AND GAMEID = ? AND TEAM = ?",
[
team_strenght, form_score, last_seasons_score,
player_score, seasonYear, serie, gameVector[j][0],
lineups[0][5]
])
#Check score away team
[team_strenght, form_score, last_seasons_score, player_score
] = calculate_team_strength(lineups[0][6], lineups[0][4], "",
c)
c.execute(
"SELECT * FROM TEAMSCORE WHERE GAMEDATE = ? AND TEAM = ?",
[lineups[0][4], lineups[0][6]])
chk = c.fetchall()
if len(chk) == 0:
c.execute(
"INSERT INTO TEAMSCORE (SEASONID, SERIE, GAMEID, GAMEDATE, TEAM, SCORE, FORM_SCORE, LAST_SEASONS_SCORE, PLAYER_SCORE) VALUES (?,?,?,?,?,?,?,?,?)",
[
seasonYear, serie, gameVector[j][0], lineups[0][4],
lineups[0][6], team_strenght, form_score,
last_seasons_score, player_score
])
else:
c.execute(
"UPDATE TEAMSCORE SET SCORE = ?, FORM_SCORE = ?, LAST_SEASONS_SCORE = ?, PLAYER_SCORE = ? WHERE SEASONID = ? AND SERIE = ? AND GAMEID = ? AND TEAM = ?",
[
team_strenght, form_score, last_seasons_score,
player_score, seasonYear, serie, gameVector[j][0],
lineups[0][6]
])
print("Score updated")
conn.commit()
# Update pre-game info
for j in range(0, len(gameVector)):
c.execute("SELECT GAMEID FROM EXP_SHOTS_TABLE WHERE GAMEID = ?",
[gameVector[j][0]])
check = c.fetchall()
if len(check) == 0:
c.execute("SELECT HOMETEAM, AWAYTEAM FROM STATS WHERE GAMEID = ?",
[gameVector[j][0]])
teams = c.fetchall()
create_pre_match_analysis(dateVector[j][0], seasonID, serie,
teams[0][0], teams[0][1],
gameVector[j][0], c, conn)
def get_stats(id, gamedate):
gameUrl = "http://stats.swehockey.se/Game/LineUps/" + str(id)
response = urllib.urlopen(gameUrl)
page_source = str(response.read())
page_source = page_source.replace("\\xc3\\xa5", "å")
page_source = page_source.replace("\\xc3\\xa4", "ä")
page_source = page_source.replace("\\xc3\\xa9", "é")
page_source = page_source.replace("\\xc3\\xb6", "ö")
page_source = page_source.replace("\\xc3\\x84", "Ä")
page_source = page_source.replace("\\xc3\\x85", "Å")
page_source = page_source.replace("\\xc3\\x96", "Ö")
team_string = ""
num_vect = []
pct_vect = []
period_list = []
output = []
for j in range(1, len(page_source) - 10):
if page_source[j] == ">":
exNum = get_isolated_number(page_source, j)
if exNum == 0 or exNum >= 1:
num_vect.append(exNum)
exPct = get_isolated_percent(page_source, j)
if exPct >= 0:
pct_vect.append(exPct)
if page_source[j] == "(":
period_vector = get_period_stats(page_source, j)
if period_vector != []:
period_list.append(period_vector)
if page_source[j:j + 17] == "\\xc2\\xa0-\\xc2\\xa0":
isf = 0
s1 = 0
s2 = 0
for k in range(1, 35):
if isf == 0 and j - k > 0:
if page_source[j - k] == ">":
isf = 1
s1 = j - k
isf = 0
for k in range(1, 60):
if isf == 0 and j + k < len(page_source):
if page_source[j + k] == "<":
isf = 1
s2 = j + k
if isnumber(page_source[s1 + 1:j]) and isnumber(page_source[j + 17:s2]):
home_score = page_source[s1 + 1:j]
away_score = page_source[j + 17:s2]
else:
home_team = page_source[s1 + 1:j]
away_team = page_source[j + 17:s2]
home_shots = num_vect[0]
away_shots = num_vect[1]
home_saves = num_vect[2]
away_saves = num_vect[3]
home_penalty = num_vect[4]
away_penalty = num_vect[5]
hp1score = period_list[1][0]
hp2score = period_list[1][2]
hp3score = period_list[1][4]
ap1score = period_list[1][1]
ap2score = period_list[1][3]
ap3score = period_list[1][5]
if len(period_list[1]) > 6:
hp4score = period_list[1][6]
else:
hp4score = 0
if len(period_list[1]) > 7:
ap4score = period_list[1][7]
else:
ap4score = 0
hp1shots = 0
hp2shots = 0
hp3shots = 0
ap1shots = 0
ap2shots = 0
ap3shots = 0
try:
hp1shots = period_list[0][0]
except IndexError:
pass
try:
hp2shots = period_list[0][1]
except IndexError:
pass
try:
hp3shots = period_list[0][2]
except IndexError:
pass
try:
ap1shots = period_list[2][0]
except IndexError:
pass
try:
ap2shots = period_list[2][1]
except IndexError:
pass
try:
ap3shots = period_list[2][2]
except IndexError:
pass
if len(period_list[0]) > 3:
hp4shots = period_list[0][3]
else:
hp4shots = 0
if len(period_list[2]) > 3:
ap4shots = period_list[2][3]
else:
ap4shots = 0
ap1saves = hp1shots - hp1score
ap2saves = hp2shots - hp2score
ap3saves = hp3shots - hp3score
ap4saves = hp4shots - hp4score
hp1saves = ap1shots - ap1score
hp2saves = ap2shots - ap2score
hp3saves = ap3shots - ap3score
hp4saves = ap4shots - ap4score
hp1penalty = period_list[5][0]
hp2penalty = period_list[5][1]
hp3penalty = period_list[5][2]
ap1penalty = period_list[6][0]
ap2penalty = period_list[6][1]
ap3penalty = period_list[6][2]
if len(period_list[5]) > 3:
hp4penalty = period_list[5][3]
else:
hp4penalty = 0
if len(period_list[6]) > 3:
ap4penalty = period_list[6][3]
else:
ap4penalty = 0
output.append(id)
output.append(gamedate)
output.append(home_team)
output.append(away_team)
output.append(home_score)
output.append(away_score)
output.append(home_shots)
output.append(away_shots)
output.append(home_saves)
output.append(away_saves)
output.append(home_penalty)
output.append(away_penalty)
output.append(hp1score)
output.append(hp2score)
output.append(hp3score)
output.append(hp4score)
output.append(ap1score)
output.append(ap2score)
output.append(ap3score)
output.append(ap4score)
output.append(hp1shots)
output.append(hp2shots)
output.append(hp3shots)
output.append(hp4shots)
output.append(ap1shots)
output.append(ap2shots)
output.append(ap3shots)
output.append(ap4shots)
output.append(hp1saves)
output.append(hp2saves)
output.append(hp3saves)
output.append(hp4saves)
output.append(ap1saves)
output.append(ap2saves)
output.append(ap3saves)
output.append(ap4saves)
output.append(hp1penalty)
output.append(hp2penalty)
output.append(hp3penalty)
output.append(hp4penalty)
output.append(ap1penalty)
output.append(ap2penalty)
output.append(ap3penalty)
output.append(ap4penalty)
if len(pct_vect) > 4:
home_pp = pct_vect[4]
else:
home_pp = 0
if len(pct_vect) > 5:
away_pp = pct_vect[5]
else:
away_pp = 0
output.append(home_pp)
output.append(away_pp)
return output
def OCRattacks(cV, min, c, matchID):
h1 = (cV[2] - cV[0]) * 0.52
h2 = (cV[2] - cV[0]) * 0.99
v1 = (cV[3] - cV[1]) * 0.50
v2 = (cV[3] - cV[1]) * 0.59
im1 = ImageGrab.grab(bbox=(cV[0] + h1, cV[1] + v1, cV[0] + h2, cV[1] + v2))
bigimage1 = im1.resize((int((h2 - h1) * 5), int((v2 - v1) * 5)),
Image.NEAREST)
image_string1 = pytesseract.image_to_string(
bigimage1, config='-psm 7 -c tessedit_char_whitelist=0123456789')
savestring = "./pics/Attacks/attack_string.jpg"
bigimage1 = bigimage1.convert("RGB")
bigimage1.save(savestring, "JPEG")
nums = get_all_numbers(image_string1)
for i in range(0, len(nums)):
if isnumber(nums[i]) == True:
nums[i] = int(nums[i])
else:
nums[i] = -1
ap = [-1, -1, -1, -1, -1, -1]
if len(nums) == 6:
for i in range(0, 6):
ap[i] = nums[i]
elif len(nums) == 9:
ap[0] = nums[0]
ap[1] = nums[2]
ap[2] = nums[3]
ap[3] = nums[5]
ap[4] = nums[6]
ap[5] = nums[8]
elif len(nums) == 7 or len(nums) == 8:
ap[0] = nums[0]
for i in range(1, 2):
if test_number(nums[i], 3, min, c, matchID) > -1:
ap[1] = nums[i]
for i in range(2, 3):
if test_number(nums[i], 4, min, c, matchID) > -1:
ap[2] = nums[i]
for i in range(3, 5):
if test_number(nums[i], 5, min, c, matchID) > -1:
ap[3] = nums[i]
for i in range(4, 7):
if len(nums) > i:
if test_number(nums[i], 6, min, c, matchID) > -1:
ap[4] = nums[i]
for i in range(5, 7):
if len(nums) > i:
if test_number(nums[i], 7, min, c, matchID) > -1:
ap[5] = nums[i]
ap[0] = test_number(ap[0], 2, min, c, matchID)
ap[1] = test_number(ap[1], 3, min, c, matchID)
ap[2] = test_number(ap[2], 4, min, c, matchID)
ap[3] = test_number(ap[3], 5, min, c, matchID)
ap[4] = test_number(ap[4], 6, min, c, matchID)
ap[5] = test_number(ap[5], 7, min, c, matchID)
if ap[4] + ap[5] != 100:
ap[4] = -1
ap[5] = -1
return ap
def get_time(cV):
xV = [(cV[2] - cV[0]) * 0.27, (cV[2] - cV[0]) * 0.22]
yV = [(cV[3] - cV[1]) * 0.30, (cV[3] - cV[1]) * 0.245]
im1 = ImageGrab.grab(bbox=(cV[0] + xV[0], cV[1] + yV[0], cV[0] + xV[1],
cV[1] + yV[1]))
bigimage1 = im1.resize((int((xV[0] - xV[1]) * 2), int(
(yV[0] - yV[1]) * 2)), Image.NEAREST)
bigimage2 = im1.resize((int((xV[0] - xV[1]) * 3), int(
(yV[0] - yV[1]) * 3)), Image.NEAREST)
bigimage3 = im1.resize((int((xV[0] - xV[1]) * 5), int(
(yV[0] - yV[1]) * 5)), Image.NEAREST)
image_string1 = pytesseract.image_to_string(bigimage1, config='-psm 7')
image_string2 = pytesseract.image_to_string(bigimage2, config='-psm 7')
image_string3 = pytesseract.image_to_string(bigimage3, config='-psm 7')
bigimage3 = bigimage3.convert("RGB")
bigimage3.save("pic1n.jpg", "JPEG")
[min1, min2, min3] = ["-1", "-1", "-1"]
[sec1, sec2, sec3] = ["-1", "-1", "-1"]
if ":" in image_string1:
i = image_string1.index(':')
min1 = image_string1[0:i]
sec1 = image_string1[i + 1:len(image_string1) + 1]
elif len(image_string1) == 5:
min1 = image_string1[0:2]
sec1 = image_string1[3:5]
if ":" in image_string2:
i = image_string2.index(':')
min2 = image_string2[0:i]
sec2 = image_string2[i + 1:len(image_string2) + 1]
elif len(image_string2) == 5:
min2 = image_string2[0:2]
sec2 = image_string2[3:5]
if ":" in image_string3:
i = image_string3.index(':')
min3 = image_string3[0:i]
sec3 = image_string3[i + 1:len(image_string3) + 1]
elif len(image_string3) == 5:
min3 = image_string3[0:2]
sec3 = image_string3[3:5]
min = -1
sec = -1
if min1 == min2 and min2 == min3:
if isnumber(min1):
min = int(min1)
elif min1 == min2:
if isnumber(min1):
min = int(min1)
elif min1 == min3:
if isnumber(min1):
min = int(min1)
elif min2 == min3:
if isnumber(min2):
min = int(min2)
if min > -1:
if sec1 == sec2 and sec2 == sec3:
if isnumber(sec1):
sec = int(sec1)
elif sec1 == sec2:
if isnumber(sec1):
sec = int(sec1)
elif sec1 == sec3:
if isnumber(sec1):
sec = int(sec1)
elif sec2 == sec3:
if isnumber(sec2):
sec = int(sec2)
else:
if isnumber(sec1):
sec = int(sec1)
elif isnumber(sec2):
sec = int(sec2)
elif isnumber(sec3):
sec = int(sec3)
return [min, sec]
