Hi Jazzy, thanks for replying.
I'm using PythonWin 2.7.2(pywin32 build 217)
I click open file and open get_rands and then all I see is the code.
import urllib2, re, datetime, win32com.client, sys, os.path, decimal
usragent = "Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.0.5; .NET CLR 2.0.50727; ffco7) Gecko/2008120122 Firefox/3.0.5"
xlpath = "%s\\%s" % (os.path.dirname(sys.argv[0]) , "rands.xlsx")
tooday = datetime.date.today()
tomorrow = tooday+datetime.timedelta(days=1)
#href="meeting.asp?meeting=13493"> Cowra</a>
match_mtgs = re.compile("href=\"meeting.asp\?meeting=(\d+)\"> (.*?)</a>")
match_races = re.compile("<a href=\"worksheet.asp\?raceno=(\d+)&meetingid=\d+")
#>Race 1<br>13:15 <font style="font-size:11px;color:#999999;">(local)</font><br><div id="timezone">13:15</div><select class='nfinput' name="timezone" style="width: 70px;font-size: 9px;" onchange="javascript:adjClock('timezone', 3, 15, this.value);">
match_rn_lt_GMT = re.compile(">Race (\d\d?)<br>(\d\d?:\d\d).*?onchange=\"javascript:adjClock\('timezone', (\d\d?), (\d\d?), this\.value\);")
def stripMarkUp(src):
src = re.sub(r"<[^>]*>", '', src)
src = re.sub(r" ", '', src)
return src.strip()
def addRace(raceList, buf, trk, rn_st):
#('1', '13:15', '3', '15')
gmt = "%02i:%02i" % (int(rn_st[2]), int(rn_st[3]))
tmpList = []
trs = buf.split("<tr")
ths = trs[1].split("<th")
assert stripMarkUp("< " + ths[2]) == "TAB"
assert stripMarkUp("< " + ths[3]) == "HORSE"
assert stripMarkUp("< " + ths[18]) == "DIV"
for tr in trs[2:]:
tds = tr.split("<td")
if len(tds) > 18:
div = stripMarkUp("< " + tds[18])
if div != "scr":
decDiv = decimal.Decimal(re.sub("\$|,", '', div))
tmpList.append((decDiv, stripMarkUp("< " + tds[2]), stripMarkUp("< " + tds[3])))
tmpList.sort()
#[(Decimal('4.00'), '8', 'Ohmy Dubai'), (Decimal('5.00'), '1', 'Cyclone Bella'),
favs = 1
for i in range(1, len(tmpList)):
if tmpList[i][0] == tmpList[0][0]:
favs += 1
else:
break
for i in range(favs):
raceList.append((gmt, trk, int(rn_st[0]), rn_st[1], tmpList[i][1], tmpList[i][2], tmpList[i][0]))
print tmpList[i]
url = "http://www.racingandsports.com.au/form-guide/"
txheaders = ****"User-agent" : usragent****
print xlpath
print tooday
req = urllib2.Request(url, None, txheaders)
buf = urllib2.urlopen(req).read()
raceList = []
###buf = open("rands.html", "rb").read()
#fp = open("rands.html", "wb")
#fp.write(buf)
#fp.close()
i = buf.find(tooday.strftime("%A"))
j = buf.find(tomorrow.strftime("%A"))
buf2 = buf[i:j]
mtgs = match_mtgs.findall(buf2)
for m in mtgs:
print m
url = "http://www.racingandsports.com.au/form-guide/worksheet.asp?meetingid=%s" % (m[0],)
req = urllib2.Request(url, None, txheaders)
buf = urllib2.urlopen(req).read()
rn_st = match_rn_lt_GMT.findall(buf)
#[('1', '13:15', '3', '15')]
fp = open("%s-%s-R%02i.html" % (tooday.strftime("%Y-%m-%d"), m[1], int(rn_st[0][0])), "wb")
fp.write(buf)
fp.close()
races = match_races.findall(buf)
k = buf.find("class='mainHeader'")
i = buf.find("<table", k-100, k)
j = buf.find("</table", i)
addRace(raceList, buf[i:j], m[1], rn_st[0])
#print races
#['2', '3', '4', '5', '6', '7', '8']
for r in races:
url = "http://www.racingandsports.com.au/form-guide/worksheet.asp?raceno=%s&meetingid=%s" % (r, m[0],)
req = urllib2.Request(url, None, txheaders)
buf = urllib2.urlopen(req).read()
rn_st = match_rn_lt_GMT.findall(buf)
#[('1', '13:15', '3', '15')]
fp = open("%s-%s-R%02i.html" % (tooday.strftime("%Y-%m-%d"), m[1], int(rn_st[0][0])), "wb")
fp.write(buf)
fp.close()
k = buf.find("class='mainHeader'")
i = buf.find("<table", k-100, k)
j = buf.find("</table", i)
addRace(raceList, buf[i:j], m[1], rn_st[0])
raceList.sort()
xl = win32com.client.Dispatch("Excel.Application")
xl.Visible = 1
try:
wb=xl.Workbooks.Open(xlpath)
except:
wb=xl.Workbooks.Add()
wb.SaveAs(xlpath)
#wb=xl.Workbooks.Open(r"rands.xls")
ws = wb.Worksheets("Sheet1")
r = 1
while 1:
if ws.Range("A%i" % (r,)).Value is None:
r += 1
if ws.Range("A%i" % (r,)).Value is None:
break
r += 1
#[('03:22', 'Eagle Farm', 3, '13:22', '5', 'Golden Hut', Decimal('4.20')), ('03:22', 'Eagle Farm', 3, '13:22', '9', 'Western Run', Decimal('4.20')), ('05:12', 'Belmont Park', 2, '13:12', '8', 'Ohmy Dubai', Decimal('4.00'))]
for rl in raceList:
ws.Range("A%i" % (r,)).Value = tooday
ws.Range("B%i" % (r,)).Value = rl[0]
ws.Range("C%i" % (r,)).Value = rl[1]
ws.Range("D%i" % (r,)).Value = rl[2]
ws.Range("E%i" % (r,)).Value = rl[3]
ws.Range("F%i" % (r,)).Value = rl[4]
ws.Range("G%i" % (r,)).Value = rl[5]
ws.Range("H%i" % (r,)).Value = rl[6]
r += 1
wb.Save()
Not sure what I'm doing wrong.