PSS: Switch over to regex based parsing

This commit is contained in:
Worros 2010-09-10 13:22:52 +08:00
parent 7c8965486f
commit 7f61fba314
2 changed files with 9 additions and 120 deletions

View File

@ -81,123 +81,12 @@ class PokerStarsSummary(TourneySummary):
re_DateTime = re.compile("\[(?P<Y>[0-9]{4})\/(?P<M>[0-9]{2})\/(?P<D>[0-9]{2})[\- ]+(?P<H>[0-9]+):(?P<MIN>[0-9]+):(?P<S>[0-9]+)")
re_Entries = re.compile("[0-9]+")
re_Prizepool = re.compile("\$[0-9]+\.[0-9]+")
re_BuyInFee = re.compile("(?P<BUYIN>[0-9]+\.[0-9]+).*(?P<FEE>[0-9]+\.[0-9]+)")
re_FPP = re.compile("(?P<FPP>[0-9]+)\sFPP")
#note: the dollar and cent in the below line are currency-agnostic
re_Added = re.compile("(?P<DOLLAR>[0-9]+)\.(?P<CENT>[0-9]+)\s(?P<CURRENCY>[A-Z]+)(\sadded\sto\sthe\sprize\spool\sby\sPokerStars)")
re_DateTimeET = re.compile("(?P<Y>[0-9]{4})\/(?P<M>[0-9]{2})\/(?P<D>[0-9]{2})[\- ]+(?P<H>[0-9]+):(?P<MIN>[0-9]+):(?P<S>[0-9]+)")
re_GameInfo = re.compile(u""".+(?P<LIMIT>No\sLimit|Limit|LIMIT|Pot\sLimit)\s(?P<GAME>Hold\'em|Razz|RAZZ|7\sCard\sStud|7\sCard\sStud\sHi/Lo|Omaha|Omaha\sHi/Lo|Badugi|Triple\sDraw\s2\-7\sLowball|5\sCard\sDraw)""")
def parseSummary(self):
lines=self.summaryText.splitlines()
self.tourNo = self.re_TourNo.findall(lines[0])[0] #ignore game and limit type as thats not recorded
result=self.re_GameInfo.search(lines[0])
result=result.groupdict()
self.gametype['limitType']=self.limits[result['LIMIT']]
self.gametype['category']=self.games[result['GAME']][1]
if lines[1].find("$")!=-1: #TODO: move this into a method and call that from PokerStarsToFpdb.py:269 if hand.buyinCurrency=="USD" etc.
self.currency="USD"
elif lines[1].find(u"")!=-1:
self.currency="EUR"
elif lines[1].find("FPP")!=-1:
self.currency="PSFP"
else:
raise FpdbParseError(_("didn't recognise buyin currency in:")+lines[1])
if self.currency=="USD" or self.currency=="EUR":
result=self.re_BuyInFee.search(lines[1])
result=result.groupdict()
self.buyin=int(100*Decimal(result['BUYIN']))
self.fee=int(100*Decimal(result['FEE']))
elif self.currency=="PSFP":
result=self.re_FPP.search(lines[1])
result=result.groupdict()
self.buyin=int(Decimal(result['FPP']))
self.fee=0
currentLine=2
self.entries = self.re_Entries.findall(lines[currentLine])[0]
currentLine+=1 #note that I chose to make the code keep state (the current line number)
#as that means it'll fail rather than silently skip potentially valuable information
#print "after entries lines[currentLine]", lines[currentLine]
result=self.re_Added.search(lines[currentLine])
if result:
result=result.groupdict()
self.added=100*int(Decimal(result['DOLLAR']))+int(Decimal(result['CENT']))
self.addedCurrency=result['CURRENCY']
currentLine+=1
else:
self.added=0
self.addedCurrency="NA"
#print "after added/entries lines[currentLine]", lines[currentLine]
result=self.re_Prizepool.findall(lines[currentLine])
if result:
self.prizepool = result[0]
self.prizepool = self.prizepool[1:-3]+self.prizepool[-2:]
currentLine+=1
#print "after prizepool lines[currentLine]", lines[currentLine]
useET=False
result=self.re_DateTime.search(lines[currentLine])
if not result:
print _("in not result starttime")
useET=True
result=self.re_DateTimeET.search(lines[currentLine])
result=result.groupdict()
datetimestr = "%s/%s/%s %s:%s:%s" % (result['Y'], result['M'],result['D'],result['H'],result['MIN'],result['S'])
self.startTime= datetime.datetime.strptime(datetimestr, "%Y/%m/%d %H:%M:%S") # also timezone at end, e.g. " ET"
self.startTime = HandHistoryConverter.changeTimezone(self.startTime, "ET", "UTC")
currentLine+=1
if useET:
result=self.re_DateTimeET.search(lines[currentLine])
else:
result=self.re_DateTime.search(lines[currentLine])
if result:
result=result.groupdict()
datetimestr = "%s/%s/%s %s:%s:%s" % (result['Y'], result['M'],result['D'],result['H'],result['MIN'],result['S'])
self.endTime= datetime.datetime.strptime(datetimestr, "%Y/%m/%d %H:%M:%S") # also timezone at end, e.g. " ET"
self.endTime = HandHistoryConverter.changeTimezone(self.endTime, "ET", "UTC")
currentLine+=1
if lines[currentLine].find("Tournament is still in progress")!=-1:
currentLine+=1
for i in range(currentLine,len(lines)-2): #lines with rank and winnings info
if lines[i].find(":")==-1:
break
result=self.re_Player.search(lines[i])
result=result.groupdict()
rank=result['RANK']
name=result['NAME']
winnings=result['WINNINGS']
if winnings:
winnings=int(100*Decimal(winnings))
else:
winnings=0
if result['STILLPLAYING']:
#print "stillplaying"
rank=None
winnings=None
self.addPlayer(rank, name, winnings, self.currency, None, None, None)#TODO: currency, ko/addon/rebuy count -> need examples!
#end def parseSummary
def parseSummaryFile(self):
m = self.re_TourneyInfo.search(self.summaryText)
if m == None:
tmp = self.summaryText[0:200]
log.error(_("parseSummaryFile: Unable to recognise Tourney Info: '%s'") % tmp)
log.error(_("parseSummaryFile: Raising FpdbParseError"))
log.error(_("parseSummary: Unable to recognise Tourney Info: '%s'") % tmp)
log.error(_("parseSummary: Raising FpdbParseError"))
raise FpdbParseError(_("Unable to recognise Tourney Info: '%s'") % tmp)
#print "DEBUG: m.groupdict(): %s" % m.groupdict()
@ -222,8 +111,8 @@ class PokerStarsSummary(TourneySummary):
m = self.re_Currency.search(self.summaryText)
if m == None:
log.error(_("parseSummaryFile: Unable to locate currency"))
log.error(_("parseSummaryFile: Raising FpdbParseError"))
log.error(_("parseSummary: Unable to locate currency"))
log.error(_("parseSummary: Raising FpdbParseError"))
raise FpdbParseError(_("Unable to locate currency"))
#print "DEBUG: m.groupdict(): %s" % m.groupdict()

View File

@ -125,11 +125,11 @@ class TourneySummary(object):
self.sym = None
if builtFrom=="IMAP":
self.parseSummary()
self.insertOrUpdate()
elif builtFrom == "file":
self.parseSummaryFile()
self.insertOrUpdate()
# Fix line endings?
pass
self.parseSummary()
self.insertOrUpdate()
#end def __init__
def __str__(self):