Mercurial > traipse_dev
view orpg/dieroller/wodex.py @ 107:ded1e7a25f8d alpha
Traipse Alpha 'OpenRPG' {091007-01}
Traipse is a distribution of OpenRPG that is designed to be easy to setup and go.
Traipse also makes it easy for developers to work on code without fear of sacrifice.
'Ornery-Orc' continues the trend of 'Grumpy' and adds fixes to the code. 'Ornery-Orc's
main goal is to offer more advanced features and enhance the productivity of the user.
Update Summary:
{091006}
00:
Adds Bookmarks (Alpha) with cool Smiley Star and Plus Symbol images!
03:
Changes made to the map for increased portability. SnowDog has changes planned in Core,
though.
Added an initial push to the BCG. Not much to see, just shows off how it is re-writing
Main code.
{091007}
00:
New images added to Plugin Control Panel for Auto Start.
01:
Attempting to fix Deprecation warning from Main.
author | sirebral |
---|---|
date | Wed, 07 Oct 2009 19:12:53 -0500 |
parents | 449a8900f9ac |
children | bf799efe7a8a |
line wrap: on
line source
## a vs die roller as used by WOD games #!/usr/bin/env python # Copyright (C) 2000-2001 The OpenRPG Project # # openrpg-dev@lists.sourceforge.net # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2 of the License, or # (at your option) any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. # -- # # File: wodex.py # Original Author: Darloth # Maintainer: # Original Version: 1.0 # # Description: A modified form of the World of Darkness die roller to # conform to ShadowRun rules-sets, then modified back to the WoD for # the new WoD system. Thanks to the ORPG team # for the original die rollers. # Much thanks to whoever wrote the original shadowrun roller (akoman I believe) from die import * __version__ = "$Id: wodex.py,v 1.9 2007/05/06 16:42:55 digitalxero Exp $" class wodex(std): def __init__(self,source=[]): std.__init__(self,source) def vs(self,actualtarget=6): return oldwodVs(self,actualtarget,(6)) def wod(self,actualtarget=8): return newwodVs(self,actualtarget,(8)) def exalt(self, actualtarget=7): return exaltVs(self, actualtarget) def exaltDmg(self, actualtarget=7): return exaltDmg(self, actualtarget) def vswide(self,actualtarget=6,maxtarget=10): #wide simply means it reports TNs from 2 to a specified max. return oldwodVs(self,actualtarget,2,maxtarget) class oldwodVs(std): def __init__(self,source=[],actualtarget=6,mintn=2,maxtn=10): std.__init__(self, source) if actualtarget > 10: actualtarget = 10 if mintn > 10: mintn = 10 if maxtn > 10: maxtn = 10 if actualtarget < 2: self.target = 2 else: self.target = actualtarget #if the target number is higher than max (Mainly for wide rolls) then increase max to tn if actualtarget > maxtn: maxtn = actualtarget if actualtarget < mintn: mintn = actualtarget #store minimum for later use as well, also in result printing section. if mintn < 2: self.mintn = 2 else: self.mintn = mintn self.maxtn = maxtn #store for later use in printing results. (Yeah, these comments are now disordered) # WoD etc uses d10 but i've left it so it can roll anything openended # self.openended(self[0].sides) #count successes, by looping through each die, and checking it against the currently set TN #1's subtract successes. def __sum__(self): s = 0 for r in self.data: if r >= self.target: s += 1 elif r == 1: s -= 1 return s #a modified sum, but this one takes a target argument, and is there because otherwise it is difficult to loop through #tns counting successes against each one without changing target, which is rather dangerous as the original TN could #easily be lost. 1s subtract successes from everything. def xsum(self,curtarget): s = 0 for r in self.data: if r >= curtarget: s += 1 elif r == 1: s -= 1 return s def __str__(self): if len(self.data) > 0: myStr = "[" + str(self.data[0]) for a in self.data[1:]: myStr += "," myStr += str(a) myStr += "] Results: " #cycle through from mintn to maxtn, summing successes for each separate TN for targ in range(self.mintn,self.maxtn+1): if (targ == self.target): myStr += "<b>" myStr += "(" + str(self.xsum(targ)) + " vs " + str(targ) + ") " if (targ == self.target): myStr += "</b>" else: myStr = "[] = (0)" return myStr class newwodVs(std): def __init__(self,source=[],actualtarget=8,mintn=8,maxtn=8): std.__init__(self, source) if actualtarget > 30: actualtarget = 30 if mintn > 10: mintn = 10 if maxtn > 10: maxtn = 10 if actualtarget < 2: self.target = 2 else: self.target = actualtarget #if the target number is higher than max (Mainly for wide rolls) then increase max to tn if actualtarget > maxtn: maxtn = actualtarget if actualtarget < mintn: mintn = actualtarget #store minimum for later use as well, also in result printing section. if mintn < 2: self.mintn = 2 else: self.mintn = mintn self.maxtn = maxtn #store for later use in printing results. (Yeah, these comments are now disordered) # WoD etc uses d10 but i've left it so it can roll anything openended # self.openended(self[0].sides) #a modified sum, but this one takes a target argument, and is there because otherwise it is difficult to loop through #tns counting successes against each one without changing target, which is rather dangerous as the original TN could #easily be lost. 1s subtract successes from original but not re-rolls. def xsum(self,curtarget,subones=1): s = 0 done = 1 for r in self.data: if r >= curtarget: s += 1 elif ((r == 1) and (subones == 1)): s -= 1 if r == 10: done = 0 subones = 0 self.append(di(10)) if done == 1: return s else: return self.xsum(0) def openended(self,num): if num <= 1: self done = 1 for i in range(len(self.data)): if self.data[i].lastroll() == num: self.data[i].extraroll() done = 0 if done: return self else: return self.openended(num) def __str__(self): if len(self.data) > 0: myStr = "[" + str(self.data[0]) for a in self.data[1:]: myStr += "," myStr += str(a) myStr += "] Results: " #cycle through from mintn to maxtn, summing successes for each separate TN for targ in range(self.mintn,self.maxtn+1): if (targ == self.target): myStr += "<b>" myStr += "(" + str(self.xsum(targ)) + " vs " + str(targ) + ") " if (targ == self.target): myStr += "</b>" else: myStr = "[] = (0)" return myStr class exaltVs(std): def __init__(self, source=[], actualtarget=7): std.__init__(self, source) if actualtarget > 10: actualtarget = 10 if actualtarget < 2: self.target = 2 else: self.target = actualtarget def xsum(self, target): s = 0 for r in self.data: if r >= target: s += 1 if r == 10: s += 1 return s def __str__(self): if len(self.data) > 0: myStr = str(self.data) myStr += " Results: " succ = self.xsum(self.target) if succ == 0 and 1 in self.data: myStr += 'BOTCH!' elif succ == 0: myStr += str(succ) + " Failure" elif succ == 1: myStr += str(succ) + " Success" else: myStr += str(succ) + " Successes" return myStr class exaltDmg(std): def __init__(self, source=[], actualtarget=7): std.__init__(self, source) if actualtarget > 10: actualtarget = 10 if actualtarget < 2: self.target = 2 else: self.target = actualtarget def xsum(self, target): s = 0 for r in self.data: if r >= target: s += 1 return s def __str__(self): if len(self.data) > 0: myStr = str(self.data) myStr += " Results: " succ = self.xsum(self.target) if succ == 0 and 1 in self.data: myStr += 'BOTCH!' elif succ == 0: myStr += str(succ) + " Failure" elif succ == 1: myStr += str(succ) + " Success" else: myStr += str(succ) + " Successes" return myStr