CTD in espionage code

[Archid]

Just had a CTD in my current game that I have traced back to a K-Mod update. CvGameCoreUtils::getEspionageModifier() will return 0 if neither the spy owner nor the target have any espionage points. This is used as a multiplier in CvPlayer::getEspionageMissionCostModifier() which eventually ends up as a denominator in CvUnityAI::AI_getEspionageTargetValue() causing a divide by 0 error.

Looking at the code if neither side has any espionage points the value of ESPIONAGE_SPENDING_MULTIPLIER from the global defines should be used. The fix is very simple by wrapping the numerator values in a std::max in a similar way that the denominator already is:

Spoiler :

int getEspionageModifier(TeamTypes eOurTeam, TeamTypes eTargetTeam)
{
	FAssert(eOurTeam != eTargetTeam);
	FAssert(eOurTeam != BARBARIAN_TEAM);
	// FAssert(eTargetTeam != BARBARIAN_TEAM); // K-Mod note. This is possible for legitimate reasons (although, the result is never important...)

	// K-Mod. Scale the points modifier based on the teams' population. (Note ESPIONAGE_SPENDING_MULTIPLIER is 100 in the default xml.)
	const CvTeam& kOurTeam = GET_TEAM(eOurTeam);
	const CvTeam& kTargetTeam = GET_TEAM(eTargetTeam);

	int iPopScale = 5 * GC.getWorldInfo(GC.getMapINLINE().getWorldSize()).getTargetNumCities();
	int iTargetPoints = 10 * kTargetTeam.getEspionagePointsEver() / std::max(1, iPopScale + kTargetTeam.getTotalPopulation(false));
	int iOurPoints = 10 * kOurTeam.getEspionagePointsEver() / std::max(1, iPopScale + kOurTeam.getTotalPopulation(false));

	return GC.getDefineINT("ESPIONAGE_SPENDING_MULTIPLIER") * [COLOR="Red"][B]std::max(1, [/B][/COLOR](2 * iTargetPoints + iOurPoints)[COLOR="Red"][B])[/B][/COLOR] / std::max(1, iTargetPoints + 2 * iOurPoints);
	// K-Mod end
}

 

[karadoc]

I think it would be reasonable to use the base value if neither team had any espionage points - but that never happens in K-Mod anyway, because the Palace gives espionage. I'll make your suggested change anyway since it might help some mods.

By the way, the same issue exists in the original BtS code.