Battle Test Runner Failsafes/Improvements

[ghoulslash]

1. Failsafe check on assigned move PP. If a mon has a moveset with a single move, PP in move slot 2+ will be 0, so some tests will fail with a hard-to-track issue
2. TryMessage has been improved to be more similar to a StringCopy function/simpler to read. It also will skip over ext ctrl codes for colors. The loop will now basically skip over any space, \p, \n etc code as well as colors and just compare relevant characters.

[AlexOn1ine]

PR doesn't build

[ghoulslash]

PR doesn't build

looks like it was a weird CI issue. builds now

[mrgriffin]

We should prefer static inline functions to macros wherever possible.

Suggested change

	#define IS_SKIPOVER_CHAR(char) (char == CHAR_SPACE \
	|| char == CHAR_PROMPT_SCROLL \
	|| char == CHAR_PROMPT_CLEAR \
	|| char == CHAR_NEWLINE)
	static inline bool32 IsSkipoverChar(u32 char)
	{
	return char == CHAR_SPACE
	|| char == CHAR_PROMPT_SCROLL
	|| char == CHAR_PROMPT_CLEAR
	|| char == CHAR_NEWLINE;
	}

[mrgriffin]

Doesn't this essentially ignore all whitespace inside the pattern and output? e.g. "Hello World" would match "HelloWorld" and vice-versa?

The intent of the original code was that a space in the pattern would match any whitespace in the actual output, so I think this change should say something like "try to match 1 or more skipover characters in event->pattern, and if so match 1 or more skipover characters in string"?

Example incorrectly-passing test (no engine code modified):

WILD_BATTLE_TEST("Pokemon gain exp after catching a Pokemon")
{
    u8 level = 0;

    PARAMETRIZE { level = 50; }
    PARAMETRIZE { level = MAX_LEVEL; }

    GIVEN {
        PLAYER(SPECIES_WOBBUFFET) { Level(level); }
        OPPONENT(SPECIES_CATERPIE) { HP(1); }
    } WHEN {
        TURN { USE_ITEM(player, ITEM_ULTRA_BALL); }
    } SCENE {
-       MESSAGE("You used Ultra Ball!");
+       MESSAGE("YouusedUltraBall!");
        ANIMATION(ANIM_TYPE_SPECIAL, B_ANIM_BALL_THROW, player);
        if (level != MAX_LEVEL) {
            EXPERIENCE_BAR(player);
        }
    }
}

Example incorrectly passing code (no tests modified):

-static const u8 sText_PlayerUsedItem[] = _("You used\n{B_LAST_ITEM}!");
+static const u8 sText_PlayerUsedItem[] = _("Youused{B_LAST_ITEM}!");

[mrgriffin]

I think event->pattern[k] == EOS cannot occur, because the previous line is if (string[j] != event->pattern[k]) break;, so if string[k] != EOS && event->pattern[k] == EOS we should have already breaked?

If it could occur, I think this code would be wrong because we shouldn't exit this code with a success if we get to the end of event->pattern without also getting to the end of string.

[mrgriffin]

Is this choice arbitrary? If so I'd probably use 1 or 0x7F (max) to signal the arbitrariness more directly.

[hedara90]

From testing it with the cases I could remember that this would fix it works as advertised.
I don't think this is supposed to fix the various forced switch peculiarities, but I can't find any regressions with those at least.
This test fails without this PR and passes with it.

SINGLE_BATTLE_TEST("Species with only 1 learnable move can use more moves in tests")
{
    GIVEN {
        PLAYER(SPECIES_BELDUM);
        OPPONENT(SPECIES_WOBBUFFET);
    } WHEN {
        TURN { MOVE(player, MOVE_TACKLE); }
        TURN { MOVE(player, MOVE_GROWL); }
        TURN { MOVE(player, MOVE_SURF); }
        TURN { MOVE(player, MOVE_STRENGTH); }
    } SCENE {
        MESSAGE("Beldum used Tackle!");
        MESSAGE("Beldum used Growl!");
        MESSAGE("Beldum used Surf!");
        MESSAGE("Beldum used Strength!");
    }
}