aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorMartin Kiewitz2013-11-18 00:52:04 +0100
committerMartin Kiewitz2013-11-18 00:52:04 +0100
commit52ec143e6d08b35179397e3f28b3a8bfef9e69f3 (patch)
treee5a0c552eddf18e2f4ca6a69a9a62346b105bbd1
parente80eef3bcd6d0a1f89ec6dc0e945c68f6ff90cb2 (diff)
downloadscummvm-rg350-52ec143e6d08b35179397e3f28b3a8bfef9e69f3.tar.gz
scummvm-rg350-52ec143e6d08b35179397e3f28b3a8bfef9e69f3.tar.bz2
scummvm-rg350-52ec143e6d08b35179397e3f28b3a8bfef9e69f3.zip
SCI: added script patch to fix sq5 toolbox bug
fixes bug #6457
-rw-r--r--engines/sci/engine/script_patches.cpp62
1 files changed, 62 insertions, 0 deletions
diff --git a/engines/sci/engine/script_patches.cpp b/engines/sci/engine/script_patches.cpp
index c636f161e3..4a829c2457 100644
--- a/engines/sci/engine/script_patches.cpp
+++ b/engines/sci/engine/script_patches.cpp
@@ -1738,6 +1738,65 @@ SciScriptPatcherEntry sq1vgaSignatures[] = {
{ 58, "Sarien armory droid zapping ego first time", 1, 0, 0, sq1vgaSignatureEgoShowsCard, sq1vgaPatchEgoShowsCard },
SCI_SIGNATUREENTRY_TERMINATOR};
+// ===========================================================================
+// The toolbox in sq5 is buggy. When you click on the upper part of the "put
+// in inventory"-button (some items only - for example the hole puncher - at the
+// upper left), points will get awarded correctly and the item will get put into
+// the player's inventory, but you will then get a "not here" message and the
+// item will also remain to be the current mouse cursor.
+// The bug report also says that items may get lost. I wasn't able to reproduce
+// that part.
+// This is caused by the mouse-click event getting reprocessed (which wouldn't
+// be a problem by itself) and during this reprocessing coordinates are not
+// processed the same as during the first click (script 226 includes a local
+// subroutine, which checks coordinates in a hardcoded way w/o port-adjustment).
+// Because of this, the hotspot for the button is lower than it should be, which
+// then results in the game thinking that the user didn't click on the button
+// and also results in the previously mentioned message.
+// This happened in Sierra SCI as well (of course).
+// We fix it by combining state 0 + 1 of takeTool::changeState and so stopping
+// the event to get reprocessed. This was the only way possible, because everything
+// else is done in SCI system scripts and I don't want to touch those.
+// Applies to at least: English/German/French PC floppy
+// Responsible method: takeTool::changeState
+// Fixes bug #6457
+const uint16 sq5SignatureToolboxFix[] = {
+ 0x31, 0x13, // bnt [check for state 1]
+ SIG_MAGICDWORD,
+ 0x38, SIG_UINT16 + 0xaa, 0x00, // pushi 00aa
+ 0x39, 0x05, // pushi 05
+ 0x39, 0x16, // pushi 16
+ 0x76, // push0
+ 0x39, 0x03, // pushi 03
+ 0x76, // push0
+ 0x7c, // pushSelf
+ 0x81, 0x5b, // lag 5b
+ 0x4a, 0x0e, // send 0e
+ 0x32, SIG_UINT16 + 0x88, 0x00, // jmp [end-of-method]
+ 0x3c, // dup
+ 0x35, 0x01, // ldi 01
+ 0x1a, // eq?
+ 0x31, 0x28, // bnt [check for state 2]
+ SIG_END
+};
+
+const uint16 sq5PatchToolboxFix[] = {
+ 0x31, 0x41, // bnt [check for state 2]
+ PATCH_ADDTOOFFSET +16, // skip to jmp offset
+ 0x35, 0x01, // ldi 01
+ 0x65, 0x14, // aTop [state]
+ 0x36, 0x00, 0x00, // ldi 0000 (waste 3 bytes)
+ 0x35, 0x00, // ldi 00 (waste 2 bytes)
+ PATCH_END
+};
+
+// script, description, signature patch
+SciScriptPatcherEntry sq5Signatures[] = {
+ { 226, "toolbox fix", 1, 0, 0, sq5SignatureToolboxFix, sq5PatchToolboxFix },
+ SCI_SIGNATUREENTRY_TERMINATOR
+};
+
+
// will actually patch previously found signature area
void Script::patcherApplyPatch(const SciScriptPatcherEntry *patchEntry, byte *scriptData, const uint32 scriptSize, int32 signatureOffset, const bool isMacSci11) {
const uint16 *patchData = patchEntry->patchData;
@@ -2132,6 +2191,9 @@ void Script::patcherProcessScript(uint16 scriptNr, byte *scriptData, const uint3
case GID_SQ4:
signatureTable = sq4Signatures;
break;
+ case GID_SQ5:
+ signatureTable = sq5Signatures;
+ break;
default:
break;
}