continously opening and closing the language.bnd file is rather inefficient. It's about 400k, though...
svn-id: r8158
This commit is contained in:
parent
877ca1b859
commit
3efdd3ad6b
2 changed files with 26 additions and 38 deletions
|
@ -745,8 +745,8 @@ void Scumm::initCharset(int charsetno) {
|
|||
}
|
||||
|
||||
int indexCompare(const void *p1, const void *p2) {
|
||||
const struct langIndexNode *i1 = (const struct langIndexNode *) p1;
|
||||
const struct langIndexNode *i2 = (const struct langIndexNode *) p2;
|
||||
const LangIndexNode *i1 = (const LangIndexNode *) p1;
|
||||
const LangIndexNode *i2 = (const LangIndexNode *) p2;
|
||||
|
||||
return strcmp(i1->tag, i2->tag);
|
||||
}
|
||||
|
@ -797,23 +797,32 @@ void Scumm::loadLanguageBundle() {
|
|||
// instead, but the extra overhead in the node structure would
|
||||
// easily have doubled the memory consumption of the index.
|
||||
|
||||
_languageIndex = (struct langIndexNode *) malloc(_languageStrCount * sizeof(struct langIndexNode));
|
||||
_languageIndex = (LangIndexNode *)calloc(_languageStrCount, sizeof(LangIndexNode));
|
||||
|
||||
ptr = _languageBuffer;
|
||||
|
||||
for (i = 0; i < _languageStrCount; i++) {
|
||||
int j;
|
||||
|
||||
for (j = 0; j < 8 && !isspace(ptr[j]); j++)
|
||||
_languageIndex[i].tag[j] = toupper(ptr[j]);
|
||||
// First 8 chars in the line give the string ID / 'tag'
|
||||
for (j = 0; j < 8 && !isspace(*ptr); j++, ptr++)
|
||||
_languageIndex[i].tag[j] = toupper(*ptr);
|
||||
_languageIndex[i].tag[j] = 0;
|
||||
ptr += (j + 1);
|
||||
|
||||
// After that follows a single space which we skip
|
||||
assert(isspace(*ptr));
|
||||
ptr++;
|
||||
|
||||
// Then comes the translated string: we record an offset to that.
|
||||
_languageIndex[i].offset = ptr - _languageBuffer;
|
||||
|
||||
//skip:
|
||||
// Skip over newlines (and turn them into null bytes)
|
||||
ptr = strpbrk(ptr, "\n\r");
|
||||
if (ptr == NULL)
|
||||
break;
|
||||
while (*ptr == '\n' || *ptr == '\r')
|
||||
ptr++;
|
||||
*ptr++ = 0;
|
||||
}
|
||||
|
||||
// Conceptually, it may be more elegant to construct the
|
||||
|
@ -821,8 +830,7 @@ void Scumm::loadLanguageBundle() {
|
|||
// start. However, this is less error-prone and likely to be
|
||||
// much more optimized than anything I might implement.
|
||||
|
||||
qsort(_languageIndex, _languageStrCount, sizeof(struct langIndexNode), indexCompare);
|
||||
free(_languageBuffer);
|
||||
qsort(_languageIndex, _languageStrCount, sizeof(LangIndexNode), indexCompare);
|
||||
}
|
||||
|
||||
_existLanguageFile = true;
|
||||
|
@ -833,8 +841,8 @@ void Scumm::translateText(const byte *text, byte *trans_buff) {
|
|||
|
||||
if ((text[0] == '/') && _existLanguageFile) {
|
||||
if (_gameId == GID_CMI) {
|
||||
struct langIndexNode target;
|
||||
struct langIndexNode *found = NULL;
|
||||
LangIndexNode target;
|
||||
LangIndexNode *found = NULL;
|
||||
|
||||
// copy name from text /..../
|
||||
for (l = 0; (l < 8) && (text[l + 1] != '/'); l++)
|
||||
|
@ -849,41 +857,21 @@ void Scumm::translateText(const byte *text, byte *trans_buff) {
|
|||
// we can't find translations for these.
|
||||
|
||||
if (strcmp(target.tag, "PU_M001") != 0 && strcmp(target.tag, "PU_M002") != 0)
|
||||
found = (struct langIndexNode *)bsearch(&target, _languageIndex, _languageStrCount, sizeof(struct langIndexNode), indexCompare);
|
||||
found = (LangIndexNode *)bsearch(&target, _languageIndex, _languageStrCount, sizeof(LangIndexNode), indexCompare);
|
||||
if (found != NULL) {
|
||||
File file;
|
||||
|
||||
file.open("language.tab", getGameDataPath());
|
||||
if (file.isOpen()) {
|
||||
byte *ptr = trans_buff;
|
||||
byte c;
|
||||
|
||||
file.seek(found->offset, SEEK_SET);
|
||||
for (;;) {
|
||||
c = file.readByte();
|
||||
if (c == 10 || c == 13) {
|
||||
*ptr = 0;
|
||||
break;
|
||||
} else
|
||||
*ptr++ = c;
|
||||
}
|
||||
file.close();
|
||||
return;
|
||||
} else {
|
||||
// Some evil person removed the language file?
|
||||
_existLanguageFile = false;
|
||||
}
|
||||
strcpy((char *)trans_buff, _languageBuffer + found->offset);
|
||||
return;
|
||||
}
|
||||
} else if (_gameId == GID_DIG) {
|
||||
// FIXME: This code really could stand a rewrite
|
||||
// It's rather inefficient and if a string isn't found it'll read past the
|
||||
// end of the _languageBuffer.
|
||||
char name[20], tmp[500], tmp2[20], num_s[20], number[4], enc;
|
||||
char name[12+1], tmp[500], tmp2[20], num_s[20], number[4], enc;
|
||||
int32 num, j, k, r, pos = 0;
|
||||
char *buf = _languageBuffer;
|
||||
|
||||
// copy name from text /..../
|
||||
for (l = 0; (l < 20) && (text[l + 1] != '.'); l++) {
|
||||
for (l = 0; (l < 12) && (text[l + 1] != '.'); l++) {
|
||||
name[l] = text[l + 1];
|
||||
}
|
||||
name[l] = 0;
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue