|
a/src/common/textsplit.cpp |
|
b/src/common/textsplit.cpp |
|
... |
|
... |
| 88 |
|
88 |
|
| 89 |
char wild[] = "*?[]";
|
89 |
char wild[] = "*?[]";
|
| 90 |
for (i = 0; i < strlen(wild); i++)
|
90 |
for (i = 0; i < strlen(wild); i++)
|
| 91 |
charclasses[int(wild[i])] = WILD;
|
91 |
charclasses[int(wild[i])] = WILD;
|
| 92 |
|
92 |
|
| 93 |
char special[] = ".@+-,#'_\n\r\f";
|
93 |
char special[] = ".@+-#'_\n\r\f";
|
| 94 |
for (i = 0; i < strlen(special); i++)
|
94 |
for (i = 0; i < strlen(special); i++)
|
| 95 |
charclasses[int(special[i])] = special[i];
|
95 |
charclasses[int(special[i])] = special[i];
|
| 96 |
|
96 |
|
| 97 |
for (i = 0; i < sizeof(unipunc) / sizeof(int); i++) {
|
97 |
for (i = 0; i < sizeof(unipunc) / sizeof(int); i++) {
|
| 98 |
spunc.insert(unipunc[i]);
|
98 |
spunc.insert(unipunc[i]);
|
|
... |
|
... |
| 458 |
goto SPACE;
|
458 |
goto SPACE;
|
| 459 |
}
|
459 |
}
|
| 460 |
break;
|
460 |
break;
|
| 461 |
|
461 |
|
| 462 |
case '.':
|
462 |
case '.':
|
| 463 |
case ',':
|
|
|
| 464 |
{
|
463 |
{
|
| 465 |
// Need a little lookahead here. At worse this gets the end null
|
464 |
// Need a little lookahead here. At worse this gets the end null
|
| 466 |
int nextc = it[it.getCpos()+1];
|
465 |
int nextc = it[it.getCpos()+1];
|
| 467 |
int nextwhat = whatcc(nextc);
|
466 |
int nextwhat = whatcc(nextc);
|
| 468 |
if (m_inNumber) {
|
467 |
if (m_inNumber) {
|
| 469 |
// we're eliminating 132.jpg here. Good idea ?
|
468 |
if (nextwhat != DIGIT)
|
| 470 |
if (nextwhat != DIGIT && nextc != 'e' && nextc != 'E')
|
|
|
| 471 |
goto SPACE;
|
469 |
goto SPACE;
|
| 472 |
m_wordLen += it.appendchartostring(m_span);
|
470 |
m_wordLen += it.appendchartostring(m_span);
|
| 473 |
curspanglue = cc;
|
471 |
curspanglue = cc;
|
| 474 |
break;
|
472 |
break;
|
| 475 |
} else {
|
473 |
} else {
|