TCString CompileRegexp(TCString Regexp, int bAddAsUsualSubstring, int ID)
{
TCString Result(_T(""));
sPcreCompileData s = {0};
int NewID = PcreCompileData.AddElem(s);
PcreCompileData[NewID].ID = ID;
if (!bAddAsUsualSubstring)
{
const char *Err;
int ErrOffs;
int Flags = PCRE_CASELESS;
if (Regexp[0] == '/')
{
TCString OrigRegexp = Regexp;
Regexp = Regexp.Right(Regexp.GetLen() - 1);
TCHAR *pRegexpEnd = (TCHAR*)Regexp + Regexp.GetLen();
TCHAR *p = _tcsrchr(Regexp.GetBuffer(), '/');
if (!p)
{
Regexp = OrigRegexp;
} else
{
*p = 0;
Flags = 0;
while (++p < pRegexpEnd)
{
switch (*p) {
case 'i':
Flags |= PCRE_CASELESS;
break;
case 'm':
Flags |= PCRE_MULTILINE;
break;
case 's':
Flags |= PCRE_DOTALL;
break;
case 'x':
Flags |= PCRE_EXTENDED;
break;
case 'A':
Flags |= PCRE_ANCHORED;
break;
case 'f':
Flags |= PCRE_FIRSTLINE;
break;
case 'D':
Flags |= PCRE_DOLLAR_ENDONLY;
break;
case 'U':
Flags |= PCRE_UNGREEDY;
break;
case 'X':
Flags |= PCRE_EXTRA;
break;
default:
// Result += LogMessage(Translate("Warning, unknown pattern modifier '%c':\n"), *p );
break;
}
}
}
Regexp.ReleaseBuffer();
}
PcreCompileData[NewID].pPcre = pcre16_compile(Regexp, PCRE_UTF8 | PCRE_NO_UTF8_CHECK | Flags, &Err, &ErrOffs, NULL);
if (PcreCompileData[NewID].pPcre) {
PcreCompileData[NewID].pExtra = NULL;
PcreCompileData[NewID].pExtra = pcre16_study(PcreCompileData[NewID].pPcre, 0, &Err);
}
else {
// Result += LogMessage(TranslateT("Syntax error in regexp\n%s\nat offset %d: %s."), (TCHAR*)Regexp, ErrOffs, (TCHAR*)ANSI2TCHAR(Err)) + _T("\n\n");
PcreCompileData[NewID].Pattern = Regexp;
}
}
else PcreCompileData[NewID].Pattern = Regexp;
return Result;
}
static int regression_tests(void)
{
struct regression_test_case *current = regression_test_cases;
const char *error;
const char *cpu_info;
int i, err_offs;
int is_successful, is_ascii_pattern, is_ascii_input;
int total = 0;
int successful = 0;
int counter = 0;
#ifdef SUPPORT_PCRE8
pcre *re8;
pcre_extra *extra8;
int ovector8_1[32];
int ovector8_2[32];
int return_value8_1, return_value8_2;
int utf8 = 0, ucp8 = 0;
int disabled_flags8 = 0;
#endif
#ifdef SUPPORT_PCRE16
pcre16 *re16;
pcre16_extra *extra16;
int ovector16_1[32];
int ovector16_2[32];
int return_value16_1, return_value16_2;
int utf16 = 0, ucp16 = 0;
int disabled_flags16 = 0;
int length16;
#endif
/* This test compares the behaviour of interpreter and JIT. Although disabling
utf or ucp may make tests fail, if the pcre_exec result is the SAME, it is
still considered successful from pcre_jit_test point of view. */
#ifdef SUPPORT_PCRE8
pcre_config(PCRE_CONFIG_JITTARGET, &cpu_info);
#else
pcre16_config(PCRE_CONFIG_JITTARGET, &cpu_info);
#endif
printf("Running JIT regression tests\n");
printf(" target CPU of SLJIT compiler: %s\n", cpu_info);
#ifdef SUPPORT_PCRE8
pcre_config(PCRE_CONFIG_UTF8, &utf8);
pcre_config(PCRE_CONFIG_UNICODE_PROPERTIES, &ucp8);
if (!utf8)
disabled_flags8 |= PCRE_UTF8;
if (!ucp8)
disabled_flags8 |= PCRE_UCP;
printf(" in 8 bit mode with utf8 %s and ucp %s:\n", utf8 ? "enabled" : "disabled", ucp8 ? "enabled" : "disabled");
#endif
#ifdef SUPPORT_PCRE16
pcre16_config(PCRE_CONFIG_UTF16, &utf16);
pcre16_config(PCRE_CONFIG_UNICODE_PROPERTIES, &ucp16);
if (!utf16)
disabled_flags16 |= PCRE_UTF8;
if (!ucp16)
disabled_flags16 |= PCRE_UCP;
printf(" in 16 bit mode with utf16 %s and ucp %s:\n", utf16 ? "enabled" : "disabled", ucp16 ? "enabled" : "disabled");
#endif
while (current->pattern) {
/* printf("\nPattern: %s :\n", current->pattern); */
total++;
if (current->start_offset & F_PROPERTY) {
is_ascii_pattern = 0;
is_ascii_input = 0;
} else {
is_ascii_pattern = check_ascii(current->pattern);
is_ascii_input = check_ascii(current->input);
}
error = NULL;
#ifdef SUPPORT_PCRE8
re8 = NULL;
if (!(current->start_offset & F_NO8))
re8 = pcre_compile(current->pattern,
current->flags & ~(PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART | disabled_flags8),
&error, &err_offs, tables(0));
extra8 = NULL;
if (re8) {
error = NULL;
extra8 = pcre_study(re8, PCRE_STUDY_JIT_COMPILE, &error);
if (!extra8) {
printf("\n8 bit: Cannot study pattern: %s\n", current->pattern);
pcre_free(re8);
re8 = NULL;
}
if (!(extra8->flags & PCRE_EXTRA_EXECUTABLE_JIT)) {
printf("\n8 bit: JIT compiler does not support: %s\n", current->pattern);
pcre_free_study(extra8);
pcre_free(re8);
re8 = NULL;
}
} else if (((utf8 && ucp8) || is_ascii_pattern) && !(current->start_offset & F_NO8))
printf("\n8 bit: Cannot compile pattern: %s\n", current->pattern);
#endif
#ifdef SUPPORT_PCRE16
if ((current->flags & PCRE_UTF8) || (current->start_offset & F_FORCECONV))
convert_utf8_to_utf16(current->pattern, regtest_buf, NULL, REGTEST_MAX_LENGTH);
else
copy_char8_to_char16(current->pattern, regtest_buf, REGTEST_MAX_LENGTH);
re16 = NULL;
if (!(current->start_offset & F_NO16))
re16 = pcre16_compile(regtest_buf,
current->flags & ~(PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART | disabled_flags16),
&error, &err_offs, tables(0));
extra16 = NULL;
if (re16) {
error = NULL;
extra16 = pcre16_study(re16, PCRE_STUDY_JIT_COMPILE, &error);
if (!extra16) {
printf("\n16 bit: Cannot study pattern: %s\n", current->pattern);
pcre16_free(re16);
re16 = NULL;
}
if (!(extra16->flags & PCRE_EXTRA_EXECUTABLE_JIT)) {
printf("\n16 bit: JIT compiler does not support: %s\n", current->pattern);
pcre16_free_study(extra16);
pcre16_free(re16);
re16 = NULL;
}
} else if (((utf16 && ucp16) || is_ascii_pattern) && !(current->start_offset & F_NO16))
printf("\n16 bit: Cannot compile pattern: %s\n", current->pattern);
#endif
counter++;
if ((counter & 0x3) != 0) {
#ifdef SUPPORT_PCRE8
setstack8(NULL);
#endif
#ifdef SUPPORT_PCRE16
setstack16(NULL);
#endif
}
#ifdef SUPPORT_PCRE8
return_value8_1 = -1000;
return_value8_2 = -1000;
for (i = 0; i < 32; ++i)
ovector8_1[i] = -2;
for (i = 0; i < 32; ++i)
ovector8_2[i] = -2;
if (re8) {
setstack8(extra8);
return_value8_1 = pcre_exec(re8, extra8, current->input, strlen(current->input), current->start_offset & OFFSET_MASK,
current->flags & (PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART), ovector8_1, 32);
return_value8_2 = pcre_exec(re8, NULL, current->input, strlen(current->input), current->start_offset & OFFSET_MASK,
current->flags & (PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART), ovector8_2, 32);
}
#endif
#ifdef SUPPORT_PCRE16
return_value16_1 = -1000;
return_value16_2 = -1000;
for (i = 0; i < 32; ++i)
ovector16_1[i] = -2;
for (i = 0; i < 32; ++i)
ovector16_2[i] = -2;
if (re16) {
setstack16(extra16);
if ((current->flags & PCRE_UTF8) || (current->start_offset & F_FORCECONV))
length16 = convert_utf8_to_utf16(current->input, regtest_buf, regtest_offsetmap, REGTEST_MAX_LENGTH);
else
length16 = copy_char8_to_char16(current->input, regtest_buf, REGTEST_MAX_LENGTH);
return_value16_1 = pcre16_exec(re16, extra16, regtest_buf, length16, current->start_offset & OFFSET_MASK,
current->flags & (PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART), ovector16_1, 32);
return_value16_2 = pcre16_exec(re16, NULL, regtest_buf, length16, current->start_offset & OFFSET_MASK,
current->flags & (PCRE_NOTBOL | PCRE_NOTEOL | PCRE_NOTEMPTY | PCRE_NOTEMPTY_ATSTART), ovector16_2, 32);
}
#endif
/* If F_DIFF is set, just run the test, but do not compare the results.
Segfaults can still be captured. */
is_successful = 1;
if (!(current->start_offset & F_DIFF)) {
#if defined SUPPORT_PCRE8 && defined SUPPORT_PCRE16
if (utf8 == utf16 && !(current->start_offset & F_FORCECONV)) {
/* All results must be the same. */
if (return_value8_1 != return_value8_2 || return_value8_1 != return_value16_1 || return_value8_1 != return_value16_2) {
printf("\n8 and 16 bit: Return value differs(%d:%d:%d:%d): [%d] '%s' @ '%s'\n",
return_value8_1, return_value8_2, return_value16_1, return_value16_2,
total, current->pattern, current->input);
is_successful = 0;
} else if (return_value8_1 >= 0) {
return_value8_1 *= 2;
/* Transform back the results. */
if (current->flags & PCRE_UTF8) {
for (i = 0; i < return_value8_1; ++i) {
if (ovector16_1[i] >= 0)
ovector16_1[i] = regtest_offsetmap[ovector16_1[i]];
if (ovector16_2[i] >= 0)
ovector16_2[i] = regtest_offsetmap[ovector16_2[i]];
}
}
for (i = 0; i < return_value8_1; ++i)
if (ovector8_1[i] != ovector8_2[i] || ovector8_1[i] != ovector16_1[i] || ovector8_1[i] != ovector16_2[i]) {
printf("\n8 and 16 bit: Ovector[%d] value differs(%d:%d:%d:%d): [%d] '%s' @ '%s' \n",
i, ovector8_1[i], ovector8_2[i], ovector16_1[i], ovector16_2[i],
total, current->pattern, current->input);
is_successful = 0;
}
}
} else {
#endif /* SUPPORT_PCRE8 && SUPPORT_PCRE16 */
/* Only the 8 bit and 16 bit results must be equal. */
#ifdef SUPPORT_PCRE8
if (return_value8_1 != return_value8_2) {
printf("\n8 bit: Return value differs(%d:%d): [%d] '%s' @ '%s'\n",
return_value8_1, return_value8_2, total, current->pattern, current->input);
is_successful = 0;
} else if (return_value8_1 >= 0) {
return_value8_1 *= 2;
for (i = 0; i < return_value8_1; ++i)
if (ovector8_1[i] != ovector8_2[i]) {
printf("\n8 bit: Ovector[%d] value differs(%d:%d): [%d] '%s' @ '%s'\n",
i, ovector8_1[i], ovector8_2[i], total, current->pattern, current->input);
is_successful = 0;
}
}
#endif
#ifdef SUPPORT_PCRE16
if (return_value16_1 != return_value16_2) {
printf("\n16 bit: Return value differs(%d:%d): [%d] '%s' @ '%s'\n",
return_value16_1, return_value16_2, total, current->pattern, current->input);
is_successful = 0;
} else if (return_value16_1 >= 0) {
return_value16_1 *= 2;
for (i = 0; i < return_value16_1; ++i)
if (ovector16_1[i] != ovector16_2[i]) {
printf("\n16 bit: Ovector[%d] value differs(%d:%d): [%d] '%s' @ '%s'\n",
i, ovector16_1[i], ovector16_2[i], total, current->pattern, current->input);
is_successful = 0;
}
}
#endif
#if defined SUPPORT_PCRE8 && defined SUPPORT_PCRE16
}
#endif /* SUPPORT_PCRE8 && SUPPORT_PCRE16 */
}
if (is_successful) {
#ifdef SUPPORT_PCRE8
if (!(current->start_offset & F_NO8) && ((utf8 && ucp8) || is_ascii_input)) {
if (return_value8_1 < 0 && !(current->start_offset & F_NOMATCH)) {
printf("8 bit: Test should match: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
if (return_value8_1 >= 0 && (current->start_offset & F_NOMATCH)) {
printf("8 bit: Test should not match: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
}
#endif
#ifdef SUPPORT_PCRE16
if (!(current->start_offset & F_NO16) && ((utf16 && ucp16) || is_ascii_input)) {
if (return_value16_1 < 0 && !(current->start_offset & F_NOMATCH)) {
printf("16 bit: Test should match: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
if (return_value16_1 >= 0 && (current->start_offset & F_NOMATCH)) {
printf("16 bit: Test should not match: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
}
#endif
}
if (is_successful)
successful++;
#ifdef SUPPORT_PCRE8
if (re8) {
pcre_free_study(extra8);
pcre_free(re8);
}
#endif
#ifdef SUPPORT_PCRE16
if (re16) {
pcre16_free_study(extra16);
pcre16_free(re16);
}
#endif
/* printf("[%d-%d|%d-%d]%s", ovector8_1[0], ovector8_1[1], ovector16_1[0], ovector16_1[1], (current->flags & PCRE_CASELESS) ? "C" : ""); */
printf(".");
fflush(stdout);
current++;
}
tables(1);
#ifdef SUPPORT_PCRE8
setstack8(NULL);
#endif
#ifdef SUPPORT_PCRE16
setstack16(NULL);
#endif
if (total == successful) {
printf("\nAll JIT regression tests are successfully passed.\n");
return 0;
} else {
printf("\nSuccessful test ratio: %d%% (%d failed)\n", successful * 100 / total, total - successful);
return 1;
}
}
