/// A return value of true means all is well (even if no replacements were performed), false
/// indicates an unrecoverable error.
bool regex_replacer_t::replace_matches(const wchar_t *arg) {
if (regex.code == 0) {
// pcre2_compile() failed
return false;
}
uint32_t options = PCRE2_SUBSTITUTE_OVERFLOW_LENGTH | PCRE2_SUBSTITUTE_EXTENDED |
(opts.all ? PCRE2_SUBSTITUTE_GLOBAL : 0);
size_t arglen = wcslen(arg);
PCRE2_SIZE bufsize = (arglen == 0) ? 16 : 2 * arglen;
wchar_t *output = (wchar_t *)malloc(sizeof(wchar_t) * bufsize);
int pcre2_rc;
bool done = false;
while (!done) {
if (output == NULL) {
DIE_MEM();
}
PCRE2_SIZE outlen = bufsize;
pcre2_rc = pcre2_substitute(regex.code, PCRE2_SPTR(arg), arglen,
0, // start offset
options, regex.match,
0, // match context
PCRE2_SPTR(replacement.c_str()), PCRE2_ZERO_TERMINATED,
(PCRE2_UCHAR *)output, &outlen);
if (pcre2_rc != PCRE2_ERROR_NOMEMORY || bufsize >= outlen) {
done = true;
} else {
bufsize = outlen;
// cppcheck-suppress memleakOnRealloc
output = (wchar_t *)realloc(output, sizeof(wchar_t) * bufsize);
}
}
bool rc = true;
if (pcre2_rc < 0) {
string_error(streams, _(L"%ls: Regular expression substitute error: %ls\n"), argv0,
pcre2_strerror(pcre2_rc).c_str());
rc = false;
} else {
if (!opts.quiet) {
streams.out.append(output);
streams.out.append(L'\n');
}
total_replaced += pcre2_rc;
}
free(output);
return rc;
}
char *preg_replace(char *re, char *replacement, char *subject) {
int rc;
pcre2_code *compiled_re = get_compiled_re(re);
PCRE2_SPTR pcre2_subject = (PCRE2_SPTR)subject;
size_t subject_length = strlen((char *)subject);
PCRE2_SPTR pcre2_replacement = (PCRE2_SPTR)replacement;
size_t replacement_length = strlen((char *)replacement);
PCRE2_UCHAR output[256];
size_t output_length = 256;
rc = pcre2_substitute(
compiled_re,
pcre2_subject,
subject_length,
0,
PCRE2_SUBSTITUTE_GLOBAL,
NULL,
NULL,
pcre2_replacement,
replacement_length,
output,
&output_length
);
if (rc < 0) {
switch(rc) {
case PCRE2_ERROR_NOMEMORY:
printf("Output buffer not large enough\n"); break;
case PCRE2_ERROR_BADREPLACEMENT:
printf("Invalid replacement string %s\n", replacement); break;
default:
printf("Unknown error %d \n", rc); break;
}
exit(1);
}
return strndup((char *)output, output_length);
}
tb_char_t const* tb_regex_replace(tb_regex_ref_t self, tb_char_t const* cstr, tb_size_t size, tb_size_t start, tb_char_t const* replace_cstr, tb_size_t replace_size, tb_size_t* plength)
{
// check
tb_regex_t* regex = (tb_regex_t*)self;
tb_assert_and_check_return_val(regex && regex->code && cstr && replace_cstr, tb_null);
// done
tb_char_t const* result = tb_null;
do
{
// clear length first
if (plength) *plength = 0;
// end?
tb_check_break(start < size);
// init options
#ifdef __tb_debug__
tb_uint32_t options = 0;
#else
tb_uint32_t options = PCRE2_NO_UTF_CHECK;
#endif
if (regex->mode & TB_REGEX_MODE_GLOBAL) options |= PCRE2_SUBSTITUTE_GLOBAL;
// init buffer
if (!regex->buffer_data)
{
regex->buffer_maxn = tb_max(size + replace_size + 64, 256);
regex->buffer_data = (PCRE2_UCHAR*)tb_malloc_bytes(regex->buffer_maxn);
}
tb_assert_and_check_break(regex->buffer_data);
// done
tb_long_t ok = -1;
PCRE2_SIZE length = 0;
while (1)
{
// replace it
length = (PCRE2_SIZE)regex->buffer_maxn;
ok = pcre2_substitute(regex->code, (PCRE2_SPTR)cstr, (PCRE2_SIZE)size, (PCRE2_SIZE)start, options, tb_null, tb_null, (PCRE2_SPTR)replace_cstr, (PCRE2_SIZE)replace_size, regex->buffer_data, &length);
// no space?
if (ok == PCRE2_ERROR_NOMEMORY)
{
// grow buffer
regex->buffer_maxn <<= 1;
regex->buffer_data = (PCRE2_UCHAR*)tb_ralloc_bytes(regex->buffer_data, regex->buffer_maxn);
tb_assert_and_check_break(regex->buffer_data);
}
// failed
else if (ok < 0)
{
#if defined(__tb_debug__) && !defined(TB_CONFIG_OS_WINDOWS)
// get error info
PCRE2_UCHAR info[256];
pcre2_get_error_message(ok, info, sizeof(info));
// trace
tb_trace_d("replace failed at offset %lu: error: %ld, %s\n", start, ok, info);
#endif
// end
break;
}
else break;
}
// check
tb_check_break(ok > 0);
tb_assert_and_check_break(length < regex->buffer_maxn);
// end
regex->buffer_data[length] = '\0';
// trace
tb_trace_d(" replace: [%lu]: %s", length, regex->buffer_data);
// save length
if (plength) *plength = (tb_size_t)length;
// ok
result = (tb_char_t const*)regex->buffer_data;
} while (0);
// ok?
return result;
}
bool replace_matches(const wchar_t *arg)
{
// A return value of true means all is well (even if no replacements
// were performed), false indicates an unrecoverable error.
if (regex.code == 0)
{
// pcre2_compile() failed
return false;
}
uint32_t options = opts.all ? PCRE2_SUBSTITUTE_GLOBAL : 0;
size_t arglen = wcslen(arg);
PCRE2_SIZE bufsize = (arglen == 0) ? 16 : 2 * arglen;
wchar_t *output = (wchar_t *)malloc(sizeof(wchar_t) * bufsize);
if (output == 0)
{
DIE_MEM();
}
int pcre2_rc = 0;
for (;;)
{
PCRE2_SIZE outlen = bufsize;
pcre2_rc = pcre2_substitute(
regex.code,
PCRE2_SPTR(arg),
arglen,
0, // start offset
options,
regex.match,
0, // match context
PCRE2_SPTR(replacement.c_str()),
PCRE2_ZERO_TERMINATED,
(PCRE2_UCHAR *)output,
&outlen);
if (pcre2_rc == PCRE2_ERROR_NOMEMORY)
{
if (bufsize < MAX_REPLACE_SIZE)
{
bufsize = std::min(2 * bufsize, MAX_REPLACE_SIZE);
output = (wchar_t *)realloc(output, sizeof(wchar_t) * bufsize);
if (output == 0)
{
DIE_MEM();
}
continue;
}
string_error(streams, _(L"%ls: Replacement string too large\n"), argv0);
free(output);
return false;
}
break;
}
bool rc = true;
if (pcre2_rc < 0)
{
string_error(streams, _(L"%ls: Regular expression substitute error: %ls\n"),
argv0, pcre2_strerror(pcre2_rc).c_str());
rc = false;
}
else
{
if (!opts.quiet)
{
streams.out.append(output);
streams.out.append(L'\n');
}
total_replaced += pcre2_rc;
}
free(output);
return rc;
}
