64 lines
2.4 KiB
C++
64 lines
2.4 KiB
C++
#include <libregexis024sol/special_terminals.h>
|
|
#include <libregexis024sol/sol_misc_base.h>
|
|
#include <assert.h>
|
|
|
|
namespace regexis024 {
|
|
uint32_t read_hex(REGEX_IS024_MeaningContext& ctx, int sz){
|
|
uint32_t res = 0;
|
|
for (int i = 0; i < sz; i++){
|
|
int32_t ch = peep(ctx);
|
|
if ('0' <= ch && ch <= '9')
|
|
res = ((res << 4) | ((uint32_t)ch - '0'));
|
|
else if ('a' <= ch && ch <= 'z')
|
|
res = ((res << 4) | ((uint32_t)ch - 'a' + 10));
|
|
else if ('A' <= ch && ch <= 'Z')
|
|
res = ((res << 4) | ((uint32_t)ch - 'A' + 10));
|
|
else{
|
|
report(ctx, "escape backslash expression: bad unicode code");
|
|
return 0;
|
|
}
|
|
readChar(ctx);
|
|
}
|
|
return res;
|
|
}
|
|
|
|
void unicode_in_bs_case(REGEX_IS024_MeaningContext &ctx, bool &ret_is_multicode, codeset_t &ret_set, int sz){
|
|
ret_is_multicode = false;
|
|
readChar(ctx);
|
|
uint32_t hc = read_hex(ctx, sz); // Might create an error
|
|
ret_set = codeset_of_one_char(hc);
|
|
}
|
|
|
|
void
|
|
backslash_expression_parsing_try_regular(REGEX_IS024_MeaningContext &ctx, const CommonCodesets& cc,
|
|
bool &ret_is_multicode, codeset_t &ret_set)
|
|
{
|
|
int32_t leader = peep(ctx);
|
|
if (ctx.error)
|
|
return;
|
|
#define block(l, b, E) case l: ret_is_multicode = b; ret_set = E; readChar(ctx); break;
|
|
switch (leader) {
|
|
block('s', false, codeset_of_one_char(U' '))
|
|
block('t', false, codeset_of_one_char(U'\t'))
|
|
block('n', false, codeset_of_one_char(U'\n'))
|
|
block('r', false, codeset_of_one_char(U'\r'))
|
|
block('e', true, cc.spaces);
|
|
block('E', true, invert_set(cc.spaces))
|
|
block('w', true, cc.word_constituents);
|
|
block('W', true, invert_set(cc.word_constituents));
|
|
case 'u':
|
|
unicode_in_bs_case(ctx, ret_is_multicode, ret_set, 4);
|
|
break;
|
|
case 'U':
|
|
unicode_in_bs_case(ctx, ret_is_multicode, ret_set, 8);
|
|
break;
|
|
default:
|
|
if (leader >= 0){
|
|
ret_is_multicode = false;
|
|
ret_set = codeset_of_one_char(leader);
|
|
} else {
|
|
report(ctx, "backslash in the wrong place");
|
|
}
|
|
}
|
|
}
|
|
} |