weaselab/weaseljson
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Use table for hex values

Browse Source
This commit is contained in:
Andrew Noyes
2025-06-23 21:52:17 -04:00
parent 2299904557
commit 9803364adb
2 changed files with 25 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
src/parser3.h
View File

@@ -514,16 +514,12 @@ inline PRESERVE_NONE WeaselJsonStatus n_string_following_escape(Parser3 *self,
inline PRESERVE_NONE WeaselJsonStatus t_hex(Parser3 *self, char *buf, inline PRESERVE_NONE WeaselJsonStatus t_hex(Parser3 *self, char *buf,
char *bufEnd) { char *bufEnd) {
self->utf8Codepoint <<= 4; auto hexVal = tables.hex[uint8_t(*buf)];
if (('0' <= *buf && *buf <= '9')) { if (hexVal < 0) [[unlikely]] {
self->utf8Codepoint |= *buf - '0';
} else if ('a' <= *buf && *buf <= 'f') {
self->utf8Codepoint |= 10 + *buf - 'a';
} else if ('A' <= *buf && *buf <= 'F') {
self->utf8Codepoint |= 10 + *buf - 'A';
} else [[unlikely]] {
return WeaselJson_REJECT; return WeaselJson_REJECT;
} }
self->utf8Codepoint <<= 4;
self->utf8Codepoint |= hexVal;
++buf; ++buf;
self->pop(); self->pop();
MUSTTAIL return Parser3::keepGoing(self, buf, bufEnd); MUSTTAIL return Parser3::keepGoing(self, buf, bufEnd);
@@ -531,16 +527,12 @@ inline PRESERVE_NONE WeaselJsonStatus t_hex(Parser3 *self, char *buf,
inline PRESERVE_NONE WeaselJsonStatus t_hex2(Parser3 *self, char *buf, inline PRESERVE_NONE WeaselJsonStatus t_hex2(Parser3 *self, char *buf,
char *bufEnd) { char *bufEnd) {
self->utf8Codepoint <<= 4; auto hexVal = tables.hex[uint8_t(*buf)];
if (('0' <= *buf && *buf <= '9')) { if (hexVal < 0) [[unlikely]] {
self->utf8Codepoint |= *buf - '0';
} else if ('a' <= *buf && *buf <= 'f') {
self->utf8Codepoint |= 10 + *buf - 'a';
} else if ('A' <= *buf && *buf <= 'F') {
self->utf8Codepoint |= 10 + *buf - 'A';
} else [[unlikely]] {
return WeaselJson_REJECT; return WeaselJson_REJECT;
} }
self->utf8Codepoint <<= 4;
self->utf8Codepoint |= hexVal;
++buf; ++buf;
// Write codepoint in utf-8 if there's room in the user provided buffer. If // Write codepoint in utf-8 if there's room in the user provided buffer. If
@@ -599,16 +591,12 @@ inline PRESERVE_NONE WeaselJsonStatus t_hex2(Parser3 *self, char *buf,
inline PRESERVE_NONE WeaselJsonStatus t_hex3(Parser3 *self, char *buf, inline PRESERVE_NONE WeaselJsonStatus t_hex3(Parser3 *self, char *buf,
char *bufEnd) { char *bufEnd) {
self->utf8Codepoint <<= 4; auto hexVal = tables.hex[uint8_t(*buf)];
if (('0' <= *buf && *buf <= '9')) { if (hexVal < 0) [[unlikely]] {
self->utf8Codepoint |= *buf - '0';
} else if ('a' <= *buf && *buf <= 'f') {
self->utf8Codepoint |= 10 + *buf - 'a';
} else if ('A' <= *buf && *buf <= 'F') {
self->utf8Codepoint |= 10 + *buf - 'A';
} else [[unlikely]] {
return WeaselJson_REJECT; return WeaselJson_REJECT;
} }
self->utf8Codepoint <<= 4;
self->utf8Codepoint |= hexVal;
++buf; ++buf;
if (!(0xdc00 <= self->utf8Codepoint && self->utf8Codepoint <= 0xdfff)) if (!(0xdc00 <= self->utf8Codepoint && self->utf8Codepoint <= 0xdfff))

13
src/tables.h
View File

@@ -17,9 +17,22 @@ constexpr inline struct Tables {
unescape['f'] = '\f'; unescape['f'] = '\f';
unescape['\\'] = '\\'; unescape['\\'] = '\\';
unescape['/'] = '/'; unescape['/'] = '/';
for (int i = 0; i < 256; ++i) {
hex[i] = -1;
}
for (int i = '0'; i <= '9'; ++i) {
hex[i] = i - '0';
}
for (int i = 'a'; i <= 'f'; ++i) {
hex[i] = 10 + i - 'a';
}
for (int i = 'A'; i <= 'F'; ++i) {
hex[i] = 10 + i - 'A';
}
} }
bool whitespace[256]{}; bool whitespace[256]{};
char unescape[256]{}; char unescape[256]{};
int8_t hex[256]{};
} tables; } tables;
// See https://gist.github.com/pervognsen/218ea17743e1442e59bb60d29b1aa725 for // See https://gist.github.com/pervognsen/218ea17743e1442e59bb60d29b1aa725 for