mirror of
https://github.com/GrammaticalFramework/gf-core.git
synced 2026-04-09 04:59:31 -06:00
optimize the reading of strings in the grammar loader.
This commit is contained in:
@@ -253,7 +253,6 @@ gu_in_f64be(GuIn* in, GuExn* err)
|
||||
return gu_decode_double(gu_in_u64le(in, err));
|
||||
}
|
||||
|
||||
|
||||
static void
|
||||
gu_in_fini(GuFinalizer* fin)
|
||||
{
|
||||
|
||||
@@ -5,6 +5,7 @@
|
||||
#include <gu/string.h>
|
||||
#include <gu/utf8.h>
|
||||
#include <gu/assert.h>
|
||||
#include <stdlib.h>
|
||||
|
||||
const GuString gu_empty_string = { 1 };
|
||||
|
||||
@@ -165,6 +166,43 @@ gu_string_write(GuString s, GuOut* out, GuExn* err)
|
||||
gu_out_bytes(out, src, sz, err);
|
||||
}
|
||||
|
||||
GuString
|
||||
gu_string_read(size_t len, GuPool* pool, GuIn* in, GuExn* err)
|
||||
{
|
||||
uint8_t* buf = alloca(len*4);
|
||||
uint8_t* p = buf;
|
||||
for (size_t i = 0; i < len; i++) {
|
||||
gu_in_utf8_buf(&p, in, err);
|
||||
}
|
||||
return gu_utf8_string(buf, p-buf, pool);
|
||||
}
|
||||
|
||||
GuString
|
||||
gu_string_read_latin1(size_t len, GuPool* pool, GuIn* in, GuExn* err)
|
||||
{
|
||||
if (len < GU_MIN(sizeof(GuWord), 128)) {
|
||||
GuWord w = 0;
|
||||
for (size_t n = 0; n < len; n++) {
|
||||
w = w << 8 | gu_in_u8(in, err);
|
||||
}
|
||||
w = w << 8 | (len << 1) | 1;
|
||||
return (GuString) { w };
|
||||
}
|
||||
uint8_t* p = NULL;
|
||||
if (len < 256) {
|
||||
p = gu_malloc_aligned(pool, 1 + len, 2);
|
||||
p[0] = (uint8_t) len;
|
||||
} else {
|
||||
p = gu_malloc_prefixed(pool, gu_alignof(size_t),
|
||||
sizeof(size_t), 1, 1 + len);
|
||||
((size_t*) p)[-1] = len;
|
||||
p[0] = 0;
|
||||
}
|
||||
|
||||
gu_in_bytes(in, &p[1], len, err);
|
||||
return (GuString) { (GuWord) (void*) p };
|
||||
}
|
||||
|
||||
GuString
|
||||
gu_format_string_v(const char* fmt, va_list args, GuPool* pool)
|
||||
{
|
||||
|
||||
@@ -34,6 +34,12 @@ gu_string_copy(GuString string, GuPool* pool);
|
||||
void
|
||||
gu_string_write(GuString string, GuOut* out, GuExn* err);
|
||||
|
||||
GuString
|
||||
gu_string_read(size_t len, GuPool* pool, GuIn* in, GuExn* err);
|
||||
|
||||
GuString
|
||||
gu_string_read_latin1(size_t len, GuPool* pool, GuIn* in, GuExn* err);
|
||||
|
||||
GuIn*
|
||||
gu_string_in(GuString string, GuPool* pool);
|
||||
|
||||
|
||||
@@ -72,8 +72,8 @@ fail:
|
||||
return 0;
|
||||
}
|
||||
|
||||
extern inline void
|
||||
gu_out_utf8(GuUCS ucs, GuOut* out, GuExn* err);
|
||||
extern inline GuUCS
|
||||
gu_in_utf8(GuIn* in, GuExn* err);
|
||||
|
||||
static size_t
|
||||
gu_advance_utf8(GuUCS ucs, uint8_t* buf)
|
||||
@@ -121,5 +121,46 @@ gu_out_utf8_(GuUCS ucs, GuOut* out, GuExn* err)
|
||||
}
|
||||
}
|
||||
|
||||
extern inline GuUCS
|
||||
gu_in_utf8(GuIn* in, GuExn* err);
|
||||
extern inline void
|
||||
gu_out_utf8(GuUCS ucs, GuOut* out, GuExn* err);
|
||||
|
||||
void
|
||||
gu_in_utf8_buf(uint8_t** buf, GuIn* in, GuExn* err)
|
||||
{
|
||||
uint8_t* p = *buf;
|
||||
|
||||
uint8_t c = gu_in_u8(in, err);
|
||||
if (!gu_ok(err)) {
|
||||
return;
|
||||
}
|
||||
*(p++) = c;
|
||||
int len = (c < 0x80 ? 0 :
|
||||
c < 0xc2 ? -1 :
|
||||
c < 0xe0 ? 1 :
|
||||
c < 0xf0 ? 2 :
|
||||
c < 0xf5 ? 3 :
|
||||
-1);
|
||||
if (len < 0) {
|
||||
goto fail;
|
||||
} else if (len == 0) {
|
||||
*buf = p;
|
||||
return;
|
||||
}
|
||||
static const uint8_t mask[4] = { 0x7f, 0x1f, 0x0f, 0x07 };
|
||||
// If reading the extra bytes causes EOF, it is an encoding
|
||||
// error, not a legitimate end of character stream.
|
||||
GuExn* tmp_err = gu_exn(err, GuEOF, NULL);
|
||||
gu_in_bytes(in, p, len, tmp_err);
|
||||
if (tmp_err->caught) {
|
||||
goto fail;
|
||||
}
|
||||
if (!gu_ok(err)) {
|
||||
return;
|
||||
}
|
||||
*buf = p;
|
||||
return;
|
||||
|
||||
fail:
|
||||
gu_raise(err, GuUCSExn);
|
||||
return;
|
||||
}
|
||||
|
||||
@@ -29,7 +29,12 @@ gu_out_utf8(GuUCS ucs, GuOut* out, GuExn* err)
|
||||
}
|
||||
}
|
||||
|
||||
// Helper functions used in other modules
|
||||
|
||||
GuUCS
|
||||
gu_utf8_decode(const uint8_t** utf8);
|
||||
|
||||
void
|
||||
gu_in_utf8_buf(uint8_t** buf, GuIn* in, GuExn* err);
|
||||
|
||||
#endif // GU_UTF8_H_
|
||||
|
||||
@@ -91,38 +91,15 @@ pgf_read_len(PgfReader* rdr)
|
||||
static PgfCId
|
||||
pgf_read_cid(PgfReader* rdr)
|
||||
{
|
||||
GuPool* tmp_pool = gu_new_pool();
|
||||
GuStringBuf* sbuf = gu_string_buf(tmp_pool);
|
||||
GuOut* out = gu_string_buf_out(sbuf);
|
||||
|
||||
size_t len = pgf_read_len(rdr);
|
||||
for (size_t i = 0; i < len; i++) {
|
||||
// CIds are in latin-1
|
||||
GuUCS ucs = gu_in_u8(rdr->in, rdr->err);
|
||||
gu_out_utf8(ucs, out, rdr->err);
|
||||
}
|
||||
GuString str = gu_string_buf_freeze(sbuf, rdr->opool);
|
||||
gu_pool_free(tmp_pool);
|
||||
return str;
|
||||
return gu_string_read_latin1(len, rdr->opool, rdr->in, rdr->err);
|
||||
}
|
||||
|
||||
static GuString
|
||||
pgf_read_string(PgfReader* rdr)
|
||||
{
|
||||
GuPool* tmp_pool = gu_new_pool();
|
||||
GuStringBuf* sbuf = gu_string_buf(tmp_pool);
|
||||
GuOut* out = gu_string_buf_out(sbuf);
|
||||
|
||||
GuLength len = pgf_read_len(rdr);
|
||||
|
||||
for (size_t i = 0; i < len; i++) {
|
||||
GuUCS ucs = gu_in_utf8(rdr->in, rdr->err);
|
||||
gu_out_utf8(ucs, out, rdr->err);
|
||||
}
|
||||
GuString str = gu_string_buf_freeze(sbuf, rdr->opool);
|
||||
gu_pool_free(tmp_pool);
|
||||
|
||||
return str;
|
||||
return gu_string_read(len, rdr->opool, rdr->in, rdr->err);
|
||||
}
|
||||
|
||||
static void
|
||||
|
||||
Reference in New Issue
Block a user