GitHub/gf-core
1
0
Fork 1
mirror of https://github.com/GrammaticalFramework/gf-core.git synced 2026-05-08 18:52:50 -06:00
Code Issues Packages Projects Releases Wiki Activity

restore the sharing of sequences. Shrinks the grammar by ~45%

Browse Source
This commit is contained in:
Krasimir Angelov
2022-01-08 19:49:42 +01:00
parent cd2c6aa32a
commit 00f857559d
31 changed files with 882 additions and 353 deletions
Download Patch File Download Diff File Expand all files Collapse all files

48
src/runtime/c/pgf/text.cxx
View File

@@ -5,7 +5,7 @@ int textcmp(PgfText *t1, PgfText *t2)
{
for (size_t i = 0; ; i++) {
if (i >= t1->size)
return (i - t2->size);
return -(i < t2->size);
if (i >= t2->size)
return 1;
@@ -16,6 +16,48 @@ int textcmp(PgfText *t1, PgfText *t2)
}
}
PGF_INTERNAL
void texticmp(PgfText *t1, PgfText *t2, int res[2])
{
const uint8_t *s1 = (uint8_t*) &t1->text;
const uint8_t *e1 = s1+t1->size;
const uint8_t *s2 = (uint8_t*) &t2->text;
const uint8_t *e2 = s2+t2->size;
for (;;) {
if (s1 >= e1) {
res[0] = (res[1] = -(s2 < e2));
return;
}
if (s2 >= e2) {
res[0] = (res[1] = 1);
return;
}
uint32_t ucs1 = pgf_utf8_decode(&s1);
uint32_t ucs1i = pgf_utf8_to_upper(ucs1);
uint32_t ucs2 = pgf_utf8_decode(&s2);
uint32_t ucs2i = pgf_utf8_to_upper(ucs2);
if (ucs1i > ucs2i) {
res[0] = (res[1] = 1);
return;
}
else if (ucs1i < ucs2i) {
res[0] = (res[1] = -1);
return;
}
else if (res[1] == 0) {
if (ucs1 > ucs2)
res[1] = 1;
else if (ucs1 < ucs2)
res[1] = -1;
}
}
}
PGF_INTERNAL
bool textstarts(PgfText *t, PgfText *prefix)
{
@@ -49,7 +91,7 @@ ref<PgfText> textdup_db(PgfText *t1)
}
PGF_API uint32_t
pgf_utf8_decode(const uint8_t** src_inout)
pgf_utf8_decode(const uint8_t **src_inout)
{
const uint8_t* src = *src_inout;
uint8_t c = src[0];
@@ -74,7 +116,7 @@ pgf_utf8_decode(const uint8_t** src_inout)
}
PGF_API void
pgf_utf8_encode(uint32_t ucs, uint8_t** buf)
pgf_utf8_encode(uint32_t ucs, uint8_t **buf)
{
uint8_t* p = *buf;
if (ucs < 0x80) {