github_mirrors/vectorscan
3
0
Fork 1
mirror of https://github.com/VectorCamp/vectorscan.git synced 2025-06-28 16:41:01 +03:00
Code Issues Packages Projects Releases Wiki Activity

reduce character classes for hwlm accel

Browse Source
This commit is contained in:
Alex Coyte 2016-07-01 14:04:55 +10:00 committed by Matthew Barr
parent 69933edf68
commit c58424cab3
1 changed files with 32 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
src/hwlm/hwlm_build.cpp
View File

@ -347,6 +347,25 @@ void filterLits(const vector<hwlmLiteral> &lits, hwlm_group_t expected_groups,
} }
} }
static
bool litGuardedByCharReach(const CharReach &cr, const hwlmLiteral &lit,
u32 max_offset) {
for (u32 i = 0; i <= max_offset && i < lit.s.length(); i++) {
unsigned char c = lit.s[i];
if (lit.nocase) {
if (cr.test(mytoupper(c)) && cr.test(mytolower(c))) {
return true;
}
} else {
if (cr.test(c)) {
return true;
}
}
}
return false;
}
static static
void findForwardAccelScheme(const vector<hwlmLiteral> &lits, void findForwardAccelScheme(const vector<hwlmLiteral> &lits,
hwlm_group_t expected_groups, AccelAux *aux) { hwlm_group_t expected_groups, AccelAux *aux) {
@ -364,26 +383,33 @@ void findForwardAccelScheme(const vector<hwlmLiteral> &lits,
return; return;
} }
/* look for shufti/truffle */
vector<CharReach> reach(MAX_ACCEL_OFFSET, CharReach()); vector<CharReach> reach(MAX_ACCEL_OFFSET, CharReach());
for (const auto &lit : lits) { for (const auto &lit : lits) {
if (!(lit.groups & expected_groups)) { if (!(lit.groups & expected_groups)) {
continue; continue;
} }
for (u32 i = 0; i < MAX_ACCEL_OFFSET && i < lit.s.length(); i++) { for (u32 i = 0; i < MAX_ACCEL_OFFSET; i++) {
unsigned char c = lit.s[i]; CharReach &reach_i = reach[i];
if (litGuardedByCharReach(reach_i, lit, i)) {
continue;
}
unsigned char c = i < lit.s.length() ? lit.s[i] : lit.s.back();
if (lit.nocase) { if (lit.nocase) {
reach[i].set(mytoupper(c)); reach_i.set(mytoupper(c));
reach[i].set(mytolower(c)); reach_i.set(mytolower(c));
} else { } else {
reach[i].set(c); reach_i.set(c);
} }
} }
} }
u32 min_count = ~0U; u32 min_count = ~0U;
u32 min_offset = ~0U; u32 min_offset = ~0U;
for (u32 i = 0; i < min_len; i++) { for (u32 i = 0; i < MAX_ACCEL_OFFSET; i++) {
size_t count = reach[i].count(); size_t count = reach[i].count();
DEBUG_PRINTF("offset %u is %s (reach %zu)\n", i, DEBUG_PRINTF("offset %u is %s (reach %zu)\n", i,
describeClass(reach[i]).c_str(), count); describeClass(reach[i]).c_str(), count);
@ -392,7 +418,6 @@ void findForwardAccelScheme(const vector<hwlmLiteral> &lits,
min_offset = i; min_offset = i;
} }
} }
assert(min_offset <= min_len);
if (min_count > MAX_SHUFTI_WIDTH) { if (min_count > MAX_SHUFTI_WIDTH) {
DEBUG_PRINTF("FAIL: min shufti with %u chars is too wide\n", min_count); DEBUG_PRINTF("FAIL: min shufti with %u chars is too wide\n", min_count);