Move dfa's to tables.h
This commit is contained in:
@@ -179,7 +179,7 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
|
||||
json_utf8 = json_utf8 && c >= 0x20 && c != '"' && c != '\\';
|
||||
}
|
||||
if (json_utf8) {
|
||||
parser3::Utf8Dfa dfa;
|
||||
Utf8Dfa dfa;
|
||||
auto result = dfa.scan((const char *)data, (const char *)data + size);
|
||||
bool ok = result == (const char *)data + size && dfa.accept();
|
||||
bool valid = simdjson::validate_utf8(s.data(), s.size());
|
||||
|
||||
626
src/parser3.h
626
src/parser3.h
@@ -16,632 +16,6 @@
|
||||
|
||||
namespace parser3 {
|
||||
|
||||
// See https://gist.github.com/pervognsen/218ea17743e1442e59bb60d29b1aa725 for
|
||||
// an explanation of this cycle/byte dfa implementation.
|
||||
//
|
||||
// Recognizes json number syntax. As a regex:
|
||||
// -?([0-9]|[1-9][0-9]*)(\.[0-9]+)?((e|E)(-|\+)?[0-9]+)?
|
||||
struct NumDfa {
|
||||
constexpr static uint64_t table[256] = {
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x36000ull,
|
||||
0x0ull,
|
||||
0x36600ull,
|
||||
0x12480000000000ull,
|
||||
0x0ull,
|
||||
0x780aa47b091ec00ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0xc30c000000000ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0xc30c000000000ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
};
|
||||
// Restore this dfa to its start state
|
||||
void reset() { state = 6; }
|
||||
// Return true if this dfa is in an accept state. You probably want to call
|
||||
// scan until the match ends first.
|
||||
bool accept() const {
|
||||
return (state & 63) == 30 || (state & 63) == 36 || (state & 63) == 48 ||
|
||||
(state & 63) == 42;
|
||||
}
|
||||
// clang-format off
|
||||
#ifdef __x86_64__
|
||||
__attribute__((target_clones("default", "bmi2")))
|
||||
#endif
|
||||
// Return value either points to the first byte which does not match, or bufEnd.
|
||||
// Leaves the dfa in the last state of the match.
|
||||
const char *scan(const char *buf, const char *bufEnd) {
|
||||
// clang-format on
|
||||
auto state_ = state;
|
||||
for (;;) {
|
||||
constexpr int kStride = 16;
|
||||
if (bufEnd - buf < kStride) [[unlikely]] {
|
||||
while (buf != bufEnd) {
|
||||
uint64_t row = table[uint8_t(*buf)];
|
||||
auto prev = state_;
|
||||
state_ = (row >> (state_ & 63)) & 63;
|
||||
if (state_ == 0) {
|
||||
state_ = prev;
|
||||
break;
|
||||
}
|
||||
++buf;
|
||||
}
|
||||
state = state_;
|
||||
return buf;
|
||||
}
|
||||
uint8_t prev[kStride + 1];
|
||||
prev[0] = state_;
|
||||
for (int i = 0; i < kStride; ++i) {
|
||||
uint64_t row = table[uint8_t(*buf)];
|
||||
prev[i + 1] = row >> (prev[i] & 63);
|
||||
if ((prev[i + 1] & 63) == 0) {
|
||||
state = prev[i];
|
||||
return buf;
|
||||
}
|
||||
++buf;
|
||||
}
|
||||
state_ = prev[kStride];
|
||||
}
|
||||
}
|
||||
|
||||
private:
|
||||
uint64_t state = 6;
|
||||
};
|
||||
|
||||
// Recognizes sequences of valid utf8 characters except 0-0x20, double quote,
|
||||
// and backslash
|
||||
struct Utf8Dfa {
|
||||
constexpr static uint64_t table[256] = {
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x0ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x0ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x2a000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x24000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0xc000000000000ull,
|
||||
0x6000000000000ull,
|
||||
0x6000000000000ull,
|
||||
0x6000000000000ull,
|
||||
0x12000000000000ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
};
|
||||
// Restore this dfa to its start state
|
||||
void reset() { state = 48; }
|
||||
// Return true if this dfa is in an accept state. You probably want to call
|
||||
// scan until the match ends first.
|
||||
bool accept() const { return (state & 63) == 48; }
|
||||
// clang-format off
|
||||
#ifdef __x86_64__
|
||||
__attribute__((target_clones("default", "bmi2")))
|
||||
#endif
|
||||
// Return value either points to the first byte which does not match, or bufEnd.
|
||||
// Leaves the dfa in the last state of the match.
|
||||
const char *scan(const char *buf, const char *bufEnd) {
|
||||
// clang-format on
|
||||
auto state_ = state;
|
||||
for (;;) {
|
||||
constexpr int kStride = 16;
|
||||
if (bufEnd - buf < kStride) [[unlikely]] {
|
||||
while (buf != bufEnd) {
|
||||
uint64_t row = table[uint8_t(*buf)];
|
||||
auto prev = state_;
|
||||
state_ = (row >> (state_ & 63)) & 63;
|
||||
if (state_ == 0) {
|
||||
state_ = prev;
|
||||
break;
|
||||
}
|
||||
++buf;
|
||||
}
|
||||
state = state_;
|
||||
return buf;
|
||||
}
|
||||
uint8_t prev[kStride + 1];
|
||||
prev[0] = state_;
|
||||
for (int i = 0; i < kStride; ++i) {
|
||||
uint64_t row = table[uint8_t(*buf)];
|
||||
prev[i + 1] = row >> (prev[i] & 63);
|
||||
if ((prev[i + 1] & 63) == 0) {
|
||||
state = prev[i];
|
||||
return buf;
|
||||
}
|
||||
++buf;
|
||||
}
|
||||
state_ = prev[kStride];
|
||||
}
|
||||
}
|
||||
|
||||
private:
|
||||
uint64_t state = 48;
|
||||
};
|
||||
|
||||
typedef PRESERVE_NONE WeaselJsonStatus (*Continuation)(struct Parser3 *,
|
||||
char *buf, char *bufEnd);
|
||||
|
||||
|
||||
628
src/tables.h
628
src/tables.h
@@ -1,5 +1,7 @@
|
||||
#pragma once
|
||||
|
||||
#include <cstdint>
|
||||
|
||||
constexpr inline struct Tables {
|
||||
|
||||
constexpr Tables() {
|
||||
@@ -19,3 +21,629 @@ constexpr inline struct Tables {
|
||||
bool whitespace[256]{};
|
||||
char unescape[256]{};
|
||||
} tables;
|
||||
|
||||
// See https://gist.github.com/pervognsen/218ea17743e1442e59bb60d29b1aa725 for
|
||||
// an explanation of this cycle/byte dfa implementation.
|
||||
//
|
||||
// Recognizes json number syntax. As a regex:
|
||||
// -?([0-9]|[1-9][0-9]*)(\.[0-9]+)?((e|E)(-|\+)?[0-9]+)?
|
||||
struct NumDfa {
|
||||
constexpr static uint64_t table[256] = {
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x36000ull,
|
||||
0x0ull,
|
||||
0x36600ull,
|
||||
0x12480000000000ull,
|
||||
0x0ull,
|
||||
0x780aa47b091ec00ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x780aa47aa91ea80ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0xc30c000000000ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0xc30c000000000ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
};
|
||||
// Restore this dfa to its start state
|
||||
void reset() { state = 6; }
|
||||
// Return true if this dfa is in an accept state. You probably want to call
|
||||
// scan until the match ends first.
|
||||
bool accept() const {
|
||||
return (state & 63) == 30 || (state & 63) == 36 || (state & 63) == 48 ||
|
||||
(state & 63) == 42;
|
||||
}
|
||||
// clang-format off
|
||||
#ifdef __x86_64__
|
||||
__attribute__((target_clones("default", "bmi2")))
|
||||
#endif
|
||||
// Return value either points to the first byte which does not match, or bufEnd.
|
||||
// Leaves the dfa in the last state of the match.
|
||||
const char *scan(const char *buf, const char *bufEnd) {
|
||||
// clang-format on
|
||||
auto state_ = state;
|
||||
for (;;) {
|
||||
constexpr int kStride = 16;
|
||||
if (bufEnd - buf < kStride) [[unlikely]] {
|
||||
while (buf != bufEnd) {
|
||||
uint64_t row = table[uint8_t(*buf)];
|
||||
auto prev = state_;
|
||||
state_ = (row >> (state_ & 63)) & 63;
|
||||
if (state_ == 0) {
|
||||
state_ = prev;
|
||||
break;
|
||||
}
|
||||
++buf;
|
||||
}
|
||||
state = state_;
|
||||
return buf;
|
||||
}
|
||||
uint8_t prev[kStride + 1];
|
||||
prev[0] = state_;
|
||||
for (int i = 0; i < kStride; ++i) {
|
||||
uint64_t row = table[uint8_t(*buf)];
|
||||
prev[i + 1] = row >> (prev[i] & 63);
|
||||
if ((prev[i + 1] & 63) == 0) {
|
||||
state = prev[i];
|
||||
return buf;
|
||||
}
|
||||
++buf;
|
||||
}
|
||||
state_ = prev[kStride];
|
||||
}
|
||||
}
|
||||
|
||||
private:
|
||||
uint64_t state = 6;
|
||||
};
|
||||
|
||||
// Recognizes sequences of valid utf8 characters except 0-0x20, double quote,
|
||||
// and backslash
|
||||
struct Utf8Dfa {
|
||||
constexpr static uint64_t table[256] = {
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x0ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x0ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x30000000000000ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x18630780780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x1863001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x60063001e780ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x18000000000000ull,
|
||||
0x2a000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x24000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0x1e000000000000ull,
|
||||
0xc000000000000ull,
|
||||
0x6000000000000ull,
|
||||
0x6000000000000ull,
|
||||
0x6000000000000ull,
|
||||
0x12000000000000ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
0x0ull,
|
||||
};
|
||||
// Restore this dfa to its start state
|
||||
void reset() { state = 48; }
|
||||
// Return true if this dfa is in an accept state. You probably want to call
|
||||
// scan until the match ends first.
|
||||
bool accept() const { return (state & 63) == 48; }
|
||||
// clang-format off
|
||||
#ifdef __x86_64__
|
||||
__attribute__((target_clones("default", "bmi2")))
|
||||
#endif
|
||||
// Return value either points to the first byte which does not match, or bufEnd.
|
||||
// Leaves the dfa in the last state of the match.
|
||||
const char *scan(const char *buf, const char *bufEnd) {
|
||||
// clang-format on
|
||||
auto state_ = state;
|
||||
for (;;) {
|
||||
constexpr int kStride = 16;
|
||||
if (bufEnd - buf < kStride) [[unlikely]] {
|
||||
while (buf != bufEnd) {
|
||||
uint64_t row = table[uint8_t(*buf)];
|
||||
auto prev = state_;
|
||||
state_ = (row >> (state_ & 63)) & 63;
|
||||
if (state_ == 0) {
|
||||
state_ = prev;
|
||||
break;
|
||||
}
|
||||
++buf;
|
||||
}
|
||||
state = state_;
|
||||
return buf;
|
||||
}
|
||||
uint8_t prev[kStride + 1];
|
||||
prev[0] = state_;
|
||||
for (int i = 0; i < kStride; ++i) {
|
||||
uint64_t row = table[uint8_t(*buf)];
|
||||
prev[i + 1] = row >> (prev[i] & 63);
|
||||
if ((prev[i + 1] & 63) == 0) {
|
||||
state = prev[i];
|
||||
return buf;
|
||||
}
|
||||
++buf;
|
||||
}
|
||||
state_ = prev[kStride];
|
||||
}
|
||||
}
|
||||
|
||||
private:
|
||||
uint64_t state = 48;
|
||||
};
|
||||
|
||||
@@ -309,7 +309,7 @@ TEST_CASE("bench5") {
|
||||
}
|
||||
|
||||
TEST_CASE("num dfa") {
|
||||
parser3::NumDfa dfa;
|
||||
NumDfa dfa;
|
||||
std::string match =
|
||||
"111111111111111111111111111111111111111111111111111111111111111111111111"
|
||||
"111111111111111111111111111111111111111111111111111111111111111111111111"
|
||||
@@ -339,7 +339,7 @@ const char *utf8str =
|
||||
"💩💩💩💩💩💩💩💩💩💩💩💩💩💩💩💩💩💩💩💩💩💩💩💩";
|
||||
|
||||
TEST_CASE("utf8 dfa") {
|
||||
parser3::Utf8Dfa dfa;
|
||||
Utf8Dfa dfa;
|
||||
std::string match = utf8str;
|
||||
auto *buf = dfa.scan(match.data(), match.data() + match.size());
|
||||
CHECK(buf == match.data() + match.size());
|
||||
|
||||
Reference in New Issue
Block a user