// Ugly but simple.
string make_pattern() {
std::ostringstream oss;
oss << "^[";
for (size_t i = reach.find_first(); i != CharReach::npos;
i = reach.find_next(i)) {
oss << "\\x" << std::hex << std::setw(2) << std::setfill('0')
<< (unsigned)(i & 0xff) << std::dec;
}
oss << "]{" << min << "," << max << "}";
return oss.str();
}
/** Calculate the stop alphabet for each depth from 0 to MAX_STOP_DEPTH. Then
* build an eight-bit mask per character C, with each bit representing the
* depth before the location of character C (if encountered) that the NFA would
* be in a predictable start state. */
vector<u8> findLeftOffsetStopAlphabet(const NGHolder &g, som_type som) {
const depth max_depth(MAX_STOP_DEPTH);
const InitDepths depths(g);
const map<NFAVertex, BoundedRepeatSummary> no_vertices;
vector<CharReach> reach(MAX_STOP_DEPTH);
for (auto v : vertices_range(g)) {
if (is_special(v, g)) {
continue;
}
CharReach v_cr;
if (som == SOM_NONE) {
v_cr = reduced_cr(v, g, no_vertices);
} else {
v_cr = g[v].char_reach;
}
u32 d = min(max_depth, depths.maxDist(g, v));
for (u32 i = 0; i < d; i++) {
reach[i] |= v_cr;
}
}
#ifdef DEBUG
for (u32 i = 0; i < MAX_STOP_DEPTH; i++) {
DEBUG_PRINTF("depth %u, stop chars: ", i);
describeClass(stdout, ~reach[i], 20, CC_OUT_TEXT);
printf("\n");
}
#endif
vector<u8> stop(N_CHARS, 0);
for (u32 i = 0; i < MAX_STOP_DEPTH; i++) {
CharReach cr = ~reach[i]; // invert reach for stop chars.
const u8 mask = 1U << i;
for (size_t c = cr.find_first(); c != cr.npos; c = cr.find_next(c)) {
stop[c] |= mask;
}
}
return stop;
}
vector<u8> findLeftOffsetStopAlphabet(const CastleProto &castle,
UNUSED som_type som) {
const depth max_width = findMaxWidth(castle);
DEBUG_PRINTF("castle has reach %s and max width %s\n",
describeClass(castle.reach()).c_str(),
max_width.str().c_str());
const CharReach escape = ~castle.reach(); // invert reach for stop chars.
u32 d = min(max_width, depth(MAX_STOP_DEPTH));
const u8 mask = verify_u8((1U << d) - 1);
vector<u8> stop(N_CHARS, 0);
for (size_t c = escape.find_first(); c != escape.npos;
c = escape.find_next(c)) {
stop[c] |= mask;
}
return stop;
}
TEST(ng_charreach, count) {
CharReach cr;
cr.set(1);
cr.set(2);
cr.set('a');
cr.set('Z');
cr.set('m');
cr.set('~');
cr.set(210);
size_t n = cr.find_first();
ASSERT_FALSE(n == CharReach::npos);
unsigned int i = 0;
while (n != CharReach::npos) {
i++;
n = cr.find_next(n);
}
ASSERT_EQ(i, cr.count());
}