static TrieNode<CharT, BucketT>*
pseudo_sample(unsigned char** strings, size_t n)
{
debug()<<__func__<<"(): sampling "<<n/8192<<" strings ...\n";
size_t max_nodes = (sizeof(CharT) == 1) ? 5000 : 2000;
TrieNode<CharT, BucketT>* root = new TrieNode<CharT, BucketT>;
for (size_t i=0; i < n; i += 8192) {
unsigned char* str = strings[i];
size_t depth = 0;
TrieNode<CharT, BucketT>* node = root;
while (true) {
CharT c = get_char<CharT>(str, depth);
if (is_end(c)) break;
depth += sizeof(CharT);
node->extend(c+1);
if (not node->is_trie(c)) {
node->_buckets[c] = new TrieNode<CharT, BucketT>;
make_trie(node->_buckets[c]);
if (--max_nodes==0) goto finish;
}
node = node->get_node(c);
assert(node);
}
}
finish:
return root;
}
static TrieNode<CharT, BucketT>*
random_sample(unsigned char** strings, size_t n)
{
const size_t sample_size = n/8192;
debug()<<__PRETTY_FUNCTION__<<" sampling "<<sample_size<<" strings\n";
size_t max_nodes = (sizeof(CharT) == 1) ? 5000 : 2000;
TrieNode<CharT, BucketT>* root = new TrieNode<CharT, BucketT>;
for (size_t i=0; i < sample_size; ++i) {
unsigned char* str = strings[size_t(drand48()*n)];
size_t depth = 0;
TrieNode<CharT, BucketT>* node = root;
while (true) {
CharT c = get_char<CharT>(str, depth);
if (is_end(c)) break;
depth += sizeof(CharT);
node->extend(c+1);
if (not node->is_trie(c)) {
node->_buckets[c] = new TrieNode<CharT, BucketT>;
make_trie(node->_buckets[c]);
if (--max_nodes==0) goto finish;
}
node = node->get_node(c);
assert(node);
}
}
finish:
return root;
}