bool LccrFeaturizer::AccumulateFeatureCountFromLine(
trie_t &trie, const std::vector<std::vector<std::string>> &textfeats) {
for (const std::vector<std::string> &posfeat : textfeats) {
for (const std::string &feat : posfeat) {
trie.update(feat.c_str(), feat.size(), 1);
}
}
return true;
}
void LccrFeaturizer::FilterFeatureWithCount(trie_t &raw, trie_t &filtered,
int cutoff) {
namespace signal = boost::signals2::detail;
size_t from(0), p(0);
signal::auto_buffer<char, signal::store_n_bytes<LccrFeaturizer::MAXFEATLEN>>
buffer(LccrFeaturizer::MAXFEATLEN, '\0');
int featcnt;
for (featcnt = raw.begin(from, p); featcnt != trie_t::CEDAR_NO_PATH;
featcnt = raw.next(from, p)) {
if (featcnt >= cutoff) {
raw.suffix(buffer.data(), p, from);
int key = filtered.num_keys();
filtered.update(buffer.data(), p, 0) = key;
}
}
}
void makeIndex (trie_t & trie, props_t & props, bool & abort, Config const & cfg)
{
try
{
props_t tmp_props;
tmp_props.reserve(1024);
std::map<tstring, int, icasecompare> tmp_propmap;
for (SearchLocationInfo const & info : cfg.m_locations)
{
if (abort)
return;
SearchDirectory(
info.m_dir_path, info.m_includes, info.m_excludes, info.m_recursive, info.m_follow_symlinks
, TEXT("")
, [] (tstring const & fname, tstring const & cmp) { return true; }
, [&trie, &tmp_props, &tmp_propmap] (tstring const & fname, tstring const & fpath)
{
std::map<tstring, int>::iterator it = tmp_propmap.find(fname);
if (it == tmp_propmap.end())
{
tstring fname_lwr = fname;
boost::algorithm::to_lower(fname_lwr);
tmp_props.push_back(Props(fname_lwr, fpath));
trie_t::result_type const id = static_cast<trie_t::result_type>(tmp_props.size() - 1);
tmp_propmap[fname_lwr] = id;
//dbg_printf("insert: fname=%s fpath=%s idx=%i\n", fname_lwr.c_str(), fpath.c_str(), id);
trie.update(fname_lwr.c_str(), fname_lwr.length(), id);
}
else
{
tmp_props[it->second].m_fpath.push_back(fpath);
//dbg_printf("update: fname=%s fpath=%s idx=%i\n", fname.c_str(), fpath.c_str(), it->second);
}
}
, abort);
}
props = tmp_props;
}
catch (std::regex_error const & e)
{
dbg_printf("Exception caught: %s", e.what());
}
//printf("keys: %ld\n", trie.num_keys ());
//printf("size: %ld\n", trie.size ());
}