array<string>
tt_analyze (string family) {
array<string> r;
font fn= tt_font (family, 10, 1200);
font_metric fnm= tt_font_metric (family, 10, 1200);
//cout << "Analyzing " << family << "\n";
get_glyph_fatal= false;
analyze_range (fn, fnm, r, family);
analyze_special (fn, fnm, r);
analyze_major (fn, fnm, r);
//analyze_trace (fn, fnm, r);
get_glyph_fatal= true;
//cout << " -> " << r << "\n";
cout << r << " " << family << "\n";
return r;
}
int generic_morpho::analyze(string_piece form, guesser_mode guesser, vector<tagged_lemma>& lemmas) const {
lemmas.clear();
if (form.len) {
// Generate all casing variants if needed (they are different than given form).
string form_uclc; // first uppercase, rest lowercase
string form_lc; // all lowercase
generate_casing_variants(form, form_uclc, form_lc);
// Start by analysing using the dictionary and all casing variants.
dictionary.analyze(form, lemmas);
if (!form_uclc.empty()) dictionary.analyze(form_uclc, lemmas);
if (!form_lc.empty()) dictionary.analyze(form_lc, lemmas);
if (!lemmas.empty()) return NO_GUESSER;
// Then call analyze_special to handle numbers, punctuation and symbols.
analyze_special(form, lemmas);
if (!lemmas.empty()) return NO_GUESSER;
// For the statistical guesser, use all casing variants.
if (guesser == GUESSER && statistical_guesser) {
if (form_uclc.empty() && form_lc.empty())
statistical_guesser->analyze(form, lemmas, nullptr);
else {
morpho_statistical_guesser::used_rules used_rules; used_rules.reserve(3);
statistical_guesser->analyze(form, lemmas, &used_rules);
if (!form_uclc.empty()) statistical_guesser->analyze(form_uclc, lemmas, &used_rules);
if (!form_lc.empty()) statistical_guesser->analyze(form_lc, lemmas, &used_rules);
}
}
if (!lemmas.empty()) return GUESSER;
}
lemmas.emplace_back(string(form.str, form.len), unknown_tag);
return -1;
}