bool Dict::AcceptableChoice(const WERD_CHOICE& best_choice, XHeightConsistencyEnum xheight_consistency) { float CertaintyThreshold = stopper_nondict_certainty_base; int WordSize; if (stopper_no_acceptable_choices) return false; if (best_choice.length() == 0) return false; bool no_dang_ambigs = !best_choice.dangerous_ambig_found(); bool is_valid_word = valid_word_permuter(best_choice.permuter(), false); bool is_case_ok = case_ok(best_choice, getUnicharset()); if (stopper_debug_level >= 1) { const char *xht = "UNKNOWN"; switch (xheight_consistency) { case XH_GOOD: xht = "NORMAL"; break; case XH_SUBNORMAL: xht = "SUBNORMAL"; break; case XH_INCONSISTENT: xht = "INCONSISTENT"; break; default: xht = "UNKNOWN"; } tprintf("\nStopper: %s (word=%c, case=%c, xht_ok=%s=[%g,%g])\n", best_choice.unichar_string().string(), (is_valid_word ? 'y' : 'n'), (is_case_ok ? 'y' : 'n'), xht, best_choice.min_x_height(), best_choice.max_x_height()); } // Do not accept invalid words in PASS1. if (reject_offset_ <= 0.0f && !is_valid_word) return false; if (is_valid_word && is_case_ok) { WordSize = LengthOfShortestAlphaRun(best_choice); WordSize -= stopper_smallword_size; if (WordSize < 0) WordSize = 0; CertaintyThreshold += WordSize * stopper_certainty_per_char; } if (stopper_debug_level >= 1) tprintf("Stopper: Rating = %4.1f, Certainty = %4.1f, Threshold = %4.1f\n", best_choice.rating(), best_choice.certainty(), CertaintyThreshold); if (no_dang_ambigs && best_choice.certainty() > CertaintyThreshold && xheight_consistency < XH_INCONSISTENT && UniformCertainties(best_choice)) { return true; } else { if (stopper_debug_level >= 1) { tprintf("AcceptableChoice() returned false" " (no_dang_ambig:%d cert:%.4g thresh:%g uniform:%d)\n", no_dang_ambigs, best_choice.certainty(), CertaintyThreshold, UniformCertainties(best_choice)); } return false; } }
// Update hyphen_word_, and copy the given DawgPositionVectors into // hyphen_active_dawgs_. void Dict::set_hyphen_word(const WERD_CHOICE &word, const DawgPositionVector &active_dawgs) { if (hyphen_word_ == NULL) { hyphen_word_ = new WERD_CHOICE(word.unicharset()); hyphen_word_->make_bad(); } if (hyphen_word_->rating() > word.rating()) { *hyphen_word_ = word; // Remove the last unichar id as it is a hyphen, and remove // any unichar_string/lengths that are present. hyphen_word_->remove_last_unichar_id(); hyphen_active_dawgs_ = active_dawgs; } if (hyphen_debug_level) { hyphen_word_->print("set_hyphen_word: "); } }