bool Dict::absolute_garbage(const WERD_CHOICE &word,
const UNICHARSET &unicharset) {
if (word.length() < kMinAbsoluteGarbageWordLength) return false;
int num_alphanum = 0;
for (int x = 0; x < word.length(); ++x) {
num_alphanum += (unicharset.get_isalpha(word.unichar_id(x)) ||
unicharset.get_isdigit(word.unichar_id(x)));
}
return (static_cast<float>(num_alphanum) /
static_cast<float>(word.length()) < kMinAbsoluteGarbageAlphanumFrac);
}
int Dict::case_ok(const WERD_CHOICE &word, const UNICHARSET &unicharset) {
int state = 0;
int x;
for (x = 0; x < word.length(); ++x) {
UNICHAR_ID ch_id = word.unichar_id(x);
if (unicharset.get_isupper(ch_id))
state = case_state_table[state][1];
else if (unicharset.get_islower(ch_id))
state = case_state_table[state][2];
else if (unicharset.get_isdigit(ch_id))
state = case_state_table[state][3];
else
state = case_state_table[state][0];
if (state == -1) return false;
}
return state != 5; // single lower is bad
}
BOOL8 Tesseract::non_0_digit(const UNICHARSET& ch_set, UNICHAR_ID unichar_id) {
return ch_set.get_isdigit(unichar_id) && !ch_set.eq(unichar_id, "0");
}