void
PinyinPhraseLib::find_phrases_impl (PhraseVector &pv,
const PinyinPhraseOffsetVector::iterator &begin,
const PinyinPhraseOffsetVector::iterator &end,
const PinyinKeyVector::const_iterator &key_begin,
const PinyinKeyVector::const_iterator &key_pos,
const PinyinKeyVector::const_iterator &key_end)
{
if (begin == end) return;
if (key_pos == key_begin) {
for (PinyinPhraseOffsetVector::iterator i=begin; i!=end; i++) {
if (valid_pinyin_phrase (i->first, i->second) &&
get_phrase (i->first).is_enable ())
pv.push_back (get_phrase (i->first));
}
return;
}
std::sort (begin, end, PinyinPhraseLessThanByOffsetSP (this, m_pinyin_key_less, key_pos-key_begin));
std::pair<PinyinPhraseOffsetVector::iterator, PinyinPhraseOffsetVector::iterator> it =
std::equal_range (begin, end, *key_pos,
PinyinPhraseLessThanByOffsetSP (this, m_pinyin_key_less, key_pos-key_begin));
return find_phrases_impl (pv, it.first, it.second, key_begin, key_pos-1, key_end);
}
static void handle_phrase_space()
{
static U8 s_first_tap_was_on_empty_step;
U8 phrase = get_phrase();
release_space_handler_p = release_space_handler;
if (g_marking) return;
switch (g_space_tap_count)
{
case 1:
s_first_tap_was_on_empty_step = (0xffu == phrase);
if (s_first_tap_was_on_empty_step)
{
set_phrase(last_input_phrase[EDIT_CH]);
}
else
{
last_input_phrase[EDIT_CH] = phrase;
}
break;
case 2:
new_or_clone(s_first_tap_was_on_empty_step);
break;
}
}
bool
PinyinPhraseLib::insert_pinyin_phrase_into_index (uint32 phrase_index, uint32 pinyin_index)
{
if (!valid_pinyin_phrase (phrase_index, pinyin_index))
return false;
uint32 len = get_phrase (phrase_index).length();
if (len <= 0) return false;
PinyinKey key = get_pinyin_key (pinyin_index);
PinyinPhraseTable::iterator ptit=
std::lower_bound (m_phrases[len-1].begin (), m_phrases[len-1].end (), key, PinyinKeyExactLessThan ());
if (ptit != m_phrases[len-1].end () && PinyinKeyExactEqualTo () (*ptit,key)) {
ptit->get_vector ().push_back (PinyinPhraseOffsetPair (phrase_index, pinyin_index));
} else {
PinyinPhraseEntry entry (key);
entry.get_vector ().push_back (PinyinPhraseOffsetPair (phrase_index, pinyin_index));
if (ptit != m_phrases [len-1].end () &&
ptit >= m_phrases [len-1].begin () &&
m_phrases[len-1].size () > 0) {
m_phrases[len-1].insert (ptit, entry);
} else {
m_phrases[len-1].push_back (entry);
}
}
return true;
}
void
PinyinPhraseLib::dump_content (std::ostream &os, int minlen, int maxlen)
{
PinyinPhraseLessThanByOffset less_op (this, m_pinyin_key_less);
if (minlen < 1) minlen = 1;
if (maxlen > SCIM_PHRASE_MAX_LENGTH) maxlen = SCIM_PHRASE_MAX_LENGTH;
for (int i = minlen; i <= maxlen; ++ i) {
PinyinPhraseOffsetVector offsets;
for (PinyinPhraseTable::iterator tit = m_phrases [i-1].begin (); tit != m_phrases [i-1].end (); ++ tit) {
PinyinPhraseOffsetVector::iterator begin = tit->get_vector ().begin ();
PinyinPhraseOffsetVector::iterator end = tit->get_vector ().end ();
offsets.insert (offsets.end (), begin, end);
}
std::sort (offsets.begin (), offsets.end (), less_op);
for (PinyinPhraseOffsetVector::iterator oit = offsets.begin (); oit != offsets.end (); ++ oit) {
bool before = false, after = false;
os << get_phrase (oit->first).frequency () << "\t";
if (oit > offsets.begin () && get_phrase ((oit-1)->first) == get_phrase (oit->first)) before = true;
if (oit < offsets.end () - 1 && get_phrase ((oit+1)->first) == get_phrase (oit->first)) after = true;
if (before || after) os << "+";
else os << "-";
os << utf8_wcstombs (get_phrase (oit->first).get_content ());
os << " =";
for (unsigned int j = 0; j < get_phrase (oit->first).length (); ++ j)
os << " " << get_pinyin_key (oit->second + j);
os << "\n";
}
}
}
U8 get_next_phrase()
{
U8 retval;
if (CUR_ROW_CHAIN == 0xf)
return 0xffu;
++CUR_ROW_CHAIN;
retval = get_phrase();
--CUR_ROW_CHAIN;
return retval;
}
U8 get_prev_phrase()
{
U8 retval;
if (!CUR_ROW_CHAIN)
return 0xffu;
--CUR_ROW_CHAIN;
retval = get_phrase();
++CUR_ROW_CHAIN;
return retval;
}
int
bos_reader::get_phrase_str (char **next, char *buf, char delim) const
{
trim_left (next);
char *start = *next;
if (get_phrase (next, delim))
return -2;
//printf ("NAMEMMMM: %s\n", start);
if (*start == '*' || *start == '\0')
return 0;
strcpy (buf, start);
return 0;
}
int
bos_reader::get_phrase_int (char **next, t_long *i, char delim) const
{
trim_left (next);
char *start = *next;
long t;
if (get_phrase (next, delim))
return -2;
if (*start == '*' || *start == '\0')
return 0;
if (sscanf (start, "%ld", &t) < 1)
{
fprintf (stderr, "Error: can not read int from %s\n", start);
return -1;
}
*i = (t_long)t;
return 0;
}
static void new_or_clone(U8 do_new)
{
U8 phrase;
if (do_new)
{
// No phrase on step, alloc new.
phrase = next_new_phrase();
}
else
{
// Clone existing chain.
phrase = clone_phrase(get_phrase());
}
if (phrase == 0xffu)
return;
last_input_phrase[EDIT_CH] = phrase;
alloc_phrase(phrase);
set_phrase(phrase);
}
int
bos_reader::get_phrase_double (char **next, t_double *d, char delim) const
{
trim_left (next);
char *start = *next;
double t;
if (get_phrase (next, delim))
{
//printf ("get_phrase kkkkkk\n");
return -2;
}
if (*start == '*' || *start == '\0')
return 0;
if (sscanf (start, "%lf", &t) < 1)
{
fprintf (stderr, "Error: can not read double from %s\n", start);
return -1;
}
*d = (t_double)t;
return 0;
}
inline static grn_cell *
get_token(grn_ctx *ctx, grn_query *q)
{
grn_cell *token = NIL;
grn_operator op = q->default_op;
{
int weight = DEFAULT_WEIGHT, prefixp = 0, mode = -1, option = 0;
skip_space(ctx, q);
if (q->cur_expr >= q->max_exprs ||
q->cur_cell >= q->max_cells ||
q->cur >= q->str_end) { return NIL; }
switch (*q->cur) {
case '\0' :
return NIL;
case GRN_QUERY_PARENR :
q->cur++;
return NIL;
case GRN_QUERY_QUOTEL :
q->cur++;
if ((token = get_phrase(ctx, q)) == NULL) {
return NIL;
}
break;
case GRN_QUERY_PREFIX :
q->cur++;
token = get_op(q, op, weight);
break;
case GRN_QUERY_AND :
q->cur++;
token = op_new(q, GRN_OP_AND, weight, mode, option);
break;
case GRN_QUERY_BUT :
q->cur++;
token = op_new(q, GRN_OP_BUT, weight, mode, option);
break;
case GRN_QUERY_ADJ_INC :
q->cur++;
if (weight < 127) { weight++; }
token = op_new(q, GRN_OP_ADJUST, weight, mode, option);
break;
case GRN_QUERY_ADJ_DEC :
q->cur++;
if (weight > -128) { weight--; }
token = op_new(q, GRN_OP_ADJUST, weight, mode, option);
break;
case GRN_QUERY_ADJ_NEG :
q->cur++;
token = op_new(q, GRN_OP_ADJUST, -1, mode, option);
break;
case GRN_QUERY_PARENL :
q->cur++;
token = get_expr(ctx, q);
break;
default :
if ((token = get_word(ctx, q, &prefixp)) &&
token->u.b.value[0] == 'O' &&
token->u.b.value[1] == 'R' &&
token->u.b.size == 2) {
cell_del(q);
q->cur_expr--;
token = op_new(q, GRN_OP_OR, weight, mode, option);
}
break;
}
}
return cons(q, token, NIL);
}
static void release_space_handler()
{
alloc_phrase(get_phrase());
release_space_handler_p = 0;
}
void
PinyinPhraseLib::refine_pinyin_lib ()
{
PinyinKeyVector tmp_pinyin_lib;
PinyinKeyVector::const_iterator result;
PinyinKeyVector::const_iterator vit_begin;
PinyinKeyVector::const_iterator vit_end;
uint32 len;
uint32 pinyin_offset;
tmp_pinyin_lib.reserve (m_pinyin_lib.size () + 1);
for (int i=SCIM_PHRASE_MAX_LENGTH-1; i>=0; i--) {
for (PinyinPhraseTable::iterator tit=m_phrases[i].begin(); tit!=m_phrases[i].end(); tit++) {
for (PinyinPhraseOffsetVector::iterator vit=tit->get_vector ().begin();
vit!=tit->get_vector ().end(); vit++) {
len = get_phrase (vit->first).length ();
if (len > 0) {
vit_begin = m_pinyin_lib.begin () + vit->second;
vit_end = vit_begin + len;
for (result = tmp_pinyin_lib.begin ();
result != tmp_pinyin_lib.end ();
result ++) {
uint32 j;
for (j=0; j< len && result + j < tmp_pinyin_lib.end (); j++) {
if (!m_pinyin_key_equal (*(result+j), *(vit_begin + j)))
break;
}
if (j == len)
break;
}
/*
result = std::find_end (tmp_pinyin_lib.begin (),
tmp_pinyin_lib.end (),
vit_begin,
vit_end,
m_pinyin_key_equal);
*/
if (result != tmp_pinyin_lib.end ())
pinyin_offset = result - tmp_pinyin_lib.begin ();
else {
pinyin_offset = tmp_pinyin_lib.size ();
for (uint32 j=0; j<len; j++)
tmp_pinyin_lib.push_back (get_pinyin_key (vit->second + j));
}
vit->second = pinyin_offset;
}
std::cout << "." << std::flush;
}
}
}
std::cout << "\n";
m_pinyin_lib = tmp_pinyin_lib;
}