int match_line_with_keyword(const char *line, int line_length, const char *keyword, MatchMode mode)
{
MYLOG("line_length %d", line_length);
wchar_t line_char,keyword_char;
int match_hanzi_count = 0;
utf8vector line_vector = utf8vector_create(line, line_length);
utf8vector keyword_vector = utf8vector_create(keyword, -1);
int keyword_length = utf8vector_uni_count(keyword_vector);
int keyword_index = 0;
wchar_t *keyword_uni = malloc(sizeof(wchar_t) * keyword_length);
while ((keyword_char = utf8vector_next_unichar(keyword_vector)) != '\0')
{
keyword_uni[keyword_index] = keyword_char;
keyword_index ++;
}
//printf("keyword length %d\n", keyword_length);
int match_rt = 1;
keyword_index = 0;
while((line_char = utf8vector_next_unichar(line_vector)) != '\0'
&& keyword_index < keyword_length)
{
keyword_char = keyword_uni[keyword_index];
//printf("keyword index %d\n", keyword_index);
if (pinyin_ishanzi(line_char))
{
if (pinyin_ishanzi(keyword_char))
{
if (line_char != keyword_char)
{
match_rt = 0;
break;
}
}
else if (pinyin_isabc(keyword_char))
{
keyword_char = pinyin_lowercase(keyword_char);
//printf("keyword_char %d\n", keyword_char);
const char **pinyins;
int count = pinyin_get_pinyins_by_unicode(line_char, &pinyins);
if (mode == MatchModeFirstLetter)
{
int finded = 0;
for (int i = 0; i < count; i++)
{
//printf("pinyin0 %d\n", pinyins[i][0]);
if (keyword_char == pinyins[i][0])
{
finded = 1;
//printf("matched !!!!\n");
break;
}
}
if (finded == 0)
match_rt = 0;
else
match_hanzi_count ++;
}
else if (mode == MatchModeFull)
{
int finded = 0;
for (int i = 0; i < count; i++)
{
int kindex_start = keyword_index;
const char *pinyin = pinyins[i];
int j = 0;
char pinyin_char;
while ((pinyin_char = pinyin[j]) != '\0' && kindex_start < keyword_length)
{
if (pinyin_char != pinyin_lowercase(keyword_uni[kindex_start]))
{
break;
}
j++;
kindex_start ++;
}
int matched = (pinyin_char == '\0');
if (matched)
{
finded = 1;
keyword_index = kindex_start - 1;
break;
}
}
if (finded == 0)
match_rt = 0;
else
match_hanzi_count ++;
}
free(pinyins);
if (match_rt == 0)
break;
}
else
{
//printf("not hanzi or abc %d\n", keyword_char);
match_rt = 0;
break;
}
}
else
{
if (line_char != keyword_char)
{
match_rt = 0;
break;
}
}
keyword_index ++;
}
//keyword.length > line.length
if (match_rt == 1 && keyword_index < keyword_length)
match_rt = 0;
free(keyword_uni);
utf8vector_free(line_vector);
utf8vector_free(keyword_vector);
if (match_rt == 0)
return -1;
else
return match_hanzi_count;
}
void hz2py(const char *line,
int line_length,
int add_blank,
int polyphone_support,
int first_letter_only,
int convert_double_char,
int show_tones)
{
wchar_t uni_char;
wchar_t last_uni_char = 0;
const char *utf8;
int utf8_length;
utf8vector line_vector = utf8vector_create(line, line_length);
while((uni_char = utf8vector_next_unichar_with_raw(line_vector, &utf8, &utf8_length)) != '\0')
{
if (pinyin_ishanzi(uni_char))
{
const char **pinyins = NULL;
int print_count = 0;
int count = pinyin_get_pinyins_by_unicode(uni_char, &pinyins);
if (count == 0)
{
printf("%.*s", utf8_length, utf8);
}
else
{
char *tones = NULL;
if (show_tones)
pinyin_get_tones_by_unicode(uni_char, &tones);
// add blank
if (add_blank && last_uni_char != 0 && !pinyin_ishanzi(last_uni_char)) printf(" ");
for (int i = 0; i < count; i++)
{
if (first_letter_only)
{
if (show_tones)
{
if (print_count > 0)
printf("|");
printf("%c", pinyins[i][0]);
print_count ++;
}
else
{
int has_print = 0;
char c = pinyins[i][0];
for (int j = 0; j < i; j ++)
{
if (pinyins[j][0] == c)
{
has_print = 1;
break;
}
}
if (! has_print)
{
// fprintf(stderr, "I am here, i =%d \n", i);
if (print_count > 0)
printf("|");
printf("%c", pinyins[i][0]);
print_count++;
}
else
{
continue;
}
}
}
else
{
if (show_tones)
{
if (print_count > 0)
printf("|");
printf("%s", pinyins[i]);
print_count ++;
}
else
{
int has_print = 0;
char *s = (char *)pinyins[i];
for (int j = 0; j < i; j ++)
{
if (strcmp(pinyins[j], s) == 0)
{
has_print = 1;
break;
}
}
if (!has_print)
{
if (print_count > 0)
printf("|");
printf("%s", pinyins[i]);
print_count ++;
}
else
{
continue;
}
}
}
if (show_tones)
printf("%d", tones[i]);
if (!polyphone_support)
break;
}
if (add_blank) printf(" ");
free(tones);
}
free(pinyins);
}
else
{
if (convert_double_char && uni_char > 65280 && uni_char < 65375)
{
printf("%c", uni_char - 65248);
}
else if (convert_double_char && uni_char == 12288)
{
printf("%c", 32);
}
else
{
printf("%.*s", utf8_length, utf8);
}
}
last_uni_char = uni_char;
}
printf("\n");
utf8vector_free(line_vector);
}
