int main(void) {
string_t temp;
string_t string1 = string_new("hello world");
string_t string2 = string_new("another");
string_t string3 = string_new("a third");
printf("Testing new()...\n");
assert(strcmp("hello world", string1.bytes) == 0);
printf("Testing at()...\n");
assert(*string_at(string1, 4) == 'o');
assert(*string_at(string1, -1) == 'd');
assert(string_at(string1, 100) == NULL);
assert(string_at(string1, -100) == NULL);
printf("Testing cmp()...\n");
assert(string_cmp(string1, string1) == 0);
assert(string_cmp(string1, string2) > 0);
assert(string_cmp(string2, string1) < 0);
printf("Testing eq()...\n");
assert(string_eq(string1, string1));
assert(! string_eq(string1, string2));
printf("Testing cat()...\n");
temp = string_cat(string1, string2);
assert(temp.length == string1.length + string2.length);
assert(strncmp(string1.bytes, temp.bytes, string1.length) == 0);
assert(strcmp(string2.bytes, temp.bytes + string1.length) == 0);
string_free(temp);
printf("Testing slice()...\n");
temp = string_slice(string1, 1, 6);
assert(strcmp(temp.bytes, "ello"));
string_free(temp);
temp = string_slice(string1, 1, -1);
assert(strcmp(temp.bytes, "ello worl") == 0);
string_free(temp);
temp = string_slice(string1, -5, -2);
assert(strcmp(temp.bytes, "wor") == 0);
string_free(temp);
temp = string_slice(string1, -100, 100);
assert(strcmp(temp.bytes, "hello world") == 0);
string_free(temp);
printf("Testing split()...\n");
string_split_t split = string_split_init(string1, 'o');
temp = string_split_next(&split);
assert(strcmp(temp.bytes, "hell") == 0);
string_free(temp);
temp = string_split_next(&split);
assert(strcmp(temp.bytes, " w") == 0);
string_free(temp);
temp = string_split_next(&split);
assert(strcmp(temp.bytes, "rld") == 0);
string_free(temp);
temp = string_split_next(&split);
assert(! temp.bytes);
string_free(temp);
printf("Testing join()...\n");
string_t parts[] = {string1, string2, string3};
temp = string_join(parts, 3, ' ');
assert(strcmp(temp.bytes, "hello world another a third") == 0);
string_free(temp);
// string_t[] parts = {s1, s2, s3};
// int joined = join(parts, sizeof(parts) / sizeof(*parts));
string_free(string1);
string_free(string2);
string_free(string3);
printf("All tests passed.\n");
return 0;
}
/**
* load all the valid wors from a specified lexicon file .
*
* @param dic friso dictionary instance (A hash array)
* @param lex the lexicon type
* @param lex_file the path of the lexicon file
* @param length the maximum length of the word item
*/
FRISO_API void friso_dic_load(
friso_t friso,
friso_config_t config,
friso_lex_t lex,
fstring lex_file,
uint_t length )
{
FILE * _stream;
char __char[1024], _buffer[512];
fstring _line;
string_split_entry sse;
fstring _word;
char _sbuffer[512];
fstring _syn;
friso_array_t sywords;
uint_t _fre;
if ( ( _stream = fopen( lex_file, "rb" ) ) != NULL )
{
while ( ( _line = file_get_line( __char, _stream ) ) != NULL )
{
//clear up the notes
//make sure the length of the line is greater than 1.
//like the single '#' mark in stopwords dictionary.
if ( _line[0] == '#' && strlen(_line) > 1 ) continue;
//handle the stopwords.
if ( lex == __LEX_STOPWORDS__ )
{
//clean the chinese words that its length is greater than max length.
if ( ((int)_line[0]) < 0 && strlen( _line ) > length ) continue;
friso_dic_add( friso->dic, __LEX_STOPWORDS__,
string_copy_heap( _line, strlen(_line) ), NULL );
continue;
}
//split the fstring with '/'.
string_split_reset( &sse, "/", _line);
if ( string_split_next( &sse, _buffer ) == NULL ) continue;
//1. get the word.
_word = string_copy_heap( _buffer, strlen(_buffer) );
if ( string_split_next( &sse, _buffer ) == NULL )
{
//normal lexicon type,
//add them to the dictionary directly
friso_dic_add( friso->dic, lex, _word, NULL );
continue;
}
/*
* filter out the words that its length is larger
* than the specified limit.
* but not for __LEX_ECM_WORDS__ and english __LEX_STOPWORDS__
* and __LEX_CEM_WORDS__.
*/
if ( ! ( lex == __LEX_ECM_WORDS__ || lex == __LEX_CEM_WORDS__ )
&& strlen( _word ) > length )
{
FRISO_FREE(_word);
continue;
}
//2. get the synonyms words.
_syn = NULL;
if ( strcmp( _buffer, "null" ) != 0 )
_syn = string_copy( _buffer, _sbuffer, strlen(_buffer) );
//3. get the word frequency if it available.
_fre = 0;
if ( string_split_next( &sse, _buffer ) != NULL )
_fre = atoi( _buffer );
/**
* Here:
* split the synonyms words with mark ","
* and put them in a array list if the synonyms is not NULL
*/
sywords = NULL;
if ( config->add_syn && _syn != NULL )
{
string_split_reset( &sse, ",", _sbuffer );
sywords = new_array_list_with_opacity(5);
while ( string_split_next( &sse, _buffer ) != NULL )
{
if ( strlen(_buffer) > length ) continue;
array_list_add( sywords,
string_copy_heap(_buffer, strlen(_buffer)) );
}
sywords = array_list_trim( sywords );
}
//4. add the word item
friso_dic_add_with_fre(
friso->dic, lex, _word, sywords, _fre );
}
fclose( _stream );
} else {
printf("Warning: Fail to open lexicon file %s\n", lex_file);
}
}
