markpron.cpp

wince平台下的一个录音机

#include <stdio.h>
#include "markpron.h"
#include "util.h"

//dict format:
//"<word>\t<syll>"
//or "<word> <syll>"

StringStringHash g_hash_word2syll;
StringStringHash g_hash_syll2mono;

int load_dict_ex(const char *file_dict, StringStringHash *hash)
{
	FILE *fp = NULL;
	char *p;
	char line[MAX_LINE_LEN+1];
	string word;
	string syll;


	hash->clear();
	
	fp = fopen(file_dict, "r");
	if( !fp )
	{
		errMsg("cannot open file %s", file_dict);
		return -1;
	}

	while( fgets(line, MAX_LINE_LEN, fp) )
	{
		if( '#' == line[0] )
			continue;
		
		p = strstr(line, "\t");

		if( !p )
			p = strstr(line, " ");
		if( p )
		{

#if 0
			if( len < MAX_WORD_LEN )
			{
				len = p - line;
				memset(tmp_buf, 0, MAX_WORD_LEN);
				strncpy(tmp_buf, line, len);				

			}else{
				wrnMsg("word in dict(%s) line %d exceed max len(%d), ignored:\n%s\n", file_dict, line_id, MAX_WORD_LEN, line);
			}
#else
				//stl implementation
			*p = 0;
			p++;
			if( '\n' == p[strlen(p)-1] )
				p[strlen(p)-1] = 0;

			word = line;
			syll = p;
			hash->insert( pair<string, string>(word, syll) );
#endif
		}else{
			errMsg("Cannot find <tab> or <space> between word and syll in dict %s\n  line:%s", file_dict, line);
		}
	}
	fclose(fp);

	return 0;
}


int load_word2syll_dict(const char *file_dict)
{
	return load_dict_ex(file_dict, &g_hash_word2syll);
}

int load_syll2mono_dict(const char *file_dict)
{
	return load_dict_ex(file_dict, &g_hash_syll2mono);
}

int load_dict(const char *file_word2syll, const char *file_syll2mono)
{
	if( -1 == load_word2syll_dict(file_word2syll) 
		|| -1 == load_syll2mono_dict(file_syll2mono) )
		return -1;
	else
		return 0;
}
// find pronounce of the word
// for left to right order
int lookup_dict(const char *word, char *pron, StringStringHash *hash)
{
	iStringStringHash itr;
	if( !word || !strlen(word) || !hash || !pron)
		return -1;

	pron[0] = 0;
	
	char w[1024];
	int pos = 0;
	while(pos < strlen(word) ){

		int len = strlen(word+pos);
		while(len > 0)
		{
			memset(w, 0, 1024);
			strncpy(w, word+pos, len); 
			itr = hash->find(w);
			if( hash->end() != itr )
			{
				if( strlen(pron))
					strcat(pron, " ");
				strcat(pron, itr->second.c_str() );
				
				if( ' ' == pron[strlen(pron)-1] )
					pron[strlen(pron)-1] = 0;

				break;
			}else{
				if( word[len-1] > 0 )
				{
					len --;
				}else{
					len -= 2;
				}
			}
		}
		pos += strlen(w);
	}

	return 0;
}

// find pronounce of the word
// for left to right order
int word2syll(const char *word, char *pron)
{
	return lookup_dict(word, pron, &g_hash_word2syll);
}

// find pronounce of the word
// for left to right order
int syll2mono(const char *word, char *pron)
{
	return lookup_dict(word, pron, &g_hash_syll2mono);
}

// find pronounce of the word
// for left to right order
// for asr application
int word2syll_sil(const char *word, char *pron)
{
	char p[1024];

	if( 0 == word2syll(word, p) )
	{
		sprintf(pron, "sil %s sil", p);
	}
	return 0;
}

// find mono-phone of the word
int word2mono(const char *word, char *pron)
{
	char p[1024];

	if( 0 == lookup_dict(word, p, &g_hash_word2syll) )
	{
		return lookup_dict(p, pron, &g_hash_syll2mono);
	}
	return -1;
}

// find mono-phone of the word
int word2mono_sil(const char *word, char *pron)
{
	char p[1024];
	
	if( 0 == word2mono(word, p) )
	{
		sprintf(pron, "sil %s sil", p);
		return 0;
	}
	return -1;
}

void unload_dict()
{
	g_hash_word2syll.clear();
	g_hash_syll2mono.clear();
	
	return;
}