int main(int argc,char **argv)
{
int ret;
int i=0;
int total[COUNT];
int code_num = 0;
huffman_tree root = NULL;
HuffmanCode huffcode = NULL;
memset(total,0,128*sizeof(int));
if(0 != (ret = getTotalMs(compressFile,total)))
{
printf("ERROR:get Total Message Failed\n");
}else if(0 != create_huffman_tree(root,total,COUNT))
{
printf("ERROR:create huffman tree Failed.\n");
}else if(0 != huffman_encode(huffcode,root,COUNT,code_num))
{
printf("ERROR:Generate Code Failed\n");
}else
{
//free(root);
//free(huffcode);
}
printf("code Count :%d\n",code_num);
for(i=0;i<code_num;i++)
{
printf("%d:%c code is:",i,huffcode[i].character);
printf("%s\n",huffcode[i].code);
}
system("pause");
return 0;
}
int32_t load_huffman_table(uint32_t *cpt, FILE *f, huff_table_t *ht)
{
printhead("%5u : nb codes : ",*cpt);
uint32_t init = *cpt;
uint32_t tabnb[16];
for (uint32_t i = 0; i < 16; i++) {
uint8_t oct = read_byte(cpt,f);
printhead("%x ",oct);
tabnb[i] = oct;
}
printhead("\n%5u : symboles : ┌─────────────────────────┐\n",*cpt);
uint8_t **tabsym = create_huffman_table(tabnb,cpt,f);
create_huffman_tree(tabnb,tabsym,ht);
for (uint32_t i = 0; i < 16; i++) {
free(tabsym[i]);
}
free(tabsym);
return *cpt - init;
}
void Word2Vec::build_vocab(vector<vector<string>> &sentences)
{
unordered_map<string, int> word_cn;
for(auto& sentence: sentences)
for(auto& w: sentence)
if(word_cn.count(w) > 0)
word_cn[w]++;
else
word_cn[w] = 1;
for(auto kv: word_cn)
{
if(kv.second < min_count)
continue;
Word *w = new Word(0, kv.second, kv.first);
vocab.push_back(w);
vocab_hash[w->text] = WordP(w);
}
//update word index
size_t vocab_size = vocab.size();
sort(vocab.begin(), vocab.end(), comp);
for(uint32_t i = 0; i < vocab_size; i++)
{
vocab[i]->index = i;
idx2word.push_back(vocab[i]->text);
}
if(train_method == "hs")
create_huffman_tree();
if(negative)
make_table();
precalc_sampling();
}
