void write_header(std::vector<int> freqs, BitOutputStream out)
{
int size = 0;
int char_length = 0;
for(int i = 0; i < freqs.size(); i++)
{
if(freqs[i])
{
size++;
char_length += freqs[i];
}
}
out.writeByte(size);
out.writeInt(char_length);
for(int i = 0; i < freqs.size(); i++)
{
if(freqs[i])
{
out.writeByte(i);
out.writeInt(freqs[i]);
}
}
}
int main(int argc,char **argv)
{
//Check for arguments
if(argc != 3) {
cout << "Invalid number of arguments.\n"
<< "Usage: ./compress <input filename> <output filename>.\n";
return -1;
}
//Check for file names
if(!strcmp(argv[1],argv[2])) {
cerr << "Error: Same name for input and output files.\n";
return -1;
}
/** 1. Open the input file for reading. */
ifstream in;
in.open(argv[1],ios::binary);
/**
* 2. Read bytes from the file, counting the number of occurrences of
* each byte value; then close the file.
*/
vector<int> count(256,0);
int ch;
int isempty = 1;
while(1) {
ch = in.get();
if(!in.good()) break; // failure
count[ch]++; // read a char, count it
isempty = 0;
}
if(isempty) { // empty file
ofstream out;
out.open(argv[2],ios::binary);
out.close();
return 0;
}
if(!in.eof()) { // loop stopped for some bad reason...
cerr << "There was a problem, sorry." << endl;
return -1;
}
in.close();
/**
* 3. Use these byte counts to construct a Huffman coding tree. Each
* unique byte with a non-zero count will be a leaf node in the Huffman tree.
*/
HCTree Huffman;
Huffman.build(count);
/** 4. Open the output file for writing. */
ofstream out;
out.open(argv[2],ios::binary);
/**
* 5. Write enough information to the output file to enable the coding
* tree to be reconstructed when the file is read by your uncompress program.
*/
BitOutputStream os = BitOutputStream(out);
int realcount = 0;
for(int i=0;i<count.size();i++) {
if (count[i]) realcount++; //count for asciis that appear
}
os.writeInt(realcount); //print the number of asciis that appear
os.writeByte('\n');
for(int i=0;i<count.size();i++) {
if (count[i]) {
os.writeInt(i);
os.writeByte(' ');
os.writeInt(count[i]);
os.writeByte('\n');
}
}
/** 6. Open input file for reading again. */
in.open(argv[1],ios::binary);
/**
* 7. Using the Huffman coding tree, translate each byte from the input
* file into its code, and append these codes as a sequence of bits to
* the output file, after the header.
*/
char symbol;
while(1) {
symbol = in.get();
if(!in.good()) break;
Huffman.encode(symbol,os);
}
os.flush();
/** 8. Close the input and output files. */
in.close();
out.close();
return 0;
}
int main(int argc, char** argv){
if(argc != 3){
cerr << "One file doesn't exist, sorry." << endl;
return -1;
}
if(strcmp(argv[1],argv[2])==0){
cerr << "Output file has the same name as input file, sorry." << endl;
return -1;
}
int valid = 0;
int onlychar;
int total = 0;
int isEmpty = 1;
ifstream in;
in.open(argv[1], ios::binary);
ofstream out;
out.open(argv[2], ios::binary);
if(!in.good()){
cerr << "The input file can't be opened or read from, sorry." << endl;
return -1;
}
if(!out.good()){
cerr << "The output file can't be written, sorry." << endl;
return -1;
}
vector<int> count = vector<int>(256,0);
unsigned char ch;
while(1){
ch = in.get();
if(!in.good()) break;
isEmpty = 0;
count[ch]++;
total++;
}
if(!in.eof()){
cerr << "There was a problem, sorry." << endl;
return -1;
}
if(isEmpty == 1){
cerr << "The file is empty, sorry." << endl;
return -1;
}
in.close();
// ofstream out;
// out.open(argv[2], ios::binary);
BitOutputStream* a = new BitOutputStream(out);
a->writeInt(total);
for(int i=0;i<256;i++){
if(count[i]){
valid++;
onlychar = i;
}
}
if(valid==1){
a->writeByte(1);
a->writeByte(onlychar);
out.close();
return 0;
}else{
a->writeByte(0);
}
// for(int i=0;i<256;i++){
// cout << i << " " << count[i] << endl;
// }
HCTree* tree = new HCTree();
tree->build(count);
for(int i=0;i<256;i++){
a->writeByte(tree->leaf[i]);
}
for(int i=0;i<255;i++){
a->writeByte(tree->intnode[i]);
}
for(int i=0;i<32;i++){
a->writeByte(tree->leafchildbit[i]);
}
for(int i=0;i<32;i++){
a->writeByte(tree->intnodechildbit[i]);
}
in.open(argv[1], ios::binary);
//cout << "test" << endl;
while(1){
ch = in.get();
if(!in.good()) break;
tree->encode(ch, *a);
}
if(!in.eof()){
cerr << "There was a problem, sorry." << endl;
return -1;
}
if(a->get_buf_index()!=0){
a->flush();
}
in.close();
out.close();
return 0;
}