/****************************************************************************
* Function : EncodeLZSSByFile
* Description: This function will read an input file and write an output
* file encoded according to the traditional LZSS algorithm.
* This algorithm encodes strings as 16 bits (a 12 bit offset
* + a 4 bit length).
* Parameters : fpIn - pointer to the open binary file to encode
* fpOut - pointer to the open binary file to write encoded
* output
* Effects : fpIn is encoded and written to fpOut. Neither file is
* closed after exit.
* Returned : EXIT_SUCCESS or EXIT_FAILURE
****************************************************************************/
int EncodeLZSSByFile(FILE *fpIn, FILE *fpOut)
{
bit_file_t *bfpOut;
encoded_string_t matchData;
unsigned int i, c;
unsigned int len; /* length of string */
/* head of sliding window and lookahead */
unsigned int windowHead, uncodedHead;
/* use stdin if no input file */
if (fpIn == NULL)
{
fpIn = stdin;
}
if (fpOut == NULL)
{
/* use stdout if no output file */
bfpOut = MakeBitFile(stdout, BF_WRITE);
}
else
{
/* convert output file to bitfile */
bfpOut = MakeBitFile(fpOut, BF_WRITE);
}
windowHead = 0;
uncodedHead = 0;
/* Window Size : 2^12 same as offset */
/************************************************************************
* Fill the sliding window buffer with some known vales. DecodeLZSS must
* use the same values. If common characters are used, there's an
* increased chance of matching to the earlier strings.
************************************************************************/
memset(slidingWindow, ' ', WINDOW_SIZE * sizeof(unsigned char));
/* MAX_CODED : 2 to 17 because we cant have 0 to 1 */
/************************************************************************
* Copy MAX_CODED bytes from the input file into the uncoded lookahead
* buffer.
************************************************************************/
for (len = 0; len < MAX_CODED && (c = getc(fpIn)) != EOF; len++)
{
uncodedLookahead[len] = c;
}
if (len == 0)
{
return (EXIT_SUCCESS); /* inFile was empty */
}
/* Look for matching string in sliding window */
InitializeSearchStructures();
matchData = FindMatch(windowHead, uncodedHead);
/* now encoded the rest of the file until an EOF is read */
while (len > 0)
{
if (matchData.length > len)
{
/* garbage beyond last data happened to extend match length */
matchData.length = len;
}
if (matchData.length <= MAX_UNCODED)
{
/* not long enough match. write uncoded flag and character */
BitFilePutBit(UNCODED, bfpOut);
BitFilePutChar(uncodedLookahead[uncodedHead], bfpOut);
matchData.length = 1; /* set to 1 for 1 byte uncoded */
}
else
{
unsigned int adjustedLen;
/* adjust the length of the match so minimun encoded len is 0*/
adjustedLen = matchData.length - (MAX_UNCODED + 1);
/* match length > MAX_UNCODED. Encode as offset and length. */
BitFilePutBit(ENCODED, bfpOut);
BitFilePutBitsInt(bfpOut, &matchData.offset, OFFSET_BITS,
sizeof(unsigned int));
BitFilePutBitsInt(bfpOut, &adjustedLen, LENGTH_BITS,
sizeof(unsigned int));
}
/********************************************************************
* Replace the matchData.length worth of bytes we've matched in the
* sliding window with new bytes from the input file.
********************************************************************/
i = 0;
while ((i < matchData.length) && ((c = getc(fpIn)) != EOF))
{
/* add old byte into sliding window and new into lookahead */
ReplaceChar(windowHead, uncodedLookahead[uncodedHead]);
uncodedLookahead[uncodedHead] = c;
windowHead = Wrap((windowHead + 1), WINDOW_SIZE);
uncodedHead = Wrap((uncodedHead + 1), MAX_CODED);
i++;
}
/* handle case where we hit EOF before filling lookahead */
while (i < matchData.length)
{
ReplaceChar(windowHead, uncodedLookahead[uncodedHead]);
/* nothing to add to lookahead here */
windowHead = Wrap((windowHead + 1), WINDOW_SIZE);
uncodedHead = Wrap((uncodedHead + 1), MAX_CODED);
len--;
i++;
}
/* find match for the remaining characters */
matchData = FindMatch(windowHead, uncodedHead);
}
/* we've decoded everything, free bitfile structure */
BitFileToFILE(bfpOut);
return (EXIT_SUCCESS);
}
/*
* aazip - compress files using a transform based compression system
*/
int main(int argc, char** argv)
{
FILE* f;
bit_file_t* of;
char* infile,*outfile;
uint8_t* input,*lupdate,*bwt,lumode;
int32_t I,osize,opt;
uint32_t size;
mode_t lupdate_alg;
float ient,oent;
uint64_t cost,tstart,tstop,elapsed;
/* parse command line parameter */
opt = GETOPT_FINISHED;
if (argc <= 1) {
print_usage(argv[0]);
exit(EXIT_FAILURE);
}
while ((opt = getopt(argc, argv, "m:h")) != GETOPT_FINISHED) {
switch (opt) {
case 'm':
if (strcmp(optarg, "simple") == 0) lupdate_alg = SIMPLE;
else if (strcmp(optarg, "mtf") == 0) lupdate_alg = MTF;
else if (strcmp(optarg, "fc") == 0) lupdate_alg = FC;
else if (strcmp(optarg, "wfc") == 0) lupdate_alg = WFC;
else if (strcmp(optarg, "timestamp") == 0) lupdate_alg = TS;
else fatal("ERROR: mode <%s> unknown!\n", optarg);
break;
case 'h':
default:
print_usage(argv[0]);
exit(EXIT_FAILURE);
}
}
/* read input file name */
if (optind < argc) infile = argv[optind];
else {
print_usage(argv[0]);
exit(EXIT_FAILURE);
}
/* read input file */
f = safe_fopen(infile,"r");
size = safe_filesize(f);
input = (uint8_t*) safe_malloc(size+1);
if (fread(input,1,size,f)!=(size_t)size) {
fatal("read input file.");
}
safe_fclose(f);
input[size] = 0;
/* TODO calculate input entropy */
ient = 0.0f;
/* perform bwt */
bwt = (uint8_t*) safe_malloc(size);
tstart = gettime();
bwt = transform_bwt(input,size,bwt,&I);
/* peform list update */
switch (lupdate_alg) {
case SIMPLE:
fprintf(stdout,"ALGORITHM: simple\n");
lupdate = lupdate_simple(bwt,size,input,&cost);
break;
case MTF:
fprintf(stdout,"ALGORITHM: move to front\n");
lupdate = lupdate_movetofront(bwt,size,input,&cost);
break;
case FC:
fprintf(stdout,"ALGORITHM: frequency count\n");
lupdate = lupdate_freqcount(bwt,size,input,&cost);
break;
case WFC:
fprintf(stdout,"ALGORITHM: weighted frequency count\n");
lupdate = lupdate_wfc(bwt,size,input,&cost);
break;
case TS:
fprintf(stdout,"ALGORITHM: timestamp\n");
lupdate = lupdate_timestamp(bwt,size,input,&cost);
break;
default:
fatal("unkown list update algorithm.");
}
fprintf(stdout,"INPUT: %s (%d bytes)\n",infile,size);
fprintf(stdout,"COST: %lu\n",cost);
/* TODO calculate entropy after list update*/
oent = 0.0f;
/* write output */
outfile = safe_strcat(infile,".aazip");
/* create bit file for writing */
of = BitFileOpen(outfile, BF_WRITE);
/* write aa zip header */
BitFilePutChar('A', of);
BitFilePutChar('A', of);
/* write I */
BitFilePutBitsInt(of,&I,32,sizeof(uint32_t));
/* write lupdate mode */
lumode = lupdate_alg;
BitFilePutBitsInt(of,&lumode,8,sizeof(uint8_t));
fprintf(stderr,"I %d lumode %d\n",I,lumode);
/* perform huffman coding */
encode_huffman(lupdate,size,of);
tstop = gettime();
elapsed = tstop - tstart;
fprintf(stdout,"TIME: %.3f s\n",(float)elapsed/1000000);
/* flush and get file stats */
BitFileFlushOutput(of,0);
f = BitFileToFILE(of);
osize = ftell(f);
fprintf(stdout,"OUTPUT: %s\n",outfile);
fprintf(stdout,"ENTROPY: %.2f bps / %.2f bps\n",ient,oent);
fprintf(stdout,"COMPRESSION: %.2f\n",((float)osize/(float)size)*100);
/* clean up*/
safe_fclose(f);
free(input);
free(bwt);
return (EXIT_SUCCESS);
}
/****************************************************************************
* Function : HuffmanDecodeFile
* Description: This routine reads a Huffman coded file and writes out a
* decoded version of that file.
* Parameters : inFile - Open file pointer for file to decode
* outFile - Open file pointer for file receiving decoded data
* Effects : Huffman encoded file is decoded
* Returned : 0 for success, -1 for failure. errno will be set in the
* event of a failure. Either way, inFile and outFile will
* be left open.
****************************************************************************/
int HuffmanDecodeFile(FILE *inFile, FILE *outFile)
{
huffman_node_t *huffmanArray[NUM_CHARS]; /* array of all leaves */
huffman_node_t *huffmanTree;
huffman_node_t *currentNode;
int i, c;
bit_file_t *bInFile;
/* validate input and output files */
if ((NULL == inFile) || (NULL == outFile))
{
errno = ENOENT;
return -1;
}
bInFile = MakeBitFile(inFile, BF_READ);
if (NULL == bInFile)
{
perror("Making Input File a BitFile");
return -1;
}
/* allocate array of leaves for all possible characters */
for (i = 0; i < NUM_CHARS; i++)
{
if ((huffmanArray[i] = AllocHuffmanNode(i)) == NULL)
{
/* allocation failed clear existing allocations */
for (i--; i >= 0; i--)
{
free(huffmanArray[i]);
}
inFile = BitFileToFILE(bInFile);
return -1;
}
}
/* populate leaves with frequency information from file header */
if (0 != ReadHeader(huffmanArray, bInFile))
{
for (i = 0; i < NUM_CHARS; i++)
{
free(huffmanArray[i]);
}
inFile = BitFileToFILE(bInFile);
return -1;
}
/* put array of leaves into a huffman tree */
if ((huffmanTree = BuildHuffmanTree(huffmanArray, NUM_CHARS)) == NULL)
{
FreeHuffmanTree(huffmanTree);
inFile = BitFileToFILE(bInFile);
return -1;
}
/* now we should have a tree that matches the tree used on the encode */
currentNode = huffmanTree;
while ((c = BitFileGetBit(bInFile)) != EOF)
{
/* traverse the tree finding matches for our characters */
if (c != 0)
{
currentNode = currentNode->right;
}
else
{
currentNode = currentNode->left;
}
if (currentNode->value != COMPOSITE_NODE)
{
/* we've found a character */
if (currentNode->value == EOF_CHAR)
{
/* we've just read the EOF */
break;
}
fputc(currentNode->value, outFile); /* write out character */
currentNode = huffmanTree; /* back to top of tree */
}
}
/* clean up */
inFile = BitFileToFILE(bInFile); /* make file normal again */
FreeHuffmanTree(huffmanTree); /* free allocated memory */
return 0;
}
/****************************************************************************
* Function : HuffmanEncodeFile
* Description: This routine genrates a huffman tree optimized for a file
* and writes out an encoded version of that file.
* Parameters : inFile - Open file pointer for file to encode (must be
* rewindable).
* outFile - Open file pointer for file receiving encoded data
* Effects : File is Huffman encoded
* Returned : 0 for success, -1 for failure. errno will be set in the
* event of a failure. Either way, inFile and outFile will
* be left open.
****************************************************************************/
int HuffmanEncodeFile(FILE *inFile, FILE *outFile)
{
huffman_node_t *huffmanTree; /* root of huffman tree */
code_list_t codeList[NUM_CHARS]; /* table for quick encode */
bit_file_t *bOutFile;
int c;
/* validate input and output files */
if ((NULL == inFile) || (NULL == outFile))
{
errno = ENOENT;
return -1;
}
bOutFile = MakeBitFile(outFile, BF_WRITE);
if (NULL == bOutFile)
{
perror("Making Output File a BitFile");
return -1;
}
/* build tree */
if ((huffmanTree = GenerateTreeFromFile(inFile)) == NULL)
{
outFile = BitFileToFILE(bOutFile);
return -1;
}
/* build a list of codes for each symbol */
/* initialize code list */
for (c = 0; c < NUM_CHARS; c++)
{
codeList[c].code = NULL;
codeList[c].codeLen = 0;
}
if (0 != MakeCodeList(huffmanTree, codeList))
{
outFile = BitFileToFILE(bOutFile);
return -1;
}
/* write out encoded file */
/* write header for rebuilding of tree */
WriteHeader(huffmanTree, bOutFile);
/* read characters from file and write them to encoded file */
rewind(inFile); /* start another pass on the input file */
while((c = fgetc(inFile)) != EOF)
{
BitFilePutBits(bOutFile,
BitArrayGetBits(codeList[c].code),
codeList[c].codeLen);
}
/* now write EOF */
BitFilePutBits(bOutFile,
BitArrayGetBits(codeList[EOF_CHAR].code),
codeList[EOF_CHAR].codeLen);
/* free the code list */
for (c = 0; c < NUM_CHARS; c++)
{
if (codeList[c].code != NULL)
{
BitArrayDestroy(codeList[c].code);
}
}
/* clean up */
outFile = BitFileToFILE(bOutFile); /* make file normal again */
FreeHuffmanTree(huffmanTree); /* free allocated memory */
return 0;
}
/***************************************************************************
* Function : main
* Description: This function demonstrates the usage of each of the bit
* bit file functions.
* Parameters : argc - the number command line arguments (not used)
* Parameters : argv - array of command line arguments (not used)
* Effects : Writes bit file, reads back results, printing them to
* stdout.
* Returned : EXIT_SUCCESS
***************************************************************************/
int main(int argc, char *argv[])
{
bit_file_t *bfp;
FILE *fp;
int i, numCalls, value;
if (argc < 2)
{
numCalls = NUM_CALLS;
}
else
{
numCalls = atoi(argv[1]);
}
/* create bit file for writing */
bfp = BitFileOpen("testfile", BF_WRITE);
if (bfp == NULL)
{
perror("opening file");
return (EXIT_FAILURE);
}
/* write chars */
value = (int)'A';
for (i = 0; i < numCalls; i++)
{
printf("writing char %c\n", value);
if(BitFilePutChar(value, bfp) == EOF)
{
perror("writing char");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
value++;
}
/* write single bits */
value = 0;
for (i = 0; i < numCalls; i++)
{
printf("writing bit %d\n", value);
if(BitFilePutBit(value, bfp) == EOF)
{
perror("writing bit");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
value = 1 - value;
}
/* write ints as bits */
value = 0x11111111;
for (i = 0; i < numCalls; i++)
{
printf("writing bits %0X\n", (unsigned int)value);
if(BitFilePutBits(bfp, &value,
(unsigned int)(8 * sizeof(int))) == EOF)
{
perror("writing bits");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
value += 0x11111111;
}
/* close bit file */
if (BitFileClose(bfp) != 0)
{
perror("closing file");
return (EXIT_FAILURE);
}
else
{
printf("closed file\n");
}
/* reopen file for appending */
bfp = BitFileOpen("testfile", BF_APPEND);
if (bfp == NULL)
{
perror("opening file");
return (EXIT_FAILURE);
}
/* append some chars */
value = (int)'A';
for (i = 0; i < numCalls; i++)
{
printf("appending char %c\n", value);
if(BitFilePutChar(value, bfp) == EOF)
{
perror("appending char");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
value++;
}
/* write some bits from an integer */
value = 0x111;
for (i = 0; i < numCalls; i++)
{
printf("writing 12 bits from an integer %03X\n", (unsigned int)value);
if(BitFilePutBitsInt(bfp, &value, 12, sizeof(value)) == EOF)
{
perror("writing bits from an integer");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
value += 0x111;
}
/* convert to normal file */
fp = BitFileToFILE(bfp);
if (fp == NULL)
{
perror("converting to stdio FILE");
return (EXIT_FAILURE);
}
else
{
printf("converted to stdio FILE\n");
}
/* append some chars */
value = (int)'a';
for (i = 0; i < numCalls; i++)
{
printf("appending char %c\n", value);
if(fputc(value, fp) == EOF)
{
perror("appending char to FILE");
if (fclose(fp) == EOF)
{
perror("closing stdio FILE");
}
return (EXIT_FAILURE);
}
value++;
}
/* close file */
if (fclose(fp) == EOF)
{
perror("closing stdio FILE");
return (EXIT_FAILURE);
}
/* now read back writes */
/* open bit file */
bfp = BitFileOpen("testfile", BF_READ);
if (bfp == NULL)
{
perror("reopening file");
return (EXIT_FAILURE);
}
/* read chars */
for (i = 0; i < numCalls; i++)
{
value = BitFileGetChar(bfp);
if(value == EOF)
{
perror("reading char");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
else
{
printf("read %c\n", value);
}
}
/* read single bits */
for (i = 0; i < numCalls; i++)
{
value = BitFileGetBit(bfp);
if(value == EOF)
{
perror("reading bit");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
else
{
printf("read bit %d\n", value);
}
}
/* read ints as bits */
for (i = 0; i < numCalls; i++)
{
if(BitFileGetBits(bfp, &value, (unsigned int)(8 * sizeof(int))) == EOF)
{
perror("reading bits");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
else
{
printf("read bits %0X\n", (unsigned int)value);
}
}
if (BitFileByteAlign(bfp) == EOF)
{
fprintf(stderr, "failed to align file\n");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
else
{
printf("byte aligning file\n");
}
/* read appended characters */
for (i = 0; i < numCalls; i++)
{
value = BitFileGetChar(bfp);
if(value == EOF)
{
perror("reading char");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
else
{
printf("read %c\n", value);
}
}
/* read some bits into an integer */
for (i = 0; i < numCalls; i++)
{
value = 0;
if(BitFileGetBitsInt(bfp, &value, 12, sizeof(value)) == EOF)
{
perror("reading bits from an integer");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
else
{
printf("read 12 bits into an integer %03X\n", (unsigned int)value);
}
}
/* convert to stdio FILE */
fp = BitFileToFILE(bfp);
if (fp == NULL)
{
perror("converting to stdio FILE");
return (EXIT_FAILURE);
}
else
{
printf("converted to stdio FILE\n");
}
/* read append some chars */
value = (int)'a';
for (i = 0; i < numCalls; i++)
{
value = fgetc(fp);
if(value == EOF)
{
perror("stdio reading char");
if (0 != BitFileClose(bfp))
{
perror("closing bitfile");
}
return (EXIT_FAILURE);
}
else
{
printf("stdio read %c\n", value);
}
}
/* close file */
if (fclose(fp) == EOF)
{
perror("closing stdio FILE");
return (EXIT_FAILURE);
}
return(EXIT_SUCCESS);
}