writing 5 bits at the time to a file

[zambak]

Hi

I have assignment for some wierd compression alghoritam that will read
in from a file convert characters to 5 bit codes and then write out
compressed
version of the original file. For example if input file contains string
"AMIR"
and the codes are A=0, M=12,I=8,R=17 i am supposed to write out 3 byte file

The problem? How do I figure out shifting because I can only write out bytes
and not
bits...I figure out I would "create" a byte and write it but the codes will
overlap.
In my example I would have something like this in binary

00000001 00010001 00010000

where first 5 bits are hex 0, second 5 bits are 12 and so on...
Sometimes one created byte is based on codes of 3 different input
characters.
I have a loop in which I read a char at the time from input file, create a
int code and now I have to figure out
an alghoritam that would create a byte ( a char) that I can write out to
output file

// get char at the time from input
while ((curr_chr = getc(in_file)) != EOF)
{
// get code for it
code = getCode(curr_chr);

// shifting code ......


// write created byte...
fwrite(&wbyte, 1, 1, out_file);
}


Any help is much appreciated....

Thnaks
Amir

 

[zambak]

Hi
So are you saying create an array of 5 chars like

unsigned char glob[5];

then read in 8 characters at the time , shift manipulate them into the glob
and then
write out the glob array? And if I run out of the input while getting 8
chars at the time
then just padd the rest of the glob array with zeroes....

Is that what you are suggesting?

Thanks
Amir

 

[osmium]

I have assignment for some wierd compression alghoritam that will read
in from a file convert characters to 5 bit codes and then write out
compressed
version of the original file. For example if input file contains string
"AMIR"
and the codes are A=0, M=12,I=8,R=17 i am supposed to write out 3 byte file

The problem? How do I figure out shifting because I can only write out bytes
and not
bits...I figure out I would "create" a byte and write it but the codes will
overlap.
In my example I would have something like this in binary

00000001 00010001 00010000

where first 5 bits are hex 0, second 5 bits are 12 and so on...
Sometimes one created byte is based on codes of 3 different input
characters.
I have a loop in which I read a char at the time from input file, create a
int code and now I have to figure out
an alghoritam that would create a byte ( a char) that I can write out to
output file

// get char at the time from input
while ((curr_chr = getc(in_file)) != EOF)
{
// get code for it
code = getCode(curr_chr);

// shifting code ......


// write created byte...
fwrite(&wbyte, 1, 1, out_file);
}

I may not fully understand your problem, but what else is new?

It osunds like you want to write five byte "globs" since 40 is the least
common multiple of 8 and 5. Add pad bits as needed and some method of
identifying them, perhaps a four-byte "field count" as a prefix in the file
or an unused five bit code.

 

[zambak]

HI

Thank you.....That seems to work just fine....

Amir

So are you saying create an array of 5 chars like

unsigned char glob[5];

then read in 8 characters at the time , shift manipulate them into the glob
and then
write out the glob array? And if I run out of the input while getting 8
chars at the time
then just padd the rest of the glob array with zeroes....

Is that what you are suggesting?

Yes, something like that is what I had in mind.

 

[Simon Biber]

Thank you.....That seems to work just fine....

I'm glad you managed to get it working... it took me some
debugging to get all the shifts and masks correct.

This basically works; it doesn't know exactly when the input
has ended so you get a variable number (0 to 7) of extra
'a' characters (code value zero) at the end of the decoded
file.

It loses capitalisation as there's not room in 5 bits to
store both uppercase and lowercase letters.

#include <stdio.h>
#include <stdlib.h>
#include <ctype.h>
#include <string.h>

const char *set = "abcdefghijklmnopqrstuvwxyz .,?!\n"; /* 32 chars */

unsigned char encodechar(unsigned char c)
{
char *p = strchr(set, tolower(c));
if(p == NULL) /* Not in code set */
{
if(isprint(c))
fprintf(stderr, "Error: Cannot encode character '%c'\n", (int)c);
else
fprintf(stderr, "Error: Cannot encode character %u\n", (unsigned int)c);
return 0;
}
return p - set;
}

unsigned char decodechar(unsigned char c)
{
if(c > 31)
{
fprintf(stderr, "Error: Got code %d\n", c);
return 0;
}
return set[c];
}

/*
Encode n characters (n <= 8)
from inbuf (up to 8 characters)
to outbuf (5 8-bit bytes)

Bit packing:
where '(' is most significant bit and
')' is least significant bit.
from (---)(---)(---)(---)(---)(---)(---)(---)
to (------)(------)(------)(------)(------)
*/
void encodebuf(unsigned char *outbuf, unsigned char *inbuf, size_t n)
{
memset(outbuf, 0, 5);
for(size_t i = 0; i < n; i++) inbuf = encodechar(inbuf);
if(n >= 1) outbuf[0] |= inbuf[0] << 3;
if(n >= 2) outbuf[0] |= inbuf[1] >> 2, outbuf[1] |= (inbuf[1] & 3) << 6;
if(n >= 3) outbuf[1] |= inbuf[2] << 1;
if(n >= 4) outbuf[1] |= inbuf[3] >> 4, outbuf[2] |= (inbuf[3] & 15) << 4;
if(n >= 5) outbuf[2] |= inbuf[4] >> 1, outbuf[3] |= (inbuf[4] & 1) << 7;
if(n >= 6) outbuf[3] |= inbuf[5] << 2;
if(n >= 7) outbuf[3] |= inbuf[6] >> 3, outbuf[4] |= (inbuf[6] & 7) << 5;
if(n >= 8) outbuf[4] |= inbuf[7];
}

/* Decode 8 characters from inbuf (5 8-bit bytes) to 8-byte outbuf

from (------)(------)(------)(------)(------)
to (---)(---)(---)(---)(---)(---)(---)(---)
*/
void decodebuf(unsigned char *outbuf, unsigned char *inbuf)
{
memset(outbuf, 0, 8);
outbuf[0] = decodechar( inbuf[0] >> 3);
outbuf[1] = decodechar((inbuf[0] & 7) << 2 | inbuf[1] >> 6);
outbuf[2] = decodechar((inbuf[1] >> 1) & 31);
outbuf[3] = decodechar((inbuf[1] & 1) << 4 | inbuf[2] >> 4);
outbuf[4] = decodechar((inbuf[2] & 15) << 1 | inbuf[3] >> 7);
outbuf[5] = decodechar((inbuf[3] >> 2) & 31);
outbuf[6] = decodechar((inbuf[3] & 3) << 3 | inbuf[4] >> 5);
outbuf[7] = decodechar( inbuf[4] & 31);
}

void encode(FILE *fpin, FILE *fpout)
{
unsigned char inbuf[8] = {0};
unsigned char outbuf[5] = {0};
size_t n;
while((n = fread(inbuf, 1, 8, fpin)) != 0)
{
encodebuf(outbuf, inbuf, n);
fwrite(outbuf, 1, 5, fpout);
}
}

void decode(FILE *fpin, FILE *fpout)
{
unsigned char inbuf[5] = {0};
unsigned char outbuf[8] = {0};
size_t n;
while((n = fread(inbuf, 1, 5, fpin)) != 0)
{
decodebuf(outbuf, inbuf);
fwrite(outbuf, 1, 8, fpout);
}
}

int main(int argc, char **argv)
{
if(argc != 4)
{
fprintf(stderr, "Usage: requires three arguments:\n");
fprintf(stderr, " encode or decode\n");
fprintf(stderr, " input filename\n");
fprintf(stderr, " output filename\n");
}
else
{
if(strcmp(argv[1], "encode") == 0)
{
FILE *fpin = fopen(argv[2], "r");
if(fpin == NULL)
{
fprintf(stderr, "Error opening file %s for text reading\n", argv[2]);
exit(EXIT_FAILURE);
}
FILE *fpout = fopen(argv[3], "wb");
if(fpout == NULL)
{
fprintf(stderr, "Error opening file %s for binary writing\n", argv[3]);
exit(EXIT_FAILURE);
}
encode(fpin, fpout);
fclose(fpin);
fclose(fpout);
}
else if(strcmp(argv[1], "decode") == 0)
{
FILE *fpin = fopen(argv[2], "rb");
if(fpin == NULL)
{
fprintf(stderr, "Error opening file %s for binary reading\n", argv[2]);
exit(EXIT_FAILURE);
}
FILE *fpout = fopen(argv[3], "w");
if(fpout == NULL)
{
fprintf(stderr, "Error opening file %s for text writing\n", argv[3]);
exit(EXIT_FAILURE);
}
decode(fpin, fpout);
fclose(fpin);
fclose(fpout);
}
else
{
fprintf(stderr, "Incorrect argument.\n");
}
}
return 0;
}

 

[osmium]

So are you saying create an array of 5 chars like

unsigned char glob[5];

then read in 8 characters at the time , shift manipulate them into the glob
and then
write out the glob array? And if I run out of the input while getting 8
chars at the time
then just padd the rest of the glob array with zeroes....

Is that what you are suggesting?

Yes, something like that is what I had in mind.