"Parsing" a PPM header...

[Robert Harp]

Need help here...

Writing a program that reads in a PPM image, modifies it, and spits it back out.

I can make it work for a regular PPM header, like this one:

P6
690 461 255

but there are some that are like this:

P6
# CREATOR: GIMP PNM Filter Version 1.1
600 399
255

This is a school assignment and people in my class frequent these forums, so I can't post my complete code. The file needs to account for ALL types of PPM headers, not just the first example.

As I said, I can't copy my entire code here, but here's the snippet that reads in PPM header data..

        if (fscanf(inFile, "P6 %d %d", &width, &height) != 2)
        {
            fprintf(stderr, "Invalid image size.\n");
            exit(1);
        }

any ideas?

[jimblumberg]

I'd recommend you retrieve the complete lines with fgets() then you can test the strings to see if:

1. The first line contains "P6".
2. If the line starts with a '#' that indicates a comment line that should be skipped.
3. If the line starts with numeric data you can use sscanf() to process the line and if you don't get all three parameters read the next line, then process it for the rest of the parameters.


Jim

[Barney McGrew]

Start by reading the specification for the file format: PPM Format Specification

As for your code, I suggest beginning with something like this:

#include <stdio.h>

int main(void)
{
    char buf[8192];

    while (scanf("%8191s", buf) == 1)        // read next token
        if (buf[0] != '#')
            printf("\"%s\"\n", buf);
        else if (scanf("%*[^\n]") == EOF)    // eat the line
            break;
    return 0;
}

That will go through stdin, discarding comments, and reading and emitting tokens. For further information you should look up the documentation for the *scanf family as well as some other functions you might find useful, like strcmp and strtol.

EDIT: Depending on how thoroughly you want to support the format, you may need to search for a '#' in every token. According to the specification:

Before the whitespace character that delimits the raster, any characters from a "#" through the next carriage return or newline character, is a comment and is ignored. Note that this is rather unconventional, because a comment can actually be in the middle of what you might consider a token. Note also that this means if you have a comment right before the raster, the newline at the end of the comment is not sufficient to delimit the raster.