程序读取文件，但在不应跳过的行时“跳过”行

Question

So basically I have a program that reads in a specific format of a file as shown:

P3
# CREATOR: GIMP PNM Filter Version 1.1
400 530
255
189
165
181
181
156
...

My program reads in the first line as the format of the file, then any comments starting with a '#', then the width and height, then the max rgb value and then any other rgb values.

NOTE: the part reading in the rgb values is incomplete so ignore this for my question. Now to my issue..

The output i want is as shown as the exact file format, but the result i get is as follows:

P3
# CREATOR: GIMP PNM Filter Version 1.1
255 189
165

As you can see, it completely skips the width and height values and uses the next 2 values in place of it for some reason...

Here is my code:

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

#define COMMENT_LENGTH 256
#define COMMENT_ARRAY_SIZE 10
#define MAX_PIXEL_HEIGHT 480
#define MAX_PIXEL_WIDTH 640

typedef struct PPM {
    char format[2]; //2 letter code for PPM format
    char comments[COMMENT_LENGTH][COMMENT_ARRAY_SIZE]; //comment array
    int width; //number of columns
    int height; //number of rows
    int max; //maximum colour value (usually 255)
    int rgbPixels[MAX_PIXEL_WIDTH * MAX_PIXEL_HEIGHT][3]; //integers between 0 and max for pixel i's RGB values
}PPM;

struct PPM * getPPM(FILE *fd);

int main(int argc, char **argv) {
    FILE *file = fopen("ape.ppm", "r");
    if(file == NULL) return 0;
    else {
        struct PPM *newPPM = getPPM(file);

        return 0;
    }
}

struct PPM * getPPM(FILE *fd) {
    if(fd == NULL) {
        return NULL;
    }

    struct PPM *newPPMFile = (PPM *) malloc(sizeof(PPM));

    fscanf(fd, "%2s\n", newPPMFile->format);
    printf("%2s\n", newPPMFile->format);

    //check for comments
    int i = 0;
    char str[COMMENT_LENGTH];
    while(i < COMMENT_ARRAY_SIZE && fgets(str, COMMENT_LENGTH, fd) != NULL) {
        if(str[0] != '#') {
            break;
        } else {
            strcpy(newPPMFile->comments[i], str);
            printf("%s", newPPMFile->comments[i]);
        }
    }

    //read width and height
    fscanf(fd, "%d %d", &newPPMFile->width, &newPPMFile->height);
    printf("%d %d\n", newPPMFile->width, newPPMFile->height);

    //read max
    fscanf(fd, "%d", &newPPMFile->max);
    printf("%d\n", newPPMFile->max);

    //read rgb data in rgb array (INCOMPLETE)


    //close file
    fclose(fd);

    return newPPMFile;
};

Answer 1

There are a couple things here:

while(i < COMMENT_ARRAY_SIZE && fgets(str, COMMENT_LENGTH, fd) != NULL) {
    if(str[0] != '#') {
        break;

The position for the next read (offset) has already advanced, if you break or not. So naturally, it'll continue at the advanced offset, like you experienced here.

Also,

(PPM *) malloc(sizeof(PPM));

Casting the result of malloc() is unnecessary and potentially masks errors.

Answer 2

fgets reads the full line into the buffer. It it doesn't begin with a hash mark, you lose the data on that line.

A solution is to use fscanf throughout. fscanf stops scanning when the pattern isn't matched or if a conversion can't be made. So if you enforce a hash mark before the comment, but the line doesn't begin with a hash mark, the rest of the line is untouched:

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

int main(void)
{
    int width, height, max;
    char buf[80];                             // temporary buffer
    FILE *fd = fopen("kk", "r");

    if(fd == NULL) return 1;                  // TODO: Error message

    fscanf(fd, "%79s", buf);
    if (strcmp(buf, "P3") != 0) return 1;     // TODO: Error message

    while (fscanf(fd, " # %79[^\n]%*[^\n]", buf) == 1) {
        printf("Comment: %s\n", buf);
    }

    fscanf(fd, "%d %d", &width, &height);     // TODO: Check success
    printf("%d x %d\n", width, height);

    fscanf(fd, "%d", &max);                   // TODO: Check success
    printf("max: %d\n", max);

    fclose(fd);    
    return 0;
}

There are some things to keep in mind, though:

• fscanf does not care or know about new lines; new-line characters are just white space. That means that your first hash mark could be on the same line with the file format P3 .
• What is an advantage in reading the comments may be a pitfall when you read the values: The code above doesn't check whether a conversion was made. That's not good, especially that if you read EOF or a string that isn't an integer, fscanf gets stuck.
• The format %79[^\\n] will read only up to 79 characters. If the comment line is longer, the file position will not be on the new-line character, this srewing up the reading of subsequent lines. To circumvent that, you can add %*[^\\n] after that: This will enforce that you read to the next new-line character, but it won't store the results. ( * means "don't store result" in scanf formats.) That way you get up to the first 80 characters of each comment line and also ensure that the line is read completely if it is longer.

Answer 3

1. Insufficient space @errikos .

    // char format[2]; //2 letter code for PPM format char format[2 + 1]; //2 letter code for PPM format ... // 2 is the max width of non-white-space characters to scan fscanf(fd, "%2s\\n", newPPMFile->format); 

2. Check the return values form fscanf()

    // fscanf(fd, "%d %d", &newPPMFile->width, &newPPMFile->height); if (fscanf(fd, "%d %d", &newPPMFile->width, &newPPMFile->height) != 2) { fprintf(stderr, "Error reading width & height\\n"); } 

3. Code mixes fscanf() with fgets() . Tis often is a problem as many formats used to not consume that trailing '\\n`` and then a following fgets() simply reads just that one character. Reccomedn only using simply reads just that one character. Reccomedn only using fgets()`.

4. Possibly other important issues too.

Answer 4

here is the root of the problem.

in the loop reading comments:

when the line is read (when reading comments) that does not start with a #, then the code has already read width+height values from the input file.

So the code needs to parse those values from the already read line.