Reputation: 131
I have a problem with reading pixels from bmp file. It might be something with padding at the end of row or the base64 padding. I have no clue. I've been struggling with this for some days and can't move on because the next task requires this one to be solved. I only share important parts of the code, since reading bmp header worked fine (tests had 0 failures).
bmp.c
struct pixel* read_data(FILE* stream, const struct bmp_header* header){
if(stream == NULL || header == NULL){
return 0;
}
// w == 1 && p == 1; w == 2 && p == 2; w == 3 && p == 3; w == 4 && p == 0
int padding = header->width % 4;
int num_of_pixels = header->width * header->height;
struct pixel* Pixel[num_of_pixels];
fseek(stream, 54, SEEK_SET); //move 54B (header size)
int index_p = 0;
for(int i = 0; i < header->height; i++){
for(int j = 0; j < header->width; j++){
Pixel[index_p] = malloc(sizeof(struct pixel));
fread(&(Pixel[index_p]->blue), 1, 1, stream);
fread(&(Pixel[index_p]->green), 1, 1, stream);
fread(&(Pixel[index_p]->red), 1, 1, stream);
index_p++;
}
fseek(stream, padding, SEEK_CUR); //padding at the end of row
}
return *Pixel;
}
bmp.h
struct pixel {
uint8_t blue;
uint8_t green;
uint8_t red;
//uint8_t alpha;
} __attribute__((__packed__));
/**
* Read the pixels
*
* Reads the data (pixels) from stream representing the image. If the stream
* is not open or header is not provided, returns `NULL`.
*
* @param stream opened stream, where the image data are located
* @param header the BMP header structure
* @return the pixels of the image or `NULL` if stream or header are broken
*/
struct pixel* read_data(FILE* stream, const struct bmp_header* header);
header if needed (basically we use only 24bit color)
struct bmp_header{
uint16_t type; // "BM" (0x42, 0x4D)
uint32_t size; // file size
uint16_t reserved1; // not used (0)
uint16_t reserved2; // not used (0)
uint32_t offset; // offset to image data (54B)
uint32_t dib_size; // DIB header size (40B)
uint32_t width; // width in pixels
uint32_t height; // height in pixels
uint16_t planes; // 1
uint16_t bpp; // bits per pixel (24)
uint32_t compression; // compression type (0/1/2) 0
uint32_t image_size; // size of picture in bytes, 0
uint32_t x_ppm; // X Pixels per meter (0)
uint32_t y_ppm; // X Pixels per meter (0)
uint32_t num_colors; // number of colors (0)
uint32_t important_colors; // important colors (0)
} __attribute__((__packed__));
main.c I do not need to assign any variables to called functions because we have a program for testing this, I just have to call them in main
int main(){
struct bmp_header* header;
FILE *stream = fopen("./assets/square.2x3.bmp", "rb");
header = read_bmp_header(stream);
read_data(stream, header);
read_bmp(stream);
struct bmp_image* image;
image = malloc(sizeof(struct bmp_image));
free_bmp_image(image);
fclose(stream);
return 0;
}
testing (there are more tests, but this should be enough)
1:
FILE* stream = "Qk0+AAAAAAAAADYAAAAoAAAAAgAAAAEAAAABABgAAAAAAAgAAAAjLgAAIy4AAAAAAAAAAAAA/wAAAP8AAAA="; // base64 encoded stream
struct bmp_header* header = read_bmp_header(stream);
fseek(stream, offset, SEEK_SET);
Assertion 'read_data(stream, header) == "/wAAAP8A"' failed. [got "/wAAFctV"]
2:
FILE* stream = "Qk1GAAAAAAAAADYAAAAoAAAAAgAAAAIAAAABABgAAAAAABAAAAAjLgAAIy4AAAAAAAAAAAAA/wAAAAAAAAAAAP8A/wAAAA=="; // base64 encoded stream
struct bmp_header* header = read_bmp_header(stream);
fseek(stream, offset, SEEK_SET);
Assertion 'read_data(stream, header) == "/wAAAAAAAAD/AP8A"' failed. [got "/wAAAAAAAAAAAAAA"]
So after the "==" is expected result and in the brackets is my the result from my code. As I mentioned, it might be something with padding, since it starts well but doesn't end well. Thanks for help.
Upvotes: 0
Views: 6070
Reputation: 941
Short Answer: Set padding to (4-((3*width)%4))%4
Long answer:
Your code included:
int padding = header->width % 4;
//Some lines of code
fseek(stream, padding, SEEK_CUR);
In a bitmap, padding is added until each row is a multiple of 4 bytes. You took padding as width % 4
.
First off, each pixel takes 3 bytes(for rgb). So it should be (3*width)%4
. Next, we need to subtract it from 4 bytes (Since padding is 4-pixels occupied
). So padding would be 4-((3*width)%4)
. Another small modification, if (3*width)%4==0
then padding would come to be 4 (whereas, we expect it to be 0). So we take another mod4 just to be sure
So padding would come out to be (4-((3*width)%4))%4
EDIT:
As pointed out by user Craig Estey in the comments, its better to use sizeof(struct pixel) instead of 3
Upvotes: 2