I'm not sure it's going to be possible to do anything in PHP.

There is no error reported from the C api to ImageMagick with the test program below.

If there's no error reported, there's no steps we can take.

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <malloc.h>
#include <wand/MagickWand.h>
    

int main(int argc,char **argv) {
    MagickWand *magick_wand;
    MagickBooleanType status;
    char *exception_text;
    ExceptionType severity;

    MagickWandGenesis();
    magick_wand = NewMagickWand();
    status = MagickReadImage(magick_wand, "./images/pages_invalid_gif.gif");

    if (status == MagickFalse) {
        printf("Failed to MagickReadImage");
        return -1;
    }
    exception_text = MagickGetException(magick_wand, &severity);

    if (exception_text == NULL) {
        printf("exception_text is NULL\n");
    }
    else {
        printf("exception_text is %s\n", exception_text);
    }
    
    printf("Severity is %d\n", severity);

    MagickSetImageFormat(magick_wand, "png");

    status = MagickWriteImages(magick_wand, "./73263.png", MagickTrue);
    if (status == MagickFalse) {
        printf("Failed to MagickWriteImages");
        return -1;
    }

    MagickWandTerminus();
 
    return(0);
}

With ImageMagick 6, I don't get any exception:

exception_text is 
Severity is 0

With ImageMagick 7 (currently 7.0.3-7) on the other hand, I get the exception below, but no errors from the PHP program I've posted above:

exception_text is InvalidColormapIndex `./pages.gif' @ error/colormap-private.h/ConstrainColormapIndex/34
Severity is 425

Ah, awesome. I didn't try it with IM7 - but if it is reporting that something is wrong, then we can do something about it.

I'll probably need to do this at at least a minor version, as it will probably 'break' 'working' code for some people, who are unaware that they are opening 'invalid' files.

Additionally, there is probably an argument for allowing the calling code to be able to continue past certain errors, which needs thinking about.

I've made a note about it here - https://github.com/mkoppanen/imagick/issues/164

In this specific case, breaking working code may not be desirable, because imagick does load a somewhat acceptable representation of the file. The default behaviour should be to flag the error, but not to abort processing of the corrupt image.