The script is present and the link does work!
It just took a little time for the web site to recieve the script which I build from the perl I previously developed on the other discussion. Basically put the perl in a file, map command line arguments to the called convert command.
It could do with a bit more work as it currently buffers one image in memory before calling the internal "convert" part of the pipeline. It could for example start the convert command as soon as it sees the start of the image, then close it when it sees the start of the next image without buffering the whole image, just one line (row) of ascii PPM text.
Of course with some extra smarts it could actually read the PPM header and then read and transfer the exact amount of RAW (binary) PPM data. In that case it will know exactly when the image ends and the next (should begin).
The IDEAL method would be to use mogrify with some smarts to only read ONE PbmPlus image in from a pipeline at a time, without closing the input pipeline, so when ready it can read the next one.
However knowing the source of mogrify, this may NOT fit with the implementation as it currently stands. That is: read first image argument found, apply all command line arguments that is not to do with reading an image, write result, then loop to read next image found on command line. It would need to be modified to handle multiple 'single' images from one image read argument. Coders would also need to modify to return one image at a time too.
My solution was a wrapper work-around, and will actually allow full "convert" style arguments.
In the future I hope to create a morgify that can handle "convert" type arguments, though that will need some sort of separation on 'arguments to loop on, including secondary reads', and the 'images to read one at a time'.
Sorry this is an area I have been looking at with a view to IM v7 improvements. EG: read image operations from files, pipelines and sockets, with image processing operations being REALLY applied as seen.