When using PHP and ImageMagick to resize and name images recursively, how to best write new files based on the existing filename and extension

Question

I've got a directory of 'source' images that I'm saving out multiple resized files. I'd like all jpg and png files in a given directory to be resized and saved with new names based on their original file names (a la filename-small.jpg, filename-medium.jpg, etc.) I'm using ImageMagick.

The regex stuff to get all the files was found in another question here on SO and has helped—but I don't totally get what all is going on here.

What I have currently plops the original file extension in the middle of the file name. While I understand why this is happening, I'm not sure the best approach to fix this.

How can I get this to to be filename-large.jpg instead of filename.jpg-large.jpg?

Here is what I have currently:

<?php

    $directory = new RecursiveDirectoryIterator('source/assets/uploads/test');
    $iterator = new RecursiveIteratorIterator($directory);
    $images = new RegexIterator($iterator, '/^.+(.jpe?g|.png)$/i', RecursiveRegexIterator::GET_MATCH);

    $widths = [
        'small' => 400,
        'medium' => 800,
        'large' => 1000,
        'xlarge' => 1600,
    ];
    
    foreach($images as $image => $value) {

        // Set the original filename w/o the extension
        $originalFilename = $image;

        // set the extension based on the original
        // ideally conform the extension jpeg to jpg
        $originalExtension = '.jpg';
        
        // create a new Imagick instance
        $newImage = new Imagick($image);

        // strip out some junk
        $newImage->stripImage();

        // write a new file for each width 
        // and name it based on the original filename
        foreach ($widths as $key => $value) {
            // using 0 in the second $arg will keep the same image ratio
            $newImage->resizeImage($value,0, imagick::FILTER_LANCZOS, 0.9);
            $newImage->writeImage($originalFilename . '-' . $key . $originalExtension);
        }
    }

Try `/^.+\.(?<!-small.|-xlarge.|-large.|-medium.)(jpe?g|png)$/i` pattern, to exclude all qualifying strings with `$widths` suffixes. — Wiktor Stribiżew, Feb 03 '21 at 17:58
Or, `/^(.+?)(?:-(?:small|x?large|medium))?\.(?:jpe?g|png)$/i` — Wiktor Stribiżew, Feb 03 '21 at 18:17
@WiktorStribiżew I like where this is headed—this might help in the future when new images are added to the repo and I want to update only those. However—I could also see wanting to update 'large' to 1200 though. — Nathan Gross, Feb 03 '21 at 18:51
Perhaps longterm would be better to just name the files based on the image size value rather than the abstracted key. If I need a new size, adding it would be less of a problem. — Nathan Gross, Feb 03 '21 at 19:33

Karl Wilbur · Accepted Answer · 2021-02-03T18:54:28.187

Nathan,

You are capturing that extension with the parens. You want to capture the file name, not just the extension. You can use non-caputring parens ((?:)) for grouping the extension variations.

$images = new RegexIterator($iterator, '/^(.+)\.(?:jpe?g|png)$/i', RecursiveRegexIterator::GET_MATCH);

I also escaped the literal dots (.) before the extensions.

Alternatively, you could just split up the extension and filename from the $image variable:

preg_match('/^(.+)\.([^.]+)$/', $image, $matches);
// Set the original filename w/o the extension
$originalFilename = $matches[1];
// Set the original extension
$originalExtension = $matches[2];

Lastly, FWIW, were I to do this I would not use the RegexIterator but tather just iterate the directory and just use preg_match only on each filename.

$directory = dir('source/assets/uploads/test');
while(false !== ($entry = $directory->read())) {
    $result = preg_match('/^(.+)\.(jpe?g|png)$/i', $entry, $matches);
    if($result === false) {
      die('ERROR: `preg_match` failed on '.$entry);
    } else if($result === 0) {
      // not an image file
      continue;
    }
    // Set the original filename w/o the extension
    $originalFilename = $matches[1];
    // Set the original extension
    $originalExtension = $matches[2];

    // ... do the other things on the file
}

Refs:

EDIT:

For named references inside of $matches you can use this modified regexp:

preg_match('/^(?P<filename>.+)\.(?P<extension>jpe?g|png)$/i', $entry, $matches);

The ?P<key> is a "placeholder" that assigns key as a reference to that caputred match.

You are still going to get $matches back as an array, not an object, but then you can access the values with that placeholder as the index.

// Set the original filename w/o the extension
$originalFilename = $matches['filename'];
// Set the original extension
$originalExtension = $matches['extension'];

FWIW, the numeric indexes are still there too.

This is very helpful—thank you! To make this even more clear for my inexperienced programming brain, is it possible to capture the data in the array in a more key/value type of way? So rather than $matches[2], something like $matches->extension? — Nathan Gross, Feb 03 '21 at 18:48

When using PHP and ImageMagick to resize and name images recursively, how to best write new files based on the existing filename and extension

1 Answers1

EDIT: