Did you want to also provide a map for short names that could work both ways?
Typically mapping 😓 to :sweat:?

@alamirault you review the PR a bit too soon :) It was a WIP (I pushed in order to backup my work).

@nicolas-grekas

Typically mapping 😓 to :sweat: ?

That's already the case, or I missed something ?

    public function testTransliterate()
    {
        $tr = EmojiTransliterator::getInstance('en');

        $this->assertSame(
            'a grinning cat, black cat, and a lion go to national park️... smiling face with heart-eyes party popper yellow heart',
            $tr->transliterate('a 😺, 🐈‍⬛, and a 🦁 go to 🏞️... 😍 🎉 💛')
        );
    }

@stof Thanks for the review, I've addressed your comments.

I think the PR is now ready.

BTW, looking at the maps, I'm wondering: shouldn't eg en be applied also for en_ca (after it likely)? en_ca looks very short compared to en.

I missed that point!

What do you recommend? we can merge en into en_ca?

What do you recommend? we can merge en into en_ca?

that might increase the size of maps, better run the str_replace twice when needed (with the php-based implem)

we can merge en into en_ca

I went this way actually, not much to merge. The PR on your fork is up to date.

@nicolas-grekas Feel free to merge this PR, and then you can open a new PR to improve the situation if it needs to

@lyrixx Let's finish this PR before merging it. I don't see why we would merge something that we know is not finished yet.

Especially when this means bloating the git history :)

So I think the best way It to merge the PR of @nicolas-grekas (in my PR) (I'll do that in few minutes)

So we can discuss about it. But Unfortunately, I don't understand the new code, and I miss some time to dig into it.

I talked a bit with @nicolas-grekas on slack, and we decided to share our thought on the subject here.

Nikos rewrite all the code to be able to rely on Opcache preloading, in order to increase the performance.

To be honest, I don't like this new implementation (that explains my previous^2 comment). So the rest of this comment might be subjective. But I let you judge :)

The previous code was easier to understand, maintain and work with (see exception message for instance) 😄 :

• old: https://github.com/lyrixx/symfony/blob/intl-trans-emoji-back/src/Symfony/Component/Intl/Transliterator/EmojiTransliterator.php
• new: https://github.com/lyrixx/symfony/blob/intl-trans-emoji/src/Symfony/Component/Intl/Transliterator/EmojiTransliterator.php

About the performance, I wrote this Unit Test in both branch

    public function testBench()
    {
        $t = microtime(true);

        for ($i=0; $i < 1000; $i++) {
            $tr = EmojiTransliterator::create('en'); // I also tried with this line outside the loop. Same result ($tr is cached internally) 
            $tr->transliterate('un 😺, 🐈‍⬛, et a 🦁 vont au 🏞️');
        }

        dump((microtime(true) - $t));
    }

Here are the results:

Previous code: 0.0145s
New code: 0.088s

It looks like the previous code is 6 times faster than the new code when using only one transliterator instance. It's legit to me, since the new code does not use the translitator anymore but strtr.

But what about many transliterator? To know that, I used the following command that run for all locales

./phpunit src/Symfony/Component/Intl/Tests/Transliterator/EmojiTransliteratorTest.php --filter testAllTransliterator --repeat 100

old code 00:01.746
new code: 00:02.592

So here the old code is also faster

PHP 8.1.7 (cli) (built: Jun 25 2022 08:13:46) (NTS)
Copyright (c) The PHP Group
Zend Engine v4.1.7, Copyright (c) Zend Technologies
    with Zend OPcache v8.1.7, Copyright (c), by Zend Technologies
    with blackfire v1.77.0~linux-x64-non_zts81, https://blackfire.io, by Blackfire

/etc/php/8.1/cli/conf.d/10-opcache.ini,
opcache.blacklist_filename => no value => no value
opcache.consistency_checks => 0 => 0
opcache.dups_fix => Off => Off
opcache.enable => On => On
opcache.enable_cli => On => On
opcache.enable_file_override => Off => Off
opcache.error_log => no value => no value
opcache.file_cache => no value => no value
opcache.file_cache_consistency_checks => On => On
opcache.file_cache_only => Off => Off
opcache.file_update_protection => 2 => 2
opcache.force_restart_timeout => 180 => 180
opcache.huge_code_pages => Off => Off
opcache.interned_strings_buffer => 8 => 8
opcache.jit => tracing => tracing
opcache.jit_bisect_limit => 0 => 0
opcache.jit_blacklist_root_trace => 16 => 16
opcache.jit_blacklist_side_trace => 8 => 8
opcache.jit_buffer_size => 0 => 0
opcache.jit_debug => 0 => 0
opcache.jit_hot_func => 127 => 127
opcache.jit_hot_loop => 64 => 64
opcache.jit_hot_return => 8 => 8
opcache.jit_hot_side_exit => 8 => 8
opcache.jit_max_exit_counters => 8192 => 8192
opcache.jit_max_loop_unrolls => 8 => 8
opcache.jit_max_polymorphic_calls => 2 => 2
opcache.jit_max_recursive_calls => 2 => 2
opcache.jit_max_recursive_returns => 2 => 2
opcache.jit_max_root_traces => 1024 => 1024
opcache.jit_max_side_traces => 128 => 128
opcache.jit_prof_threshold => 0.005 => 0.005
opcache.lockfile_path => /tmp => /tmp
opcache.log_verbosity_level => 1 => 1
opcache.max_accelerated_files => 15000 => 15000
opcache.max_file_size => 0 => 0
opcache.max_wasted_percentage => 5 => 5
opcache.memory_consumption => 128 => 128
opcache.opt_debug_level => 0 => 0
opcache.optimization_level => 0x7FFEBFFF => 0x7FFEBFFF
opcache.preferred_memory_model => no value => no value
opcache.preload => no value => no value
opcache.preload_user => no value => no value
opcache.protect_memory => Off => Off
opcache.record_warnings => Off => Off
opcache.restrict_api => no value => no value
opcache.revalidate_freq => 2 => 2
opcache.revalidate_path => Off => Off
opcache.save_comments => On => On
opcache.use_cwd => On => On
opcache.validate_permission => Off => Off
opcache.validate_root => Off => Off
opcache.validate_timestamps => On => On

Thanks for challenging my proposal @lyrixx
I figured out that using strtr() was way faster than using str_replace() for the use case.
I also figured out that we can skip creating a slicing transliterator most of the time.
The PHP version is now faster again :)

My bench: ab -n 100 http://localhost:8080/ with the script below.

Results with 100 transliterations:

• before: 75 req/s
• after: 390 req/s

Results with 1000 transliterations:

• before: 40 req/s
• after: 70 req/s

<?php
  
use Symfony\Component\Intl\Transliterator\EmojiTransliterator;

require __DIR__.'/vendor/autoload.php';

$tr = EmojiTransliterator::create('en');
//$tr = EmojiTransliterator::getInstance('en');

for ($i=0; $i < 100; $i++) {
    $tr->transliterate('un 😺, 🐈<200d>⬛, et a 🦁 vont au 🏞️ ');
}

echo $tr->transliterate('un 😺, 🐈<200d>⬛, et a 🦁 vont au 🏞️ ');

how would we link AsciiSlugger + EmojiTransliterator? is it desired?

how would we link AsciiSlugger + EmojiTransliterator? is it desired?

In a follow up PR, and yes, I think it's desired.

Thank you @lyrixx.