tigitz
feature: initial PSR-6 cache implementation
Description
This pull request introduces a caching system based on PSR-6 contract to improve the performance.
[!NOTE]
Initial implementation introduces file system caching adapter only
Motivation and context
After some tests and profiling the performance of this package there's a bottleneck in my application when repeatedly calling aspell->check(). Although this cache feature does not address first-run, the completion time of subsequent calls are significantly improved (~16 seconds to ~1.5 seconds). The integration approach was to cache the output of the Aspell process.
How has this been tested?
make tests-dox
make phpstan
make phpcs
Cache implementation
<?php
use PhpSpellcheck\Cache\FileCache;
// Method 1: Using constructor
$cache = new FileCache(
namespace: 'myapp', // Cache namespace
defaultLifetime: 3600, // 1 hour default TTL
directory: null // Optional custom directory
);
// Method 2: Using static creator
$cache = FileCache::create('myapp', 3600);
// Core Methods
// Store item
$item = $cache->getItem('cache-key');
$item->set($value);
$cache->save($item);
// Retrieve item
$item = $cache->getItem('cache-key');
if ($item->isHit()) {
$value = $item->get();
}
// Delete item
$cache->deleteItem('cache-key');
// Clear all items
$cache->clear();
// Deferred storage
$cache->saveDeferred($item);
$cache->commit();
Cache spellchecker usage
<?php
use PhpSpellcheck\Cache\FileCache;
use PhpSpellcheck\Spellchecker\Aspell;
use PhpSpellcheck\Spellchecker\CacheableSpellchecker;
// Create cache instance
$cache = FileCache::create(
namespace: 'Aspell',
defaultLifetime: 3600
);
// Create cached spellchecker
$spellchecker = new CacheableSpellchecker(
cache: $cache,
spellchecker: Aspell::create()
);
// Check spelling - first time hits spellchecker
$misspellings = $spellchecker->check('Hello wurld');
// Second check - returns cached results
$misspellings = $spellchecker->check('Hello wurld');
Checklist:
Go over all the following points before making your PR:
- [x] I have read the CONTRIBUTING document.
- [x] My pull request addresses exactly one patch/feature.
- [x] I have created a branch for this patch/feature.
- [x] I have added tests to cover my changes.
- [x] If my change requires a change to the documentation, I have updated it accordingly.
If you're unsure about any of these, don't hesitate to ask. We're here to help!
Hi @tigitz I've put together an initial approach to caching Aspell output. Let me know your thoughts and if this is the right direction to go here in terms of caching. Thanks!
As an aside phpcs seems to be behaving differently local vs CI. With native_function_invocation failures on CI - no idea why this is, I'm assuming it's using the same config file.
Hey @chr15k , I've updated the master branch to be up to date with latest php version and deps, it introduced some changes, could you please rebase before going further on this please.
Hey @chr15k , I've updated the master branch to be up to date with latest php version and deps, it introduced some changes, could you please rebase before going further on this please.
@tigitz Done - thanks for your feedback, I'll switch out of draft status when I've finished the updated PSR implementation.
Thank you for the initial implementation.
After some thoughts, I suggest using
CacheableSpellchecker(SpellcheckerInterface, CacheInterface)to flexibly decorate any compliant Spellchecker. Additionally, using PSR'sCacheInterfacewould better support broader standards than Symfony's contracts.I'll fix the CI issues on master meanwhile, don't worry about it
@tigitz I was wondering if you might have any suggestions for handling generators when caching. In this case, I’ve implemented SpellcheckerInterface for CacheableSpellchecker, which requires the check method to return iterable<PhpSpellcheck\MisspellingInterface>.
Since we're working with a generator, I’ve been caching the results by converting it to an array with iterator_to_array($misspellings). This works fine, but when fetching from the cache, I need to convert it back to the appropriate format to satisfy the contract.
I was originally considering caching the raw output from Aspell directly, as that seems more efficient. However, I was trying to avoid modifying the underlying spellchecker implementation.
Do you have any thoughts or suggestions for a cleaner or more efficient approach?
@chr15k Good point, my first idea would be this:
<?php
namespace PhpSpellcheck\Spellchecker;
use Psr\Cache\CacheItemPoolInterface;
class CacheableSpellchecker implements SpellcheckerInterface
{
public function __construct(
private readonly CacheItemPoolInterface $cache,
private readonly SpellcheckerInterface $spellchecker
) {
}
public function check(
string $text,
array $languages = [],
array $context = []
): iterable {
$cacheKey = md5(serialize([$this->spellchecker, $text, $languages, $context]));
$cacheItem = $this->cache->getItem($cacheKey);
if ($cacheItem->isHit()) {
yield from $cacheItem->get();
return;
}
$misspellings = iterator_to_array($this->spellchecker->check($text, $languages, $context));
$this->cache->save($cacheItem->set($misspellings));
yield from $misspellings;
}
public function getSupportedLanguages(): iterable
{
$cacheKey = md5(serialize([$this->spellchecker]));
$cacheItem = $this->cache->getItem($cacheKey);
if ($cacheItem->isHit()) {
yield from $cacheItem->get();
return;
}
$languages = iterator_to_array($this->spellchecker->getSupportedLanguages());
$this->cache->save($cacheItem->set($languages));
yield from $languages;
}
}
It's using "psr/cache": "^3.0" and we could implement a FileCache like this:
<?php
namespace PhpSpellcheck\Cache;
use Psr\Cache\CacheItemInterface;
use Psr\Cache\CacheItemPoolInterface;
use RuntimeException;
class FileCache implements CacheItemPoolInterface
{
private string $cacheFile;
private array $cache = [];
private array $deferred = [];
public function __construct(?string $cacheFile = null)
{
$this->cacheFile = $cacheFile ?? $this->getDefaultCacheFile();
$this->loadCache();
}
public function getItem(string $key): CacheItemInterface
{
return new FileCacheItem($key, $this->cache[$key] ?? null, isset($this->cache[$key]));
}
public function getItems(array $keys = []): iterable
{
return array_map(fn($key) => $this->getItem($key), $keys);
}
public function hasItem(string $key): bool
{
return isset($this->cache[$key]);
}
public function clear(): bool
{
$this->cache = [];
$this->deferred = [];
return $this->saveCache();
}
public function deleteItem(string $key): bool
{
unset($this->cache[$key]);
return $this->saveCache();
}
public function deleteItems(array $keys): bool
{
foreach ($keys as $key) {
unset($this->cache[$key]);
}
return $this->saveCache();
}
public function save(CacheItemInterface $item): bool
{
$this->cache[$item->getKey()] = $item->get();
return $this->saveCache();
}
public function saveDeferred(CacheItemInterface $item): bool
{
$this->deferred[$item->getKey()] = $item->get();
return true;
}
public function commit(): bool
{
foreach ($this->deferred as $key => $value) {
$this->cache[$key] = $value;
}
$this->deferred = [];
return $this->saveCache();
}
private function loadCache(): void
{
if (!file_exists($this->cacheFile)) {
$this->cache = [];
return;
}
$data = include $this->cacheFile;
if (!is_array($data)) {
throw new RuntimeException('Invalid cache file format');
}
$this->cache = $data;
}
private function saveCache(): bool
{
$dir = dirname($this->cacheFile);
if (!is_dir($dir) && !mkdir($dir, 0777, true) && !is_dir($dir)) {
throw new RuntimeException(sprintf('Directory "%s" could not be created', $dir));
}
$tmpFile = $this->cacheFile . '.tmp';
$content = '<?php return ' . var_export($this->cache, true) . ';';
if (file_put_contents($tmpFile, $content) === false) {
return false;
}
if (!rename($tmpFile, $this->cacheFile)) {
unlink($tmpFile);
return false;
}
return true;
}
private function getDefaultCacheFile(): string
{
return '/.php-spellchecker.cache';
}
}