Optimize collected data size (#168)

Author: janedbal

src/Collector/BufferedUsageCollector.php

@@ -32,7 +32,7 @@ private function tryFlushBuffer(
             return $data === []
                 ? null
                 : array_map(
-                    static fn (CollectedUsage $usage): string => $usage->serialize(),
+                    static fn (CollectedUsage $usage): string => $usage->serialize($scope->getFile()),
                     $data,
                 );
         }

src/Graph/CollectedUsage.php

@@ -52,7 +52,14 @@ public function concretizeMixedUsage(string $className): self
         );
     }
 
-    public function serialize(): string
+    /**
+     * Scope file is passed to optimize transferred data size (and thus result cache size)
+     * - PHPStan itself transfers all collector data along with scope file
+     * - thus if our data match those already-transferred ones, lets omit those
+     *
+     * @see https://github.com/phpstan/phpstan-src/blob/2fe4e0f94e75fe8844a21fdb81799f01f0591dfe/src/Analyser/FileAnalyser.php#L198
+     */
+    public function serialize(string $scopeFile): string
     {
         $origin = $this->usage->getOrigin();
         $memberRef = $this->usage->getMemberRef();
@@ -63,7 +70,7 @@ public function serialize(): string
             'o' => [
                     'c' => $origin->getClassName(),
                     'm' => $origin->getMethodName(),
-                    'f' => $origin->getFile(),
+                    'f' => $origin->getFile() === $scopeFile ? '_' : $origin->getFile(),
                     'l' => $origin->getLine(),
                     'p' => $origin->getProvider(),
                     'n' => $origin->getNote(),
@@ -82,7 +89,7 @@ public function serialize(): string
         }
     }
 
-    public static function deserialize(string $data): self
+    public static function deserialize(string $data, string $scopeFile): self
     {
         try {
             /** @var array{e: string|null, t: MemberType::*, o: array{c: string|null, m: string|null, f: string|null, l: int|null, p: string|null, n: string|null}, m: array{c: string|null, m: string, d: bool}} $result */
@@ -92,7 +99,14 @@ public static function deserialize(string $data): self
         }
 
         $memberType = $result['t'];
-        $origin = new UsageOrigin($result['o']['c'], $result['o']['m'], $result['o']['f'], $result['o']['l'], $result['o']['p'], $result['o']['n']);
+        $origin = new UsageOrigin(
+            $result['o']['c'],
+            $result['o']['m'],
+            $result['o']['f'] === '_' ? $scopeFile : $result['o']['f'],
+            $result['o']['l'],
+            $result['o']['p'],
+            $result['o']['n'],
+        );
         $exclusionReason = $result['e'];
 
         $usage = $memberType === MemberType::CONSTANT

src/Rule/DeadCodeRule.php

@@ -158,10 +158,10 @@ public function processNode(
         $memberUseData = array_merge_recursive($methodCallData, $providedUsagesData, $constFetchData);
         unset($methodCallData, $providedUsagesData, $constFetchData);
 
-        foreach ($memberUseData as $usesPerFile) {
+        foreach ($memberUseData as $file => $usesPerFile) {
             foreach ($usesPerFile as $useStrings) {
                 foreach ($useStrings as $useString) {
-                    $collectedUsage = CollectedUsage::deserialize($useString);
+                    $collectedUsage = CollectedUsage::deserialize($useString, $file);
                     $memberUsage = $collectedUsage->getUsage();
 
                     if ($memberUsage->getMemberRef()->getClassName() === null) {

tests/Graph/SerializationTest.php

@@ -10,18 +10,31 @@ class SerializationTest extends TestCase
     /**
      * @dataProvider provideData
      */
-    public function testSerialization(CollectedUsage $expected, string $serialized): void
+    public function testSerialization(string $filePath, CollectedUsage $expected, string $serialized): void
     {
-        self::assertSame($serialized, $expected->serialize());
-        self::assertEquals($expected, CollectedUsage::deserialize($serialized));
+        self::assertSame($serialized, $expected->serialize($filePath));
+        self::assertEquals($expected, CollectedUsage::deserialize($serialized, $filePath));
     }
 
     /**
-     * @return iterable<array{CollectedUsage, string}>
+     * @return iterable<string, array{string, CollectedUsage, string}>
      */
     public static function provideData(): iterable
     {
-        yield [
+        yield 'path optimized' => [
+            '/app/index.php',
+            new CollectedUsage(
+                new ClassConstantUsage(
+                    new UsageOrigin('Clazz', 'method', '/app/index.php', 7, null, null),
+                    new ClassConstantRef(null, 'CONSTANT', true),
+                ),
+                'excluder',
+            ),
+            '{"e":"excluder","t":2,"o":{"c":"Clazz","m":"method","f":"_","l":7,"p":null,"n":null},"m":{"c":null,"m":"CONSTANT","d":true}}',
+        ];
+
+        yield 'path differs' => [
+            '/app/different.php',
             new CollectedUsage(
                 new ClassConstantUsage(
                     new UsageOrigin('Clazz', 'method', '/app/index.php', 7, null, null),