Added output of benchmark tests into github step summary (#601)

* Added output of benchmark tests into github step summary * Reworked benchmarks execution * Fixed extractors benchmarks * Fixed failing tests * Replaced iterator_to_array with foreach in benchmarks * Fixed benchmarks summary pages markdown * Workflow summary formatting adjustments
flow-php · Oct 16, 2023 · 314b941 · 314b941
1 parent 681571c
commit 314b941
Show file tree

Hide file tree

Showing 13 changed files with 200 additions and 83 deletions.
diff --git a/.github/workflows/benchmark-tag.yml b/.github/workflows/benchmark-tag.yml
@@ -95,14 +95,12 @@ jobs:
           restore-keys: |
             php-${{ matrix.php-version }}-${{ env.PHPBENCH_TAG }}-phpbench-
 
-      - name: Adding main headline
-        run: |
-            echo '# Flow PHP - Benchmark - ${{ env.PHPBENCH_TAG }}' >> $GITHUB_STEP_SUMMARY
-            echo '```' >> $GITHUB_STEP_SUMMARY
-
       - name: "Benchmark"
-        run: "composer test:benchmark -- --tag=${{ env.PHPBENCH_TAG }} --progress=none >> $GITHUB_STEP_SUMMARY"
-
-      - name: Adding footer
         run: |
-            echo '```' >> $GITHUB_STEP_SUMMARY
+          echo '# Flow PHP - Benchmark - ${{ env.PHPBENCH_TAG }}' >> $GITHUB_STEP_SUMMARY
+          echo ' ' >> $GITHUB_STEP_SUMMARY
+          echo '---' >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+          composer test:benchmark -- --tag=${{ env.PHPBENCH_TAG }} --progress=none >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+
diff --git a/.github/workflows/test-benchmark.yml b/.github/workflows/test-benchmark.yml
@@ -3,6 +3,7 @@ name: Benchmark Suite
 on:
   pull_request:
     paths:
+      - '.github/workflows/**'
       - 'src/adapter/**'
       - 'src/core/**'
       - 'src/lib/**'
@@ -57,9 +58,42 @@ jobs:
         uses: "actions/cache@v3"
         with:
           path: "var/phpbench"
-          key: "php-${{ matrix.php-version }}-phpbench"
+          key: "php-${{ matrix.php-version }}-1.x-phpbench"
           restore-keys: |
-            php-${{ matrix.php-version }}-phpbench-
+            php-${{ matrix.php-version }}-1.x-phpbench-
 
-      - name: "Benchmark"
-        run: "composer test:benchmark -- --ref=original --progress=plain"
+      - name: Adding main headline
+        run: |
+          echo '# Flow PHP - Benchmark' >> $GITHUB_STEP_SUMMARY
+          echo ' ' >> $GITHUB_STEP_SUMMARY
+          echo '---' >> $GITHUB_STEP_SUMMARY
+
+      - name: "Benchmark Extractors"
+        run: |
+          echo ' ' >> $GITHUB_STEP_SUMMARY
+          echo '## Extractors' >> $GITHUB_STEP_SUMMARY
+          echo ' ' >> $GITHUB_STEP_SUMMARY
+          echo '---' >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+          composer test:benchmark -- --ref=1.x --progress=none --group=extractor >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+
+      - name: "Benchmark Transformers"
+        run: |
+          echo ' ' >> $GITHUB_STEP_SUMMARY
+          echo '## Transformers' >> $GITHUB_STEP_SUMMARY
+          echo ' ' >> $GITHUB_STEP_SUMMARY
+          echo '---' >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+          composer test:benchmark -- --ref=1.x --progress=none --group=transformer >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+
+      - name: "Benchmark Entry Factories"
+        run: |
+          echo ' ' >> $GITHUB_STEP_SUMMARY
+          echo '## Entry Factory' >> $GITHUB_STEP_SUMMARY
+          echo ' ' >> $GITHUB_STEP_SUMMARY
+          echo '---' >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+          composer test:benchmark -- --ref=1.x --progress=none --group=entry_factory >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
diff --git a/...apter/etl-adapter-avro/tests/Flow/ETL/Adapter/Avro/Tests/Benchmark/AvroExtractorBench.php b/...apter/etl-adapter-avro/tests/Flow/ETL/Adapter/Avro/Tests/Benchmark/AvroExtractorBench.php
@@ -2,16 +2,21 @@
 
 namespace Flow\ETL\Adapter\Avro\Tests\Benchmark;
 
+use Flow\ETL\Config;
 use Flow\ETL\DSL\Avro;
+use Flow\ETL\FlowContext;
+use PhpBench\Attributes\Groups;
 use PhpBench\Attributes\Iterations;
 use PhpBench\Attributes\Revs;
 
-#[Iterations(5)]
+#[Iterations(3)]
+#[Groups(['extractor'])]
 final class AvroExtractorBench
 {
-    #[Revs(1000)]
-    public function bench_extract() : void
+    #[Revs(5)]
+    public function bench_extract_10k() : void
     {
-        Avro::from(__DIR__ . '/../Fixtures/orders_flow.avro');
+        foreach (Avro::from(__DIR__ . '/../Fixtures/orders_flow.avro')->extract(new FlowContext(Config::default())) as $rows) {
+        }
     }
 }
diff --git a/src/adapter/etl-adapter-csv/tests/Flow/ETL/Adapter/CSV/Tests/Benchmark/CSVExtractorBench.php b/src/adapter/etl-adapter-csv/tests/Flow/ETL/Adapter/CSV/Tests/Benchmark/CSVExtractorBench.php
@@ -2,16 +2,21 @@
 
 namespace Flow\ETL\Adapter\CSV\Tests\Benchmark;
 
+use Flow\ETL\Config;
 use Flow\ETL\DSL\CSV;
+use Flow\ETL\FlowContext;
+use PhpBench\Attributes\Groups;
 use PhpBench\Attributes\Iterations;
 use PhpBench\Attributes\Revs;
 
-#[Iterations(5)]
+#[Iterations(3)]
+#[Groups(['extractor'])]
 final class CSVExtractorBench
 {
-    #[Revs(1000)]
-    public function bench_extract() : void
+    #[Revs(5)]
+    public function bench_extract_10k() : void
     {
-        CSV::from(__DIR__ . '/../Fixtures/orders_flow.csv');
+        foreach (CSV::from(__DIR__ . '/../Fixtures/orders_flow.csv')->extract(new FlowContext(Config::default())) as $rows) {
+        }
     }
 }
diff --git a/...apter/etl-adapter-json/tests/Flow/ETL/Adapter/JSON/Tests/Benchmark/JsonExtractorBench.php b/...apter/etl-adapter-json/tests/Flow/ETL/Adapter/JSON/Tests/Benchmark/JsonExtractorBench.php
@@ -2,16 +2,21 @@
 
 namespace Flow\ETL\Adapter\JSON\Tests\Benchmark;
 
+use Flow\ETL\Config;
 use Flow\ETL\DSL\Json;
+use Flow\ETL\FlowContext;
+use PhpBench\Attributes\Groups;
 use PhpBench\Attributes\Iterations;
 use PhpBench\Attributes\Revs;
 
-#[Iterations(5)]
+#[Iterations(3)]
+#[Groups(['extractor'])]
 final class JsonExtractorBench
 {
-    #[Revs(1000)]
-    public function bench_extract() : void
+    #[Revs(5)]
+    public function bench_extract_10k() : void
     {
-        Json::from(__DIR__ . '/../Fixtures/orders_flow.json');
+        foreach (Json::from(__DIR__ . '/../Fixtures/orders_flow.json')->extract(new FlowContext(Config::default())) as $rows) {
+        }
     }
 }
diff --git a/...-adapter-parquet/tests/Flow/ETL/Adapter/Parquet/Tests/Benchmark/ParquetExtractorBench.php b/...-adapter-parquet/tests/Flow/ETL/Adapter/Parquet/Tests/Benchmark/ParquetExtractorBench.php
@@ -2,16 +2,21 @@
 
 namespace Flow\ETL\Adapter\Parquet\Tests\Benchmark;
 
+use Flow\ETL\Config;
 use Flow\ETL\DSL\Parquet;
+use Flow\ETL\FlowContext;
+use PhpBench\Attributes\Groups;
 use PhpBench\Attributes\Iterations;
 use PhpBench\Attributes\Revs;
 
-#[Iterations(5)]
+#[Iterations(3)]
+#[Groups(['extractor'])]
 final class ParquetExtractorBench
 {
-    #[Revs(1000)]
-    public function bench_extract() : void
+    #[Revs(5)]
+    public function bench_extract_10k() : void
     {
-        Parquet::from(__DIR__ . '/../Fixtures/orders_flow.parquet');
+        foreach (Parquet::from(__DIR__ . '/../Fixtures/orders_flow.parquet')->extract(new FlowContext(Config::default())) as $rows) {
+        }
     }
 }
diff --git a/...apter/etl-adapter-text/tests/Flow/ETL/Adapter/Text/Tests/Benchmark/TextExtractorBench.php b/...apter/etl-adapter-text/tests/Flow/ETL/Adapter/Text/Tests/Benchmark/TextExtractorBench.php
@@ -2,16 +2,21 @@
 
 namespace Flow\ETL\Adapter\Text\Tests\Benchmark;
 
+use Flow\ETL\Config;
 use Flow\ETL\DSL\Text;
+use Flow\ETL\FlowContext;
+use PhpBench\Attributes\Groups;
 use PhpBench\Attributes\Iterations;
 use PhpBench\Attributes\Revs;
 
-#[Iterations(5)]
+#[Iterations(3)]
+#[Groups(['extractor'])]
 final class TextExtractorBench
 {
-    #[Revs(1000)]
+    #[Revs(5)]
     public function bench_extract() : void
     {
-        Text::from(__DIR__ . '/../Fixtures/annual-enterprise-survey-2019-financial-year-provisional-csv.csv');
+        foreach (Text::from(__DIR__ . '/../Fixtures/annual-enterprise-survey-2019-financial-year-provisional-csv.csv')->extract(new FlowContext(Config::default())) as $rows) {
+        }
     }
 }
diff --git a/src/adapter/etl-adapter-xml/tests/Flow/ETL/Adapter/XML/Tests/Benchmark/XmlExtractorBench.php b/src/adapter/etl-adapter-xml/tests/Flow/ETL/Adapter/XML/Tests/Benchmark/XmlExtractorBench.php
@@ -2,16 +2,21 @@
 
 namespace Flow\ETL\Adapter\XML\Tests\Benchmark;
 
+use Flow\ETL\Config;
 use Flow\ETL\DSL\XML;
+use Flow\ETL\FlowContext;
+use PhpBench\Attributes\Groups;
 use PhpBench\Attributes\Iterations;
 use PhpBench\Attributes\Revs;
 
-#[Iterations(5)]
+#[Iterations(3)]
+#[Groups(['extractor'])]
 final class XmlExtractorBench
 {
-    #[Revs(1000)]
+    #[Revs(5)]
     public function bench_extract() : void
     {
-        XML::from(__DIR__ . '/../Fixtures/simple_items.xml');
+        foreach (XML::from(__DIR__ . '/../Fixtures/simple_items.xml')->extract(new FlowContext(Config::default())) as $rows) {
+        }
     }
 }
diff --git a/src/core/etl/src/Flow/ETL/Rows.php b/src/core/etl/src/Flow/ETL/Rows.php
@@ -4,6 +4,7 @@
 
 namespace Flow\ETL;
 
+use function Flow\ETL\DSL\array_to_rows;
 use Flow\ETL\DSL\Entry;
 use Flow\ETL\Exception\InvalidArgumentException;
 use Flow\ETL\Exception\RuntimeException;
@@ -40,28 +41,7 @@ public function __construct(Row ...$rows)
 
     public static function fromArray(array $data, EntryFactory $entryFactory = new NativeEntryFactory()) : self
     {
-        /** @var array<Row> $rows */
-        $rows = [];
-
-        foreach ($data as $entries) {
-            if (!\is_array($entries)) {
-                throw new InvalidArgumentException('Rows expects nested array data structure: array<array<mixed>>');
-            }
-
-            $row = [];
-
-            /**
-             * @var string $entryName
-             * @var mixed $entryValue
-             */
-            foreach ($entries as $entryName => $entryValue) {
-                $row[] = $entryFactory->create($entryName, $entryValue);
-            }
-
-            $rows[] = Row::create(...$row);
-        }
-
-        return new self(...$rows);
+        return array_to_rows($data, $entryFactory);
     }
 
     public function __serialize() : array

diff --git a/src/core/etl/tests/Flow/ETL/Tests/Benchmark/EntryFactory/NativeEntryFactoryBench.php b/src/core/etl/tests/Flow/ETL/Tests/Benchmark/EntryFactory/NativeEntryFactoryBench.php
@@ -0,0 +1,76 @@
+<?php declare(strict_types=1);
+
+namespace Flow\ETL\Tests\Benchmark\EntryFactory;
+
+use function Flow\ETL\DSL\array_to_rows;
+use Faker\Factory;
+use Flow\ETL\Row\Factory\NativeEntryFactory;
+use PhpBench\Attributes\BeforeMethods;
+use PhpBench\Attributes\Groups;
+use PhpBench\Attributes\Iterations;
+use PhpBench\Attributes\Revs;
+
+#[Iterations(5)]
+#[Groups(['entry_factory'])]
+final class NativeEntryFactoryBench
+{
+    private array $rowsArray = [];
+
+    public function setUp() : void
+    {
+        $faker = Factory::create();
+
+        $this->rowsArray = \array_map(
+            static fn (int $i) : array => [
+                'order_id' => $faker->uuid,
+                'created_at' => $faker->dateTimeThisYear->format(\DateTimeInterface::RFC3339),
+                'updated_at' => $faker->dateTimeThisMonth->format(\DateTimeInterface::RFC3339),
+                'cancelled_at' => ($i % 10) === 0 ? $faker->dateTimeThisMonth->format(\DateTimeInterface::RFC3339) : null,
+                'active' => !(($i % 20) === 0),
+                'total_price' => $faker->randomFloat(2, 0, 500),
+                'discount' => $faker->randomFloat(2, 0, 50),
+                'customer' => [
+                    'name' => $faker->firstName,
+                    'last_name' => $faker->lastName,
+                    'email' => $faker->email,
+                ],
+                'address' => [
+                    'street' => $faker->streetAddress,
+                    'city' => $faker->city,
+                    'zip' => $faker->postcode,
+                    'country' => $faker->country,
+                    'location' => [
+                        'lat' => $faker->latitude,
+                        'lng' => $faker->longitude,
+                    ],
+                ],
+                'notes' => \array_map(
+                    static fn ($i) => $faker->sentence,
+                    \range(1, $faker->numberBetween(1, 5))
+                ),
+            ],
+            \range(1, 10_000)
+        );
+    }
+
+    #[BeforeMethods(['setUp'])]
+    #[Revs(5)]
+    public function bench_10k_rows() : void
+    {
+        array_to_rows($this->rowsArray, new NativeEntryFactory());
+    }
+
+    #[BeforeMethods(['setUp'])]
+    #[Revs(5)]
+    public function bench_1k_rows() : void
+    {
+        array_to_rows(\array_slice($this->rowsArray, 0, 1_000), new NativeEntryFactory());
+    }
+
+    #[BeforeMethods(['setUp'])]
+    #[Revs(5)]
+    public function bench_5k_rows() : void
+    {
+        array_to_rows(\array_slice($this->rowsArray, 0, 5_000), new NativeEntryFactory());
+    }
+}
diff --git a/...re/etl/tests/Flow/ETL/Tests/Benchmark/Transformer/EntryExpressionEvalTransformerBench.php b/...re/etl/tests/Flow/ETL/Tests/Benchmark/Transformer/EntryExpressionEvalTransformerBench.php
@@ -9,10 +9,12 @@
 use Flow\ETL\Row;
 use Flow\ETL\Rows;
 use Flow\ETL\Transformer\EntryExpressionEvalTransformer;
+use PhpBench\Attributes\Groups;
 use PhpBench\Attributes\Iterations;
 use PhpBench\Attributes\Revs;
 
 #[Iterations(5)]
+#[Groups(['transformer'])]
 final class EntryExpressionEvalTransformerBench
 {
     #[Revs(1000)]