Refactor benchmark setup and enhance CSV generation logging (#112)

This refactor splits the benchmark setup into two separate jobs for
better handling of individual configurations: one for Docker-based
benchmarks and another for PHP-based benchmarks. The jobs now support
testing with varying numbers of columns. In addition, the CSV generation
process now includes better-riched logs for easier debugging.

.github/workflows/benchmark.yml

@@ -10,41 +10,95 @@
 # @see        https://github.com/JBZoo/Csv-Blueprint
 #
 
-name: Stress Test
+name: Speed Test
 
 on:
-  pull_request:
-    branches:
-      - '*'
-  push:
-    branches:
-      - 'master'
+  workflow_dispatch:
+    inputs:
+      branch:
+        description: 'Branch Name'
+        required: true
+        default: 'master'
 
 jobs:
-  stress-test:
-    name: Benchmark
+  docker-benchmark:
+    name: Docker
     runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        version: [ latest, local ]
+        columns: [ 1, 5, 10, 20 ]
+    env:
+      BENCH_COLS: ${{ matrix.columns }}
+      DOCKER_IMAGE: jbzoo/csv-blueprint:${{ matrix.version }}
     steps:
       - name: Checkout code
         uses: actions/checkout@v4
         with:
           fetch-depth: 0
+          ref: ${{ github.event.inputs.branch }}
+          # ref: ${{ github.event.pull_request.head.ref }}
 
-      - name: Setup PHP
-        uses: shivammathur/setup-php@v2
-
+      # We need it to build benchmark tool. See ./tests/Benchmarks
       - name: Build project
         run: make build --no-print-directory
 
-      - name: Create random huge CSV files
-        run: make bench-create-csv --no-print-directory
-
-      - name: Building Docker Image
+      - name: Build local image
+        if: matrix.version == 'local'
         uses: docker/build-push-action@v5
         with:
           context: .
           push: false
-          tags: jbzoo/csv-blueprint:local
+          tags: ${{ env.DOCKER_IMAGE }}
+
+      - name: Pull latest image
+        if: matrix.version == 'latest'
+        run: docker pull ${{ env.DOCKER_IMAGE }}
+
+      - name: 🔥 Benchmark 🔥
+        run: make bench-create-csv bench-docker --no-print-directory
+
+
+  php-benchmark:
+    name: Phar
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        version: [ latest, local ]
+        columns: [ 1, 5, 10, 20 ]
+    env:
+      BENCH_COLS: ${{ matrix.columns }}
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          ref: ${{ github.event.inputs.branch }}
+          # ref: ${{ github.event.pull_request.head.ref }}
+
+      - name: Setup PHP
+        uses: shivammathur/setup-php@v2
+        with:
+          php-version: 8.3
+          coverage: none
+          ini-values: opcache.enable_cli=1, opcache.jit=1255
+
+      - name: Build local Phar file
+        if: matrix.version == 'local'
+        run: |
+          make build-prod build-phar-file --no-print-directory
+          ./build/csv-blueprint.phar
+
+      - name: Download latest Phar file
+        if: matrix.version == 'latest'
+        run: |
+          wget https://github.com/JBZoo/Csv-Blueprint/releases/latest/download/csv-blueprint.phar -O ./build/csv-blueprint.phar
+          chmod +x ./build/csv-blueprint.phar
+          ./build/csv-blueprint.phar
+
+      # We need it to build benchmark tool. See ./tests/Benchmarks
+      - name: Build project
+        run: make build --no-print-directory
 
-      - name: 🔥 Benchmark with Docker 🔥
-        run: make bench-docker --no-print-directory
+      - name: 🔥 Benchmark 🔥
+        run: make bench-create-csv bench-phar --no-print-directory

.github/workflows/main.yml

@@ -198,7 +198,7 @@ jobs:
 
 
   verify-phar-binary:
-    name: Verify PHAR
+    name: Verify Phar
     runs-on: ubuntu-latest
     strategy:
       matrix:

.github/workflows/publish.yml

@@ -18,7 +18,7 @@ on:
 
 jobs:
   phar:
-    name: Publish PHAR
+    name: Publish Phar
     runs-on: ubuntu-latest
     steps:
       - name: Checkout code
@@ -36,10 +36,10 @@ jobs:
       - name: Build project in production mode
         run: make build-prod build-phar-file --no-print-directory
 
-      - name: 🎨 Test PHAR file
+      - name: 🎨 Test Phar file
         run: ./build/csv-blueprint.phar --ansi -vvv
 
-      - name: Upload PHAR to the release
+      - name: Upload Phar to the release
         uses: softprops/action-gh-release@v2
         with:
           token: ${{ secrets.GITHUB_TOKEN }}

Makefile

@@ -19,6 +19,7 @@ endif
 DOCKER_IMAGE     ?= jbzoo/csv-blueprint:local
 CMD_VALIDATE     := validate:csv --ansi
 BLUEPRINT        := COLUMNS=300 time $(PHP_BIN) ./csv-blueprint $(CMD_VALIDATE)
+BLUEPRINT_PHAR   := COLUMNS=300 time $(PHP_BIN) ./build/csv-blueprint.phar $(CMD_VALIDATE)
 BLUEPRINT_DOCKER := time docker run --rm  --workdir=/parent-host -v .:/parent-host $(DOCKER_IMAGE) $(CMD_VALIDATE)
 
 VALID_CSV       := --csv='./tests/fixtures/demo.csv'
@@ -37,7 +38,7 @@ build-prod: ##@Project Build project in production mode
 	@rm -f `pwd`/ci-report-converter
 	@make build-version
 
-build-phar-file: ##@Project Build PHAR file
+build-phar-file: ##@Project Build Phar file
 	curl -L "https://github.com/box-project/box/releases/download/4.5.1/box.phar" -o ./build/box.phar
 	@make build-version
 	@php ./build/box.phar --version
@@ -95,13 +96,12 @@ BENCH_SCHEMAS     := --schema='./tests/Benchmarks/benchmark-*.yml'
 BENCH_FLAGS       := --debug --profile --report=text -vvv
 
 
-bench-all: ##@Benchmarks Run all benchmarks
+bench: ##@Benchmarks Run all benchmarks
 	@make bench-create-csv
-	@make docker-build
 	@make bench-docker
 
 bench-create-csv: ##@Benchmarks Create CSV file
-	$(call title,"Benchmark - Create CSV file")
+	$(call title,"Benchmark - Create CSV file. Cols=${BENCH_COLS} Rows=${BENCH_ROWS_SRC}_000")
 	@mkdir -pv ./build/bench/
 	@rm -fv    ./build/bench/*.csv
 	@time bash ./tests/Benchmarks/create-csv.sh
@@ -112,6 +112,11 @@ bench-docker: ##@Benchmarks Run CSV file with Docker
 	-$(BLUEPRINT_DOCKER) $(BENCH_CSV) $(BENCH_SCHEMAS) $(BENCH_FLAGS)
 
 
-bench-php: ##@Benchmarks Run CSV file with PHP binary
-	$(call title,"Benchmark - CSV file with PHP binary")
+bench-phar: ##@Benchmarks Run CSV file with Phar
+	$(call title,"Benchmark - CSV file with Phar")
+	-$(BLUEPRINT_PHAR) $(BENCH_CSV) $(BENCH_SCHEMAS) $(BENCH_FLAGS)
+
+
+bench-php: ##@Benchmarks Run CSV file with classic PHP binary
+	$(call title,"Benchmark - CSV file with classic PHP binary")
 	-$(BLUEPRINT) $(BENCH_CSV) $(BENCH_SCHEMAS) $(BENCH_FLAGS)

src/Rules/Ruleset.php

@@ -56,12 +56,7 @@ public function validateRuleSet(array|string $cellValue, int $line, int $linesTo
             $errors->addError($rule->validate($cellValue, $line));
 
             if ($linesToAggregate > 0) {
-                Utils::debug(
-                    "  {$rule->getRuleCode()} - "
-                    . '<blue>'
-                    . \number_format($linesToAggregate / (\microtime(true) - $startTimer))
-                    . '</blue> l/s',
-                );
+                Utils::debugSpeed("  {$rule->getRuleCode()} -", $linesToAggregate, $startTimer);
             }
         }
 

src/Utils.php

@@ -61,6 +61,12 @@ public static function debug(int|string $message): void
         }
     }
 
+    public static function debugSpeed(string $messPrefix, int $lines, float $startTimer): void
+    {
+        $kiloLines = \round(($lines / (\microtime(true) - $startTimer)) / 1000);
+        self::debug("{$messPrefix} <blue>" . \number_format($kiloLines) . 'K</blue> lines/sec');
+    }
+
     public static function kebabToCamelCase(string $input): string
     {
         return \str_replace(' ', '', \ucwords(\str_replace(['-', '_'], ' ', $input)));

src/Validators/ValidatorCsv.php

@@ -165,12 +165,7 @@ private function validateLines(bool $quickStop = false): ErrorSuite
                 }
             }
             Utils::debug("{$messPrefix} Lines <yellow>" . \number_format($lineCounter) . '</yellow>');
-            Utils::debug(
-                "{$messPrefix} <yellow>Speed:cell</yellow> "
-                . '<blue>'
-                . \number_format($lineCounter / (\microtime(true) - $startTimer))
-                . '</blue> l/s',
-            );
+            Utils::debugSpeed($messPrefix, $lineCounter, $startTimer);
 
             if ($isAggRules) { // Time optimization
                 $errors->addErrorSuit($colValidator->validateList($columValues, $lineCounter));

tests/Benchmarks/create-csv.sh

@@ -25,6 +25,10 @@ for i in {1..1000}; do
     cat ./build/bench/${BENCH_COLS}_${BENCH_ROWS_SRC}.csv >> $BENCH_CSV_PATH
 done
 
+echo "----"
+echo $BENCH_CSV_PATH
+head $BENCH_CSV_PATH
+
 echo "----"
 echo "File size : $(du -h $BENCH_CSV_PATH)"
 echo "Rows count: $(wc -l $BENCH_CSV_PATH)"