misc: refactor artifact size metrics

Author: 0marperez

.github/actions/artifact-size-metrics/build-and-upload/action.yml

@@ -0,0 +1,24 @@
+name: Calculate Artifact Size
+description: Calculates size for JVM artifacts
+
+inputs:
+  artifact-dirs:
+    description: List of directories for which artifact sizes will be calculated
+    required: true
+    # TODO: Additional inputs
+
+runs:
+  using: composite
+  steps:
+    - name: Checkout repository
+      uses: actions/checkout@v4
+
+    - name: Calculate artifact sizes
+      shell: bash
+      run: |
+        chmod +x ./src/main.sh
+        chmod +x ./src/metrics.sh
+        chmod +x ./src/s3.sh
+        chmod +x ./src/cloudwatch.sh
+        
+        ./src/main.sh

.github/actions/artifact-size-metrics/show-results/action.yml renamed to .github/actions/artifact-size-metrics/download-and-process/action.yml

@@ -1,25 +1,31 @@
-name: Artifact Size Metrics - Show Results
-description: Posts a new artifact analysis comment on the PR
+name: Download and process artifact size metrics
+description: TODO
+
 inputs:
-  working-directory:
-    required: false
-    description: The working directory to use for reading the previously-generated report.
-    default: ''
+  download:
+    description: TODO
+    default: "false"
+    # TODO: Additional inputs
+
 runs:
-  using: "composite"
+  using: composite
   steps:
-    - name: Post Artifact Analysis Comment
+    - name: Download and process artifact size metrics
+      shell: bash
+      run: |
+        chmod +x ./src/main.sh
+        chmod +x ./src/s3.sh
+        chmod +x ./src/compare.sh
+        chmod +x ./src/comment.sh
+        
+        ./src/main.sh
+
+    - name: Post artifact analysis comment # TODO: MOVE TO SCRIPT IF POSSIBLE
       uses: actions/github-script@v7
       with:
         script: |
-          const workingDirectory = '${{ inputs.working-directory }}'
-          if (workingDirectory) {
-            process.chdir(workingDirectory)
-          }
-
           const fs = require('node:fs')
-          const prNumber = context.issue.number ?? process.env.SDK_PR
-
+          const prNumber = context.issue.number ?? process.env.SDK_PR # TOSO - CHANGE - THIS WON'T WORK FOR EXTERNAL CONTRIBUTORS BTW - IF EXTERNAL CONTRIBUTOR JUST FAIL
           const prInfo = await github.rest.pulls.get({
             owner: context.repo.owner,
             repo: context.repo.repo,
@@ -83,4 +89,14 @@ runs:
                 clientMutationId
               }
             }`)
-          }
+          }
+
+    - name: Fail if large size increase
+      shell: bash
+      if: ${{ !contains(github.event.pull_request.labels.*.name, 'acknowledge-artifact-size-increase') }}
+      run: |
+        if [ "$LARGE_DIFF" == "true" ]; then
+            echo An artifact increased in size by more than allowed or a new artifact was created.
+            echo If this is expected please add the 'acknowledge-artifact-size-increase' label to this pull request.
+            exit 1
+        fi

.github/actions/artifact-size-metrics/utils/build-and-upload/cloudwatch.sh

@@ -0,0 +1,41 @@
+# Upload the artifact size metrics to cloudwatch
+uploadToCloudwatch() {
+  set -o pipefail
+
+  METRICS_FILE="metrics.csv"      # path to your CSV file
+  PROJECT_REPO_NAME=$1
+  NAMESPACE="Artifact Size Metrics"
+
+  # Read CSV, skipping header
+  tail -n +2 "$METRICS_FILE" | while IFS=',' read -r artifactName artifactSize; do
+      artifactName=$(echo "$artifactName" | xargs) # trim spaces
+      artifactSize=$(echo "$artifactSize" | xargs)
+
+      # Build JSON for CloudWatch
+      metric_json=$(jq -n \
+          --arg name "$PROJECT_REPO_NAME-$artifactName" \
+          --arg value "$artifactSize" \
+          --arg project "$PROJECT_REPO_NAME" \
+          '{
+              MetricName: $name,
+              Timestamp: (now | todate),
+              Unit: "Bytes",
+              Value: ($value | tonumber),
+              Dimensions: [
+                  { Name: "Project", Value: $project }
+              ]
+          }'
+      )
+
+      METRICS+=("$metric_json")
+  done
+
+  # Send metrics in chunks of 1000
+  chunk_size=1000
+  for ((i=0; i<${#METRICS[@]}; i+=chunk_size)); do
+      chunk=("${METRICS[@]:i:chunk_size}")
+      aws cloudwatch put-metric-data \
+          --namespace "$NAMESPACE" \
+          --metric-data "$(printf '%s\n' "${chunk[@]}" | jq -s '.')"
+  done
+}

.github/actions/artifact-size-metrics/utils/build-and-upload/main.sh

@@ -0,0 +1,37 @@
+#!/bin/bash
+
+source ./metrics.sh
+source ./cloudwatch.sh
+source ../s3.sh
+source ../setup.sh
+
+setup
+
+# Build
+if [ "$GITHUB_REPOSITORY" = "aws-sdk-kotlin" ]; then
+    # FIXME: Enable K/N builds
+    ./gradlew build -Paws.kotlin.native=false build --parallel --max-workers 16
+else
+    ./gradlew build
+fi
+
+# Move artifacts that will be published to staging dir (build/m2)
+./gradlew publish
+
+# Calculate size for artifacts in staging dir
+getArtifactSizes
+
+# Upload size metrics
+if [ "$UPLOAD" == "true" ]; then
+  if [ "$RELEASE_METRICS" == "true" ]; then
+    # For record-keeping
+    uploadToS3 "$GITHUB_REPOSITORY"-v"$IDENTIFIER".csv
+    uploadToS3 "$GITHUB_REPOSITORY"-latest.csv
+
+    # For display in our OPS dashboard
+    uploadToCloudwatch "$GITHUB_REPOSITORY"
+  else
+    # For downstream consumption in pull requests
+    uploadToS3 [TEMP]"$GITHUB_REPOSITORY"-pull-request-"$IDENTIFIER".csv
+  fi
+fi

.github/actions/artifact-size-metrics/utils/build-and-upload/metrics.sh

@@ -0,0 +1,14 @@
+# Gets artifact size metrics from staging dir
+getArtifactSizes() {
+  output="build/reports/metrics/artifact-size-metrics.csv"
+
+  # Write CSV header
+  echo "Jar File,Size (Bytes)" > "$output"
+
+  # Find all JARs (exclude sources and javadoc)
+  # TODO: Calculate KN artifacts sizes
+  find build/m2 -type f -name "*.jar" ! -name "*-sources.jar" ! -name "*-javadoc.jar" | while read -r jar; do
+      size=$(stat -c%s "$jar")
+      echo "\"$jar\",$size" >> "$output"
+  done
+}

.github/actions/artifact-size-metrics/utils/download-and-process/compare.sh

@@ -0,0 +1,40 @@
+# Compares two metrics files, saves results into a file, returns "true" if large diff was found
+compareMetrics() {
+    local pr_file="$1"
+    local release_file="$2"
+    local output_file="$3"
+
+    # Create header for the Markdown table
+    echo "| Artifact | Pull Request (bytes) | Latest Release (bytes) | Delta (bytes) | Delta (percentage) |" > "$output_file"
+    echo "|----------|--------------------|----------------------|---------------|------------------|" >> "$output_file"
+
+    local flag=false
+
+    # Read PR file line by line (skip header)
+    tail -n +2 "$pr_file" | while IFS=',' read -r artifact pr_size; do
+        # Trim whitespace
+        artifact=$(echo "$artifact" | xargs)
+        pr_size=$(echo "$pr_size" | xargs)
+
+        # Find corresponding artifact in release file
+        release_size=$(awk -F',' -v art="$artifact" 'NR>1 && $1 ~ art {gsub(/ /,"",$2); print $2}' "$release_file")
+
+        # Skip if artifact not found in release file
+        [ -z "$release_size" ] && continue
+
+        # Compute delta and percentage
+        delta=$((pr_size - release_size))
+        abs_delta=${delta#-}
+        percent=$((100 * abs_delta / release_size))
+
+        # Add row to Markdown table
+        echo "| $artifact | $pr_size | $release_size | $delta | $percent% |" >> "$output_file"
+
+        # Check if delta > 5%
+        if [ "$percent" -gt 5 ]; then
+            flag=true
+        fi
+    done
+
+    $flag && echo "true"
+}

.github/actions/artifact-size-metrics/utils/download-and-process/main.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+
+source ./compare.sh
+source ../s3.sh
+source ../setup.sh
+setup
+
+# Get metrics from pull request and from latest release
+if [ "$DOWNLOAD" == "true" ]; then
+  downloadFromS3 [TEMP]"$GITHUB_REPOSITORY"-pull-request-"$IDENTIFIER".csv ./current.csv
+fi
+downloadFromS3 "$GITHUB_REPOSITORY"-latest.csv ./latest.csv
+
+# Compare metrics and save if large diff was found
+export LARGE_DIFF=$(compareMetrics current.csv latest.csv comparisson.md)

.github/actions/artifact-size-metrics/utils/s3.sh

@@ -0,0 +1,12 @@
+# Owned by: aws-kotlin-sdk+ci
+S3_ARTIFACT_SIZE_METRICS_BUCKET="artifact-size-metrics"
+
+# Uploads metrics to the metrics bucket under the specified file name
+uploadToS3() {
+  aws s3 cp "$1" s3://"$S3_ARTIFACT_SIZE_METRICS_BUCKET"/"$2"
+}
+
+# Downloads metrics from the metrics bucket to the specified local file
+downloadFromS3() {
+  aws s3 cp s3://"$S3_ARTIFACT_SIZE_METRICS_BUCKET"/"$1" "$2"
+}

.github/actions/artifact-size-metrics/utils/setup.sh

@@ -0,0 +1,5 @@
+setup() {
+  # Exit if non zero exit code or if env var is missing
+  set -u
+  set -e
+}