added bench command

.github/workflows/command-bench.yml

@@ -0,0 +1,108 @@
+name: Command Bench
+
+on: 
+  workflow_dispatch: 
+    inputs:
+      pr:
+        description: Number of the Pull Request
+        required: true
+      benchmark:
+        description: Pallet benchmark
+        type: choice
+        required: true
+        options:
+          - substrate-pallet
+          - polkadot-pallet
+          - cumulus-assets
+          - cumulus-collectives
+          - cumulus-coretime
+          - cumulus-bridge-hubs
+          - cumulus-contracts
+          - cumulus-glutton
+          - cumulus-starters
+          - cumulus-people
+          - cumulus-testing
+      subcommand:
+        description: Subcommand
+        type: choice
+        required: true
+        options:
+          - pallet
+          - xcm
+      runtime:
+        description: Runtime
+        type: choice
+        options:
+          - dev
+          - rococo
+          - westend
+          - asset-hub-westend
+          - asset-hub-rococo
+          - collectives-westend
+          - coretime-rococo
+          - coretime-westend
+          - bridge-hub-rococo
+          - bridge-hub-westend
+          - contracts-rococo
+          - glutton-westend
+          - glutton-westend-dev-1300
+          - seedling
+          - shell
+          - people-westend
+          - people-rococo
+          - penpal
+          - rococo-parachain
+      pallet:
+        description: Pallet
+        type: string
+        default: pallet_name
+      target_dir:
+        description: Target directory
+        type: choice
+        options:
+          - substrate
+          - polkadot
+          - cumulus
+
+jobs:
+  set-image:     
+    runs-on: ubuntu-latest
+    outputs:
+      IMAGE: ${{ steps.set_image.outputs.IMAGE }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - id: set_image
+        run: cat .github/env >> $GITHUB_OUTPUT
+  cmd-bench:
+    needs: [set-image]
+    runs-on: ubuntu-latest
+    container:
+      image: ${{ needs.set-image.outputs.IMAGE }}
+    steps:
+      - name: Download repo
+        uses: actions/checkout@v4
+      - name: Install gh cli
+        id: gh
+        uses: ./.github/actions/set-up-gh
+        with:
+          pr-number: ${{ inputs.pr }}
+          GH_TOKEN: ${{ github.token }}
+      - name: Run bench
+        run: |
+          "./scripts/bench.sh" "${{ inputs.benchmark }}" --runtime "${{ inputs.runtime }}" --pallet "${{ inputs.pallet }}" --target_dir "${{ inputs.target_dir }}" --subcommand "${{ inputs.subcommand }}"
+      - name: Report failure
+        if: ${{ failure() }}
+        run: gh pr comment ${{ inputs.pr }} --body "## Command failed ❌\n\nRun by @${{ github.actor }} for <code>${{ github.workflow }}</code> failed. See logs <a href=\"$RUN\">here</a>."
+        env:
+          RUN: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          GH_TOKEN: ${{ github.token }}
+      - uses: stefanzweifel/git-auto-commit-action@v5
+        with:
+          commit_message: cmd-action - ${{ github.workflow }}
+          branch: ${{ steps.gh.outputs.branch }}
+      - name: Report succeed
+        run: gh pr comment ${{ inputs.pr }} --body "## Action completed 🎉🎉\n\nRun by @${{ github.actor }} for <code>${{ github.workflow }}</code> completed 🎉. See logs <a href=\"$RUN\">here</a>."
+        env:
+          RUN: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+          GH_TOKEN: ${{ github.token }}

scripts/bench.sh

@@ -0,0 +1,116 @@
+#!/bin/bash
+# Initially based on https://github.com/paritytech/bench-bot/blob/cd3b2943d911ae29e41fe6204788ef99c19412c3/bench.js
+
+# Most external variables used in this script, such as $GH_CONTRIBUTOR, are
+# related to https://github.com/paritytech/try-runtime-bot
+
+# This script relies on $GITHUB_TOKEN which is probably a protected GitLab CI
+# variable; if this assumption holds true, it is implied that this script should
+# be ran only on protected pipelines
+
+set -eu -o pipefail
+shopt -s inherit_errexit
+
+# realpath allows to reuse the current
+BENCH_ROOT_DIR=$(realpath "$(dirname "${BASH_SOURCE[0]}")")
+
+. "$(realpath "$(dirname "${BASH_SOURCE[0]}")/command-utils.sh")"
+
+repository_name="$(basename "$PWD")"
+
+get_arg optional --target_dir "$@"
+target_dir="${out:-""}"
+
+get_arg optional --noexit "$@"
+noexit="${out:-""}"
+
+output_path="."
+
+profile="production"
+
+if [[ "$repository_name" == "polkadot-sdk" ]]; then
+  output_path="./$target_dir"
+fi
+
+cargo_run_benchmarks="cargo run --quiet --profile=${profile}"
+
+echo "Repository: $repository_name"
+echo "Target Dir: $target_dir"
+echo "Output Path: $output_path"
+
+cargo_run() {
+  echo "Running $cargo_run_benchmarks" "${args[@]}"
+
+  # if not patched with PATCH_something=123 then use --locked
+  if [[ -z "${BENCH_PATCHED:-}" ]]; then
+    cargo_run_benchmarks+=" --locked"
+  fi
+
+  $cargo_run_benchmarks "${args[@]}"
+}
+
+
+main() {
+
+  # Remove the "github" remote since the same repository might be reused by a
+  # GitLab runner, therefore the remote might already exist from a previous run
+  # in case it was not cleaned up properly for some reason
+  &>/dev/null git remote remove github || :
+
+  tmp_dirs=()
+  cleanup() {
+    exit_code=$?
+    # Clean up the "github" remote at the end since it contains the
+    # $GITHUB_TOKEN secret, which is only available for protected pipelines on
+    # GitLab
+    &>/dev/null git remote remove github || :
+    rm -rf "${tmp_dirs[@]}"
+    echo "Done, exit: $exit_code"
+    exit $exit_code
+  }
+
+  # avoid exit if --noexit is passed
+  if [ -z "$noexit" ]; then
+    trap cleanup EXIT
+  fi
+
+  # set -x
+
+  get_arg required --subcommand "$@"
+  local subcommand="${out:-""}"
+
+  case "$subcommand" in
+    runtime|pallet|xcm)
+      echo 'Running bench_pallet'
+      . "$BENCH_ROOT_DIR/lib/bench-pallet.sh" "$@"
+    ;;
+    overhead)
+      echo 'Running bench_overhead'
+      . "$BENCH_ROOT_DIR/lib/bench-overhead.sh" "$@"
+    ;;
+    all)
+      echo "Running all-$target_dir"
+      . "$BENCH_ROOT_DIR/lib/bench-all-${target_dir}.sh" "$@"
+    ;;
+    *)
+      die "Invalid subcommand $subcommand to process_args"
+    ;;
+  esac
+
+  # set +x
+
+  # in case we used diener to patch some dependency during benchmark execution,
+  # revert the patches so that they're not included in the diff
+  git checkout --quiet HEAD Cargo.toml
+
+  # Save the generated weights to GitLab artifacts in case commit+push fails
+  echo "Showing weights diff for command"
+  git diff -P | tee -a "${ARTIFACTS_DIR}/weights.patch"
+  echo "Wrote weights patch to \"${ARTIFACTS_DIR}/weights.patch\""
+
+
+  # instead of using `cargo run --locked`, we allow the Cargo files to be updated
+  # but avoid committing them. It is so `cmd_runner_apply_patches` can work
+  git restore --staged Cargo.*
+
+main "$@"