calculate-cache.sh: extract runs_on and template from workflow file

Signed-off-by: Norio Nomura <[email protected]>

Author: norio-nomura

hack/calculate-cache.sh

@@ -1,5 +1,5 @@
 #!/usr/bin/env bash
-# This script calculates the expected content size, actual cached size, and cache-keys used in caching method prior and after
+# This script calculates the expected content size, actual cached size, and cache-keys used in caching method before and after
 # implementation in https://github.com/lima-vm/lima/pull/2508
 #
 # Answer to the question in https://github.com/lima-vm/lima/pull/2508#discussion_r1699798651
@@ -19,43 +19,30 @@ arch=x86_64
 LIMA_HOME=$(mktemp -d)
 export LIMA_HOME
 
-# shellcheck disable=SC2034
-macos_12=(
-	# integration
-	examples/default.yaml
-	# vmnet
-	examples/vmnet.yaml
-	# upgrade
-	https://raw.githubusercontent.com/lima-vm/lima/v0.15.1/examples/ubuntu-lts.yaml
-)
-
-# shellcheck disable=SC2034
-ubuntu_2204=(
-	# integration-linux
-	examples/alpine.yaml
-	examples/debian.yaml
-	examples/fedora.yaml
-	examples/archlinux.yaml
-	examples/opensuse.yaml
-	examples/experimental/net-user-v2.yaml
-	examples/experimental/9p.yaml
-	examples/docker.yaml
-	examples/../hack/test-templates/alpine-9p-writable.yaml
-	examples/../hack/test-templates/test-misc.yaml
-)
-
-# shellcheck disable=SC2034
-macos_13=(
-	# vz
-	examples/experimental/vz.yaml
-	examples/fedora.yaml
-)
+# parse the workflow file and print runs-on and template
+function print_runs_on_template_from_workflow() {
+	yq -o=j "$1" | jq -r '
+		"./.github/actions/setup_cache_for_template" as $action |
+		"\\$\\{\\{\\s*(?<path>\\S*)\\s*\\}\\}" as $pattern |
+		.jobs | map_values(
+			."runs-on" as $runs_on |
+			{
+				template: .steps | map_values(select(.uses == $action)) | first |.with.template,
+				matrix: .strategy.matrix
+			} | select(.template) |
+			. + { path: .template | (if test($pattern) then sub(".*\($pattern).*";"\(.path)")|split(".") else null end) } |
+			(
+				.template as $template|
+				if .path then
+					getpath(.path)|map(. as $item|$template|sub($pattern;$item))
+				else
+					[$template]
+				end
+			) | map("\($runs_on)\t\(.)")
 
-runners=(
-	macos_12
-	ubuntu_2204
-	macos_13
-)
+		) | flatten |.[]
+	'
+}
 
 function runner_os_from_runner() {
 	# shellcheck disable=SC2249
@@ -149,50 +136,43 @@ actual_cache_sizes=$(
 		jq 'sort_by(.createdAt)|reverse|unique_by(.key)|sort_by(.key)|map({"key":.key,"value":.sizeInBytes})|from_entries'
 )
 
+workflows=(
+	.github/workflows/test.yml
+)
+
 # shellcheck disable=SC2016
-for cache_method in prior after; do
-	echo "==> expected content size, actual cached size, and cache-keys used in caching method ${cache_method} implementation in https://github.com/lima-vm/lima/pull/2508"
+echo "=> compare expected content size, actual cached size, and cache-keys used before and after the change in https://github.com/lima-vm/lima/pull/2508"
+# iterate over before and after
+for cache_method in before after; do
+	echo "==> ${cache_method}"
 	echo "content-size actual-size cache-key"
 	output_yaml=$(
-		for runner in "${runners[@]}"; do
+		for workflow in "${workflows[@]}"; do
+			print_runs_on_template_from_workflow "${workflow}"
+		done | while IFS=$'\t' read -r runner template; do
 			runner_os=$(runner_os_from_runner "${runner}")
-			declare -n ref="${runner}"
-			tepmlates_used_in_test_yml=("${ref[@]}")
-			for template in "${tepmlates_used_in_test_yml[@]}"; do
-				location_digest_size_hash=$(print_location_digest_size_hash_from_template "${template}") || continue
-				read -r location digest size hash containerd containerd_location containerd_digest containerd_size <<<"${location_digest_size_hash}"
-				if [[ ${cache_method} == prior ]]; then
-					key=${runner_os}-${hash}
-				elif [[ ${digest} == null ]]; then
-					key=image:$(basename "${location}")-url-sha256:$(echo -n "${location}" | sha256sum | cut -d' ' -f1)
+			location_digest_size_hash=$(print_location_digest_size_hash_from_template "${template}") || continue
+			read -r location digest size hash containerd containerd_location containerd_digest containerd_size <<<"${location_digest_size_hash}"
+			if [[ ${cache_method} != after ]]; then
+				key=${runner_os}-${hash}
+			elif [[ ${digest} == null ]]; then
+				key=image:$(basename "${location}")-url-sha256:$(echo -n "${location}" | sha256sum | cut -d' ' -f1)
+			else
+				key=image:$(basename "${location}")-${digest}
+			fi
+			if [[ ${containerd} == true ]]; then
+				if [[ ${cache_method} != after ]]; then
+					# previous caching method packages the containerd archive with the image
+					size=$((size + containerd_size))
 				else
-					key=image:$(basename "${location}")-${digest}
-				fi
-				if [[ ${containerd} == true ]]; then
-					if [[ ${cache_method} == prior ]]; then
-						# previous caching method packages the containerd archive with the image
-						size=$((size + containerd_size))
-					else
-						# new caching method packages the containerd archive separately
-						containerd_key=containerd:$(basename "${containerd_location}")-${containerd_digest}
-						cat <<-EOF
-							- key: ${containerd_key}
-							  template: ${template}
-							  location: ${containerd_location}
-							  digest: ${containerd_digest}
-							  size: ${containerd_size}
-						EOF
-					fi
+					# new caching method packages the containerd archive separately
+					containerd_key=containerd:$(basename "${containerd_location}")-${containerd_digest}
+					printf -- "- key: %s\n  template: %s\n  location: %s\n  digest: %s\n  size: %s\n" \
+						"${containerd_key}" "${template}" "${containerd_location}" "${containerd_digest}" "${containerd_size}"
 				fi
-				cat <<-EOF
-					- key: ${key}
-					  template: ${template}
-					  location: ${location}
-					  digest: ${digest}
-					  size: ${size}
-				EOF
-				# echo -e "- key: ${key}\n  template: ${template}\n  location: ${location}\n  size: ${size}\n  containerd: ${containerd}\n  containerd_location: ${containerd_location}\n  containerd_digest: ${containerd_digest}"
-			done
+			fi
+			printf -- "- key: %s\n  template: %s\n  location: %s\n  digest: %s\n  size: %s\n" \
+				"${key}" "${template}" "${location}" "${digest}" "${size}"
 		done
 	)
 	cat <<<"${output_yaml}" >".calculate-cache-collected-info-${cache_method}.yaml"