|
| 1 | +#!/usr/bin/env bash |
| 2 | +set -euo pipefail |
| 3 | + |
| 4 | +# Usage: ./fetch_stats_from_ci.sh |
| 5 | +# |
| 6 | +# This script is for fetching historic test stats from GitHub Actions CI. |
| 7 | +# |
| 8 | +# Requires gh with credentials. |
| 9 | +# |
| 10 | +# https://github.com/cli/cli/blob/trunk/pkg/cmd/run/view/view.go#L434 |
| 11 | + |
| 12 | +dir="$(dirname "$0")"/ci-stats |
| 13 | +mkdir -p "${dir}" |
| 14 | + |
| 15 | +pushd "${dir}" >/dev/null |
| 16 | + |
| 17 | +# Stats step name, used for filtering log. |
| 18 | +job_step_name="Print test stats" |
| 19 | + |
| 20 | +if [[ ! -f list-ci.yaml.json ]]; then |
| 21 | + gh run list -w ci.yaml -L 1000 --json conclusion,createdAt,databaseId,displayTitle,event,headBranch,headSha,name,number,startedAt,status,updatedAt,url,workflowDatabaseId,workflowName \ |
| 22 | + >list-ci.yaml.json || { |
| 23 | + rm -f list-ci.yaml.json |
| 24 | + exit 1 |
| 25 | + } |
| 26 | +fi |
| 27 | + |
| 28 | +runs="$( |
| 29 | + jq -r '.[] | select(.status == "completed") | select(.conclusion == "success" or .conclusion == "failure") | [.databaseId, .event, .displayTitle, .headBranch, .headSha] | @tsv' \ |
| 30 | + <list-ci.yaml.json |
| 31 | +)" |
| 32 | + |
| 33 | +while read -r run; do |
| 34 | + mapfile -d $'\t' -t parts <<<"${run}" |
| 35 | + parts[-1]="${parts[-1]%$'\n'}" |
| 36 | + |
| 37 | + database_id="${parts[0]}" |
| 38 | + event="${parts[1]}" |
| 39 | + display_title="${parts[2]}" |
| 40 | + head_branch="${parts[3]}" |
| 41 | + head_sha="${parts[4]}" |
| 42 | + |
| 43 | + run_jobs_file=run-"${database_id}"-"${event}"-jobs.json |
| 44 | + if [[ ! -f "${run_jobs_file}" ]]; then |
| 45 | + echo "Fetching jobs for run: ${display_title} (${database_id}, ${event}, ${head_branch})" |
| 46 | + gh run view "${database_id}" --json jobs >"${run_jobs_file}" || { |
| 47 | + rm -f "${run_jobs_file}" |
| 48 | + exit 1 |
| 49 | + } |
| 50 | + fi |
| 51 | + |
| 52 | + jobs="$( |
| 53 | + jq -r '.jobs[] | select(.name | startswith("test-go")) | select(.status == "completed") | select(.conclusion == "success" or .conclusion == "failure") | [.databaseId, .startedAt, .completedAt, .name, .url] | @tsv' \ |
| 54 | + <"${run_jobs_file}" |
| 55 | + )" |
| 56 | + |
| 57 | + while read -r job; do |
| 58 | + mapfile -d $'\t' -t parts <<<"${job}" |
| 59 | + parts[-1]="${parts[-1]%$'\n'}" |
| 60 | + |
| 61 | + job_database_id="${parts[0]}" |
| 62 | + job_started_at="${parts[1]}" |
| 63 | + job_completed_at="${parts[2]}" |
| 64 | + job_name="${parts[3]}" |
| 65 | + job_url="${parts[4]}" |
| 66 | + |
| 67 | + job_log=run-"${database_id}"-job-"${job_database_id}"-"${job_name}".log |
| 68 | + if [[ ! -f "${job_log}" ]]; then |
| 69 | + echo "Fetching log for: ${job_name} (${job_database_id}, ${job_url})" |
| 70 | + # Example log (partial). |
| 71 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:18.4063489Z ##[group]Run # Artifacts are not available after rerunning a job, |
| 72 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:18.4063872Z # Artifacts are not available after rerunning a job, |
| 73 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:18.4064188Z # so we need to print the test stats to the log. |
| 74 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:18.4064642Z go run ./scripts/ci-report/main.go gotests.json | tee gotests_stats.json |
| 75 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:18.4110112Z shell: /usr/bin/bash -e {0} |
| 76 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:18.4110364Z ##[endgroup] |
| 77 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:19.3440469Z { |
| 78 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:19.3441078Z "packages": [ |
| 79 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:19.3441448Z { |
| 80 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:19.3442927Z "name": "agent", |
| 81 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:19.3443311Z "time": 17.538 |
| 82 | + # test-go (ubuntu-latest) Print test stats 2023-04-11T03:02:19.3444048Z }, |
| 83 | + # ... |
| 84 | + gh run view --job "${job_database_id}" --log >"${job_log}" || { |
| 85 | + # Sometimes gh failes to extract ZIP, etc. :'( |
| 86 | + rm -f "${job_log}" |
| 87 | + echo "Failed to fetch log for: ${job_name} (${job_database_id}, ${job_url}), skipping..." |
| 88 | + continue |
| 89 | + } |
| 90 | + log_lines="$(wc -l "${job_log}" | awk '{print $1}')" |
| 91 | + if [[ ${log_lines} -lt 2 ]]; then |
| 92 | + # Sometimes gh returns nothing and gives no error :'( |
| 93 | + rm -f "${job_log}" |
| 94 | + echo "Log is empty for: ${job_name} (${job_database_id}, ${job_url}), skipping..." |
| 95 | + continue |
| 96 | + fi |
| 97 | + fi |
| 98 | + |
| 99 | + job_stats="$( |
| 100 | + grep "${job_name}.*${job_step_name}" "${job_log}" \ |
| 101 | + | sed -E 's/.*[0-9-]{10}T[0-9:]{8}\.[0-9]*Z //' \ |
| 102 | + | grep -E "^[{}\ ].*" |
| 103 | + )" |
| 104 | + |
| 105 | + if ! jq -e . >/dev/null 2>&1 <<<"${job_stats}"; then |
| 106 | + # Sometimes actions logs are partial when fetched via CLI :'( |
| 107 | + echo "Failed to parse stats for: ${job_name} (${job_database_id}, ${job_url}), skipping..." |
| 108 | + continue |
| 109 | + fi |
| 110 | + |
| 111 | + job_stats_file=run-"${database_id}"-job-"${job_database_id}"-"${job_name}"-stats.json |
| 112 | + jq \ |
| 113 | + --arg event "${event}" \ |
| 114 | + --arg branch "${head_branch}" \ |
| 115 | + --arg sha "${head_sha}" \ |
| 116 | + --arg started_at "${job_started_at}" \ |
| 117 | + --arg completed_at "${job_completed_at}" \ |
| 118 | + --arg display_title "${display_title}" \ |
| 119 | + --arg url "${job_url}" \ |
| 120 | + '{event: $event, branch: $branch, sha: $sha, started_at: $started_at, completed_at: $completed_at, display_title: $display_title, url: $url, stats: .}' \ |
| 121 | + <<<"${job_stats}" \ |
| 122 | + >"${job_stats_file}" |
| 123 | + done <<<"${jobs}" |
| 124 | +done <<<"${runs}" |
0 commit comments