Get commit deploys diff from particular workflow

aggregate-deploys.sh

#!/usr/bin/env bash
set -Eeuo pipefail

PAIRS=${PAIRS:-3}
EXTRA_SCAN=${EXTRA_SCAN:-10}
OUTFILE="${1:-deploy-diffs-$(date +'%Y-%m-%d').md}"

mapfile -t PROJECTS < <(find . -maxdepth 2 -mindepth 1 -type f -name "get-deploys-diff.sh" -exec dirname {} \; | sort -u)

if [[ ${#PROJECTS[@]} -eq 0 ]]; then
  echo "No projects with get-deploys-diff.sh found under $(pwd)"
  exit 1
fi

TMPDIR="$(mktemp -d)"
trap 'rm -rf "$TMPDIR"' EXIT

for proj in "${PROJECTS[@]}"; do
  proj_name="${proj#./}"
  echo "Processing $proj_name..."

  raw="$(mktemp)"
  (
    cd "$proj" || exit 1
    chmod +x ./get-deploys-diff.sh || true
    EMIT_META=1 PAIRS="$PAIRS" EXTRA_SCAN="$EXTRA_SCAN" bash ./get-deploys-diff.sh
  ) >"$raw" 2>&1 || true

  current_file=""
  while IFS= read -r line; do
    if [[ "$line" =~ ^@@@DEPLOY\| ]]; then
      current_file=""

      IFS='|' read -r _ iso runid base head <<<"$line"
      key="${iso//:/-}"
      block_path="$TMPDIR/${key}__${proj_name}__${runid}.block"
      current_file="$block_path"
      {
        echo "## === ${proj_name} ==="
      } > "$current_file"
      continue
    fi

    if [[ -n "$current_file" ]]; then
      echo "$line" >> "$current_file"
    fi
  done < "$raw"

  rm -f "$raw"
done

{
  echo "# Deploy diffs summary — generated on $(date '+%B %d, %Y %H:%M')"
  echo

  for f in $(ls -1 "$TMPDIR"/*.block 2>/dev/null | sort -r); do
    cat "$f"
    echo
  done
} > "$OUTFILE"

echo "✅ Saved summary to: $(pwd)/$OUTFILE"

get-deploys-diff.sh

#!/usr/bin/env bash
set -Eeuo pipefail

# Format ISO8601 like 2025-07-30T12:34:56Z -> "July 30, 2025"
fmt_date() {
  python3 - "$1" <<'PY'
import sys
from datetime import datetime, timezone
s = sys.argv[1]
# robust parse (handles ...Z)
dt = datetime.fromisoformat(s.replace("Z", "+00:00"))
print(f"{dt.strftime('%B')} {dt.day}, {dt.year}")
PY
}

# ====== SETTINGS ======
WORKFLOW_NAME="Deploy pipeline PROD"
WORKFLOW_FILE=".github/workflows/prod.yaml"
BRANCH="main"
PAIRS=${PAIRS:-1}
EXTRA_SCAN=${EXTRA_SCAN:-10}

# Task extraction
TASK_PATTERN="${TASK_PATTERN:-TECH-[0-9]+}"
TASK_LINK_BASE="${TASK_LINK_BASE:-https://example.atlassian.net/browse}"
# ======================

need() { command -v "$1" >/dev/null || { echo "❌ Required tool not found: $1"; exit 1; }; }
need gh; need jq; need grep; need sort; need awk

# Resolve {owner}/{repo}
if ! REPO="$(gh repo view --json nameWithOwner --jq .nameWithOwner 2>/dev/null)"; then
  REMOTE_URL="$(git config --get remote.origin.url || true)"
  if [[ -z "$REMOTE_URL" ]]; then
    echo "❌ Could not detect repository. Run inside a git repo or set REPO manually."
    exit 1
  fi
  REPO="${REMOTE_URL%.git}"
  REPO="${REPO#git@github.com:}"
  REPO="${REPO#https://github.com/}"
fi

GITHUB_HTTP_BASE="https://github.com/${REPO}"

# Check auth
if ! gh auth status -h github.com >/dev/null 2>&1; then
  echo "⚠️  GitHub CLI is not authenticated. Run: gh auth login"
fi

# Determine workflow selector
WF_SELECTOR=""
if [[ -n "$WORKFLOW_FILE" ]]; then
  WF_SELECTOR="$WORKFLOW_FILE"
elif [[ -n "$WORKFLOW_NAME" ]]; then
  list_json="$(gh workflow list --repo "$REPO" --limit 200 --json id,name,path,state 2>/dev/null || true)"
  wf_match="$(echo "$list_json" | jq -r --arg n "$WORKFLOW_NAME" '
    .[] | select((.name==$n) or (.name|test($n;"i")) or (.path|test($n;"i"))) | .path
  ' | head -n1)"
  if [[ -z "$wf_match" ]]; then
    echo "❌ Workflow not found by name: \"$WORKFLOW_NAME\""
    echo "Available workflows:"
    echo "$list_json" | jq -r '.[] | "- \(.name)  (\(.path)) [\(.state)]"'
    exit 1
  fi
  WF_SELECTOR="$wf_match"
else
  echo "❌ Provide WORKFLOW_NAME or WORKFLOW_FILE in the script settings."
  exit 1
fi

RUNS_LIMIT=$(( PAIRS + 1 + EXTRA_SCAN ))

# Fetch runs
runs_json="$(gh run list \
  --repo "$REPO" \
  --workflow "$WF_SELECTOR" \
  --branch "$BRANCH" \
  --json databaseId,headSha,displayTitle,createdAt,updatedAt,conclusion,event \
  --limit "$RUNS_LIMIT" 2>/dev/null || true)"

if [[ -z "$runs_json" || "$runs_json" == "[]" ]]; then
  echo "❌ No runs found. Check workflow/branch: $WF_SELECTOR / $BRANCH"
  exit 1
fi

# Keep successful manual deploys, sort ascending by time
runs_sorted="$(echo "$runs_json" | jq -c '
  map(select(.conclusion=="success" and .event=="workflow_dispatch"))
  | sort_by(.updatedAt) | reverse
')"

need_runs=$(( PAIRS + 1 ))
runs_slice="$(echo "$runs_sorted" | jq -c ".[0:$need_runs]")"
slice_count="$(echo "$runs_slice" | jq 'length')"

if (( slice_count < 2 )); then
  echo "Not enough successful deploys to compare (found: $slice_count)."
  exit 0
fi

for i in $(seq 0 $((slice_count-2))); do
  base_index=$((i+1))
  head_index=$i

  base_sha="$(echo "$runs_slice" | jq -r ".[$base_index].headSha")"
  head_sha="$(echo "$runs_slice" | jq -r ".[$head_index].headSha")"
  deployed_at="$(echo "$runs_slice" | jq -r ".[$head_index].updatedAt")"
  run_id="$(echo "$runs_slice" | jq -r ".[$head_index].databaseId")"
  title="$(echo "$runs_slice" | jq -r ".[$head_index].displayTitle")"

  deployed_at_fmt="$(fmt_date "$deployed_at")"

  # If aggregator asked for meta, emit machine-readable marker (single line)
  if [[ "${EMIT_META:-0}" == "1" ]]; then
    # project name is resolved by aggregator; ISO date + run id + shas
    echo "@@@DEPLOY|${deployed_at}|${run_id}|${base_sha}|${head_sha}"
  fi

  COMPARE_URL="${GITHUB_HTTP_BASE}/compare/${base_sha}...${head_sha}"

  echo ""
  echo "=== DEPLOY #$run_id – $title – $deployed_at_fmt ==="
  echo ""
  echo "Compare: ${COMPARE_URL}"
  echo "Range: $base_sha...$head_sha"
  echo "Commits:"
  comp_json="$(gh api "repos/$REPO/compare/$base_sha...$head_sha")"

  echo "$comp_json" | jq -r --arg base "$GITHUB_HTTP_BASE" '
        .commits[] |
        "- \(.sha[0:7]) \(.commit.message|split("\n")[0]) (by \(.commit.author.name)) — \($base)/commit/\(.sha)"
      '

  # Extract unique task IDs from commit messages by regex
  messages="$(echo "$comp_json" | jq -r '.commits[].commit.message')"
  # grep -oE prints only matches, one per line; sort -u makes them unique
  tasks="$(printf "%s\n" "$messages" | grep -oE "$TASK_PATTERN" | sort -u || true)"

  if [[ -n "$tasks" ]]; then
    echo ""
    echo "These tasks were closed:"
    while IFS= read -r t; do
      if [[ -n "$TASK_LINK_BASE" ]]; then
        # Plain text with URL
        echo "- $t (${TASK_LINK_BASE%/}/$t)"
      else
        echo "- $t"
      fi
    done <<< "$tasks"
  fi
done

Example of usage:

Preparation

1. We have a folder "Projects", inside that folder we have projects "proj-1", "proj-2" etc.
2. In the "Projects/" we place aggregate-deploys.sh
3. In the "proj-1", "proj-2" we place get-deploys-diff.sh

Diff for one repo

1. Go to the "proj-1"
2. run bash get-deploys-diff.sh

You will receive something like that in the console:

=== DEPLOY #16628231194 – Deploy pipeline PROD – July 30, 2025 ===

Compare: https://github.com/example/merchant-dashboard-fe/compare/9b38807715ae4143d89619c68eb6963f5dd8267f...ddab6142bcbe2dbaf5691b32852e56d7d5003767
Range: 9b38807715ae4143d89619c68eb6963f5dd8267f...ddab6142bcbe2dbaf5691b32852e56d7d5003767
Commits:

• edc5f55 feat[TECH:1355]: remove v2 from csv upload endpoint (by OtherTeamMember) — https://github.com/example/merchant-dashboard-fe/commit/edc5f55c3577cdbcbc723bd3ebc1551fb8f7077b
• 8585c86 TECH-1365 - Update toast messages in CsvUploader (by Sundwell) — https://github.com/example/merchant-dashboard-fe/commit/8585c86a0caa4475c74ebcfc0e929e67616434b4
• ddab614 Merge pull request #33 from example/feat/TECH-1355-update-csv-upload-endpoint (by Sundwell) — https://github.com/example/merchant-dashboard-fe/commit/ddab6142bcbe2dbaf5691b32852e56d7d5003767

These tasks were closed:

• TECH-1355 (https://example.atlassian.net/browse/TECH-1355)
• TECH-1365 (https://example.atlassian.net/browse/TECH-1365)

Diff for all repos inside `Projects" folder

1. Go to Projects folder
2. run bash aggregate-deploys.sh

New file in format deploy-diffs-YYYY-MM-DD.md will be generated with sorted diffs from all repos with get-deploys-diff.sh file in it