Find stale PRs where the author hasn't commented in 2 weeks

stale-prs.sh

#!/bin/bash

# Function to display help information.
help_message() {
    cat <<EOF
Usage: $0 [-r <repository>] [-d <inactivity days>] [-v]
  -r   Repository in the format owner/repo (default: kubernetes/website)
  -d   Inactivity period in days (default: 14)
  -v   Enable verbose output for troubleshooting.
EOF
}

# Default values
REPO="kubernetes/website"
INACTIVE_DAYS=14
VERBOSE=0

# If no arguments provided, show help and note defaults, then continue.
if [ $# -eq 0 ]; then
    help_message
    echo "No arguments provided. Using default values."
fi

# Parse command-line options.
while getopts "r:d:vh" opt; do
    case $opt in
        r) REPO="$OPTARG" ;;
        d) INACTIVE_DAYS="$OPTARG" ;;
        v) VERBOSE=1 ;;
        h) help_message; exit 0 ;;
        *) help_message; exit 1 ;;
    esac
done

[ $VERBOSE -eq 1 ] && echo "Repository: $REPO"
[ $VERBOSE -eq 1 ] && echo "Inactivity period (days): $INACTIVE_DAYS"

# Calculate the inactive threshold date in ISO8601 format and seconds since epoch.
INACTIVE_DATE=$(date -v-"${INACTIVE_DAYS}"d +%Y-%m-%dT%H:%M:%SZ)
INACTIVE_DATE_SEC=$(date -jf "%Y-%m-%dT%H:%M:%SZ" "$INACTIVE_DATE" +%s)

[ $VERBOSE -eq 1 ] && echo "Inactive threshold date: $INACTIVE_DATE"
[ $VERBOSE -eq 1 ] && echo "Inactive threshold (seconds): $INACTIVE_DATE_SEC"

# Array to store flagged PR details.
flagged_prs=()

# Process each PR JSON object.
process_pr() {
    local pr="$1"
    local pr_number pr_url author pr_created pr_created_sec

    pr_number=$(echo "$pr" | jq '.number')
    pr_url=$(echo "$pr" | jq -r '.url')
    author=$(echo "$pr" | jq -r '.author.login')
    pr_created=$(echo "$pr" | jq -r '.createdAt')
    pr_created_sec=$(date -jf "%Y-%m-%dT%H:%M:%SZ" "$pr_created" +%s)

    [ $VERBOSE -eq 1 ] && echo "Processing PR #$pr_number: created at $pr_created ($pr_created_sec)"

    # Skip PRs that are too recent.
    if (( pr_created_sec > INACTIVE_DATE_SEC )); then
        [ $VERBOSE -eq 1 ] && echo "Skipping PR #$pr_number: too recent (created at $pr_created)"
        return
    fi

    # Filter: only process PRs with the "language/en" label using jq.
    if [ "$(echo "$pr" | jq '[.labels[] | select(.name=="language/en")] | length')" -eq 0 ]; then
        [ $VERBOSE -eq 1 ] && echo "Skipping PR #$pr_number: not labeled as language/en."
        return
    fi

    # Check if the PR is marked as work in progress.
    if echo "$pr" | jq -r '.labels[].name' | grep -q "do-not-merge/work-in-progress"; then
        [ $VERBOSE -eq 1 ] && echo "Skipping PR #$pr_number: marked as work in progress."
        return
    fi

    local should_flag=0

    # Condition 1: Check for the "cncf-cla: no" label.
    if echo "$pr" | jq -r '.labels[].name' | grep -q "cncf-cla: no"; then
        echo "PR #$pr_number: $pr_url - Author ($author) hasn't signed the CLA and is older than ${INACTIVE_DAYS} days"
        should_flag=1
    fi

    # Condition 2: Check the author's comment/review activity.
    local issue_comments review_comments review_reviews all_comments last_comment last_comment_sec
    issue_comments=$(gh api -X GET repos/"$REPO"/issues/"$pr_number"/comments --paginate | \
        jq -r --arg AUTHOR "$author" '.[] | select(.user.login == $AUTHOR) | .created_at')
    review_comments=$(gh api -X GET repos/"$REPO"/pulls/"$pr_number"/comments --paginate | \
        jq -r --arg AUTHOR "$author" '.[] | select(.user.login == $AUTHOR) | .created_at')
    review_reviews=$(gh api -X GET repos/"$REPO"/pulls/"$pr_number"/reviews --paginate | \
        jq -r --arg AUTHOR "$author" '.[] | select(.user.login == $AUTHOR) | .submitted_at')
    
    all_comments=$(echo -e "$issue_comments\n$review_comments\n$review_reviews" | grep .)

    if [ $VERBOSE -eq 1 ]; then
        echo "Issue comments for PR #$pr_number:"
        echo "$issue_comments"
        echo "Review comments for PR #$pr_number:"
        echo "$review_comments"
        echo "Review submissions for PR #$pr_number:"
        echo "$review_reviews"
    fi

    if [ -z "$all_comments" ]; then
        echo "PR #$pr_number: $pr_url - Author ($author) has not commented since creation ($pr_created)"
        should_flag=1
    else
        last_comment=$(echo "$all_comments" | sort -r | head -n 1)
        last_comment_sec=$(date -jf "%Y-%m-%dT%H:%M:%SZ" "$last_comment" +%s)
        [ $VERBOSE -eq 1 ] && echo "Last comment for PR #$pr_number: $last_comment ($last_comment_sec)"
        if (( last_comment_sec < INACTIVE_DATE_SEC )); then
            echo "PR #$pr_number: $pr_url - Author ($author) last commented on $last_comment, which is over ${INACTIVE_DAYS} days ago"
            should_flag=1
        fi
    fi

    if (( should_flag == 1 )); then
        flagged_prs+=( "PR #$pr_number: $pr_url" )
        open -a /Applications/Firefox.app "$pr_url"
    fi
}

# Fetch open PRs with required fields—here we increase the limit.
PRs=$(gh pr list -R "$REPO" --json number,url,author,createdAt,labels --state open --limit 1000)

[ $VERBOSE -eq 1 ] && echo "Fetched PRs:" && echo "$PRs" | jq .

# Process each PR via a here-string.
while IFS= read -r pr; do
    process_pr "$pr"
done <<< "$(echo "$PRs" | jq -c '.[]')"

# Summarize flagged PRs.
if [ ${#flagged_prs[@]} -eq 0 ]; then
    echo "No PRs found meeting the criteria."
else
    echo -e "\nFlagged PRs to be checked:"
    for pr in "${flagged_prs[@]}"; do
        echo "$pr"
    done
fi

This script is written to run on a mac with gh, Firefox and jq installed.

Added check for CLA

Written with the help of ChatGPT.