dotfiles/local/bin/x-pr-comments

#!/usr/bin/env bash

# x-pr-comments - Fetch GitHub Pull Request comments and review suggestions
# Copyright (c) 2025 - Licensed under MIT
#
# Usage:
#   x-pr-comments <pr-number>                    # When run inside a git repository
#   x-pr-comments <github-pr-url>                # Direct GitHub PR URL
#   x-pr-comments -h|--help                      # Show this help
#
# Examples:
#   x-pr-comments 1                              # PR #1 in current repo
#   x-pr-comments https://github.com/user/repo/pull/1
#
# Requirements:
#   - gh CLI tool installed and authenticated
#   - Internet connection for GitHub API access
#
# Output:
#   Structured list of PR comments and file change requests with LLM processing directions

set -euo pipefail

# Colors for output
readonly RED='\033[0;31m'
readonly GREEN='\033[0;32m'
readonly YELLOW='\033[1;33m'
readonly BLUE='\033[0;34m'
readonly NC='\033[0m' # No Color

# Show usage information
show_usage()
{
  sed -n '3,20p' "$0" | sed 's/^# //' | sed 's/^#//'
}

# Log functions
log_error()
{
  echo -e "${RED}ERROR:${NC} $1" >&2
}
# Log a warning message
log_warn()
{
  echo -e "${YELLOW}WARN:${NC} $1" >&2
}
# Log an informational message
log_info()
{
  if [[ "${INFO:-0}" == "1" ]]; then
    echo -e "${GREEN}INFO:${NC} $1" >&2
  fi
}
# Log a debug message
log_debug()
{
  if [[ "${DEBUG:-0}" == "1" ]]; then
    echo -e "${BLUE}DEBUG:${NC} $1" >&2
  fi
}

# Filter out CodeRabbit comments containing "Addressed in commit"
filter_coderabbit_addressed_comments()
{
  local input_data="$1"
  local is_wrapped="$2" # true for {comments: [...]}, false for [...]

  local jq_filter='select(
    (.user.login | contains("coderabbit") | not) or
    (.body | contains("Addressed in commit") | not)
  )'

  if [[ "$is_wrapped" == "true" ]]; then
    echo "$input_data" | jq "{comments: [.comments[] | $jq_filter]}" 2> /dev/null || echo "$input_data"
  else
    echo "$input_data" | jq "[.[] | $jq_filter]" 2> /dev/null || echo "$input_data"
  fi
}

# Fetch and filter API data with consistent logging
fetch_and_filter_data()
{
  local endpoint="$1"
  local data_name="$2"
  local is_wrapped="$3" # true/false

  local data
  data=$(gh api "$endpoint" 2> /dev/null || echo "[]")

  if [[ "$is_wrapped" == "true" ]]; then
    data=$(echo "$data" | jq '{comments: .}' 2> /dev/null || echo '{"comments":[]}')
  fi

  data=$(filter_coderabbit_addressed_comments "$data" "$is_wrapped")

  local count_field="length"
  [[ "$is_wrapped" == "true" ]] && count_field=".comments | length"

  local count
  count=$(echo "$data" | jq -r "$count_field" 2> /dev/null || echo "0")
  log_debug "$data_name count: $count"

  echo "$data"
}

# Format file-specific comments grouped by review
format_grouped_review_comments()
{
  local review_comments="$1"
  local reviews="$2"
  local repo="$3"

  local count
  count=$(echo "$review_comments" | jq -r 'length' 2> /dev/null || echo "0")

  if [[ "$count" -eq 0 ]]; then
    echo "No file-specific comments found."
    return
  fi

  # Group comments by review ID and format them
  echo "$review_comments" | jq -r --argjson reviews "$reviews" '
    group_by(.pull_request_review_id) | .[] |
    . as $comments |
    ($reviews[] | select(.id == $comments[0].pull_request_review_id)) as $review |
    "### Review by \($review.user.login) (\($review.submitted_at)) [\($review.state)]

Review ID: \($review.id) - API: gh api /repos/'"$repo"'/pulls/1/reviews/\($review.id)

" + ([.[] | "
#### Comment ID: \(.id)

- **File:** \(.path)
- **Commit:** \(.commit_id)
- **Author:** \(.user.login) (\(.user.type))
- **Lines:** \(if .start_line then "\(.start_line)-\(.line // "N/A")" else "\(.line // "N/A")" end) (original: \(.original_line // "N/A"))
- **Position:** \(.position // "N/A") (original: \(.original_position // "N/A"))
- **Subject Type:** \(.subject_type // "N/A")
- **API:** gh api /repos/'"$repo"'/pulls/comments/\(.id)

**Body:**
\(.body)

"] | join("")) + "
---
"
  ' 2> /dev/null || {
    log_debug "Error grouping review comments by review ID"
    echo "Error parsing grouped review comments."
  }
}

# Check if gh CLI is available
check_dependencies()
{
  if ! command -v gh &> /dev/null; then
    log_error "GitHub CLI (gh) is not installed. Please install it first:"
    log_error "  https://cli.github.com/"
    exit 1
  fi

  if ! gh auth status &> /dev/null; then
    log_error "GitHub CLI is not authenticated. Run: gh auth login"
    exit 1
  fi
}

# Get repository info from git remote
get_repo_info()
{
  if ! git rev-parse --is-inside-work-tree &> /dev/null; then
    log_error "Not inside a git repository"
    return 1
  fi

  local remote_url
  remote_url=$(git remote get-url origin 2> /dev/null || echo "")

  if [[ -z "$remote_url" ]]; then
    log_error "No origin remote found"
    return 1
  fi

  # Parse GitHub URL (both HTTPS and SSH formats)
  if [[ "$remote_url" =~ github\.com[:/]([^/]+)/([^/]+)(\.git)?$ ]]; then
    local repo_name="${BASH_REMATCH[2]}"
    # Remove .git suffix if present
    repo_name="${repo_name%.git}"
    echo "${BASH_REMATCH[1]}/${repo_name}"
  else
    log_error "Remote URL is not a GitHub repository: $remote_url"
    return 1
  fi
}

# Parse GitHub PR URL to extract owner/repo/pr-number
parse_github_url()
{
  local url="$1"

  if [[ "$url" =~ github\.com/([^/]+)/([^/]+)/pull/([0-9]+) ]]; then
    echo "${BASH_REMATCH[1]}/${BASH_REMATCH[2]}" "${BASH_REMATCH[3]}"
  else
    log_error "Invalid GitHub PR URL format: $url"
    return 1
  fi
}

# Format output with LLM directions
format_output()
{
  local repo="$1"
  local pr_number="$2"
  local pr_info="$3"
  local review_comments="$4"
  local reviews="$5"

  # Header and instructions
  cat << EOF
# GitHub PR Comments Analysis Report

## LLM Processing Instructions

You are analyzing review comments and change requests from GitHub Pull Request #$pr_number in repository $repo.

**Your tasks:**
1. **Review Analysis**: For each review, understand the reviewer's overall feedback and concerns
2. **Change Request Validation**: Check if each file-specific comment is still relevant to the current codebase
3. **Priority Assessment**: Rank change requests by importance and impact on code quality
4. **Implementation Planning**: Create actionable tasks for addressing valid change requests
5. **Pattern Recognition**: Identify recurring issues across different reviews

**Tools to use:**
- \`find\`, \`cat\`, \`rg\` commands and available tools to examine current codebase
- File system tools to verify mentioned files exist and check current state
- \`gh pr diff $pr_number\` to see what changes are being reviewed

## Pull Request Information

EOF

  # PR information
  echo "$pr_info" | jq -r '"**Title:** \(.title)
**State:** \(.state)
**URL:** \(.url)
**Number:** '"$pr_number"'
**Repository:** '"$repo"'
"' 2> /dev/null || {
    echo "**Error:** Could not parse PR information"
    return 1
  }

  # Review Comments Section
  echo -e "\n## Review Comments and Change Requests\n"

  format_grouped_review_comments "$review_comments" "$reviews" "$repo"

  # Footer
  cat << EOF

## Next Steps for LLM Analysis

1. **Analyze and verify all comments, including nitpick**
  - Consider all comments from code reviewers as change requests
  - Analyze all "Outside diff range comments" as change requests
  - All Copilot and CodeRabbit comments and change requests MUST BE analyzed
  - Do not trust blindly the comments, validate them against current codebase

2. **Validate change requests:**
  - Check if mentioned files exist and match current state
  - Verify if suggestions are still applicable
  - Identify any already-addressed issues

3. **Generate actionable implementation plan:**
  - Prioritized list of valid change requests
  - Grouped by file/area for efficient implementation
  - Clear next steps for addressing reviewer feedback
  - All valid change requests MUST BE handled, even if low priority

4. **Use todo lists and memory tools to track progress*
  - All valid change requests MUST BE handled
  - Keep handling them until each of them are handled

5. **Identify patterns and recurring issues:**
  - Highlight common themes across reviews
  - Suggest broader improvements to code quality and practices
  - Suggest LLM instructions to avoid similar issues in future
EOF
}

# Fetch and display PR data directly
fetch_and_display_pr_data()
{
  local repo="$1"
  local pr_number="$2"

  log_info "Fetching PR #$pr_number from $repo..."

  # Get PR basic info
  local pr_info
  pr_info=$(gh pr view "$pr_number" --repo "$repo" --json title,state,url 2> /dev/null) || {
    log_error "Failed to fetch PR #$pr_number from $repo"
    return 1
  }

  # Fetch review data using helper function
  local review_comments reviews
  review_comments=$(fetch_and_filter_data "/repos/$repo/pulls/$pr_number/comments" "Review comments" "false")
  reviews=$(fetch_and_filter_data "/repos/$repo/pulls/$pr_number/reviews" "Reviews" "false")

  # Display formatted output
  format_output "$repo" "$pr_number" "$pr_info" "$review_comments" "$reviews"
}

# Main function
main()
{
  local repo=""
  local pr_number=""

  # Parse arguments
  case "${1:-}" in
    -h | --help)
      show_usage
      exit 0
      ;;
    "")
      log_error "Missing argument. Provide PR number or GitHub URL."
      show_usage
      exit 1
      ;;
    https://github.com/*)
      local parsed
      parsed=$(parse_github_url "$1") || exit 1
      read -r repo pr_number <<< "$parsed"
      ;;
    [0-9]*)
      repo=$(get_repo_info) || exit 1
      pr_number="$1"
      ;;
    *)
      log_error "Invalid argument: $1"
      show_usage
      exit 1
      ;;
  esac

  check_dependencies

  log_debug "Repository: $repo"
  log_debug "PR Number: $pr_number"

  # Fetch and display data
  fetch_and_display_pr_data "$repo" "$pr_number"
}

# Run main function with all arguments
main "$@"