Pull Request data for llm

get_pr_data.sh

#!/bin/bash
set -euo pipefail

# Check if a URL is provided
if [ -z "$1" ]; then
  echo "Usage: $0 <github-pull-request-url>"
  exit 1
fi

url="$1"
curl_opts=("-L" "-s")
if [ -n "${GH_TOKEN:-}" ]; then
  curl_opts+=("-H" "Authorization: token ${GH_TOKEN}")
fi

# Extract components from the URL
if [[ "$url" =~ https://github.com/ ]]; then
  # Public GitHub URL
  domain="github.com"
  api_base="api.github.com"
  owner=$(echo "$url" | sed -E 's|https://github.com/([^/]+)/([^/]+)/pulls/([0-9]+)|\1|')
  repo=$(echo "$url" | sed -E 's|https://github.com/([^/]+)/([^/]+)/pulls/([0-9]+)|\2|')
  pr_number=$(echo "$url" | sed -E 's|https://github.com/([^/]+)/([^/]+)/pulls/([0-9]+)|\3|')
  
  # Construct API URL for public GitHub
  base_url="https://$api_base/repos/$owner/$repo/pulls/$pr_number"
else
  # GitHub Enterprise URL
  domain=$(echo "$url" | sed -E 's|https://([^/]+)/.*|\1|')
  owner=$(echo "$url" | sed -E 's|https://[^/]+/([^/]+)/([^/]+)/pulls/([0-9]+)|\1|')
  repo=$(echo "$url" | sed -E 's|https://[^/]+/([^/]+)/([^/]+)/pulls/([0-9]+)|\2|')
  pr_number=$(echo "$url" | sed -E 's|https://[^/]+/([^/]+)/([^/]+)/pulls/([0-9]+)|\3|')
  
  # Construct API URL for GitHub Enterprise
  base_url="https://$domain/api/v3/repos/$owner/$repo/pulls/$pr_number"
fi

#base_url=$(echo "$url" | sed -E 's|https://([^/]+)/([^/]+)/([^/]+)/pull/([0-9]+)|https://\1/api/v3/repos/\2/\3/pulls/\4|')

#echo $base_url
# Fetch the pull request data
#pr_data=$(curl -L -s -H "Authorization: token ${GH_TOKEN}" "$base_url")
pr_data=$(curl "${curl_opts[@]}" "$base_url")
#echo $pr_data

# Extract relevant information
pr_date=$(echo "$pr_data" | jq -r '.created_at')
pr_author=$(echo "$pr_data" | jq -r '.user.login')
pr_title=$(echo "$pr_data" | jq -r '.title')
pr_body=$(echo "$pr_data" | jq -r '.body')
comments_url=$(echo "$pr_data" | jq -r '.comments_url')
review_comments_url=$(echo "$pr_data" | jq -r '.review_comments_url')

# Fetch comments
comments=$(curl "${curl_opts[@]}"  "$comments_url")
review_comments=$(curl "${curl_opts[@]}"  "$review_comments_url")

# Fetch the diff
diff_url="${base_url}.diff"
diff=$(curl "${curl_opts[@]}"  -H "Accept: application/vnd.github.v3.diff" "$diff_url")

# Print the structured information
echo "Pull Request Information:"
echo "-------------------------"
echo "Date: $pr_date"
echo "Author: $pr_author"
echo "Title: $pr_title"
echo "Description: $pr_body"
echo
echo "Unified Diff:"
echo "-------------"
echo "$diff"
echo
echo "Comments:"
echo "---------"

# Print comments
echo "$comments" | jq -r '.[] | "User: \(.user.login)\nDate: \(.created_at)\nComment: \(.body)\n---"'
echo
echo "Review Comments:"
echo "----------------"

# Print review comments
#echo "$review_comments" | jq -r '.[] | "User: \(.user.login)\nDate: \(.created_at)\nPath: \(.path)\nLast Line of Diff Hunk: \(.diff_hunk | split("\n")[-2])\nComment: \(.body)\n---"'
echo "$review_comments" | jq -r '
  # Group comments by their root comment ID
  group_by(if .in_reply_to_id == null then .id else .in_reply_to_id end)[] |
  # For each group, extract the diff_hunk, path, and comments details
  "Discussion on file: \(.[] | select(.in_reply_to_id == null) | .path)\n" +
  "Line: \(.[] | select(.in_reply_to_id == null) | .diff_hunk | split("\n")[-1])\n" +
  "\n---\n" +
  (map(
    "Date: \(.created_at)\n" +
    "Author: \(.user.login)\n" +
    "Comment: \(.body)\n"
  ) | join("\n---\n")) +
  "\n====================\n"
'