gh-ai-prs.sh

#!/usr/bin/env bash
# gh-ai-prs.sh
# Fetch 6 months of AI-created PR counts from GitHub Search API.
# Uses total_count per query (not actual results), so no 1,000-result cap.
# Output: CSV to stdout, progress to stderr.
#
# Tracks:
#   1. PRs created by known AI bot accounts
#   2. PRs containing AI-generated markers in title/body
#
# Usage:
#   ./gh-ai-prs.sh                          # weekly buckets, last 6 months
#   ./gh-ai-prs.sh --granularity monthly
#   ./gh-ai-prs.sh --start 2025-08-01 --end 2026-02-27
#
# Requirements: gh (authenticated), jq
# Rate limit: 30 search requests/min authenticated. Script sleeps 2.5s/request.

set -euo pipefail

# ── Config ────────────────────────────────────────────────────────────────────

GRANULARITY="weekly"
END_DATE=$(date +%Y-%m-%d)
START_DATE=$(date -d "6 months ago" +%Y-%m-%d)
SLEEP_SECS=2.5

while [[ $# -gt 0 ]]; do
  case $1 in
    --granularity) GRANULARITY="$2"; shift 2 ;;
    --start)       START_DATE="$2";  shift 2 ;;
    --end)         END_DATE="$2";    shift 2 ;;
    --sleep)       SLEEP_SECS="$2";  shift 2 ;;
    *) echo "Unknown arg: $1" >&2; exit 1 ;;
  esac
done

# ── Queries to track ─────────────────────────────────────────────────────────
# Format: "display_name|search_query"
# All queries get "is:pr" prepended and "created:FROM..TO" appended automatically.

declare -a TOOLS=(
  # ── AI Coding Agents (PR authors) ──
  "Copilot SWE Agent (author)|author:copilot-swe-agent[bot]"
  "Codex Connector (author)|author:chatgpt-codex-connector[bot]"
  "Claude (author)|author:claude[bot]"
  "Devin (author)|author:devin-ai-integration[bot]"
  "Jules (author)|author:google-labs-jules[bot]"
  "Amazon Q (author)|author:amazon-q-developer[bot]"
  "Sweep (author)|author:sweep-ai[bot]"
  "Pixeebot (author)|author:pixeebot[bot]"
  "Codeflash (author)|author:codeflash-ai[bot]"
  "Codegen (author)|author:codegen-sh[bot]"
  "Mentat (author)|author:mentatbot[bot]"
  "GitAuto (author)|author:gitauto-ai[bot]"

  # ── Code Review / Other Bots (may also create PRs) ──
  "CodeRabbit (author)|author:coderabbitai[bot]"
  "Sourcery (author)|author:sourcery-ai[bot]"
  "Gemini Code Assist (author)|author:gemini-code-assist[bot]"
  "Cursor (author)|author:cursor[bot]"

  # ── Dependency / Automation (context) ──
  "Dependabot (author)|author:dependabot[bot]"
  "Renovate (author)|author:renovate[bot]"

  # ── Body/title markers ──
  "Claude Code (body)|\"Generated with Claude Code\""
  "Claude (co-author in body)|\"Co-Authored-By: Claude\""
  "Devin (body)|\"Created by Devin\""
  "Sweep (title)|\"Sweep:\""
)

# ── Date range generation ─────────────────────────────────────────────────────

generate_ranges() {
  local granularity="$1"
  local start="$2"
  local end="$3"

  case "$granularity" in
    daily)
      local current="$start"
      while [[ "$current" < "$end" || "$current" == "$end" ]]; do
        echo "${current}|${current}"
        current=$(date -d "$current + 1 day" +%Y-%m-%d)
      done
      ;;
    weekly)
      local current
      current=$(date -d "$start - $(date -d "$start" +%u) days + 1 day" +%Y-%m-%d)
      [[ "$current" < "$start" ]] && current=$(date -d "$current + 7 days" +%Y-%m-%d)
      while [[ "$current" < "$end" || "$current" == "$end" ]]; do
        local week_end
        week_end=$(date -d "$current + 6 days" +%Y-%m-%d)
        [[ "$week_end" > "$end" ]] && week_end="$end"
        echo "${current}|${week_end}"
        current=$(date -d "$current + 7 days" +%Y-%m-%d)
      done
      ;;
    monthly)
      local current="$start"
      while [[ "$current" < "$end" || "$current" == "$end" ]]; do
        local year month last_day
        year=$(date -d "$current" +%Y)
        month=$(date -d "$current" +%m)
        last_day=$(date -d "$year-$month-01 + 1 month - 1 day" +%Y-%m-%d)
        [[ "$last_day" > "$end" ]] && last_day="$end"
        local range_start
        range_start=$(date -d "$year-$month-01" +%Y-%m-%d)
        echo "${range_start}|${last_day}"
        current=$(date -d "$year-$month-01 + 1 month" +%Y-%m-%d)
      done
      ;;
  esac
}

# ── Query function ────────────────────────────────────────────────────────────

url_encode() {
  local str="$1"
  str="${str// /+}"
  str="${str//[/\%5B}"
  str="${str//]/\%5D}"
  str="${str//:/%3A}"
  echo "$str"
}

MAX_RETRIES=5
RETRY_BASE_SECS=60

query_count() {
  local query="$1"
  local from="$2"
  local to="$3"

  local full_query="is:pr ${query} created:${from}..${to}"
  local attempt=0

  while true; do
    local result
    local exit_code=0
    result=$(gh api \
      --method GET \
      "search/issues" \
      -f q="${full_query}" \
      -f per_page=1 \
      2>&1) || exit_code=$?

    if echo "$result" | grep -qi "secondary rate limit"; then
      attempt=$(( attempt + 1 ))
      if [[ "$attempt" -gt "$MAX_RETRIES" ]]; then
        echo "ERROR: secondary rate limit — gave up after $MAX_RETRIES retries" >&2
        echo "ERROR"
        return
      fi
      local wait_secs=$(( RETRY_BASE_SECS * attempt ))
      echo "  RATE LIMITED: waiting ${wait_secs}s before retry $attempt/$MAX_RETRIES..." >&2
      sleep "$wait_secs"
      continue
    fi

    if [[ "$exit_code" -ne 0 ]]; then
      echo "ERROR: gh api failed (exit $exit_code): $result" >&2
      echo "ERROR"
      return
    fi

    local count
    count=$(echo "$result" | jq -r '.total_count // empty' 2>/dev/null)

    if [[ -z "$count" ]]; then
      local err_msg
      err_msg=$(echo "$result" | jq -r '.message // .errors[0].message // "unknown error"' 2>/dev/null || echo "unparseable response")
      echo "ERROR: API response: $err_msg" >&2
      echo "ERROR"
      return
    fi

    echo "$count"
    return
  done
}

build_search_url() {
  local query="$1"
  local from="$2"
  local to="$3"
  local encoded
  encoded=$(url_encode "is:pr ${query} created:${from}..${to}")
  echo "https://github.com/search?q=${encoded}&type=pullrequests"
}

# ── Main ──────────────────────────────────────────────────────────────────────

echo "tool,period_start,period_end,granularity,total_count,url"

mapfile -t RANGES < <(generate_ranges "$GRANULARITY" "$START_DATE" "$END_DATE")
total_requests=$(( ${#TOOLS[@]} * ${#RANGES[@]} ))
done_requests=0

est_secs=$(( total_requests * 3 ))
est_mins=$(( est_secs / 60 ))

echo "==> ${#TOOLS[@]} tools × ${#RANGES[@]} periods = ${total_requests} API calls" >&2
echo "==> Estimated time: ~${est_mins} minutes" >&2
echo "" >&2

for tool_entry in "${TOOLS[@]}"; do
  display="${tool_entry%%|*}"
  query="${tool_entry#*|}"

  echo "--- ${display} ---" >&2

  for range in "${RANGES[@]}"; do
    from="${range%%|*}"
    to="${range#*|}"

    count=$(query_count "$query" "$from" "$to")
    url=$(build_search_url "$query" "$from" "$to")
    echo "${display},${from},${to},${GRANULARITY},${count},${url}"

    if [[ "$count" == "ERROR" ]]; then
      echo "  ${from}..${to}: ERROR (see above)" >&2
    else
      echo "  ${from}..${to}: ${count}  ${url}" >&2
    fi

    done_requests=$(( done_requests + 1 ))
    if [[ "$done_requests" -lt "$total_requests" ]]; then
      sleep "$SLEEP_SECS"
    fi
  done
done

echo "" >&2
echo "==> Done. ${done_requests} requests made." >&2