#!/bin/bash
# Upwork Auto-Apply Pipeline
# Called by Clawdbot wake event when new Upwork email arrives
set -euo pipefail

SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
ACCOUNT="jake@localbosses.org"
PROCESSED_FILE="$SCRIPT_DIR/processed.json"
MIN_HOURLY=50
MIN_FIXED=1000

# Initialize processed file if missing
[ -f "$PROCESSED_FILE" ] || echo '[]' > "$PROCESSED_FILE"

echo "[pipeline] Checking for new Upwork job emails..."

# Search for unread Upwork job notification emails
EMAILS=$(gog gmail search "from:donotreply@upwork.com subject:\"New job\" is:unread" \
  --max 10 --account "$ACCOUNT" --json 2>&1)

if [ $? -ne 0 ]; then
  echo "[pipeline] ERROR: Gmail search failed: $EMAILS"
  exit 1
fi

# Parse email IDs
EMAIL_IDS=$(echo "$EMAILS" | jq -r '.[].messages[0].id // empty' 2>/dev/null)

if [ -z "$EMAIL_IDS" ]; then
  echo "[pipeline] No new Upwork job emails found."
  exit 0
fi

PROCESSED_COUNT=0
SKIPPED_COUNT=0

for MSG_ID in $EMAIL_IDS; do
  # Skip already processed
  if jq -e "index(\"$MSG_ID\")" "$PROCESSED_FILE" > /dev/null 2>&1; then
    echo "[pipeline] Already processed: $MSG_ID"
    SKIPPED_COUNT=$((SKIPPED_COUNT + 1))
    continue
  fi

  echo "[pipeline] Processing email: $MSG_ID"
  
  # Get full email
  EMAIL_BODY=$(gog gmail get "$MSG_ID" --account "$ACCOUNT" --json 2>&1)
  
  if [ $? -ne 0 ]; then
    echo "[pipeline] ERROR reading email $MSG_ID: $EMAIL_BODY"
    continue
  fi

  # Extract job URL
  JOB_URL=$(echo "$EMAIL_BODY" | jq -r '.body // .snippet // ""' | \
    grep -oP 'https://www\.upwork\.com/jobs/~\d+' | head -1)
  
  if [ -z "$JOB_URL" ]; then
    # Try plain text extraction
    JOB_URL=$(gog gmail get "$MSG_ID" --account "$ACCOUNT" 2>&1 | \
      grep -oP 'https://www\.upwork\.com/jobs/~\d+' | head -1)
  fi

  if [ -z "$JOB_URL" ]; then
    echo "[pipeline] No job URL found in email $MSG_ID, skipping"
    continue
  fi

  # Extract subject for logging
  SUBJECT=$(echo "$EMAIL_BODY" | jq -r '.subject // "unknown"' 2>/dev/null || echo "unknown")
  
  echo "[pipeline] Job URL: $JOB_URL"
  echo "[pipeline] Subject: $SUBJECT"

  # Extract rate info from email body
  RATE_INFO=$(gog gmail get "$MSG_ID" --account "$ACCOUNT" 2>&1 | \
    grep -E '^\s*(Hourly|Fixed)' | head -1 || echo "")
  
  echo "[pipeline] Rate: $RATE_INFO"

  # Quick pre-filter based on email rate info
  if echo "$RATE_INFO" | grep -qi "hourly"; then
    MAX_RATE=$(echo "$RATE_INFO" | grep -oP '\$\d+' | tail -1 | tr -d '$')
    if [ -n "$MAX_RATE" ] && [ "$MAX_RATE" -lt "$MIN_HOURLY" ]; then
      echo "[pipeline] SKIP: Max hourly rate \$$MAX_RATE < \$$MIN_HOURLY minimum"
      # Mark as processed so we don't revisit
      jq ". + [\"$MSG_ID\"]" "$PROCESSED_FILE" > "$PROCESSED_FILE.tmp" && mv "$PROCESSED_FILE.tmp" "$PROCESSED_FILE"
      SKIPPED_COUNT=$((SKIPPED_COUNT + 1))
      continue
    fi
  fi

  # Output job info for Clawdbot to process
  echo "---JOB_DATA---"
  echo "MSG_ID=$MSG_ID"
  echo "JOB_URL=$JOB_URL"
  echo "SUBJECT=$SUBJECT"
  echo "RATE_INFO=$RATE_INFO"
  echo "---END_JOB_DATA---"

  # Mark as processed
  jq ". + [\"$MSG_ID\"]" "$PROCESSED_FILE" > "$PROCESSED_FILE.tmp" && mv "$PROCESSED_FILE.tmp" "$PROCESSED_FILE"
  PROCESSED_COUNT=$((PROCESSED_COUNT + 1))

done

echo "[pipeline] Done. Processed: $PROCESSED_COUNT, Skipped: $SKIPPED_COUNT"