clawdbot-workspace/reonomy-run.sh

48 lines
1.2 KiB
Bash
Executable File

#!/bin/bash
# Reonomy Scraper Runner
# Usage: ./reonomy-run.sh [max_properties] [search_id]
# Output: reonomy-export-TIMESTAMP.csv
set -e
WORKSPACE="$HOME/.clawdbot/workspace"
MAX_PROPS="${1:-10}"
SEARCH_ID="${2:-}"
TIMESTAMP=$(date +%Y%m%d-%H%M%S)
EXPORT_CSV="$WORKSPACE/reonomy-export-${TIMESTAMP}.csv"
FRESH_JSON="$WORKSPACE/reonomy-leads-run-${TIMESTAMP}.json"
echo "=== Reonomy Scraper Run ==="
echo "Max properties: $MAX_PROPS"
echo "Search ID: ${SEARCH_ID:-default}"
echo "Output: $EXPORT_CSV"
echo ""
# Clear the leads file for a fresh run
echo '{"lastUpdated":"","searchId":"","totalLeads":0,"leads":[]}' > "$WORKSPACE/reonomy-leads-v13.json"
# Run scraper
export MAX_PROPERTIES="$MAX_PROPS"
if [ -n "$SEARCH_ID" ]; then
export REONOMY_SEARCH_ID="$SEARCH_ID"
fi
echo "Starting scraper..."
node "$WORKSPACE/reonomy-scraper-v13.js"
# Copy raw JSON
cp "$WORKSPACE/reonomy-leads-v13.json" "$FRESH_JSON"
# Convert to CSV
echo ""
echo "Converting to CSV..."
node "$WORKSPACE/reonomy-to-csv.js" "$FRESH_JSON" "$EXPORT_CSV"
# Summary
LEAD_COUNT=$(node -e "const d=JSON.parse(require('fs').readFileSync('$FRESH_JSON','utf8'));console.log(d.totalLeads)")
echo ""
echo "=== Done ==="
echo "Leads: $LEAD_COUNT"
echo "JSON: $FRESH_JSON"
echo "CSV: $EXPORT_CSV"