Bel Download File From Crm DB

Download File from CRM Database

Download files stored in the CRM PostgreSQL database and save them to the filesystem in _RESULTS_FROM_AGENT/.

When to Use This Skill

Use this skill when the user wants to:

• Download a file from the CRM database to disk
• Retrieve a document/image/PDF stored in the database
• Save a database attachment to the filesystem
• Export file data from the CRM

Trigger phrases:

• "Download the contract from the database"
• "Save the file from CRM to disk"
• "Get me the PDF from database ID 123"
• "Retrieve the attachment for company X"
• "Export the image from the database"

Do NOT use this skill if:

• The user only wants to view/open a file (use bel-open-file instead)
• The user wants to see a file from the database (use bel-show-crm-file instead)
• The file is already on disk (no download needed)

How to Use This Skill

Step 1: Use the CRM Database Skill for Schema Knowledge

To understand the database structure and query files, invoke the bel-crm-db skill first to get access to table schemas and connection details.

Step 2: Identify the File to Download

Determine the file_id from the database. This may require querying tables that link to files:

Common junction tables:

• Files linked to companies: company_site_file table
• Files linked to people: person_file table
• Files linked to events: event_file table
• Files linked to opportunities: sales_opportunity_file table

Example query to find files:

-- Find all files for a company
SELECT f.file_id, f.file_name, f.mime_type, f.file_size
FROM file f
JOIN company_site_file csf ON f.file_id = csf.file_id
WHERE csf.company_site_id = 123;

Step 3: Download the File

Execute the download script with the file_id:

python scripts/download_file_from_db.py <file_id>

Options:

# Download with custom output directory
python scripts/download_file_from_db.py 456 --output-dir /custom/path

# Quiet mode (only output file path)
python scripts/download_file_from_db.py 456 --quiet

Step 4: Report Results to User

Inform the user where the file was saved:

✅ File downloaded successfully!
   File path: _RESULTS_FROM_AGENT/contract.pdf
   Original name: contract.pdf
   MIME type: application/pdf
   Size: 245,832 bytes

Technical Details

Database Schema

The script queries the file table with this structure:

CREATE TABLE file (
    file_id SERIAL PRIMARY KEY,
    file_hash VARCHAR(64) UNIQUE NOT NULL,
    file_name VARCHAR(255) NOT NULL,
    mime_type VARCHAR(127),
    file_data BYTEA NOT NULL,
    file_size INTEGER,
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

File Storage

• Default location: _RESULTS_FROM_AGENT/
• Filename collision handling: Appends _1, _2, etc. if file exists
• Path traversal protection: Sanitizes filenames before saving

Database Connection

The script uses environment variables for database connection:

• POSTGRES_URL (preferred): Full connection string
• Or individual variables: POSTGRES_HOST, POSTGRES_PORT, POSTGRES_DB, POSTGRES_USER, POSTGRES_PASSWORD

Error Handling

Common errors:

• File not found: file_id doesn't exist in database
• Connection error: Database credentials incorrect or server unavailable
• Permission error: Insufficient write permissions for output directory
• Disk space: Insufficient space for large files

Integration with Other Skills

This skill integrates with:

• bel-crm-db: For schema knowledge and querying file metadata
• bel-open-file: To open the downloaded file after saving
• bel-show-crm-file: Orchestration skill that uses both download and open

Example Workflows

Workflow 1: Download a specific file by ID

User: "Download file ID 789 from the database"

1. Use bel-crm-db to understand schema
2. Run: python scripts/download_file_from_db.py 789
3. Report the saved file path to user

Workflow 2: Find and download a company's contract

User: "Download the contract for Acme Corp"

1. Use bel-crm-db to query:
   - Find company_site_id for "Acme Corp"
   - Find file_id via company_site_file junction table
2. Run: python scripts/download_file_from_db.py <file_id>
3. Report results to user

Workflow 3: Download multiple files

User: "Download all images for event ID 5"

1. Query event_file junction table for all file_ids
2. Loop through each file_id:
   - python scripts/download_file_from_db.py <file_id>
3. Report all downloaded file paths