pySigma-backend-duckdb 0.1.2

pySigma DuckDB backend for local Sigma rule validation against JSON logs

pySigma DuckDB Backend

This is a pySigma backend that generates DuckDB SQL queries from Sigma rules. It's designed for local validation of Sigma rules against JSON log files, making it ideal for CI/CD pipelines and regression testing.

Features

• Convert Sigma rules to DuckDB SQL queries
• Validate rules against local JSON log files
• Support for JSON arrays, single objects, and NDJSON formats
• Built-in LogIndex class for efficient log loading and querying
• ValidationResult with match counts and matched log details

Installation

pip install pySigma-backend-duckdb

Usage

Basic Query Generation

from sigma.rule import SigmaRule
from sigma.collection import SigmaCollection
from sigma.backends.duckdb import DuckDBBackend

rule = SigmaRule.from_yaml("""
    title: Suspicious PowerShell Execution
    logsource:
        category: process_creation
        product: windows
    detection:
        selection:
            CommandLine|contains: powershell
        condition: selection
""")

backend = DuckDBBackend()
queries = backend.convert(SigmaCollection([rule]))
print(queries[0])
# SELECT * FROM logs WHERE CommandLine ILIKE '%powershell%'

Validating Rules Against Local Logs

from sigma.backends.duckdb import DuckDBBackend, LogIndex

# Load logs from JSON files
index = LogIndex()
index.load_json_file("logs.json")
# Or load from a directory
index.load_directory("logs/")

# Validate a rule
backend = DuckDBBackend()
result = backend.validate_rule(rule_yaml, index)

print(f"Rule: {result.rule_title}")
print(f"Matches: {result.match_count}/{result.total_logs}")
print(f"Success: {result.success}")

for log in result.matched_logs:
    print(f"  - {log.get('CommandLine', 'N/A')}")

Directory Validation for CI

from sigma.backends.duckdb import validate_rules_directory

# Validate all rules against all logs
results = validate_rules_directory(
    rules_dir="rules/",
    logs_dir="test_logs/",
)

for result in results:
    status = "PASS" if result.has_matches else "FAIL"
    print(f"{status}: {result.rule_title} ({result.match_count} matches)")

Log Format

The backend expects logs in JSON format. By default, it uses Sysmon field names (e.g., CommandLine, Image, ParentImage). You can use custom pipelines for different schemas.

Splunk Sysmon format (default):

{
    "CommandLine": "powershell.exe -e ...",
    "Image": "C:\\Windows\\System32\\powershell.exe",
    "ParentImage": "C:\\Windows\\System32\\cmd.exe"
}

Elastic ECS format (with elastic_ecs pipeline):

{
    "process": {
        "command_line": "powershell.exe -e ...",
        "executable": "C:\\Windows\\System32\\powershell.exe"
    }
}

Pipelines

The backend includes pipelines for common log formats:

from sigma.backends.duckdb import DuckDBBackend
from sigma.backends.duckdb.pipelines import splunk_sysmon, elastic_ecs

# For Splunk with Sysmon TA (default)
backend = DuckDBBackend(processing_pipeline=splunk_sysmon())

# For Elasticsearch with ECS
backend = DuckDBBackend(processing_pipeline=elastic_ecs())

License

MIT