filter-stub-application 0.2.12

No project description provided

JSONSim

JSONSim is a synthetic filter that outputs structured JSON events without analyzing image frames. Perfect for testing and debugging pipelines that expect event streams.

Features

• Two Output Modes:
  • Echo Mode: Replays events from a static JSON file
  • Random Mode: Generates synthetic events using JSON Schema templates
• Upstream Data Forwarding: Optionally forwards non-image frames from upstream filters
• Environment Variable Configuration: Easy setup using environment variables
• Debug Logging: Comprehensive logging for troubleshooting
• Sample Data Generation: Automatically creates sample files for quick testing

Quick Start

Using the Usage Script

The easiest way to run the filter is using the provided filter_usage.py script:

# Install dependencies
make install

# Run with default settings (echo mode)
python scripts/filter_usage.py

# Run in random mode
python scripts/filter_usage.py --mode random

# Specify custom output path
python scripts/filter_usage.py --output_path ./my_events.json

Using Environment Variables

Configure the filter using environment variables:

export FILTER_DEBUG=true
export FILTER_OUTPUT_MODE=random
export FILTER_FORWARD_UPSTREAM_DATA=true
export FILTER_OUTPUT_JSON_PATH=./output/events.json
export FILTER_INPUT_JSON_EVENTS_FILE_PATH=./input/events.json
export FILTER_INPUT_JSON_TEMPLATE_FILE_PATH=./input/events_template.json
export VIDEO_INPUT=./data/sample-video.mp4
export WEBVIS_PORT=8000

python scripts/filter_usage.py

Using Make Commands

# Run with default settings
make run

# Run tests
make test

Configuration

Environment Variables

Variable	Description	Default
FILTER_DEBUG	Enable debug logging	false
FILTER_OUTPUT_MODE	Output mode (echo/random)	echo
FILTER_FORWARD_UPSTREAM_DATA	Forward upstream data	true
FILTER_OUTPUT_JSON_PATH	Output file path	./output/output.json
FILTER_INPUT_JSON_EVENTS_FILE_PATH	Input events file	./input/events.json
FILTER_INPUT_JSON_TEMPLATE_FILE_PATH	Input template file	./input/events_template.json
VIDEO_INPUT	Video source	../data/sample-video.mp4
WEBVIS_PORT	Web visualization port	8000

Input File Formats

Echo Mode - JSON Array or JSON Lines:

[
  {"id": "event_1", "type": "sensor", "value": 25.5},
  {"id": "event_2", "type": "alert", "message": "Warning"}
]

Random Mode - JSON Schema:

{
  "type": "object",
  "properties": {
    "id": {"type": "string"},
    "type": {"type": "string", "enum": ["sensor", "alert"]},
    "value": {"type": "number", "minimum": 0, "maximum": 100}
  },
  "required": ["id", "type"]
}

Requirements

To follow these instructions there are a few prerequisites. You must:

• Be authenticated to GAR:

gcloud auth login
gcloud auth application-default login

• Set your gcloud project to plainsightai-prod and configure docker to use gcloud:

gcloud config set project plainsightai-prod
gcloud auth configure-docker us-west1-docker.pkg.dev

It is assumed you will be running this on a GPU. If not then you will have to comment out the deploy: section in the docker-compose.yaml file and the unit test will fail since it compares against GPU numbers.

Install

In order to run the filter locally or build/publish the Python wheel we need to install properly:

virtualenv venv
source venv/bin/activate
make install

Advanced Usage

Custom Event Files

Create your own event files for echo mode:

# Create custom events file
cat > input/my_events.json << EOF
[
  {"id": "custom_1", "type": "sensor", "value": 42.0, "location": "zone_a"},
  {"id": "custom_2", "type": "alert", "message": "Custom alert", "severity": "high"}
]
EOF

# Run with custom events
export FILTER_INPUT_JSON_EVENTS_FILE_PATH=./input/my_events.json
python scripts/filter_usage.py

Custom Schema Templates

Create custom JSON schemas for random mode:

# Create custom schema
cat > input/my_schema.json << EOF
{
  "type": "object",
  "properties": {
    "id": {"type": "string", "pattern": "^custom_[0-9]+$"},
    "type": {"type": "string", "enum": ["sensor", "alert", "status"]},
    "value": {"type": "number", "minimum": 0, "maximum": 1000},
    "timestamp": {"type": "string", "format": "date-time"}
  },
  "required": ["id", "type", "timestamp"]
}
EOF

# Run with custom schema
export FILTER_INPUT_JSON_TEMPLATE_FILE_PATH=./input/my_schema.json
export FILTER_OUTPUT_MODE=random
python scripts/filter_usage.py

Debug Mode

Enable debug logging for detailed information:

export FILTER_DEBUG=true
python scripts/filter_usage.py

Docker Usage

Environment Variables

• Docker-compose automatically reads .env files in the same directory as the compose files. The provided .env.example file can serve as a template to create a .env file.

IMPORTANT! If your filter uses the GPU and make compose doesn't automatically add it to the docker-compose.yaml then make sure to add the following to your filter's section in the compose file:

deploy:
  resources:
    reservations:
      devices:
        - driver: nvidia
          count: all
          capabilities: [gpu]

First, build the filter docker image:

make build-image

If you changed the PIPELINE in the Makefile (if not then skip this step), then rebuild the docker-compose.yaml (you may have to tweak the generated docker-compose.yaml):

make compose

Now run it:

make run-image

Again, navigating to http://localhost:8000 will show you the video.

Testing

Run the comprehensive test suite:

# Run all tests
make test

# Run specific test categories
pytest tests/test_smoke_simple.py -v
pytest tests/test_integration_config_normalization.py -v

The test suite includes:

• Smoke Tests: Basic functionality and end-to-end testing
• Integration Tests: Configuration validation and normalization
• Unit Tests: Individual component testing

Development

VS Code Debugging

Use the provided VS Code launch configuration:

1. Open VS Code in the project directory
2. Go to Run and Debug (Ctrl+Shift+D)
3. Select "JSONSim - Usage Script"
4. Set breakpoints and start debugging

Make Commands

make install    # Install dependencies
make test       # Run tests
make debug      # Run in debug mode
make run        # Run with default settings
make build-image # Build Docker image
make compose    # Generate docker-compose.yaml

Publishing

• Ensure the VERSION file at root has a production semver tag (i.e. v1.2.3)
  • If you intend to release a non-production version such as a development, release candidate or an internal release then add a build number and a classification to your version tag (i.e. v1.2.3.4-dev, v1.2.3.0-rc or v1.2.3.47-int)
• Ensure the version tag of newest entry in RELEASE.md matches the tag in VERSION
  • Important: Our releases are documentation driven. Not updating RELEASE.md will not trigger a release. Filters cannot be merged to main unless RELEASE.md is updated. The RELEASE.md file is validated by our CI and requires version entries to be in the correct descending order.
• Simple merge to main. When a new version is detected in RELEASE.md the CI will:
  • Build and publish the docker image to the GAR OCI registry
  • Build and publish the python wheel to the GAR python registry
  • Push the docs to both production and development documentation sites