OPC Replay Server

A Python-based OPC UA replay server for replaying historical timestamped tag data from CSV or Parquet files. Perfect for testing, simulation, and development scenarios where you need to reproduce specific OPC UA data sequences.

Features

• 🔄 Replay Historical Data - CSV/Parquet files with configurable speed (1x to 100x+), offset, and loop mode
• 🏷️ Auto-Generate NodeSets - Create OPC UA address space automatically from your data files
• 🎯 Real-Time Tag Injection - HTTP REST API for overriding values on-the-fly with scheduled/delayed activation
• 🛡️ Robust & Compliant - Proper OPC UA timestamps, namespace remapping, graceful error handling

Installation

From PyPI (Recommended)

pip install opc-replay

Requires Python 3.13 or later.

From Source (For Development)

git clone https://github.com/mieslep/opc-replay.git
cd opc-replay
pip install -e .

With uv (Recommended for Development)

git clone https://github.com/mieslep/opc-replay.git
cd opc-replay
uv sync

For development setup and contribution guidelines, see CONTRIBUTING.md.

Quick Start

30-Second Demo

The easiest way to get started - auto-generate the NodeSet from your data:

# Start replay server (auto-generates NodeSet and loops continuously)
opc-replay --data examples/simple-data.csv --ts-col TS --auto-nodeset --loop

# In another terminal, monitor with the OPC client
opc-replay-client

# In another terminal, inject an override
opc-replay-inject --tag "ns=2;s=Temperature" --value 99.9 --duration 30

With Pre-Generated NodeSet

If you have an existing NodeSet XML:

opc-replay \
    --nodeset examples/simple-nodeset.xml \
    --data examples/simple-data.csv \
    --ts-col TS \
    --loop

Core Commands

opc-replay - Replay Server

Start an OPC UA server that replays historical data:

# Basic usage with auto-generated NodeSet
opc-replay --data mydata.csv --ts-col TIMESTAMP --auto-nodeset

# With existing NodeSet
opc-replay --nodeset nodeset.xml --data mydata.csv --ts-col TS

# Advanced options
opc-replay \
    --data mydata.parquet \
    --ts-col TIMESTAMP \
    --auto-nodeset \
    --speed 10 \
    --loop \
    --offset 3600 \
    --api-port 8080

Key Options:

• --data FILE - CSV or Parquet data file
• --ts-col COLUMN - Name of timestamp column
• --auto-nodeset - Auto-generate NodeSet from data (recommended)
• --nodeset FILE - Use existing NodeSet XML
• --speed N - Playback speed multiplier (default: 1.0)
• --loop - Loop playback forever
• --offset N - Skip first N seconds
• --api-port PORT - HTTP API port for tag injection (default: 8080, 0 to disable)
• --log-level LEVEL - Logging verbosity: DEBUG, INFO (default), WARNING, ERROR
• --allow-non-canonical - Allow non-canonical NodeIds without auto-conversion (advanced)

opc-replay-inject - Tag Injection

Override tag values in real-time while the server is running:

# Single tag injection
opc-replay-inject --tag "ns=2;s=Temperature" --value 99.9 --duration 30

# Delayed injection (starts after 10 seconds)
opc-replay-inject --tag "ns=2;s=Pressure" --value 200.5 --offset 10 --duration 20

# Batch injection from file
opc-replay-inject --file injections.json

# List active overrides
opc-replay-inject --list

# Clear all overrides
opc-replay-inject --clear

opc-replay-client - Monitoring Client

Connect to an OPC UA server and monitor tag changes with multiple monitoring modes and output formats:

# Subscribe to all changes (default - push-based, efficient)
opc-replay-client

# Connect to remote server
opc-replay-client --endpoint opc.tcp://192.168.1.100:4840/

# Read nodemap statistics and exit
opc-replay-client --read-nodemap --all-namespaces

# CSV output with periodic summaries
opc-replay-client --format csv --report-frequency 30 > output.csv

# JSON output
opc-replay-client --format json > output.json

# Polling mode (legacy - periodic reads)
opc-replay-client --read-mode poll --poll-interval 2.0

# Show connection statistics
opc-replay-client --show-stats --report-frequency 15

# Monitor specific namespace
opc-replay-client --namespace 2

Monitoring Modes:

• --read-mode subscribe (default) - Push-based subscriptions, efficient for thousands of tags
• --read-mode poll - Periodic polling, legacy mode

Output Formats:

• --format console (default) - Human-readable tables
• --format csv - CSV rows (timestamp,node_id,browse_name,value)
• --format json - Line-delimited JSON (JSONL)

Additional Options:

• --read-nodemap - Print nodemap statistics and exit (no monitoring)
• --all-namespaces - Browse all namespaces (ns=0,1,2,...)
• --report-frequency N - Quiet mode: show summaries every N seconds
• --show-stats - Display connection statistics (uptime, reads, changes, errors)
• --namespace N - Monitor specific namespace index
• --namespace-uri URI - Monitor by namespace URI

Data Format

Your CSV or Parquet file needs these columns:

Example CSV:

TAGNAME,TAGVALUE,DATATYPE,TS
ns=2;s=Temperature,20.5,Float,2026-01-01T10:00:00Z
ns=2;s=Pressure,101.3,Float,2026-01-01T10:00:00Z
ns=2;s=Flow,15.2,Float,2026-01-01T10:00:00Z

Auto-Conversion of NodeIds

By default, opc-replay automatically converts non-canonical TAGNAMEs to canonical OPC UA format:

# Input data with mixed formats:
TAGNAME,TAGVALUE,DATATYPE,TS
Temperature,20.5,Float,2026-01-01T10:00:00Z           # Auto-converted to: ns=2;s=Temperature
ns=2;s=Pressure,101.3,Float,2026-01-01T10:00:00Z      # Already canonical, used as-is
PET001.Flow,15.2,Float,2026-01-01T10:00:00Z           # Auto-converted to: ns=2;s=PET001.Flow

Canonical Format: ns=<namespace>;[isgb]=<identifier>

• ns= - Namespace index (e.g., ns=2)
• Identifier type: s= (string), i= (integer), g= (GUID), b= (bytestring)
• Example: ns=2;s=Tank.Level.Current

Benefits:

• Works with legacy data that uses simple tag names
• No need to pre-process your CSV files
• Proper OPC UA compliance by default

Advanced: Use --allow-non-canonical to disable auto-conversion if your OPC UA client supports non-standard NodeIds.

Use Cases

Testing & Development

• Replay specific sequences at high speed for rapid testing
• Simulate sensor failures with tag injection
• Test alarm logic by injecting extreme values

Integration Testing

• Reproduce bugs by replaying exact data sequences
• Validate HMI/SCADA behavior with known data patterns
• Load testing with continuous loop mode

Training & Demos

• Demonstrate systems without live equipment
• Interactive control via tag injection API
• Scenario replay with historical data

HTTP REST API

The built-in REST API enables programmatic control from any language:

# Inject a tag override
curl -X POST http://localhost:8080/inject \
  -H "Content-Type: application/json" \
  -d '{"tagname": "ns=2;s=Temperature", "value": 99.9, "duration_s": 30}'

# List active overrides
curl http://localhost:8080/inject

# Clear all overrides
curl -X DELETE http://localhost:8080/inject

Documentation

• examples/ - Sample data files and demonstration scripts
• CHANGELOG.md - Version history and release notes

Architecture

┌─────────────────┐
│   Data File     │  CSV/Parquet with timestamped tag values
│ (CSV/Parquet)   │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  opc-replay     │  Replay server with HTTP API
│   (Server)      │  • Auto-generates NodeSet
└────────┬────────┘  • Replays data at configurable speed
         │           • Exposes OPC UA endpoint
         │           • HTTP REST API for tag injection
         ├──────────────────────┬─────────────────────┐
         ▼                      ▼                     ▼
┌─────────────────┐    ┌─────────────────┐   ┌─────────────────┐
│ opc-replay-     │    │ opc-replay-     │   │  OPC UA Client  │
│   client        │    │   inject        │   │ (UAExpert, etc) │
└─────────────────┘    └─────────────────┘   └─────────────────┘

Requirements

• Python >= 3.13
• Dependencies: opcua, pandas, pyarrow (auto-installed)

Contributing

Contributions welcome! See CONTRIBUTING.md for development setup, testing guidelines, and how to submit pull requests.

License

See LICENSE file for details.

Credits

• Tag injection feature developed by Branislav
• Built with python-opcua