DD3-LoRa-Bridge-MultiSender/REPUBLISH_README.md

DD3 MQTT Data Republisher

Standalone Python script to recover and republish lost meter data from SD card CSV files to MQTT.

Features

• Rate-limited publishing: Sends 5 messages/second by default (configurable) to prevent MQTT broker overload
• Two modes of operation:
  • Auto-detect: Connect to InfluxDB to find gaps in recorded data
  • Manual selection: User specifies start/end time range
• Cross-platform: Works on Windows, macOS, and Linux
• CSV parsing: Reads SD card CSV export format and converts to MQTT JSON
• Interactive mode: Walks user through configuration step-by-step
• Command-line mode: Scripting and automation friendly

Installation

Prerequisites

• Python 3.7 or later

Setup

# Install dependencies
pip install -r requirements_republish.txt

Optional: InfluxDB support

To enable automatic gap detection via InfluxDB, influxdb-client will be automatically installed. If you want to use the fallback manual mode only, you can skip this (though it's included in requirements).

Usage

Interactive Mode (Recommended for first use)

python republish_mqtt.py -i

The script will prompt you for:

1. CSV file location (with auto-discovery)
2. Device ID
3. MQTT broker settings
4. Time range (manual or auto-detect from InfluxDB)

Command Line Mode

Republish a specific time range:

python republish_mqtt.py \
  -f path/to/data.csv \
  -d dd3-F19C \
  --mqtt-broker 192.168.1.100 \
  --mqtt-user admin \
  --mqtt-pass password \
  --from-time "2026-03-01" \
  --to-time "2026-03-05"

Auto-detect missing data with InfluxDB:

python republish_mqtt.py \
  -f path/to/data.csv \
  -d dd3-F19C \
  --mqtt-broker 192.168.1.100 \
  --influxdb-url http://localhost:8086 \
  --influxdb-token mytoken123 \
  --influxdb-org myorg \
  --influxdb-bucket smartmeter

Different publish rate (slower for stability):

python republish_mqtt.py \
  -f data.csv \
  -d dd3-F19C \
  --mqtt-broker localhost \
  --rate 2  # 2 messages per second instead of 5

CSV Format

The script expects CSV files exported from the SD card with this header:

ts_utc,ts_hms_local,p_w,p1_w,p2_w,p3_w,e_kwh,bat_v,bat_pct,rssi,snr,err_m,err_d,err_tx,err_last

Note: ts_hms_local is the local time (HH:MM:SS) in your configured timezone, not UTC. The ts_utc field contains the Unix timestamp in UTC.

Each row is one meter sample. The script converts these to MQTT JSON format:

{
  "id": "F19C",
  "ts": 1710076800,
  "e_kwh": "1234.56",
  "p_w": 5432,
  "p1_w": 1800,
  "p2_w": 1816,
  "p3_w": 1816,
  "bat_v": "4.15",
  "bat_pct": 95,
  "rssi": -95,
  "snr": 9.25
}

How It Works

Manual Mode (Fallback)

1. User specifies a time range (start and end timestamps)
2. Script reads CSV file
3. Filters samples within the time range
4. Publishes to MQTT topic: smartmeter/{device_id}/state
5. Respects rate limiting (5 msg/sec by default)

Auto-Detect Mode (with InfluxDB)

1. Script connects to InfluxDB
2. Queries for existing data in the specified bucket
3. Identifies gaps (time ranges with no data)
4. Shows gaps to user
5. Republishes the first (oldest) gap from CSV file
6. User can re-run to fill subsequent gaps

Rate Limiting

By default, the script publishes 5 messages per second. This is:

• Safe for most MQTT brokers (no risk of overload)
• Fast enough (fills data in < 5 minute for typical daily data)
• Adjustable with --rate parameter

Examples:

• --rate 1: 1 msg/sec (very conservative)
• --rate 5: 5 msg/sec (default, recommended)
• --rate 10: 10 msg/sec (only if broker can handle it)

Device ID

The device ID is used to determine the MQTT topic. It appears on the device display and in the CSV directory structure:

• Example: dd3-F19C
• Short ID (last 4 characters): F19C

You can use either form; the script extracts the short ID for the MQTT topic.

Time Format

Dates can be specified in multiple formats:

• 2026-03-01 (YYYY-MM-DD)
• 2026-03-01 14:30:00 (YYYY-MM-DD HH:MM:SS)
• 14:30:00 (HH:MM:SS - uses today's date)
• 14:30 (HH:MM - uses today's date)

Examples

Scenario 1: Recover data from yesterday

python republish_mqtt.py -i
# Select CSV file → dd3-F19C_2026-03-09.csv
# Device ID → dd3-F19C
# MQTT broker → 192.168.1.100
# Choose manual time selection
# From → 2026-03-09 00:00:00
# To → 2026-03-10 00:00:00

Scenario 2: Find and fill gaps automatically

python republish_mqtt.py \
  -f path/to/csv/dd3-F19C/*.csv \
  -d dd3-F19C \
  --mqtt-broker mosquitto.example.com \
  --mqtt-user admin --mqtt-pass changeme \
  --influxdb-url http://influxdb:8086 \
  --influxdb-token mytoken \
  --influxdb-org myorg

Scenario 3: Slow publishing for unreliable connection

python republish_mqtt.py -i --rate 1

Troubleshooting

"Cannot connect to MQTT broker"

• Check broker address and port
• Verify firewall rules
• Check username/password if required
• Test connectivity: ping broker_address

"No data in CSV file"

• Verify CSV file path exists
• Check that CSV has data rows (not just header)
• Ensure device ID matches CSV directory name

"InfluxDB query error"

• Verify InfluxDB is running and accessible
• Check API token validity
• Verify organization name
• Check bucket contains data

"Published 0 samples"

• CSV file may be empty
• Time range may not match any data in CSV
• Try a wider date range
• Check that CSV timestamps are in Unix format

Performance

Typical performance on a standard PC:

• CSV parsing: ~10,000 rows/second
• MQTT publishing (at 5 msg/sec): 1 day's worth of data (~2800 samples) takes ~9 minutes

For large files (multiple weeks of data), the script may take longer. This is expected and safe.

Advanced: Scripting

For automation, you can use command-line mode with environment variables or config files:

#!/bin/bash
# Recover last 3 days of data

DEVICE_ID="dd3-F19C"
CSV_DIR="/mnt/sd/dd3/$DEVICE_ID"
FROM=$(date -d '3 days ago' '+%Y-%m-%d')
TO=$(date '+%Y-%m-%d')

python republish_mqtt.py \
  -f "$(ls -t $CSV_DIR/*.csv | head -1)" \
  -d "$DEVICE_ID" \
  --mqtt-broker mqtt.example.com \
  --mqtt-user admin \
  --mqtt-pass changeme \
  --from-time "$FROM" \
  --to-time "$TO" \
  --rate 5

License

Same as DD3 project

Support

For issues or feature requests, check the project repository.