fileflow-agent 0.5.2

Scheduled file transfer agent for Local, SFTP, S3 (AWS / DO Spaces / R2 / MinIO), and HDFS (hadoopcli or WebHDFS) — with a web dashboard, processing pipeline, and dedup tracking

FileFlow Agent

A modular, scheduler-driven data transfer platform built with Python. FileFlow automates the movement of files between configurable storage backends with support for cron scheduling, processing pipelines, deduplication, backup and retention policies.

Features

• Multi-backend connectors — Local filesystem, SFTP, S3-compatible (AWS S3, DigitalOcean Spaces, Cloudflare R2, Backblaze B2, MinIO), HDFS (via hadoopcli + auto kinit -kt, or WebHDFS REST), SCP
• Per-job, fully editable connections — Each job carries its own connection block, so you can run multiple SFTPs, multiple S3 buckets across different providers, and multiple HDFS clusters side by side
• Cron scheduling — APScheduler with per-job cron expressions
• Processing pipeline — Compress (gzip / bzip2 / zip), decompress, prefix-rename, timestamp or chain a custom bash script
• Deduplication — SQLite-backed tracking prevents duplicate transfers across runs
• Reliable backup & retention — Configurable backup directories with automatic strict retention cleanup (NFS/CIFS-aware)
• Transfer verification — Size match, checksum, or existence-only checks
• Web dashboard — Responsive 'Soft UI' for realtime monitoring + form-based job editing (no YAML required); reload-on-save
• REST API — Health checks, transfer stats, job listing, and log streaming
• OIDC-published wheels — Distributed via PyPI Trusted Publishing (no long-lived tokens)

Connectors

Connector	Auth	Notes
local	n/a	Filesystem source/destination
sftp	password or key (incl. ssh-agent / ~/.ssh/id_*)	Per-job host/port/user; auto mkdir -p on remote
s3	access_key+secret (per-job or env-var fallback)	Custom endpoint_url for any S3-compatible service; UI ships with provider presets for AWS, DigitalOcean Spaces (NYC3/SFO3/AMS3/FRA1/SGP1/BLR1/SYD1), Cloudflare R2, Backblaze B2, and MinIO; multipart upload via boto3 managed transfer
hdfs	Kerberos via kinit -kt <keytab> <principal> (auto-refreshed per op), or ambient ticket cache, or unsecured ?user.name=	CLI transport (default): drives hadoopcli over stdin (copyFromLocal, copyToLocal, mkdir, ls, rm); reads core-site.xml/hdfs-site.xml from the host. WebHDFS transport (opt-in): pure REST (op=CREATE + 307 → datanode), no Hadoop client needed
scp	stub	Reserved for future implementation

Architecture

├── configs/                # YAML job definitions
│   ├── jobs.yaml
│   └── test_jobs.yaml
├── src/fileflow_agent/
│   ├── api/                # FastAPI endpoints + dashboard serving
│   ├── config/             # Pydantic models and settings loader
│   ├── connectors/         # Source/Destination connector implementations
│   ├── logging/            # Structured rotating logger
│   ├── processing/         # File processing pipeline
│   ├── scheduler/          # APScheduler integration
│   ├── services/           # Transfer, backup, retention, verification
│   ├── static/             # Dashboard frontend (HTML/CSS/JS)
│   ├── tracking/           # SQLite transfer history & deduplication
│   ├── utils/              # Checksum utilities
│   └── main.py             # Application entrypoint
├── test_*.py               # Unit and integration tests
├── .env.example
├── run.sh                  # Easy startup script
├── pyproject.toml
├── requirements.txt
└── README.md

Getting Started

Prerequisites

• Python 3.10+
• pip

Installation & Workspace Setup

FileFlow Agent is designed as a standalone global Pip library. When you install it, it gives your system a new command-line tool fileflow.

# 1. Install via Pip (In a virtual environment or globally)
pip install fileflow-agent

# 2. Initialize a secure Workspace
# This creates localized databases, configuration templates, and log directories.
fileflow init ~/my_fileflow_workspace

# 3. Start the Agent from the configured workspace
fileflow start ~/my_fileflow_workspace --port 7345

Once running, open http://localhost:7345 to access the Neumorphic monitoring dashboard.

Configuration

The fileflow init command will automatically scaffold a .env and configs/jobs.yaml in your chosen workspace directory.

1. Environment Config (~/my_fileflow_workspace/.env) Set your UI authentication credentials and global AWS/SFTP master keys if needed.

2. Job Config (~/my_fileflow_workspace/configs/jobs.yaml) (You can edit this file manually, or configure jobs entirely from the Web Dashboard without touching YAML!)

YAML job definitions look like this — everything is also editable from the dashboard, so you don't have to hand-write any of it:

jobs:
  # AWS S3 (default — endpoint omitted, region us-east-1)
  - job_id: daily_backup_aws
    enabled: true
    schedule: "0 */6 * * *"
    source:
      type: local
      path: /data/incoming
      file_pattern: "*.csv"
    destination:
      type: s3
      path: archive/csv
      connection:
        bucket: my-bucket
        region: us-east-2
        # access_key/secret_key omitted -> falls back to AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY env
    processing:
      enabled: true
      steps: [compress, timestamp]
    backup:
      enabled: true
      location: backups/daily
      retention_days: 30
    verification:
      method: size_match

  # DigitalOcean Spaces (or any S3-compatible: just set endpoint_url)
  - job_id: nightly_to_do_spaces
    enabled: true
    schedule: "0 2 * * *"
    source: { type: local, path: /var/log/app, file_pattern: "*.log.gz" }
    destination:
      type: s3
      path: app-logs
      connection:
        endpoint_url: "https://nyc2.digitaloceanspaces.com"
        region: nyc3
        bucket: my-space
    verification: { method: size_match }

  # Local -> HDFS via hadoopcli (auto kinit -kt; production-friendly)
  - job_id: events_to_hdfs
    enabled: true
    schedule: "*/5 * * * *"
    source: { type: local, path: /var/tmp, file_pattern: "*.gz" }
    destination:
      type: hdfs
      path: /external/
      connection:
        transport: cli                              # default; reads core-site.xml from the host
        principal: "[e-mail]"
        keytab: "/home/example/example.kt"        # auto kinit -kt before each upload
        krb5ccname: "FILE:/tmp/k5cc_example"
        hadoop_cli: hadoopcli
        overwrite: true                             # rm -skipTrash, then copyFromLocal
    verification: { method: size_match }

The built-in Neumorphic web dashboard provides:

View	Description
Overview	Transfer stats (total, success, failed, duplicates) and recent transfer table
Configuration	Form-based job editor — add, edit, delete jobs and reload the scheduler live
System Logs	Real-time log viewer with auto-refresh

API Endpoints

Method	Path	Description
GET	/health	Health check
GET	/jobs	List configured jobs
GET	/transfers	Recent transfer records
GET	/stats/summary	Aggregated transfer statistics
GET	/logs/recent	Recent log entries
GET	/api/config	Read raw YAML config
POST	/api/config	Save config and reload scheduler

Extending Connectors

Implement SourceConnector or DestinationConnector from connectors/base.py and register in connectors/factory.py:

from fileflow_agent.connectors.base import SourceConnector

class MySourceConnector(SourceConnector):
    def list_files(self, path, pattern=None):
        ...

    def download_file(self, remote_path, local_path):
        ...

    def get_metadata(self, remote_path):
        ...

Contributing

Contributions are welcome. Please open an issue first to discuss what you'd like to change.

1. Fork the repository (https://github.com/emoncse/fileflow)
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

This project is open source and available under the MIT License.