Filedge
Reliable file ingestion for data engineers. Filedge loads files into your warehouse with per-file atomic commits, content-hash idempotency, crash-safe retry, and a full audit trail — so a failed load never leaves you guessing what made it in.
Get started in 5 minutes See it end to end
The problem
Most ETL pipelines fail in the worst possible way: they write half the rows, then crash, leaving the destination corrupt with no record of what happened. Re-running risks double-writing the rows that already succeeded.
How Filedge is different
Three guarantees, enforced on every file:
- Atomic commits — no partial loads. Each destination connector makes a file's write retry-safe for its content hash. A crash mid-load never leaves half the rows behind.
- Content-hash idempotency — no accidental re-ingestion. Files are identified by SHA-256 of their content, not filename. Renaming a file doesn't reload it; replacing it with new content does.
- A full audit trail — always know what loaded. Every file moves through a
PENDING → PROCESSING → COMMITTED/FAILEDstate machine, stored alongside row-level provenance you can query and export.
The boundary is always the File: anything upstream — file drops, API pulls,
queues, database exports — becomes a complete File in a watched directory, and
filedge run ingests it the same audited way.
30-second example
filedge inspect data.csv --output pipeline.yaml # 1. infer the schema
# add format, dest_table, and a connector: block
filedge validate data.csv --config pipeline.yaml # 2. dry-run, no data written
filedge run --dir ./incoming --config pipeline.yaml \
--audit-db-url sqlite:///filedge.db # 3. ingest with audit trail
# Committed: 1 Failed: 0 Skipped: 0 New: 1
Prefer an interactive flow? filedge author data.csv walks you through
inference, schema review, write mode, connector, and validation, then writes a
ready-to-run pipeline folder. See the Author guide.
Destinations
Pluggable via a connector: block in pipeline.yaml:
- SQLite — local dev and lightweight deployments
- PostgreSQL — production workloads
- BigQuery — GCP data warehouses
- Databricks — Databricks SQL warehouses
- Snowflake — cloud data warehouse workloads
- DuckDB — local analytics
Where to go next
| If you want to… | Go to |
|---|---|
| Install and run your first pipeline | Getting Started |
| See the full value end to end | Tutorials |
| Get a specific job done | How-to guides |
| Look up a config option or command | Reference |
| Understand how it works under the hood | Architecture |