Add thin dataset card (ADR-0005)
Browse files
README.md
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
pretty_name: PetroDB
|
| 4 |
+
tags:
|
| 5 |
+
- petroleum
|
| 6 |
+
- reservoir-engineering
|
| 7 |
+
- parquet
|
| 8 |
+
- duckdb
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# PetroDB
|
| 12 |
+
|
| 13 |
+
Public petroleum datasets served as a partitioned **parquet** tree, designed for
|
| 14 |
+
remote columnar queries with DuckDB `httpfs` (HTTP Range pushdown).
|
| 15 |
+
|
| 16 |
+
This Hugging Face dataset repo hosts the **parquet bytes only**. It mirrors the
|
| 17 |
+
`parquet/` root and is queried directly via `resolve` URLs, which honour HTTP
|
| 18 |
+
Range requests so consumers fetch only the row groups a predicate needs.
|
| 19 |
+
|
| 20 |
+
```sql
|
| 21 |
+
INSTALL httpfs; LOAD httpfs;
|
| 22 |
+
SELECT *
|
| 23 |
+
FROM 'https://huggingface.co/datasets/sumpalabs/petrodb/resolve/main/<dataset>/...';
|
| 24 |
+
```
|
| 25 |
+
|
| 26 |
+
The human-facing landing page, per-dataset schema docs, and discovery manifests
|
| 27 |
+
live at **https://petrodb.ocortez.com**.
|
| 28 |
+
|
| 29 |
+
- `main` — public production deployment
|
| 30 |
+
- `stage` — pre-production validation branch
|
| 31 |
+
|
| 32 |
+
See ADR-0005 in the project repository for the hosting rationale.
|