Context.devOne API to scrape, enrich, and extract the internet
Promoted
File SQL
Query local & S3 files with SQL — inside VS Code.
15 followers
Query local & S3 files with SQL — inside VS Code.
15 followers
Every analyst has stared at a folder of Parquet or CSV files and thought: "I just want to SQL this." File SQL does exactly that — inside VS Code. Drop in a file path, an S3 URI, or a whole folder. Each subfolder becomes a table. Query it with DuckDB. Done. Why it stands out: Local files and S3 — one extension, one workflow, Right-click any file in Explorer → instant SQL table, Multi-tab query editor with autocomplete, Runs entirely on your machine
Does the right-click → table thing handle nested Parquet partitions automatically, or do you have to flatten the folder structure first?
@dilararuhw Thanks for checking this out. Right now, it doesn't auto-handle nested Hive-style partitions. It groups by leaf subfolder. Will take it as feedback.
Does the query results pane let you export back out to Parquet or CSV directly, or do you have to pipe through another tool for that round trip?
@resulsz4y It's already on the roadmap. Will be coming in upcoming releases — planning to add programmatic export using COPY as well. Thanks for checking this out!
Love this, being able to right-click a Parquet file and query it right in VS Code saves so much context switching. One thing that would make it even better: add a way to persist and share connection configs and saved queries per project, so my whole team can run the same SQL against the same folder structure without each person setting it up from scratch.
@ilk_l13311 Thanks for the feedback. Yeah, this is a really good point. Team setup so everyone runs the same SQL on the same folders is definitely where this needs to go next. Adding it to the roadmap.
Quick update — shipped a bunch of what you all asked for in the launch thread:
Shareable workspace config — check a .filesql file into git, everyone on the team gets the same tables and saved queries
Hive-partitioned datasets — year=/month=/day= folders are auto-detected as a single partitioned table
S3 streaming — large Parquet files are range-read directly from S3 via DuckDB's httpfs, no more full downloads
Export results — one click to save any query result as CSV or Parquet
Thanks again for all the feedback that shaped this. More coming.
Right-clicking a Parquet file in the explorer and getting an instant SQL table feels like the obvious thing that should have existed years ago. Really clean workflow for quick local analysis.
@kemalcakandg84 Thanks a lot! Yeah, that's exactly the itch I built it to scratch — was tired of spinning up a notebook every time. Really appreciate you checking it out.