How to Preview Parquet and JSON Files Locally

To preview a Parquet or JSON file locally, open Zro7 JSON / Parquet Viewer and drop it in. DuckDB-WASM reads the file's schema and row groups, shows a preview of the first rows, and lets you run SQL on top. No Python, no PySpark, no VS Code extension — just a browser tab that respects your data.

Parquet: what you get

• Schema — column names, physical types (INT64, BYTE_ARRAY), logical types (STRING, TIMESTAMP), nullability.
• Compression — SNAPPY, ZSTD, GZIP per column, plus dictionary encoding stats.
• Row groups — count and size, useful for judging whether a file is well-partitioned.
• Row preview — first 100 rows rendered as a table.
• Full SQL — run any DuckDB query against the file.

JSON / JSON-lines: what you get

• Auto-detects single JSON object, JSON array, or newline-delimited JSON-lines.
• Inferred schema — DuckDB picks a type per field, promoting to VARCHAR when it sees mixed types.
• Nested fields — accessed with dot notation: SELECT user.email FROM data.
• Flatten to CSV — COPY (SELECT ...) TO 'flat.csv'.

Why local matters here

Parquet files often come from data-warehouse exports — sales, HR, medical, financial. Uploading them to parquet-viewer.example.com to just peek at the schema is the exact anti-pattern the industry moved to Parquet to avoid. DuckDB-WASM lets you inspect any Parquet on your disk in one drag-and-drop, keeping the data governance boundary intact.

Steps

1. Open JSON / Parquet Viewer.
2. Drop the file. Schema and preview show instantly.
3. For deeper queries, click through to SQL Playground — the file is already registered.