What is Parquetbay?
Parquetbay is a free online Parquet viewer, editor and converter — and the same workbench opens CSV, TSV, JSON, NDJSON, Arrow and Excel files. It is built for analysts, developers and data teams who need to inspect or reshape a tabular dataset without installing desktop software or sending the file to a conversion service.
- Cost
- Free; no account required
- Processing
- Local, inside the browser tab
- Requirements
- JavaScript, WebAssembly and available device memory
How do you open and convert a Parquet or CSV file?
-
Choose one or more files
Drop or select a .parquet, .csv, .json, .arrow or Excel file. Parsing happens in the browser; file contents are not uploaded to Parquetbay.
-
Inspect, edit or query
Browse the virtualized grid, edit cells, find and replace values, profile columns, build undoable transform steps, or query loaded tables with DuckDB SQL.
-
Export a new file
Choose Parquet, CSV, TSV, JSON, NDJSON, Arrow or XLSX, or a text representation such as Markdown, SQL, HTML, YAML, XML or LaTeX. The original file on disk is not modified.
What the Parquet viewer shows you
A .parquet file is binary and columnar, so nothing on a normal computer displays it. Parquetbay reads the file's own schema and renders it as a table you can actually work in.
Schema and types
Every column with the type stored in the file — integers, decimals, timestamps, booleans, and nested struct, list and map columns — not types guessed from text.
Row count and paging
The total row count from the file metadata, and a grid that fetches only the rows on screen, so a file with millions of rows scrolls at the same speed as one with a hundred.
Per-column profile
Null count, distinct count, min, max and mean, the most frequent values and a histogram — the questions you would otherwise answer with a notebook.
Sort, filter, search
Sort any column, quick-filter to matching rows, and find a value across the table before deciding whether to edit, query or export.
The same viewer opens CSV, TSV, JSON, NDJSON, Arrow and Excel, which is useful when the question is "what is in this file?" rather than "convert this file". There is no separate viewer URL; this page is it.
How this differs from an upload-based converter
Most "online Parquet" sites are one of two things: a form that uploads your file to a server and returns a download, or a read-only viewer. Each row below is something you can check yourself in a minute.
| Claim | Parquetbay | How to check |
|---|---|---|
| Your file leaves your device | No | Open the browser's developer tools at the Network panel, then open and export a file: no request carries your rows |
| Export is free with no size gate or account | Yes | There is no sign-in, no plan page and no watermark; the export button downloads directly |
| Edit cells, rows and columns | Yes | Double-click a cell; every change is listed in the journal and can be undone |
| SQL over the file | DuckDB, in the tab | Open the SQL console and run a join or a window function on the loaded table |
| Excel both ways | Read .xlsx/.xls, write .xlsx | Open a workbook; each non-empty sheet becomes a tab and can be exported as Parquet |
| Works on a laptop with no install | Yes, any modern browser | The engine is DuckDB compiled to WebAssembly, fetched once and cached |
View, edit and query Parquet, CSV, JSON and Excel in one workbench
The interface covers the full local workflow: inspect a file, make auditable changes, reshape the data and download the result.
Virtualized data grid
Only the visible rows are rendered in the page, which keeps table navigation responsive without creating one DOM node per cell.
Undoable cell edits
Edit, add or delete rows and columns. Changes are recorded in a journal that can be reviewed, reverted or undone.
Transform pipeline
Filter, sort, derive, rename, cast and reshape columns with steps that can be disabled, reordered or removed.
DuckDB SQL console
Run joins, window functions, aggregates and other DuckDB SQL, then open query results as another dataset tab.
Find and replace
Search one column or the whole table and replace matching values in bulk.
Multiple files and sheets
Open multiple datasets as tabs. Excel workbooks expose their non-empty sheets as separate tabs.
Column profiling
Review types, nulls, distinct counts and distributions before deciding how to clean an unfamiliar dataset.
Format conversion
Read and write the supported tabular formats, with Snappy, Zstandard or uncompressed Parquet output.
Which file formats can Parquetbay open and export?
Parquetbay supports common analytical, interchange and spreadsheet formats. An input can be exported to another supported output format.
| Format | Extensions | Open | Export | Notes |
|---|---|---|---|---|
| Apache Parquet | .parquet · .pq | Yes✓ | Yes✓ | Snappy, Zstandard or uncompressed output |
| CSV | .csv | Yes✓ | Yes✓ | Delimiter and header detection |
| TSV | .tsv · .tab | Yes✓ | Yes✓ | Tab-separated text |
| JSON | .json | Yes✓ | Yes✓ | Structured JSON records |
| NDJSON | .ndjson · .jsonl | Yes✓ | Yes✓ | One JSON record per line |
| Apache Arrow | .arrow · .feather · .ipc | Yes✓ | Yes✓ | Arrow IPC data |
| Excel | .xlsx · .xls | Yes✓ | Yes✓ | Reads workbook sheets; exports XLSX |
Parquet and Arrow signatures, JSON structure and text delimiters are inspected when an extension is missing or misleading. Text exports also include Markdown, SQL, HTML, YAML, XML and LaTeX.
Does Parquetbay upload your files?
Selected files are not uploaded. Parquetbay reads, queries and exports their contents inside your browser using DuckDB compiled to WebAssembly. The site has no account system, analytics or advertising trackers.
Your browser still downloads the website's code and other static assets, and the DuckDB engine is loaded when it is first needed. Parquetbay does not install a service worker, so a later reload normally requires a network connection. The saved theme preference is the only application setting stored in your browser.
How large a Parquet or CSV file can a browser open?
There is no honest universal file-size limit. Capacity depends on the browser, available device memory, file encoding, column types and the operation being performed. A compressed file can require substantially more memory after decoding.
The grid renders only the visible region, and DuckDB can scan columnar data efficiently. However, edits, transforms and exports may require additional working memory. For large or sensitive jobs, start with the original file preserved, close memory-heavy tabs, and validate the exported result before relying on it.
Technology and format references
Parquetbay uses documented open data technologies rather than a proprietary file format:
- DuckDB-Wasm documentation—the analytical SQL engine used in the browser.
- Apache Parquet overview—the specification and project documentation for the columnar format.
- Apache Arrow overview—the cross-language columnar memory and interchange project.
Product information reviewed .
Frequently asked questions
What is an Apache Parquet file?
Apache Parquet is an open, column-oriented data file format designed for efficient storage and retrieval. A text editor cannot display its table directly, so it needs a compatible reader such as Parquetbay, DuckDB, Spark, pandas or Polars.
How do I view a Parquet file without installing software?
Open this page in a compatible browser, select the .parquet file and wait for the local DuckDB engine to read its schema and rows. The file contents stay on your device.
How do I convert CSV to Parquet or Parquet to CSV?
Open the source file, choose Export and select the destination format. Parquet output supports Snappy, Zstandard and uncompressed encoding; CSV output can include a header row.
How do I convert a Parquet file to CSV in Python instead?
One line with pandas: pd.read_parquet("data.parquet").to_csv("data.csv", index=False), with pyarrow installed. The Parquet to CSV page shows the pyarrow, Polars and DuckDB equivalents and when a script beats doing it here.
Can Excel open a Parquet file?
Not directly. Some Microsoft 365 builds reach Parquet through Power Query, but the simplest route is to convert Parquet to XLSX here and open the result in any version of Excel, LibreOffice or Google Sheets.
Can I edit a Parquet file online?
Yes. You can edit cells, add or delete rows and columns, find and replace values, and transform the table with pipeline steps or SQL. Export creates a new result; it does not overwrite the original file on disk.
How do I open a CSV file without Excel?
Drop the .csv file on this page. Parquetbay shows it in a spreadsheet-like grid where you can sort, filter, edit, query and export it—no installation, account or upload. The same works for TSV, JSON, NDJSON, Arrow and Parquet datasets.
Can I open a CSV file that is too large for Excel?
Often, yes. Excel cannot display more than 1,048,576 rows per sheet, while Parquetbay renders rows in a virtualized grid and queries them with DuckDB, so larger datasets can stay workable. The practical ceiling depends on your device's memory and the file's encoding.
Is my data sent to a server?
Parquetbay does not send selected file contents to a Parquetbay server. Reading, queries, edits and exports run in the browser. Network requests still retrieve the site's static application assets.
Does Parquetbay require a connection after loading?
The active page can continue processing after its required assets have loaded, but Parquetbay does not install an offline cache. Opening or reloading the site normally needs a network connection.
Does Parquetbay work with Excel files?
Yes. Parquetbay reads non-empty sheets from .xlsx and .xls workbooks and exports XLSX, so you can move data between spreadsheets, Parquet and other supported formats.
Common Parquet tasks
Jump straight to a task. Each opens this same local workbench — nothing is uploaded. Opening and viewing any supported format is done right here on this page.