Database Viewer
Open SQLite databases, CSV and TSV files, SQL dumps, JSON, Excel (.xlsx/.xls), OpenDocument and dBase (.dbf) files of any size in a fast, searchable table. Files are read in slices straight from your disk, so a 20 GB file opens as fast as a 20 KB one. Nothing is uploaded.
How to Use
- Drop a file in — or several at once, or press Paste data to paste CSV or JSON text. SQLite (.db, .sqlite), CSV/TSV, SQL dumps (.sql), JSON and JSON Lines, Excel .xlsx/.xls, OpenDocument .ods and dBase .dbf all open directly, and so do those files inside a .zip or .gz.
- Rows appear straight away. Big files keep indexing in the background while you scroll; the status line says how far it has got.
- Pick a table on the left. Type in the search box to find text in any column, or use the column filter (equals, greater than, contains, is empty, regex…). Click a column header to sort.
- Click any cell to see its full value — long text, pretty-printed JSON, or a stored image or file you can download. Double-click a cell (or press F2) to edit it; edits are highlighted and go into every export.
- Export what you see — filtered and sorted — to CSV, CSV for Excel, TSV, JSON, JSON Lines or SQL INSERT statements.
Open a database or data file of any size — SQLite, CSV, SQL dumps, JSON, Excel, dBase — and browse, search, sort and export it. The file is read in slices from your own disk, so a 20 GB export opens as fast as a small one. Nothing is uploaded.
.db and its -wal file together, or a .dbf with its .fpt/.dbt memo file.
What it opens
| Format | Extensions | What you get |
|---|---|---|
| SQLite | .db .sqlite .sqlite3 .db3 .s3db .gpkg .mbtiles .anki2 — or no extension | Every table and its row count, views and virtual tables with their SQL, indexes and triggers, WITHOUT ROWID and generated columns, stored images and files. A -wal file dropped alongside is applied. |
| Delimited text | .csv .tsv .tab .psv .txt .dat | Comma, tab, semicolon or pipe — detected. Quoted fields with commas and line breaks inside them, UTF-8, UTF-16 and legacy code pages. |
| SQL dumps | .sql .dump | mysqldump and phpMyAdmin, pg_dump (INSERT or COPY), SQLite .dump, SQL Server scripts. One table per CREATE TABLE or INSERT target. Nothing is executed. |
| JSON | .json .geojson | An array of records becomes a table; an object holding several arrays becomes several tables. Nested objects flatten into address.city columns. |
| JSON Lines | .jsonl .ndjson | One record per line — logs, analytics and data-pipeline exports. |
| Excel | .xlsx .xlsm .xls | Every sheet, with dates shown as dates rather than serial numbers, and formula results rather than formulas. |
| OpenDocument | .ods | Every sheet from LibreOffice Calc or Google Sheets exports. |
| dBase / FoxPro / Clipper | .dbf (+ .dbt / .fpt memo) | All field types, deleted records flagged rather than hidden, DOS code pages 437 and 850 decoded. |
| Compressed | .zip .gz | Unpacked in the browser, then opened as whatever is inside. |
Why size does not matter
Worked Example
The built-in CSV example. It has a header and four records but occupies six lines, because record 2 has a line break inside a quoted field. The first record, 1,"Smith, Ana",Austin,"Said ""urgent"" on the phone", still splits into four fields: a comma inside quotes does not separate fields, and a doubled quote "" means one literal quote mark. The viewer shows 4 rows and 4 columns.
Scanning a big file. A 20 GB CSV read at 200 MB per second needs 20 × 1,024 ÷ 200 ≈ 102 s for one full pass (search or sort), or about 68 s at 300 MB/s and 205 s at 100 MB/s. Scrolling is quick regardless, because the viewer jumps to the nearest of its checkpoints, one for every 64 rows.
The common mistake: counting lines. Counting lines with a text editor or wc -l gives 6 for that example, not 4 rows, because quoted line breaks do not end a record. NULL (nothing stored) and an empty string are also different things: a filter for "is empty" matches both, "is NULL" only the first.
Show Work
Formulas
Where These Formats Come From
Comma-separated text is older than personal computers and had no single standard for decades; RFC 4180 (2005) wrote down the common rules about quoting and doubled quotes. dBase files date from the early 1980s and were copied by FoxPro and Clipper, which is why .dbf files from old accounting software are still in circulation. SQLite, first released in 2000, is a single-file database whose format is documented and stable, which is why phone apps, browsers and games use it for their data.
These formats were designed for programs that read part of a file at a time, so a viewer that does the same can open files far larger than memory. Spreadsheets are different: Excel files are zip archives of XML, so they are streamed out of the archive and decoded as they arrive.
About the Database Viewer
Most online database and CSV viewers load the whole file into the browser first, which is why they stop working somewhere between 100 MB and 1 GB. This one never does. It reads your file the way a database engine reads its own storage: a slice at a time, only the parts it needs, straight from your disk. Rows appear as soon as the first few kilobytes are read, and the rest of the file is indexed in the background while you work.
Each format has its own reader, all running in a background thread so the page never freezes. SQLite databases are read page by page by walking their B-trees. Text formats (CSV, TSV, JSON Lines, SQL dumps and JSON arrays) are scanned once, quote-aware, noting where every 64th row starts; to show row 30,000,000 the viewer jumps to the nearest noted position and parses forward a few rows. dBase files have fixed-size records, so any row is one read away. Search, filters and sorting run over the whole file, not just the rows on screen, and matches appear while the scan is still going.
Everything stays in your browser. Related tools: File Inspector, JSON ⇄ CSV Converter, and SQL Formatter.
Frequently Asked Questions
Is there really no file size limit?
The tool never loads your file into memory. It reads the part it needs, when it needs it, straight from your disk, the same way a database engine does. What it keeps is a small index (one position for every 64 rows), so a 20 GB CSV costs a few megabytes of memory. The practical limits are your disk and your patience: searching or sorting has to read every row once, and reading runs at roughly 100–300 MB per second. Sorting is capped at 50 million rows, because it has to hold one sort key per row.
Is my data uploaded anywhere?
No. Everything happens inside your browser tab, in a background worker. There is no server step, so customer lists, medical exports, app databases and financial records never leave your computer. Close the tab and nothing remains.
How does it open a SQLite file, and why does mine look empty?
SQLite files have a well-documented page format: a header, then fixed-size pages arranged as B-trees, one tree per table. The viewer walks those trees directly, reading the schema from page 1 and then only the pages that hold the rows on screen. If a database looks empty or out of date it is probably in WAL (write-ahead log) mode, where recent changes sit in a second file called yourfile.db-wal. Drop both files at once and the viewer layers the log over the main file, just as SQLite would; the File tab says when a database is in WAL mode.
Can I run SQL queries?
Not yet. You can fix individual cells (double-click one, type the new value, and the edit goes into every export), but there is no query engine. Search, column filters and sorting cover most "find the row" jobs, and they work on files far too large for an in-browser SQL engine. SQL dumps are never executed either: the viewer reads the INSERT and COPY statements as data.
Which formats are not supported?
Microsoft Access (.mdb/.accdb), Parquet, Arrow/Feather, Avro and binary Excel (.xlsb) are recognised and explained but not opened. Encrypted SQLite (SQLCipher, used by many phone apps) cannot be read without its key. A .sqsh or .sqfs file is a SquashFS filesystem image, not a database; extract it with 7-Zip first, then open any database inside it here.
How do I use the Database Viewer?
Just type your numbers. The answer shows up right away — there is no button to press. Change anything and it updates by itself.
Does it cost anything or need an account?
No. The tool is completely free, there is no account to create, and it keeps working offline after the page first loads.
Is anything I type uploaded?
No. The tool works entirely on your device, so the values you enter never leave your browser.
Common Use Cases
A CSV too big for Excel
Excel stops at 1,048,576 rows and most text editors choke long before a gigabyte. This scrolls a 50-million-row export and searches it.
Look inside an app database
Browser history, messaging backups, game saves, Anki decks, GeoPackage and MBTiles maps — all SQLite underneath.
Check a SQL backup without restoring it
Browse the tables in a mysqldump, pg_dump or SQLite .dump file without installing a database server.
Rescue old dBase and FoxPro data
Read .dbf tables from discontinued accounting and inventory systems, memo fields included, with DOS code pages decoded properly.
Convert between formats
Open any supported file and export it — or just the rows you filtered — as CSV, JSON, JSON Lines or SQL.
Find the one bad row
Filter a column for empty values, out-of-range numbers or a regex pattern to find what broke an import.
Last updated: