Pilotbase
Data tools

Free tools for SQL and data work

Most run entirely in your browser, so your SQL and files stay on your machine. No account needed. Sign in to set up downtime alerts and keep a history of every run.

sqlBeautify screenshot: Format, highlight and fix SQL as you type
sqlBeautifySQL

Format, highlight and fix SQL as you type

Paste, type or drop a .sql file. Get it formatted with syntax highlighting in a light or dark editor, common mistakes fixed (smart quotes, trailing commas, misspelled keywords and functions, missing brackets) with every change highlighted, and parse errors pointed out by line.

sqlCompare screenshot: Diff two SQL files like git, after formatting both
sqlCompareSQL

Diff two SQL files like git, after formatting both

Load SQL into a left and right panel (drop files or paste), beautify both so layout noise disappears, and see a side-by-side or unified git-style diff. Completely different files show the right side all in green. Schema dumps from pg_dump or mysqldump are recognised and compared as written, with the noise that differs in every dump (restrict keys, SET lines, counters, optionally owners and grants) left out.

downtimeAlert screenshot: We check your URL, API or stats page every 30 seconds and email you when it's down
downtimeAlertMonitoring

We check your URL, API or stats page every 30 seconds and email you when it's down

Add a public URL, API or stats endpoint (like https://your-site.com/api/stats) and we check it every 30 seconds from our own server. Your email is verified with a sign-in code, and alerts can also go to another address you confirm with a code. You get one email when it goes down and one when it's back.

sqlTranslate screenshot: Convert SQL between 14 dialects
sqlTranslateSQL

Convert SQL between 14 dialects

Translate queries and DDL between MySQL, PostgreSQL, SQL Server, Oracle, SQLite, Snowflake, BigQuery, DuckDB, Redshift, Spark, Databricks, Hive, Trino and ClickHouse: functions, quoting, LIMIT/TOP and types.

explainViewer screenshot: See where a Postgres query spends its time
explainViewerSQL

See where a Postgres query spends its time

Paste EXPLAIN or EXPLAIN (ANALYZE, BUFFERS) output from PostgreSQL, as text or JSON. Get the plan as a tree with the time spent in each node (loops included), the slowest node highlighted, and warnings for bad row estimates, filters that throw rows away, sorts or hashes that spill to disk, writes whose WAL is mostly full-page images, and trigger time that no node shows, such as a foreign key whose child table has no index. Paste a second plan to compare before and after, node by node and trigger by trigger, so an index on a foreign key shows up in the trigger time.

keysetBuilder screenshot: Keyset pagination queries from your ORDER BY
keysetBuilderSQL

Keyset pagination queries from your ORDER BY

Paste the ORDER BY of a list or an API endpoint and get the keyset (seek) pagination queries for PostgreSQL, MySQL 8 or SQLite: the first page, the next page from the last row's values, the index that makes every page one index range, and a cursor to hand the client instead of page numbers. Mixed ASC and DESC and nullable columns are written out the long way, because a row comparison like (a, b) > (x, y) is only right when every column sorts the same way and none can be NULL. Every query it writes was checked on the real engines.

testData screenshot: Realistic fake rows for your tables, as one INSERT for Postgres, MySQL or SQLite
testDataSQL

Realistic fake rows for your tables, as one INSERT for Postgres, MySQL or SQLite

Paste your CREATE TABLE statements and get one INSERT ... SELECT per table that fills them with rows shaped like production: amounts and foreign keys skewed so a few customers place most orders, statuses with weights, dates busier on weekdays and growing towards today, times of day with quiet nights, names, example.com emails, unique codes and a NULL rate per column. Each column is guessed from its name and type, and every choice can be changed. The random numbers come from the row number and a seed, so the same seed gives the same rows on every run and on all three engines, and the CSV preview shows exactly what the database will get. Child tables are filled after their parents, with keys that exist. Every query it writes was run on the real engines. Runs in your browser.

statementsRanker screenshot: Find the Postgres or MySQL queries that cost the most time in total
statementsRankerSQL

Find the Postgres or MySQL queries that cost the most time in total

Run one query on pg_stat_statements, paste the result (CSV, tab-separated, JSON or psql's table output) and get every statement ranked by its share of the database's total time, not by mean time, which puts a rare slow report above the update that runs 50,000 times. Each row shows calls, mean time, rows per call, disk reads, cache hits and WAL per call, with flags for the shapes worth a second look: many one-row calls (N+1 from app code), a single-row lookup that takes milliseconds (a missing index), slow statements, heavy disk reads, sorts that spill to disk and WAL-heavy writes. Click a query for EXPLAIN to copy into explainViewer. Reads PostgreSQL 12 and older column names too. On MySQL 8, paste performance_schema's statement digest summary instead: the same ranking, with rows examined per row returned and full-scan counts to point at the missing index. Runs in your browser.

lockTree screenshot: Who is blocking whom in Postgres, and the line that frees the queue
lockTreeSQL

Who is blocking whom in Postgres, and the line that frees the queue

When writes hang, run one query on pg_stat_activity with pg_blocking_pids(), paste the result (CSV, tab-separated, JSON or psql's table output) and see the blocking tree: the root blocker at the top, every waiting session under the one it waits for, the lock mode each wants and on which table, how long each has waited and how long the root's transaction has been open. It writes the line to run on the root: pg_terminate_backend for a session idle in transaction (cancelling does nothing there), pg_cancel_backend first for one that is still running. It warns when an ALTER TABLE waiting for its lock has queued every other query behind it, and when sessions wait on each other. Runs in your browser.

sqlInspector screenshot: Check a query before you run it: what it touches and the traps in it
sqlInspectorSQL

Check a query before you run it: what it touches and the traps in it

Paste SQL from an AI chat, a colleague or an old script. It is parsed for your database's dialect (syntax errors with line and column), read back in plain words (what it reads, what it changes), and checked for the mistakes that run cleanly and still give a wrong result: UPDATE or DELETE without WHERE, sums over a join that counts rows twice, BETWEEN on timestamps, NOT IN with a subquery, whole-number division, a LEFT JOIN undone by WHERE. Every table and column is listed so you can tick them against your schema, and the EXPLAIN statement for the next step is ready to copy into explainViewer. Paste a PostgreSQL migration and it also flags the index and constraint changes that take a long lock on a big table (CREATE or DROP INDEX without CONCURRENTLY, CONCURRENTLY inside a transaction, SET NOT NULL, a foreign key or CHECK without NOT VALID), each with the lock it takes and the safer form. Paste a MySQL migration and each ALTER TABLE is marked INSTANT, INPLACE or COPY the way MySQL 8.4 runs it (a COPY stops every write until it ends), with the ALGORITHM and LOCK clause that makes MySQL refuse instead of copying.

csvProfiler screenshot: Column types, nulls, distinct values and stats for any CSV
csvProfilerData prep

Column types, nulls, distinct values and stats for any CSV

Drop a CSV or Excel file and get a profile of every column: inferred type, empty and distinct counts, min, max, mean, median, standard deviation, percentiles and top values.

csvCleaner screenshot: Trim, dedupe, fix dates and find near-duplicate rows
csvCleanerData prep

Trim, dedupe, fix dates and find near-duplicate rows

Clean a CSV in your browser: trim spaces, drop empty rows and columns, remove exact duplicates, flag near-duplicates by edit distance, normalise dates to ISO and change case. Download the result.

csvEncoding screenshot: Fix é showing as é: find a CSV's encoding and save it as UTF-8
csvEncodingData prep

Fix é showing as é: find a CSV's encoding and save it as UTF-8

Drop a CSV or text file and see how it is really encoded: UTF-8 with or without a byte order mark, UTF-16 from Excel's Unicode Text, or a Windows-1252 export that a database refuses (with the line and byte where it would stop). Text that was UTF-8, read as Windows-1252 and saved again (é as é, ’ as ’) is repaired line by line, leaving correct rows alone. Download clean UTF-8, with a byte order mark for Excel or without one for Postgres, MySQL and scripts. Runs in your browser.

csvDates screenshot: Is 03/04/2025 March or April? Read a CSV's dates and rewrite them as ISO 8601
csvDatesData prep

Is 03/04/2025 March or April? Read a CSV's dates and rewrite them as ISO 8601

Drop a CSV and each date column is read the way it was written: day/month or month/day proven by values like 15/04 or by a weekday that only fits one reading, never guessed, so a column where every value reads both ways asks you which it is (with a hint when the rows are in date order only one way). Excel serial numbers (1900 and 1904 systems, without Lotus's 29 February 1900), Unix seconds and milliseconds, two-digit years with Excel's or PostgreSQL's cut-off, times, AM/PM and offsets are handled. Download the file with YYYY-MM-DD dates, in place or as new columns, and a list of every value it couldn't decide. Runs in your browser.

csvToExcel screenshot: Turn a CSV into an Excel file that keeps leading zeros, long IDs and real dates
csvToExcelData prep

Turn a CSV into an Excel file that keeps leading zeros, long IDs and real dates

Opening a CSV in Excel quietly changes it: 00501 becomes 501, a 19-digit order ID becomes 1.84930275123457E+18 and loses its last digits, MARCH1 and 1-2 become dates, 2E5 becomes 200000. Drop the CSV here and every column is typed once from all its values: codes with leading zeros, IDs over 15 digits, phone numbers and part codes stay text exactly as written; plain numbers become numbers; ISO dates and date-times become date cells. Each column says why, what Excel would have done to it, and lets you change the type. Download an .xlsx that opens the same everywhere. Runs in your browser.

pivotTable screenshot: Group by, aggregate and pivot a CSV
pivotTableData prep

Group by, aggregate and pivot a CSV

Pick group-by columns and aggregates (count, sum, mean, median, min, max, distinct) or pivot one column into many, with a total row and column and empty cells as blank or 0. Download the result as CSV, or copy the same report as SQL for PostgreSQL, MySQL, SQLite or DuckDB.

csvDiff screenshot: Compare two CSVs by key: added, removed and changed rows
csvDiffData prep

Compare two CSVs by key: added, removed and changed rows

Load two versions of a CSV, match rows on one to three key columns, and see which rows were added, removed or changed, with the changed cells highlighted and a count of changes per column. Duplicate keys are flagged, and extra spaces, letter case or 12.50 vs 12.5 can be ignored.

excelSplit screenshot: Make a file that's too big for Excel fit in Excel
excelSplitData prep

Make a file that's too big for Excel fit in Excel

Excel shows only the first 1,048,576 rows of a file, and saving it drops the rest. Open the whole export here (CSV, TSV, Parquet or JSON lines, several GB are fine), tick the columns you need, keep only the rows you want, and get Excel files back: one file if it fits now, a file per million rows, a file per month, store or account, or a summary with counts and totals. Dates stay dates and numbers stay numbers. The file never leaves your computer: it's read from your disk and the Excel files are written inside your browser tab. No formulas or SQL needed.

bigCsv screenshot: Open a CSV too big for Excel and query it with SQL
bigCsvConvert & query

Open a CSV too big for Excel and query it with SQL

Too big for Excel? Open it here. Files of several GB open in your browser and are never uploaded: row count, column types, the first 100 rows and a profile of every column. Then run any SQL you like (GROUP BY, window functions, regex) on DuckDB and download results as a smaller CSV. No SQL? excelSplit cuts the same file into Excel files with a few clicks.

pdfTables screenshot: PDF table to Excel or CSV, in your browser
pdfTablesConvert & query

PDF table to Excel or CSV, in your browser

Drop a bank or card statement, a supplier invoice or any PDF report and get its table as an Excel file or a CSV. The PDF is read inside your browser tab and never uploaded. A table that runs over several pages comes out as one, descriptions that wrap onto a second line are joined back to their row, amounts like (12.00), 12.00 CR or £1,234.56 become real numbers and dates become real dates, so SUM and filters work in Excel straight away. Works on PDFs that carry text (the ones a bank or accounting system makes); scans and photos of pages aren't read.

monthlyBook screenshot: Shopify, Square and ads exports into one table by week or month
monthlyBookConvert & query

Shopify, Square and ads exports into one table by week or month

Drop the month's exports from your shop: Shopify orders, Square transactions and item sales, Meta Ads and Google Ads reports. Each file is recognised by its columns and you get one table by week or by month: orders, sales, shipping, tax, tips, refunds, Square fees, ad spend, ad spend per sale and what's left after refunds, fees and ads. The usual traps of merging these by hand are handled: Shopify's one row per item counts each order once, Square's items file doesn't add its tax a second time, an order placed late in the evening stays on the day the store saw it, and a file downloaded twice is counted once. Download Excel (with daily, item and check sheets) or CSV. The files are read in your browser and never uploaded.

amountSplit screenshot: Split a total across CSV rows without a rounding-cent mismatch
amountSplitData prep

Split a total across CSV rows without a rounding-cent mismatch

Drop a CSV or Excel file, choose the column that supplies each row's weight, and allocate shipping, a discount, a payment or any other total. The result works in integer minor units and gives leftover units to the largest remainders, with input order breaking exact ties, so the rounded row amounts always add back to the total. Zero weights and negative totals are supported. Download CSV or Excel, or copy the equivalent PostgreSQL, MySQL 8 or SQLite query. Runs in your browser.

erasureMap screenshot: Find every table and column that holds a user, from your CREATE TABLE statements
erasureMapSQL

Find every table and column that holds a user, from your CREATE TABLE statements

Paste your schema (pg_dump, mysqldump or SQLite .schema), pick the users table, and see every foreign key that points at it, directly or through other tables, with its ON DELETE rule and whether the child column has an index. Columns that look like a user reference or personal data but have no foreign key (user_id, actor_id, created_by, email, phone, ip_address) are listed apart. Then copy a children-first delete and anonymise script for PostgreSQL, MySQL or SQLite, plus the leftover-count queries to run afterwards. Everything is read in your browser tab.

sqlOnCsv screenshot: Run SQL queries on CSV, JSON and Excel files
sqlOnCsvConvert & query

Run SQL queries on CSV, JSON and Excel files

Load one or more files as tables and query them with SQL (joins, GROUP BY, window-free aggregates) without a database. Nothing is uploaded.

csvToSql screenshot: CREATE TABLE and INSERT statements from a CSV
csvToSqlConvert & query

CREATE TABLE and INSERT statements from a CSV

Infer column types from a CSV, change any of them (keep ZIP and phone numbers as text, a date as DATE rather than TIMESTAMP), and generate CREATE TABLE plus batched INSERT statements for PostgreSQL, MySQL, SQL Server or SQLite, with values quoted to match. European Excel exports work as they are: semicolons and the sep=; line are read, and numbers written 1.234,56 or 1,234.56 go in as plain numbers.

dataConverter screenshot: CSV, JSON, Excel, YAML and XML, any to any
dataConverterConvert & query

CSV, JSON, Excel, YAML and XML, any to any

Convert between CSV, TSV, JSON, Excel (.xlsx), YAML and XML in your browser. Nested JSON is flattened for CSV and Excel. Long IDs (over 2^53, like 19-digit snowflake IDs) stay exact and go into Excel as text.

jsonTools screenshot: Repair, format, query with JSONPath and validate with JSON Schema
jsonToolsConvert & query

Repair, format, query with JSONPath and validate with JSON Schema

Fix broken JSON (single quotes, trailing commas, comments, missing quotes), pretty-print or minify it, query it with JSONPath, and validate it against a JSON Schema. Numbers too long for JavaScript, like 19-digit IDs, are kept digit for digit.

Also from the makers of Pilotbase

b59 book · b59.link

Sell sessions. Get paid into your own account.

A booking page in your own colours: one-off sessions, packs or monthly plans, times shown in each visitor's zone, reminders, and customers who move their own bookings. Payments go straight to your own Stripe, Razorpay or Cashfree account.

Make a booking page
b59 book panel with four plans, a row of days, times shown in New York time with Kolkata time under each and a Book this time button