How to convert Excel to JSON for free

rows.page reads the tab-separated text Excel puts on the clipboard and CSV files saved from Excel. It never opens the workbook itself, so to read .xlsx files directly you use DuckDB's excel extension or pandas.

For a range of cells you don't need a new file at all. rows.page reads the paste inside your browser with DuckDB-WASM, and nothing is uploaded unless you press Share.

1. Select the cells in Excel, header row included, and copy them.
2. Open rows.page, click Paste data and paste. The cells arrive tab-separated and open as a table, with the header row detected.
3. Optional: filter with price:>20 or country:NL, sort or hide columns, or run SQL against the data table.
4. Open the Export menu and pick JSON for one array of objects keyed by the header row, or JSONL for one object per line.

Export writes the current view, so filters, sort order and hidden columns all apply, and a SQL result exports up to 1,000,000 rows. Numeric columns stay JSON numbers and empty cells become null. To grab a single record, open its row and use Copy as JSON.

rows.page can't read a workbook. Dropping an .xlsx file on the page won't show your sheet, and pushing one with curl -T returns "This looks like an Excel or zip file. Export it as CSV and push again." Save the sheet as CSV instead:

1. In Excel, choose File > Save As, set the file type to CSV UTF-8 (Comma delimited) and save. Excel saves only the active sheet.
2. Drop the .csv file on rows.page or choose it from your disk.
3. Check the column types in the Columns panel, then export JSON or JSONL.

rows.page opens CSV files up to 1.5 GB per browser tab, and a bigger file loads partially with the note "Loaded the first N rows: the file is X, more than the browser can hold." The CSV sniffer detects comma, semicolon, tab and pipe delimiters, so a semicolon-separated export opens without changes.

Pick CSV UTF-8 over plain CSV (Comma delimited). A file that isn't valid UTF-8 is read as Latin-1 with the note "Read as Latin-1 text.", and one saved as Windows-1252 with a euro sign, curly quotes or long dashes won't open at all ("Couldn't open this data").

For many workbooks, or to skip the CSV step, read the .xlsx file directly in a script.

DuckDB

DuckDB 1.2 and later reads and writes .xlsx files with the excel extension; old .xls files are not supported. This reads Sheet1 and writes a JSON array:

INSTALL excel;
LOAD excel;
COPY (SELECT * FROM read_xlsx('book.xlsx', sheet = 'Sheet1')) TO 'out.json' (FORMAT json, ARRAY true);

Leave out sheet to read the first sheet, and leave out ARRAY true to get JSONL. Number cells load as DOUBLE, so an ID of 1 is written as 1.0; select CAST(id AS BIGINT) AS id to keep whole numbers. Text cells keep their leading zeros, and date cells become ISO dates such as 2019-04-01.

pandas

pandas reads .xlsx files through openpyxl (pip install pandas openpyxl). orient='records' gives an array of objects keyed by the column names, and date_format='iso' writes ISO 8601 timestamps instead of the default epoch milliseconds such as 1554076800000:

import pandas as pd

df = pd.read_excel('book.xlsx')
df.to_json('out.json', orient='records', date_format='iso')

read_excel turns digit-only text such as 02134 into the number 2134, even when the cell is stored as Text in Excel. Pass dtype={'zip': str} for ZIP codes and IDs to keep the zeros.

Whether you paste cells or save a CSV, check these five things before exporting JSON:

• Displayed values: copying takes what the cell shows, not the stored number. A cell holding 1234.5678 formatted as currency pastes as the text $1,234.57, so clear number formats first if you need exact numbers.
• Dates: copied dates arrive in your display format. A date column shows min and max dates in the Columns panel; if yours stayed text, convert it with strptime() in SQL.
• Leading zeros: Excel drops them from cells stored as numbers, which breaks ZIP codes and IDs. If they still show in Excel (a Text column or a custom format like 00000), they survive the copy and rows.page keeps 02134 as text. If they are gone, pad them back with SELECT lpad(CAST(zip AS VARCHAR), 5, '0') AS zip, * EXCLUDE (zip, __rid) FROM data.
• One sheet at a time: a clipboard copy, a CSV Save As and each read_xlsx call all cover a single sheet.
• Merged cells: a merged range keeps its value only in the top-left cell, so the other rows come out empty or null. Unmerge and fill the values down before you copy.