DBF to JSONdBase (DBF)

Legacy database files.

dBase, FoxPro, and Clipper database tables. TabularJS reads the field definitions and the records, and returns the field types, lengths and header details alongside the rows.

A .dbf file is a database table: a header, one 32-byte descriptor per field (name, type, length, decimals), then fixed-length records. It is the table format of dBase, FoxPro and Clipper, and it is still common in GIS work, because the attribute table of every ESRI shapefile is a .dbf file.

TabularJS reads the header and the records straight from the bytes and returns a worksheet with one column per field and one row per record. A DBF file has no formulas or styles, so the result is the data plus metadata about the table.

Feature support

What TabularJS extracts from DBF files

• Columns from the field definitions: title and name hold the field name, type is numeric, text, checkbox or calendar, numeric fields with decimals get a mask, and date fields get options.format.
• Values by field type: Character as trimmed strings, Numeric and Float as numbers, Logical as true/false, Date as YYYY-MM-DD strings, Integer, Double, Currency and Autoincrement as numbers, DateTime as ISO strings. Empty fields are null.
• Deleted records (flagged with *) are skipped. Their count is in meta.deletedRecords and their positions in meta.deletedRecordIndices.
• Header details in meta: dbfVersion, lastModified, totalRecords, activeRecords, languageDriver and a fields array with the original type letter, length and decimals.
• Text fields decoded with the codepage named by the file's language driver byte (for example Windows-1252, Windows-1251 or Shift_JIS), or with the encoding option.

Convert DBF to JSON in Node.js

import tabularjs from 'tabularjs';

const result = await tabularjs('products.dbf');
const table = result.worksheets[0];

// One object per record, keyed by field name
const records = table.data.map((row) =>
  Object.fromEntries(table.columns.map((col, i) => [col.name, row[i]]))
);
console.log(records[0]);
// { ID: 1, NAME: 'Café crème', PRICE: 3.5, ACTIVE: true, CREATED: '2026-01-15' }

console.log(table.meta.fields[2]);
// { name: 'PRICE', type: 'N', typeName: 'Numeric', length: 8, decimals: 2, jsType: 'numeric' }

Read DBF files in the browser

import tabularjs from 'tabularjs';

const input = document.querySelector('input[type=file]');

input.addEventListener('change', async () => {
  // Override the codepage when the header does not declare it
  const result = await tabularjs(input.files[0], { encoding: 'windows-1251' });
  console.log(result.worksheets[0].data);
});

JSON output

{
  "worksheets": [
    {
      "worksheetName": "DBF Data",
      "data": [
        [1, "Café crème", 3.5, true, "2026-01-15"],
        [2, "Croissant", 1.2, false, "2026-01-16"]
      ],
      "columns": [
        { "title": "ID", "name": "ID", "width": 100, "type": "numeric" },
        { "title": "NAME", "name": "NAME", "width": 200, "type": "text" },
        { "title": "PRICE", "name": "PRICE", "width": 100, "type": "numeric", "decimal": "..", "mask": "#,##0.00" },
        { "title": "ACTIVE", "name": "ACTIVE", "width": 100, "type": "checkbox" },
        { "title": "CREATED", "name": "CREATED", "width": 100, "type": "calendar", "options": { "format": "YYYY-MM-DD" } }
      ],
      "meta": {
        "dbfVersion": "dBASE III",
        "totalRecords": 3,
        "activeRecords": 2,
        "deletedRecords": 1,
        "languageDriver": 87,
        "fields": [{ "name": "ID", "type": "N", "typeName": "Numeric", "length": 5, "decimals": 0, "jsType": "numeric" }],
        "deletedRecordIndices": [2]
      }
    }
  ]
}

Output for a three-record dBase III table with one deleted record. meta is trimmed: it also has dbfVersionCode, lastModified, recordLength, headerLength, flags and one fields entry per field.

Options

Limitations

• Memo fields (type M) return the block number stored in the .dbf, not the memo text: the .dbt and .fpt memo files are not read.
• General and Picture fields (types G and P) return a placeholder such as <binary:10bytes>.
• DOS codepages (cp437, cp850, cp852...) need iconv-lite, which is used in Node.js only. In the browser those text fields fall back to Latin-1 unless you pass an encoding that TextDecoder supports.
• Index files (.mdx, .cdx, .ndx) are ignored, and the worksheet is always named DBF Data.

Frequently asked questions

Why TabularJS for dBase (DBF)?

• No external spreadsheet parser dependencies: faster installs, smaller bundles.
• One API in Node.js and modern browsers.
• Output is shape-compatible withJspreadsheet: render parsed data in a live workbook with one call.