Supported Formats

ExploreMyData supports the most common tabular data formats.

FormatExtensionsNotes
CSV.csvComma-separated. Auto-detects delimiter, encoding, and headers.
TSV.tsvTab-separated. Treated as CSV with tab delimiter.
Excel.xlsx, .xlsSupports multi-sheet workbooks. Choose which sheet to import during upload. Parsed via the xlsx library.
Parquet.parquetColumnar format. Native DuckDB support - fastest to load.
JSON.jsonArray of objects or newline-delimited JSON (NDJSON).
PDF.pdfExtracts tables from text-based PDFs. Scanned PDFs supported via browser-based OCR. Password-protected PDFs supported.
XML.xmlNested elements are flattened into columns via a JSON intermediate step.
Text.txtAuto-detects delimiters (tab, comma, pipe). Plain text imported as single column.
Word.doc, .docxExtracts tables from Word documents. Multiple tables loaded as separate files.
DuckDB.duckdb, .dbNative DuckDB database files. Tables inside are listed for selection on import.
SQLite.sqlite, .sqlite3SQLite database files. Tables inside are listed for selection on import.
Arrow.arrow, .featherApache Arrow IPC files. Types load with full fidelity, no sniffing.
Avro.avroAvro container files. Dates, timestamps, and decimals convert on import.
Compressed.gz, .bz2, .zst, .zip, .tar, .tgzArchives that contain one of the formats above. The file inside is extracted and loaded.

Export Formats

After transforming your data, export in any of these formats:

FormatExtensionNotes
CSV.csvComma-separated values.
TSV.tsvTab-separated values.
Excel.xlsxExcel spreadsheet via the xlsx library. Limited to 500,000 rows.
Parquet.parquetColumnar binary format. Smallest file size for large datasets.
JSON.jsonOne array of objects, pretty-printed. Limited to 500,000 rows.
JSON Lines.jsonlNewline-delimited JSON (NDJSON). One object per line.
XML.xmlA <rows> document with one <row> per row and one element per column. Limited to 500,000 rows.
PDF.pdfFormatted table layout. Limited to 10,000 rows.
HTML.htmlSelf-contained HTML table with dark mode support. No external dependencies. Limited to 500,000 rows.

In an XML export, each element name comes from the column name. Characters other than letters, digits, underscore, dot, and hyphen become underscores, and a name that cannot start an XML name gets an underscore in front of it. A null value is written as an empty element, so every row keeps the same shape.

File Size

Files up to ~100 MB work well in most browsers. Larger files may work but will use more memory. Parquet files are the most efficient format for large datasets.

Encoding

DuckDB auto-detects file encoding. If you see garbled characters, try saving your file as UTF-8 before uploading.