> ## Documentation Index
> Fetch the complete documentation index at: https://sourcetable.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Import and export

> Import virtually any file format and export your work in multiple formats.

Sourcetable handles almost any file format you can throw at it. Upload files via drag-and-drop, the toolbar, AI chat, or Google Drive.

## Supported import formats

### Spreadsheet formats

| Format            | Extension       | What's preserved                                                             |
| ----------------- | --------------- | ---------------------------------------------------------------------------- |
| **Excel**         | `.xlsx`, `.xls` | Formulas, formatting, multiple sheets, named ranges, cell styles             |
| **CSV**           | `.csv`          | Auto-detects delimiters (comma, semicolon, pipe, tab), encoding, and headers |
| **TSV**           | `.tsv`          | Tab-separated values with automatic header detection                         |
| **Google Sheets** | via Drive       | Import directly from connected Google Drive                                  |

### Data formats

| Format      | Extension  | How it's handled                                                         |
| ----------- | ---------- | ------------------------------------------------------------------------ |
| **JSON**    | `.json`    | Nested objects are flattened into columns. Arrays become rows            |
| **JSONL**   | `.jsonl`   | Line-delimited JSON — each line becomes a row                            |
| **Parquet** | `.parquet` | Columnar format, handles multi-GB files efficiently via DuckDB streaming |
| **XML**     | `.xml`     | Parsed into tabular structure based on element hierarchy                 |

### Document formats

| Format         | Extension       | How it's handled                                                       |
| -------------- | --------------- | ---------------------------------------------------------------------- |
| **PDF**        | `.pdf`          | Extracts tabular data from tables. OCR for scanned documents           |
| **Word**       | `.docx`, `.doc` | Extracts tables and structured content                                 |
| **Plain text** | `.txt`          | AI infers structure from patterns (fixed-width, delimited, log format) |

### Specialized formats

| Format          | How it's handled                                                             |
| --------------- | ---------------------------------------------------------------------------- |
| **SQL dumps**   | Parses CREATE TABLE and INSERT statements into structured tables             |
| **Log files**   | AI recognizes common log formats (Apache, nginx, syslog) and structures them |
| **HTML tables** | Extracts `<table>` elements from HTML files                                  |

## How to import

<Steps>
  <Step title="Drag and drop">
    Drag any supported file directly into your workbook. Sourcetable auto-detects the format and imports the data.
  </Step>

  <Step title="Upload button">
    Click the upload button in the toolbar or sidebar to browse and select files.
  </Step>

  <Step title="AI chat">
    Attach a file in the AI chat to analyze it without importing to the grid. You can also ask the AI to import it.
  </Step>

  <Step title="Google Drive">
    Connect your Google Drive and import files directly from your cloud storage.
  </Step>

  <Step title="URL import via AI">
    Paste a URL in the AI chat and ask it to import the data. The AI fetches, parses, and loads the data.
  </Step>
</Steps>

## Auto-detection

When you import a file, Sourcetable automatically detects:

* **Character encoding** — UTF-8, Latin-1, Windows-1252, and other encodings
* **Delimiters** — comma, semicolon, tab, pipe, or custom delimiters
* **Headers** — whether the first row contains column names
* **Data types** — numbers, dates, currencies, percentages, text
* **Date formats** — US (MM/DD/YYYY), European (DD/MM/YYYY), ISO (YYYY-MM-DD), and other formats

## Large file handling

Sourcetable uses DuckDB for streaming large files that exceed memory:

* Files over **1,048,576 rows** are handled with streaming queries
* Parquet files up to **10 GB** can be processed directly
* The AI automatically selects the best approach based on file size

See [large file analysis](/data/large-files) for more details.

## Exporting data

Export your workbook or specific sheets:

| Format   | Best for                                                                       |
| -------- | ------------------------------------------------------------------------------ |
| **XLSX** | Sharing with Excel users — preserves formulas, formatting, and multiple sheets |
| **CSV**  | Raw data for other tools, databases, or scripts                                |
| **JSON** | Structured data for APIs and applications                                      |
| **PDF**  | Formatted reports for sharing and printing                                     |
| **Doc**  | Document format for reports and presentations                                  |

### AI-powered export

Ask the AI to export specific subsets or transformations:

* "Export only the rows where revenue is over \$10,000 as CSV"
* "Create a PDF summary report of this quarter's metrics"
* "Export this sheet as JSON with nested objects for each department"

## Connecting live data

For data that needs to stay current, use [connectors](/connectors/overview) instead of file imports. Connected data sources refresh automatically, while imported files are static snapshots.