Try a JavaScript table with a known result
Open the example and scan it before pressing the button. There is one product row: Desk lamp, SKU 00123, Units 12. Click Load remaining rows, then rescan. The page now contains three product rows, including Cable, USB-C and Notebook A5.
Export the second capture as XLSX. The expected result has three columns and three product rows beneath the header. The Units values are 12, 8, and 20. Compare the workbook with the reference download. This example keeps the earlier rows in the document as it adds new ones.
Identify how the website loads its records
Dynamic describes several different behaviours. The export method depends on what happens to earlier rows when new rows appear.
| Pattern | What happens | What to check |
|---|---|---|
| Load more / appended rows | New rows join the existing table | Rescan after loading; compare the final count |
| Pagination | Next replaces the current page of records | Check whether your capture covers one page or multiple pages |
| Virtualization | The page reuses a small set of rendered rows | Check stable record IDs; scrolling to the end alone is insufficient |
| Filtering | The table shows a selected subset | Compare with the filtered total, not the full dataset |
Capture appended rows after they finish loading
For a Load more button, load the records you need before rescanning. Wait for the last request to finish; clicking capture while a loading indicator is visible can leave out the newest rows.
The extension includes scroll-assisted capture, but a site may scroll an inner panel instead of the whole document. If rows do not appear as expected, interact with the report’s own controls. Inspect the last record as well as the count. A repeated first page can make a count look convincing.
- Set the report’s filters and record its expected total.
- Load the required rows with the website’s controls.
- Rescan and open the selected table in Studio.
- Compare the header, record count, and first and last IDs.
- Export XLSX and check the saved workbook.
Why virtualized tables can lose rows
A virtualized table can show thousands of records while keeping only the current viewport and a small buffer in the document. As you scroll, earlier rows disappear and their elements may be reused for later records. Reaching the bottom does not mean the document contains the entire dataset.
Treat a complete export from a virtualized grid as something to verify on that specific site. Use the site’s native export when available. If you collect smaller ranges, compare stable IDs across the combined file and check for gaps. Removing duplicate rows alone cannot recover records that were never captured.
The JavaScript example above does not simulate virtualization. Its three-row result demonstrates appended-row capture only; it is not evidence that every virtualized application works.
Handle paginated tables separately
A Next button often replaces the current records instead of extending them. For a few pages, separate exports can be practical. Keep the same sorting and filters while collecting them, retain a record ID, and remove repeated headers when combining the files.
For hundreds of pages, look for the site’s full export or an approved API. Table Capture & Extractor does not promise complete automatic pagination on every website. A claim of all rows should always be checked against the source total and the records themselves.
DIV and ARIA grids can still contain usable cells
A JavaScript application does not have to use an HTML table element. Table Capture & Extractor also scans ARIA table and grid structures and repeating DIV layouts. Explicit row and cell roles help describe the structure, but unusual layouts can still need repairs.
Try the DIV example below: it uses ARIA roles and contains the same three products as the dynamic example. Compare the three columns before moving to a more complicated dashboard.
Choose Excel or CSV after checking completeness
XLSX is a useful choice when you want to inspect the result in Excel. Check identifiers such as 00123 and any date-like strings. CSV stores values without spreadsheet cell types, so opening it directly may let Excel reinterpret them.
Keep the export date and filter settings with the file. If a colleague asks why it contains 150 rows instead of 500, those details are more useful than a filename that only says export.