Use the Extract and Delete Columns Tool for a specific table change
This page helps you keep only selected columns or remove selected columns while retaining every data row in its original order. It is useful when a spreadsheet export, database extract, analytics file, or command-line result has the right values but the wrong table shape. Paste a representative sample first or choose a local CSV, TSV, or text file. The source stays unchanged in the input area, and the transformed result appears separately for review.
Paste a customer export and keep name, city, and signup date for a review copy, or delete an internal notes column before sharing a non-sensitive sample. Start with a few rows whose values you already know. A small example makes it easier to confirm field boundaries and row counts before processing a larger table. The page does not upload the source or result to NexaCurrent.
Choose delimiters, headings, and the operation
Select columns with exact headings or numbered positions such as #2. Commas and new lines can separate several selections. Duplicate headings must be selected by position so the choice is unambiguous. CSV, tabs, semicolons, pipes, and one-character custom delimiters share the same quoted-field rules. A delimiter inside a quoted value remains part of that value, doubled quotation marks become one quotation mark, and a line break inside quotes remains inside the cell.
Tell the page whether the first row contains headings. With headings, you can select a unique column by name; a position such as #2 always selects the second column and is required when headings repeat. Without headings, generated positions keep the operation explicit. The page stops on inconsistent row widths instead of shifting later values into the wrong fields.
Read the preview and recipe before downloading
Count the result headings and inspect the first and last source rows. Extraction must preserve the requested order; deletion must preserve the relative order of every remaining column. The result area displays the selected operation, heading policy, source and destination delimiters, and safe-export choice. It previews no more than thirty lines so a large paste does not create an enormous page, while Copy and Download use the complete validated result.
Undo result removes the current transformation without changing the pasted source, which lets you revise one setting and try again. Reset workspace clears the source, selected file, instructions, result, warnings, and download, then restores comma input, comma output, heading mode, and safe export. Downloads use CSV for commas, TSV for tabs, and plain text for other separators.
Handle spreadsheet formulas and malformed rows safely
Spreadsheet programs can interpret text beginning with =, +, -, or @ as a formula when a delimited file is opened. Safe export is enabled by default and adds a visible apostrophe to such cells. The warning reports how many cells changed. Turn it off only when the destination treats every field as plain data and you intentionally need the original first character.
An empty source, an unclosed quote, text after a closing quote, an unexpected quotation mark, mismatched row widths, an invalid custom delimiter, or an unavailable column produces a visible error and no download. A new failed attempt clears an older result so it cannot be mistaken for fresh output. Duplicate headings remain visible and must be addressed with numbered positions.
Limits, privacy, and a practical final check
Removing a column is not anonymization. Names, free text, IDs, and combinations of remaining values can still identify a person or confidential record, so review the actual result before sharing it. Input is limited to 1,000,000 characters, 20,000 data rows, 200 columns, and 200,000 cells so an accidental paste cannot exhaust the tab. Files above that boundary should be processed with a streaming desktop or command-line tool. Browser memory can also be tighter on a phone, so use a desktop for large tables.
Before replacing an import file, compare record and column counts, inspect empty and multiline values, search for a known first and last record, and open a downloaded copy in the intended application. Keep the original until the destination accepts the result. A successful transformation confirms the rules shown on this page; it cannot verify a private schema, business rule, access policy, or downstream calculation.