CSV Column Calculator

The CSV column calculator adds a result beside each source record. Use it when an order export has unit price and quantity but no amount, when you need a discount rate from two prices, or when inventory should receive a readable stock status. Each row keeps its identity and source position. Calculating a new field does not combine customers, collapse orders, sort the file, or create extra records. Use the separate grouping tool afterward if you need totals across several rows.

Start with the synthetic order example before opening private data. The example includes a product identifier with leading zeros and a small decimal price that is useful for checking arithmetic. The result is a plain value, not an executable spreadsheet formula. Expressions are assembled with field selectors, constants and grouped operations. There is no script box, remote reference, macro engine or support for the full Excel formula language. This scope makes each calculation and its inputs available for review.

Choose a CSV, TSV or XLSX file, paste cells copied from a spreadsheet, or continue with the current workbench table. Text files have an explicit encoding step, followed by delimiter, quote and header confirmation. Review the sample, then confirm the complete import. Duplicate or empty headers must be resolved. A short record can be padded only after your explicit choice; additional fields are not silently discarded. Quoted commas, embedded newlines and trailing empty fields remain part of their cells.

For XLSX, read the workbook structure and select one worksheet area. Review the header row, first data row, hidden rows and hidden columns. The values panel distinguishes stored values, displayed text and formula caches. No formula is recalculated. A formula without a usable saved value blocks import. Use displayed text for an identifier whose formatting supplies leading zeros, after checking the preview. Workbook styles, charts, macros and formulas are not preserved in the calculated CSV result.

Open a field in the type section and choose Number only when it is a numeric input. Leave product codes, postal codes and long identifiers as Text. Boolean fields accept exactly true or false. Confirm the source decimal separator, grouping separator, negative notation and treatment of percent signs. These choices are data settings; changing interface language does not change them. Currency symbols, scientific notation and undeclared whitespace are not automatically removed. Correct those values or use the existing number normalization tool first.

In the normal example, declare price and quantity as numbers, confirm the parsing choices, and open the preset list. Select Unit price × quantity, then check that the two field selectors point to price and quantity. Name the result amount and keep two decimal places. Run the complete input. For product 00123, price 12.50 and quantity 3 produce 37.50. Price 0.10 and quantity 3 produce 0.30. The identifier still reads 00123 because the calculation never converts that source field.

Add another result column when you need a dependent calculation. A field selector can reference an earlier derived amount as well as an original field. Results run in the order shown in the column list. Moving a result before one of its dependencies creates a blocking configuration error. A deleted reference, circular reference or duplicate output name also blocks execution before any row is processed. Source references use stable internal identities, so renaming or rearranging existing source fields does not deliberately retarget them.

The discount rate preset computes original price minus current price, divided by original price. The discounted price preset instead multiplies a price by one minus a discount ratio. Presets suggest fields only when a supported name matches one column. Missing or ambiguous matches stay unselected; choose them explicitly and review every field before running. The percent operation explicitly distinguishes ten meaning ten percent from zero point one meaning ten percent. A percent sign in imported text has a separate parsing setting. Combining both conversions can divide twice, so inspect a known row before processing the whole file.

Declare stock as Number and add a Text result called stock_status. Select the stock preset. Its first branch tests whether stock equals zero and returns Out of stock. Its second branch tests whether stock is less than five and returns Low stock. The final default returns In stock. Values zero, four and eight therefore receive three different labels. Zero also satisfies the second comparison, but the first matching branch stops the search.

Each branch can combine comparisons using All match or Any match. Comparisons include numeric ordering, equal values, empty input and case sensitive text containment. A return value can be a constant, a field or an arithmetic expression of the declared result type. All branches must have the same result type. Every comparison within an evaluated group is checked for invalid input. Once a branch is selected, unused return expressions and later branches are not evaluated. This allows a deliberate zero check to guard a division without hiding an invalid numeric source.

Arithmetic uses decimal values rather than binary floating point multiplication. Inputs and intermediate values are limited to one thousand digits, with two thousand forty eight significant digits of working precision. Division that requires precision handling is rounded to forty decimal places using half even rounding and counted in the summary. Final output supports zero through forty decimal places and the visible rounding choices. Display rounding does not change the value passed to a later result column within the same run. Add an explicit rounding operation when the business rule requires intermediate rounding.

Missing cells and empty strings normally produce missing input. Whitespace is text, zero is a value, false is a boolean token, and NULL is ordinary text. None is silently replaced with zero. You can enable a missing default separately for a source field; replacements are counted once per source field per row. Invalid numbers and division by zero remain separate errors. The main table retains every source record and places an empty cell at a failed result. Exporting that table requires acknowledging the issue report; downloading problem reports remains possible without that acknowledgment.

The summary separates successful records, missing only records and records with computation errors. A record with both kinds of problem belongs to the error category so those three counts reconcile to processed records. Result cell counts describe a different scope: one record can have several result cells and several problems. The result trace shows evaluated inputs, output and selected branch. Open its source link to inspect the original record, including its file, worksheet when applicable and logical record location.

Use the sample scope for the first twenty records when exploring rules. Sample results are explicitly labeled and cannot be applied as a complete workflow step. Run the entire input for final counts and delivery. Cancel terminates the active worker. Editing rules makes an earlier result stale and disables export until another run finishes. Undo rule edit and Redo rule edit restore recent configuration changes. After applying a reviewed result, the workflow undo and redo controls restore the corresponding table.

Save rules downloads configuration without embedding the source table. The file still contains field names, fixed values and missing defaults, so review it before sharing. Loading it into another table requires the same unique source names; their order may differ. Number interpretation and source replacement approval must be confirmed again. Files and working data remain in page memory. Internal tool and language navigation keep that session, while refreshing or closing clears it. A saved rule file is neither a data backup nor cloud synchronization.

Choose the complete result, error source rows, detailed issues or rule summary as the export scope. ZIP includes the selected tool result and its reports; reports may contain source values. CSV and TSV exports use UTF-8 and can include a BOM. Spreadsheet protection prefixes formula-like text with an apostrophe and may also change legitimate negative values. For machine imports, explicitly review the risk before choosing unchanged text. Download verification reads the generated file back and compares its structure and cell values. Missing cells become empty CSV fields, so keep the issue report when that distinction matters.

Calculation rules support up to twenty derived columns, twenty branches per result, ten comparisons per branch, eight levels of expression grouping and five hundred twelve expression nodes. The worker also checks file, row, cell, report and operation budgets. A file is limited to ten megabytes; table limits include one hundred thousand records, two hundred columns and two million cells. Complex rules or detailed reports may reach a smaller effective limit. An over-limit operation stops with an explanation rather than presenting a partial file as complete.

File contents are processed locally and are not uploaded for calculation. The webpage still requests static resources and hosting may keep basic access logs. Cells are rendered as text; links inside your data are not visited. This tool does not provide tax advice, aggregate across rows, run external plugins or synchronize with a spreadsheet account. After applying results, continue with filtering, grouping or deduplication through the related tools. Inspect the downstream settings because those operations have their own rules for empty values and numeric interpretation.