CSV Row Filter

What this tool is used for

• Pulling out just the rows for one customer or region from a large export.
• Removing rows that failed a check before importing the rest.
• Reducing a big file to a manageable sample that still contains the case you are investigating.
• Splitting an export into the rows that matched a condition and those that did not.
• Filtering out test or placeholder rows before handing a file to someone else.

Frequently Asked Questions

Which comparisons are available?

Contains, equals, starts with, ends with, greater than, less than, is empty, and a regular-expression match. The text comparisons are case-insensitive; the numeric ones parse both sides as numbers, so a non-numeric cell simply fails to match.

Why did my numeric filter miss rows?

Because the cell is not a plain number. `1,234` and `$99` and `12%` all fail to parse, and a value with a trailing space usually does too. Strip the formatting first, or filter with "contains" instead when the column is really text.

What is the difference between keep and remove?

They are exact opposites over the same predicate — useful because some conditions are far easier to state as an exclusion. "Remove rows where status is empty" is clearer than enumerating every status you want to keep.

How do I combine two conditions?

Run the filter twice. Filtering for A and then for B gives you A AND B; there is no OR here on purpose, since a regular expression already covers it — `^(pending|failed)$` matches either.

Is the header row preserved?

Yes, the first row is kept regardless of whether it matches, so the output is still a valid CSV with column names. If your file genuinely has no header, treat the first data row as expendable and check it survived.

How is an empty cell treated in a comparison?

Not as zero, and not as matching everything. An empty value is its own case, which is why a numeric filter can silently exclude rows that ought to be considered — filter explicitly for blanks when they matter.

Common errors and gotchas

• Filtering on a column whose values are quoted, where the quotes are part of the comparison unless handled.
• Assuming a numeric comparison on a column read as text, where 10 sorts before 9.
• Losing the header row, which turns the first record into column names in the next tool.
• Overlooking embedded newlines inside quoted fields, which a line-based filter splits incorrectly.
• Matching a substring when you meant the whole value, so 'AT' also matches 'AUT'.