String Utilities

Because most implementations reverse by code unit, and anything outside the basic multilingual plane is stored as two of them. Splitting a string on empty and reversing it separates those halves, producing a broken character that renders as a question mark or a box. This tool reverses by grapheme — the unit a person perceives as one character — so an emoji, a flag, or a family emoji made of several joined people all survive intact. Accented letters written as a base plus a combining mark survive for the same reason.

They are real characters that render as nothing or as an ordinary space: zero-width spaces, non-breaking spaces, byte-order marks, soft hyphens. They matter because they break equality. Two strings can look pixel-identical on screen and never compare equal, producing bugs that are genuinely maddening to diagnose because the evidence is invisible. They arrive from copying out of Word, from PDFs, from web pages, and from editors that write a byte-order mark. This tool finds them, names them, and can strip them.

Yes, and that is deliberate. Reversing a hundred lines individually is a real operation you might want; reversing the entire paste including its newlines is a novelty. The same applies to trimming, padding and wrapping — each is far more useful applied per line, because the common case is a pasted list rather than a single value. Line count is always preserved, so the output lines up with the input and you can diff them if you need to.

Trimming removes whitespace from the ends of a line and leaves the middle alone. Collapsing also reduces every internal run of whitespace to a single space, so tabs and double spaces inside the line become one space each. Trim is what you want before comparing or deduplicating values; collapse is what you want when tidying prose that has been reflowed or pasted from a source with inconsistent spacing. Removing all whitespace is a third, blunter operation for when you need the characters with no gaps at all.

It should not, because padding counts graphemes rather than code units — but visual alignment also depends on the font. Most emoji and CJK characters are rendered double-width in a monospace font, so a line containing one occupies two columns while counting as a single character. The padding here is correct by character count, which is what most systems mean; if you need columns to line up visually in a terminal you need width-aware padding, which is a genuinely different calculation.

Apply them one at a time — the output feeds straight back into the input with the swap button, so trimming and then removing invisible characters is two clicks rather than two visits. That is deliberate rather than a limitation: a pipeline of text transformations applied all at once is very hard to reason about when the result is not what you expected, whereas seeing each step lets you spot exactly which one did something you did not intend.