Developer builds regex-based tool to detect and clean hidden Unicode junk in pasted text

A developer shared a practical text-cleaning utility built after repeatedly encountering pasted text that looked normal but broke search, formatting, and CMS inputs. The tool uses deterministic regex patterns to identify and count specific invisible Unicode characters — such as zero-width spaces, soft hyphens, and bidirectional control marks — by their exact code points rather than grouping them loosely. A fixed Unicode offset of 0xFEE0 powers straightforward conversion between full-width and half-width ASCII-style characters. The developer argues this character-level approach is more reliable than asking AI models to clean text, since large language models tend to struggle with precise character accounting. The utility's interface displays only non-zero character categories, making results read as a targeted diagnosis rather than an overwhelming data dump.