Text Filter

A reproducible line pipeline: filter, remove duplicates, sort, then reverse. First original lines are retained. Empty pattern disables filtering/exclusion. Literal mode matches a substring; regex uses JavaScript u or iu flags without / delimiters. Zero-width matches are supported.

Alphabetical comparison uses page-language collation. Ignore case lowercases literal/duplicate/sort keys; regex case uses JavaScript semantics. No accent removal or full multi-character case folding. Numeric order accepts complete ungrouped decimals with point/comma or exponent; invalid values report the original source line. Blank lines sort last before reversal, not as zero.

LF/CRLF/CR are recognised, final terminators do not add phantom rows. The browser pastes LF; choose export endings explicitly. Copy/download preserves full model bytes despite the LF preview. Source and selected input settings are kept in this browser; computed results are never saved. Processing updates automatically after a short pause; Enter in the filter/settings or Ctrl/⌘+Enter in the source updates immediately. Each calculation has its own cancellable worker with a 2.5-second deadline; changing an input terminates the previous worker. UTF-8 files up to 8000000 bytes are accepted. Limits: 2000000 UTF-16 units and 100000 lines. Edits invalidate old output.

If filtering, sorting or reversal moves an empty line to the end, an additional final separator preserves that line in TXT. No matching lines is shown explicitly; empty exports are disabled in that case. Copy fallback selects the existing output; manual copying follows the browser’s LF representation.