Recursive Text Search

This package is the v0.19.0 reference application. It recursively searches regular UTF-8 files using only public standard-library APIs:

nocter build
./text-search NEEDLE ROOT

ROOT must name a readable directory. Symbolic links and non-regular entries are skipped; links are never followed. A matching line is written as relative/path:line:text, with one-based line numbers. Successful output is ordered by the UTF-8 byte order of each path relative to ROOT, independently of filesystem enumeration order. Lines retain their text but omit the LF or CRLF terminator recognized by BufReader.

The process returns 0 when at least one line matches, 1 when no line matches, and 2 for invalid arguments or a recoverable traversal, input, or output failure. Usage and failures go to standard error. An input failure may occur after earlier matching output has already been written.

The application retains owned paths for regular files and pending directories so it can produce deterministic output. It closes each directory stream before descending into a child, then reads one file at a time with a reusable line destination. It never retains a complete file. Memory is therefore bounded by the discovered UTF-8 path data, directory traversal state, and the largest line encountered rather than by total file contents. Successful match output passes through one bounded BufWriter; normal completion explicitly flushes it, while an output failure remains a recoverable command failure.