Module script_bytes 

script_bytes submodule (Rust-only; see the module docs). !!! WARNING: RUST-ONLY MODULE — NO C COUNTERPART !!!

Lossless decode of RAW SCRIPT BYTES into the shell pipeline’s String form.

C zsh has no encoding requirement on script input. Src/input.c reads bytes (shingetchar, c:229) and Src/utils.c:4856 metafy escapes only the reserved IMETA bytes (Src/utils.c:4195-4201 — '\0', Meta 0x83, Marker 0xa2 and the token range Pound..Nularg); every other byte, including a lone ISO-8859-1 0xe9, travels through the lexer untouched and is written back out verbatim. That is why a legacy byte in a completion file has never bothered zsh.

Rust String is UTF-8, so a raw non-UTF-8 byte cannot be stored directly. zshrs already has one encoding for exactly this, used by the $'\xNN' decoders (src/ported/lex.rs getkeystring_dollar_quote, src/extensions/compile_zsh.rs meta_encode_byte) and decoded at every write/exec boundary by crate::ported::utils::unmetafy_str: a byte >= 0x80 becomes the two chars U+0083 (Meta) and U+00(b ^ 32). This module applies that same encoding to the bytes a script file or SHIN hands us.

The transform is the byte-level twin of Src/input.c:267 shingetline, which the interactive/stdin path already performs inline: valid UTF-8 stays real chars (the lexer is Unicode-based — src/ported/lex.rs:6171 “zshrs does not metafy, it keeps UTF-8 str”), and only bytes that cannot be part of a UTF-8 sequence are Meta-encoded. Clean-ASCII and valid-UTF-8 input therefore decode to exactly the same String read_to_string produced, so the common path is unchanged; the difference is that an invalid byte no longer kills the read.

Functions

decode_script_bytes

Decode raw script bytes into the pipeline’s String form without losing a single byte.

read_script_file

Read a script file as raw bytes and decode it losslessly.

regroup_meta_utf8

Regroup the Meta-pair bytes a $'…' decoder produced.