Skip to main content
Deno 2 is finally here 🎉️
Learn more

Safe

Safe is a Deno module which produces safe strings out of any input string.

Usage

import { safe } from "https://deno.land/x/safe/mod.ts"

// use as a regular function or a template tag.
// both print: "Welcome to \u65e5\u672c!"
console.log(safe("Welcome to 日本!"))
console.log(safe`Welcome to 日本!`)

s is an alias for safe, use it if you like:

import { s } from "https://deno.land/x/safe/mod.ts"
console.log(s("Welcome to 日本!"))
console.log(s`"Welcome to 日本!"`)

What is “safe”?

safe is a function which, given a string, returns a “safe string” representation of the input. The idea is that you should be able to print a safe string in any terminal/shell/console without worrying about bells ringing, lines being deleted, screen flashing/clearing, etc.

You also don’t lose any information in the process. The safe string technically contains the same information as the original string, but not necessarily byte-for-byte.

How does it work?

“Normal” ASCII characters, meaning any characters with a code point in the range 32–126, starting with space " " and ending with tilde "~", are returned as-is.

Any other character is returned as an escape sequence, e.g. \x1b for ESC, \u{01f404} for 🐄, etc.

However, the following code points get special treatment:

  • \\0, not \x00 (NUL) – null character
  • \\b, not \x08 (BS) – backspace
  • \\t, not \x09 (HT) – horizontal tab
  • \\n, not \x0A (LF) – line feed
  • \\v, not \x0B (VT) – vertical tab
  • \\f, not \x0C (FF) – form feed
  • \\r, not \x0D (CR) – carriage return

Details

Escape sequences in the output will always be zero-padded to two, four, or six digits, depending on the code point. Other than that, safe strings always contain their shortest escaped representation.

The main reasons are a) readability, and b) consistency. For example \x7 is NOT valid JavaScript syntax, it is always \x07. Same goes for \u404, it must be \u0404. However, \u{1f404} IS valid, and the same thing as \u{01f404}.

Escape sequences are always completely in lowercase, e.g. \uabcd, not \uABCD.

We COULD have done \u{7}, but it’s longer than \x07, and not as easily recognizable, as usually when we see \u, we think of a Unicode code point, not regular old ASCII.

Normalization

Normalization is NOT performed on the input string. Normalizing it would essentially corrupt the original representation, and you would lose information.

If this is a concern for some reason, run .normalize() on the input string before passing it to Safe.

Development

Contributions are welcome! Please open an issue or a pull request.

$ deno fmt && deno lint && deno test