transduce: Zero-Copy Isomorphic Parsing

Your code should look like what it parses.

See this example from lib.rs:

let parser = exact(&b'(') >> verbatim() << exact(&b')');
let input = b"(*)";
assert_eq!(
    parser.parse(input),
    Ok(&b'*'), // Reference to the region of input inside parentheses
);
// Or, equivalently:
assert_eq!(
    parenthesized(verbatim).parse(input),
    Ok(&b'*'),
);

This does exactly what it looks like it does.

Equivalently,

assert_eq!(parenthesized(verbatim()).parse(rawstr.chars()), Ok('*'))

Pretty-printed errors

The parse method automatically locates the error (even for user-defined parsers) and prints out gorgeous, colored Rust-style errors:

(exact(b'?') >> exact(b'?')).parse_or_panic(b"???");

...| Error while parsing:
 1 | ???
   |   ^ Unparsed input remains after parsing what should have been everything

(verbatim() & verbatim() & verbatim() & verbatim()).parse_or_panic(b"???");

...| Error while parsing:
 1 | ???
   |    ^ Reached end of input but expected an item

(verbatim() << exact(b'!')).parse_or_panic(b"???")

...| Error while parsing:
 1 | ???
   |  ^ Expected 33 but found 63

Dependencies & no_std

This crate has no dependencies. It's also entirely no_std; a std version isn't even necessary.

If (for some reason?) you want to run this on a microcontroller or just without heap allocation, the alloc feature can be disabled. The entire core still works, but a few parsers in base (those involving vectors whose length is unknown at compile time, e.g. comma_separated) will be unavailable.

Thanks

Huge shoutout to UPenn's CIS 194 and Haskell's higher-order parsing libraries I learned in 194.