safe_regex_compiler/
lib.rs

1//! [![crates.io version](https://img.shields.io/crates/v/safe-regex-compiler.svg)](https://crates.io/crates/safe-regex-compiler)
2//! [![license: Apache 2.0](https://gitlab.com/leonhard-llc/safe-regex-rs/-/raw/main/license-apache-2.0.svg)](http://www.apache.org/licenses/LICENSE-2.0)
3//! [![unsafe forbidden](https://gitlab.com/leonhard-llc/safe-regex-rs/-/raw/main/unsafe-forbidden-success.svg)](https://github.com/rust-secure-code/safety-dance/)
4//! [![pipeline status](https://gitlab.com/leonhard-llc/safe-regex-rs/badges/main/pipeline.svg)](https://gitlab.com/leonhard-llc/safe-regex-rs/-/pipelines)
5//!
6//! A regular expression compiler.
7//!
8//! If you want to use regular expressions in your software, use the
9//! [`safe_regex`](https://crates.io/crates/safe-regex) crate.
10//!
11//! # Cargo Geiger Safety Report
12//! # Changelog
13//! See [`safe_regex`](https://crates.io/crates/safe-regex) create.
14//!
15//! # Release Process
16//! 1. Edit `Cargo.toml` and bump version number.
17//! 1. Run `../release.sh`
18#![forbid(unsafe_code)]
19use crate::generator::generate;
20use safe_proc_macro2::{TokenStream, TokenTree};
21
22pub mod generator;
23pub mod parser;
24
25#[macro_export]
26macro_rules! dprintln {
27    // ($($args:tt)+) => { println!( $($args)+ ) };
28    ($($args:tt)+) => {};
29}
30
31/// Converts the bytes into an ASCII string.
32#[allow(clippy::missing_panics_doc)]
33pub fn escape_ascii(input: impl AsRef<[u8]>) -> String {
34    let mut result = String::new();
35    for byte in input.as_ref() {
36        for ascii_byte in core::ascii::escape_default(*byte) {
37            result.push_str(core::str::from_utf8(&[ascii_byte]).unwrap());
38        }
39    }
40    result
41}
42
43/// Implements the `regex!` macro.
44///
45/// # Errors
46/// Returns `Err(String)` with a human-readable description of the problem.
47pub fn impl_regex(stream: TokenStream) -> Result<TokenStream, String> {
48    // Ident { sym: regex }
49    // Punct { char: '!', spacing: Alone }
50    // Group {
51    //   delimiter: Parenthesis,
52    //   stream: TokenStream [
53    //     Literal { lit: br"a" }
54    //   ]
55    // }
56    const ERR: &str = "expected a raw byte string, like br\"abc\"";
57    dprintln!(
58        "impl_regex {:?}",
59        stream
60            .clone()
61            .into_iter()
62            .map(|tree| format!("{:?} ", tree))
63            .collect::<String>()
64    );
65    let mut stream_iter = stream.into_iter();
66    let Some(TokenTree::Literal(literal)) = stream_iter.next() else {
67        return Err(ERR.to_string());
68    };
69    if stream_iter.next().is_some() {
70        return Err(ERR.to_string());
71    }
72
73    // The compiler already parsed the literal, but does not expose its fields.
74    // So we convert the literal to a string and parse it ourselves.
75    // https://stackoverflow.com/questions/61169932/how-do-i-get-the-value-and-type-of-a-literal-in-a-rust-proc-macro
76    let literal_string = literal.to_string();
77    // println!("compiling safe_regex::regex!({})", literal_string);
78    let raw_byte_string = literal_string
79        .strip_prefix("br")
80        .ok_or_else(|| ERR.to_string())?
81        // Compiler guarantees that strings are closed.
82        .trim_start_matches('#')
83        .trim_start_matches('"')
84        .trim_end_matches('#')
85        .trim_end_matches('"');
86    // The compiler guarantees that a literal byte string contains only ASCII.
87    // > regex!(br"€"); // error: raw byte string must be ASCII
88    // Therefore, we can slice the string at any byte offset.
89    let final_node = crate::parser::parse(raw_byte_string.as_bytes())?;
90    Ok(generate(&final_node))
91}
safe_regex_compiler/lib.rs

safe_regex_compiler/
lib.rs
