ripgrep/crates/regex/src/strip.rs
Andrew Gallant 81341702af regex: push more pattern handling to matcher construction
Previously, ripgrep core was responsible for escaping regex patterns and
implementing the --line-regexp flag. This commit moves that
responsibility down into the matchers such that ripgrep just needs to
hand the patterns it gets off to the matcher builder. The builder will
then take care of escaping and all that.

This was done to make pattern construction completely owned by the
matcher builders. With the arrival regex-automata, this means we can
move to the HIR very quickly and then never move back to the concrete
syntax. We can then build our regex directly from the HIR. This overall
can save quite a bit of time, especially when searching for large
dictionaries.

We still aren't quite as fast as GNU grep when searching something on
the scale of /usr/share/dict/words, but we are basically within spitting
distance. Prior to this, we were about an order of magnitude slower.

This architecture in particular lets us write a pretty simple fast path
that avoids AST parsing and HIR translation entirely: the case where one
is just searching for a literal. In that case, we can hand construct the
HIR directly.
2023-07-05 14:04:29 -04:00

171 lines
6.2 KiB
Rust

use {
grep_matcher::LineTerminator,
regex_syntax::hir::{self, Hir, HirKind},
};
use crate::error::{Error, ErrorKind};
/// Return an HIR that is guaranteed to never match the given line terminator,
/// if possible.
///
/// If the transformation isn't possible, then an error is returned.
///
/// In general, if a literal line terminator occurs anywhere in the HIR, then
/// this will return an error. However, if the line terminator occurs within
/// a character class with at least one other character (that isn't also a line
/// terminator), then the line terminator is simply stripped from that class.
///
/// If the given line terminator is not ASCII, then this function returns an
/// error.
///
/// Note that as of regex 1.9, this routine could theoretically be implemented
/// without returning an error. Namely, for example, we could turn
/// `foo\nbar` into `foo[a&&b]bar`. That is, replace line terminators with a
/// sub-expression that can never match anything. Thus, ripgrep would accept
/// such regexes and just silently not match anything. Regex versions prior to 1.8
/// don't support such constructs. I ended up deciding to leave the existing
/// behavior of returning an error instead. For example:
///
/// ```text
/// $ echo -n 'foo\nbar\n' | rg 'foo\nbar'
/// the literal '"\n"' is not allowed in a regex
///
/// Consider enabling multiline mode with the --multiline flag (or -U for short).
/// When multiline mode is enabled, new line characters can be matched.
/// ```
///
/// This looks like a good error message to me, and even suggests a flag that
/// the user can use instead.
pub(crate) fn strip_from_match(
expr: Hir,
line_term: LineTerminator,
) -> Result<Hir, Error> {
if line_term.is_crlf() {
let expr1 = strip_from_match_ascii(expr, b'\r')?;
strip_from_match_ascii(expr1, b'\n')
} else {
strip_from_match_ascii(expr, line_term.as_byte())
}
}
/// The implementation of strip_from_match. The given byte must be ASCII.
/// This function returns an error otherwise. It also returns an error if
/// it couldn't remove `\n` from the given regex without leaving an empty
/// character class in its place.
fn strip_from_match_ascii(expr: Hir, byte: u8) -> Result<Hir, Error> {
if !byte.is_ascii() {
return Err(Error::new(ErrorKind::InvalidLineTerminator(byte)));
}
let ch = char::from(byte);
let invalid = || Err(Error::new(ErrorKind::NotAllowed(ch.to_string())));
Ok(match expr.into_kind() {
HirKind::Empty => Hir::empty(),
HirKind::Literal(hir::Literal(lit)) => {
if lit.iter().find(|&&b| b == byte).is_some() {
return invalid();
}
Hir::literal(lit)
}
HirKind::Class(hir::Class::Unicode(mut cls)) => {
if cls.ranges().is_empty() {
return Ok(Hir::class(hir::Class::Unicode(cls)));
}
let remove = hir::ClassUnicode::new(Some(
hir::ClassUnicodeRange::new(ch, ch),
));
cls.difference(&remove);
if cls.ranges().is_empty() {
return invalid();
}
Hir::class(hir::Class::Unicode(cls))
}
HirKind::Class(hir::Class::Bytes(mut cls)) => {
if cls.ranges().is_empty() {
return Ok(Hir::class(hir::Class::Bytes(cls)));
}
let remove = hir::ClassBytes::new(Some(
hir::ClassBytesRange::new(byte, byte),
));
cls.difference(&remove);
if cls.ranges().is_empty() {
return invalid();
}
Hir::class(hir::Class::Bytes(cls))
}
HirKind::Look(x) => Hir::look(x),
HirKind::Repetition(mut x) => {
x.sub = Box::new(strip_from_match_ascii(*x.sub, byte)?);
Hir::repetition(x)
}
HirKind::Capture(mut x) => {
x.sub = Box::new(strip_from_match_ascii(*x.sub, byte)?);
Hir::capture(x)
}
HirKind::Concat(xs) => {
let xs = xs
.into_iter()
.map(|e| strip_from_match_ascii(e, byte))
.collect::<Result<Vec<Hir>, Error>>()?;
Hir::concat(xs)
}
HirKind::Alternation(xs) => {
let xs = xs
.into_iter()
.map(|e| strip_from_match_ascii(e, byte))
.collect::<Result<Vec<Hir>, Error>>()?;
Hir::alternation(xs)
}
})
}
#[cfg(test)]
mod tests {
use regex_syntax::Parser;
use super::{strip_from_match, LineTerminator};
use crate::error::Error;
fn roundtrip(pattern: &str, byte: u8) -> String {
roundtrip_line_term(pattern, LineTerminator::byte(byte)).unwrap()
}
fn roundtrip_crlf(pattern: &str) -> String {
roundtrip_line_term(pattern, LineTerminator::crlf()).unwrap()
}
fn roundtrip_err(pattern: &str, byte: u8) -> Result<String, Error> {
roundtrip_line_term(pattern, LineTerminator::byte(byte))
}
fn roundtrip_line_term(
pattern: &str,
line_term: LineTerminator,
) -> Result<String, Error> {
let expr1 = Parser::new().parse(pattern).unwrap();
let expr2 = strip_from_match(expr1, line_term)?;
Ok(expr2.to_string())
}
#[test]
fn various() {
assert_eq!(roundtrip(r"[a\n]", b'\n'), "a");
assert_eq!(roundtrip(r"[a\n]", b'a'), "\n");
assert_eq!(roundtrip_crlf(r"[a\n]"), "a");
assert_eq!(roundtrip_crlf(r"[a\r]"), "a");
assert_eq!(roundtrip_crlf(r"[a\r\n]"), "a");
assert_eq!(roundtrip(r"(?-u)\s", b'a'), r"(?-u:[\x09-\x0D\x20])");
assert_eq!(roundtrip(r"(?-u)\s", b'\n'), r"(?-u:[\x09\x0B-\x0D\x20])");
assert!(roundtrip_err(r"\n", b'\n').is_err());
assert!(roundtrip_err(r"abc\n", b'\n').is_err());
assert!(roundtrip_err(r"\nabc", b'\n').is_err());
assert!(roundtrip_err(r"abc\nxyz", b'\n').is_err());
assert!(roundtrip_err(r"\x0A", b'\n').is_err());
assert!(roundtrip_err(r"\u000A", b'\n').is_err());
assert!(roundtrip_err(r"\U0000000A", b'\n').is_err());
assert!(roundtrip_err(r"\u{A}", b'\n').is_err());
assert!(roundtrip_err("\n", b'\n').is_err());
}
}