From 7ddc89e893ebb6c60af4fe92c439c4a60c9118dd Mon Sep 17 00:00:00 2001 From: David Tolnay Date: Wed, 10 Apr 2024 20:55:59 -0700 Subject: [PATCH] Remove superfluous escaping from byte, byte str, and c str literals --- library/proc_macro/src/escape.rs | 57 ++++++++++++++++++++ library/proc_macro/src/lib.rs | 51 +++++++++++++----- tests/ui/proc-macro/auxiliary/api/literal.rs | 14 ++--- 3 files changed, 101 insertions(+), 21 deletions(-) create mode 100644 library/proc_macro/src/escape.rs diff --git a/library/proc_macro/src/escape.rs b/library/proc_macro/src/escape.rs new file mode 100644 index 00000000000..87a4d1d50fd --- /dev/null +++ b/library/proc_macro/src/escape.rs @@ -0,0 +1,57 @@ +#[derive(Copy, Clone)] +pub(crate) struct EscapeOptions { + /// Produce \'. + pub escape_single_quote: bool, + /// Produce \". + pub escape_double_quote: bool, + /// Produce \x escapes for non-ASCII, and use \x rather than \u for ASCII + /// control characters. + pub escape_nonascii: bool, +} + +pub(crate) fn escape_bytes(bytes: &[u8], opt: EscapeOptions) -> String { + let mut repr = String::new(); + + if opt.escape_nonascii { + for &byte in bytes { + escape_single_byte(byte, opt, &mut repr); + } + } else { + let mut chunks = bytes.utf8_chunks(); + while let Some(chunk) = chunks.next() { + for ch in chunk.valid().chars() { + escape_single_char(ch, opt, &mut repr); + } + for &byte in chunk.invalid() { + escape_single_byte(byte, opt, &mut repr); + } + } + } + + repr +} + +fn escape_single_byte(byte: u8, opt: EscapeOptions, repr: &mut String) { + if byte == b'\0' { + repr.push_str("\\0"); + } else if (byte == b'\'' && !opt.escape_single_quote) + || (byte == b'"' && !opt.escape_double_quote) + { + repr.push(byte as char); + } else { + // Escapes \t, \r, \n, \\, \', \", and uses \x## for non-ASCII and + // for ASCII control characters. + repr.extend(byte.escape_ascii().map(char::from)); + } +} + +fn escape_single_char(ch: char, opt: EscapeOptions, repr: &mut String) { + if (ch == '\'' && !opt.escape_single_quote) || (ch == '"' && !opt.escape_double_quote) { + repr.push(ch); + } else { + // Escapes \0, \t, \r, \n, \\, \', \", and uses \u{...} for + // non-printable characters and for Grapheme_Extend characters, which + // includes things like U+0300 "Combining Grave Accent". + repr.extend(ch.escape_debug()); + } +} diff --git a/library/proc_macro/src/lib.rs b/library/proc_macro/src/lib.rs index 3d7d36b27e5..581d7e3efe3 100644 --- a/library/proc_macro/src/lib.rs +++ b/library/proc_macro/src/lib.rs @@ -43,10 +43,12 @@ pub mod bridge; mod diagnostic; +mod escape; #[unstable(feature = "proc_macro_diagnostic", issue = "54140")] pub use diagnostic::{Diagnostic, Level, MultiSpan}; +use crate::escape::{escape_bytes, EscapeOptions}; use std::ffi::CStr; use std::ops::{Range, RangeBounds}; use std::path::PathBuf; @@ -1356,40 +1358,61 @@ impl Literal { /// String literal. #[stable(feature = "proc_macro_lib2", since = "1.29.0")] pub fn string(string: &str) -> Literal { - let quoted = format!("{:?}", string); - assert!(quoted.starts_with('"') && quoted.ends_with('"')); - let symbol = "ed[1..quoted.len() - 1]; - Literal::new(bridge::LitKind::Str, symbol, None) + let escape = EscapeOptions { + escape_single_quote: false, + escape_double_quote: true, + escape_nonascii: false, + }; + let repr = escape_bytes(string.as_bytes(), escape); + Literal::new(bridge::LitKind::Str, &repr, None) } /// Character literal. #[stable(feature = "proc_macro_lib2", since = "1.29.0")] pub fn character(ch: char) -> Literal { - let quoted = format!("{:?}", ch); - assert!(quoted.starts_with('\'') && quoted.ends_with('\'')); - let symbol = "ed[1..quoted.len() - 1]; - Literal::new(bridge::LitKind::Char, symbol, None) + let escape = EscapeOptions { + escape_single_quote: true, + escape_double_quote: false, + escape_nonascii: false, + }; + let repr = escape_bytes(ch.encode_utf8(&mut [0u8; 4]).as_bytes(), escape); + Literal::new(bridge::LitKind::Char, &repr, None) } /// Byte character literal. #[stable(feature = "proc_macro_byte_character", since = "1.79.0")] pub fn byte_character(byte: u8) -> Literal { - let string = [byte].escape_ascii().to_string(); - Literal::new(bridge::LitKind::Byte, &string, None) + let escape = EscapeOptions { + escape_single_quote: true, + escape_double_quote: false, + escape_nonascii: true, + }; + let repr = escape_bytes(&[byte], escape); + Literal::new(bridge::LitKind::Byte, &repr, None) } /// Byte string literal. #[stable(feature = "proc_macro_lib2", since = "1.29.0")] pub fn byte_string(bytes: &[u8]) -> Literal { - let string = bytes.escape_ascii().to_string(); - Literal::new(bridge::LitKind::ByteStr, &string, None) + let escape = EscapeOptions { + escape_single_quote: false, + escape_double_quote: true, + escape_nonascii: true, + }; + let repr = escape_bytes(bytes, escape); + Literal::new(bridge::LitKind::ByteStr, &repr, None) } /// C string literal. #[stable(feature = "proc_macro_c_str_literals", since = "1.79.0")] pub fn c_string(string: &CStr) -> Literal { - let string = string.to_bytes().escape_ascii().to_string(); - Literal::new(bridge::LitKind::CStr, &string, None) + let escape = EscapeOptions { + escape_single_quote: false, + escape_double_quote: true, + escape_nonascii: false, + }; + let repr = escape_bytes(string.to_bytes(), escape); + Literal::new(bridge::LitKind::CStr, &repr, None) } /// Returns the span encompassing this literal. diff --git a/tests/ui/proc-macro/auxiliary/api/literal.rs b/tests/ui/proc-macro/auxiliary/api/literal.rs index a3519f5791a..7109340bb64 100644 --- a/tests/ui/proc-macro/auxiliary/api/literal.rs +++ b/tests/ui/proc-macro/auxiliary/api/literal.rs @@ -29,17 +29,17 @@ fn test_display_literal() { assert_eq!(Literal::byte_string(b"aA").to_string(), r#" b"aA" "#.trim()); assert_eq!(Literal::byte_string(b"\t").to_string(), r#" b"\t" "#.trim()); - assert_eq!(Literal::byte_string(b"'").to_string(), r#" b"\'" "#.trim()); + assert_eq!(Literal::byte_string(b"'").to_string(), r#" b"'" "#.trim()); assert_eq!(Literal::byte_string(b"\"").to_string(), r#" b"\"" "#.trim()); - assert_eq!(Literal::byte_string(b"\0").to_string(), r#" b"\x00" "#.trim()); + assert_eq!(Literal::byte_string(b"\0").to_string(), r#" b"\0" "#.trim()); assert_eq!(Literal::byte_string(b"\x01").to_string(), r#" b"\x01" "#.trim()); assert_eq!(Literal::c_string(c"aA").to_string(), r#" c"aA" "#.trim()); assert_eq!(Literal::c_string(c"\t").to_string(), r#" c"\t" "#.trim()); - assert_eq!(Literal::c_string(c"❤").to_string(), r#" c"\xe2\x9d\xa4" "#.trim()); - assert_eq!(Literal::c_string(c"\'").to_string(), r#" c"\'" "#.trim()); + assert_eq!(Literal::c_string(c"❤").to_string(), r#" c"❤" "#.trim()); + assert_eq!(Literal::c_string(c"\'").to_string(), r#" c"'" "#.trim()); assert_eq!(Literal::c_string(c"\"").to_string(), r#" c"\"" "#.trim()); - assert_eq!(Literal::c_string(c"\x7f\xff\xfe\u{333}").to_string(), r#" c"\x7f\xff\xfe\xcc\xb3" "#.trim()); + assert_eq!(Literal::c_string(c"\x7f\xff\xfe\u{333}").to_string(), r#" c"\u{7f}\xff\xfe\u{333}" "#.trim()); assert_eq!(Literal::character('a').to_string(), r#" 'a' "#.trim()); assert_eq!(Literal::character('\t').to_string(), r#" '\t' "#.trim()); @@ -52,8 +52,8 @@ fn test_display_literal() { assert_eq!(Literal::byte_character(b'a').to_string(), r#" b'a' "#.trim()); assert_eq!(Literal::byte_character(b'\t').to_string(), r#" b'\t' "#.trim()); assert_eq!(Literal::byte_character(b'\'').to_string(), r#" b'\'' "#.trim()); - assert_eq!(Literal::byte_character(b'"').to_string(), r#" b'\"' "#.trim()); - assert_eq!(Literal::byte_character(0).to_string(), r#" b'\x00' "#.trim()); + assert_eq!(Literal::byte_character(b'"').to_string(), r#" b'"' "#.trim()); + assert_eq!(Literal::byte_character(0).to_string(), r#" b'\0' "#.trim()); assert_eq!(Literal::byte_character(1).to_string(), r#" b'\x01' "#.trim()); }