Add Literal methods from proc_macro_value feature

Author: dtolnay

src/lib.rs

@@ -164,8 +164,14 @@ mod imp;
 #[cfg(span_locations)]
 mod location;
 
+#[cfg(procmacro2_semver_exempt)]
+#[allow(dead_code)]
+mod rustc_literal_escaper;
+
 use crate::extra::DelimSpan;
 use crate::marker::{ProcMacroAutoTraits, MARKER};
+#[cfg(procmacro2_semver_exempt)]
+use crate::rustc_literal_escaper::MixedUnit;
 use core::cmp::Ordering;
 use core::fmt::{self, Debug, Display};
 use core::hash::{Hash, Hasher};
@@ -182,6 +188,10 @@ use std::path::PathBuf;
 #[cfg_attr(docsrs, doc(cfg(feature = "span-locations")))]
 pub use crate::location::LineColumn;
 
+#[cfg(procmacro2_semver_exempt)]
+#[cfg_attr(docsrs, doc(cfg(procmacro2_semver_exempt)))]
+pub use crate::rustc_literal_escaper::EscapeError;
+
 /// An abstract stream of tokens, or more concretely a sequence of token trees.
 ///
 /// This type provides interfaces for iterating over token trees and for
@@ -1263,6 +1273,112 @@ impl Literal {
         self.inner.subspan(range).map(Span::_new)
     }
 
+    /// Returns the unescaped string value if this is a string literal.
+    #[cfg(procmacro2_semver_exempt)]
+    pub fn str_value(&self) -> Result<String, ConversionErrorKind> {
+        let repr = self.to_string();
+
+        if repr.starts_with('"') && repr[1..].ends_with('"') {
+            let quoted = &repr[1..repr.len() - 1];
+            let mut value = String::with_capacity(quoted.len());
+            let mut error = None;
+            rustc_literal_escaper::unescape_str(quoted, |_range, res| match res {
+                Ok(ch) => value.push(ch),
+                Err(err) => {
+                    if err.is_fatal() {
+                        error = Some(ConversionErrorKind::FailedToUnescape(err));
+                    }
+                }
+            });
+            return match error {
+                Some(error) => Err(error),
+                None => Ok(value),
+            };
+        }
+
+        if repr.starts_with('r') {
+            if let Some(raw) = get_raw(&repr[1..]) {
+                return Ok(raw.to_owned());
+            }
+        }
+
+        Err(ConversionErrorKind::InvalidLiteralKind)
+    }
+
+    /// Returns the unescaped string value (including nul terminator) if this is
+    /// a c-string literal.
+    #[cfg(procmacro2_semver_exempt)]
+    pub fn cstr_value(&self) -> Result<Vec<u8>, ConversionErrorKind> {
+        let repr = self.to_string();
+
+        if repr.starts_with("c\"") && repr[2..].ends_with('"') {
+            let quoted = &repr[2..repr.len() - 1];
+            let mut value = Vec::with_capacity(quoted.len());
+            let mut error = None;
+            rustc_literal_escaper::unescape_c_str(quoted, |_range, res| match res {
+                Ok(MixedUnit::Char(ch)) => {
+                    value.extend_from_slice(ch.get().encode_utf8(&mut [0; 4]).as_bytes());
+                }
+                Ok(MixedUnit::HighByte(byte)) => value.push(byte.get()),
+                Err(err) => {
+                    if err.is_fatal() {
+                        error = Some(ConversionErrorKind::FailedToUnescape(err));
+                    }
+                }
+            });
+            return match error {
+                Some(error) => Err(error),
+                None => {
+                    value.push(b'\0');
+                    Ok(value)
+                }
+            };
+        }
+
+        if repr.starts_with("cr") {
+            if let Some(raw) = get_raw(&repr[2..]) {
+                let mut value = Vec::with_capacity(raw.len() + 1);
+                value.extend_from_slice(raw.as_bytes());
+                value.push(b'\0');
+                return Ok(value);
+            }
+        }
+
+        Err(ConversionErrorKind::InvalidLiteralKind)
+    }
+
+    /// Returns the unescaped string value if this is a byte string literal.
+    #[cfg(procmacro2_semver_exempt)]
+    pub fn byte_str_value(&self) -> Result<Vec<u8>, ConversionErrorKind> {
+        let repr = self.to_string();
+
+        if repr.starts_with("b\"") && repr[2..].ends_with('"') {
+            let quoted = &repr[2..repr.len() - 1];
+            let mut value = Vec::with_capacity(quoted.len());
+            let mut error = None;
+            rustc_literal_escaper::unescape_byte_str(quoted, |_range, res| match res {
+                Ok(byte) => value.push(byte),
+                Err(err) => {
+                    if err.is_fatal() {
+                        error = Some(ConversionErrorKind::FailedToUnescape(err));
+                    }
+                }
+            });
+            return match error {
+                Some(error) => Err(error),
+                None => Ok(value),
+            };
+        }
+
+        if repr.starts_with("br") {
+            if let Some(raw) = get_raw(&repr[2..]) {
+                return Ok(raw.as_bytes().to_owned());
+            }
+        }
+
+        Err(ConversionErrorKind::InvalidLiteralKind)
+    }
+
     // Intended for the `quote!` macro to use when constructing a proc-macro2
     // token out of a macro_rules $:literal token, which is already known to be
     // a valid literal. This avoids reparsing/validating the literal's string
@@ -1299,6 +1415,33 @@ impl Display for Literal {
     }
 }
 
+/// Error when retrieving a string literal's unescaped value.
+#[cfg(procmacro2_semver_exempt)]
+#[derive(Debug, PartialEq, Eq)]
+pub enum ConversionErrorKind {
+    /// The literal is of the right string kind, but its contents are malformed
+    /// in a way that cannot be unescaped to a value.
+    FailedToUnescape(EscapeError),
+    /// The literal is not of the string kind whose value was requested, for
+    /// example byte string vs UTF-8 string.
+    InvalidLiteralKind,
+}
+
+// ###"..."### -> ...
+#[cfg(procmacro2_semver_exempt)]
+fn get_raw(repr: &str) -> Option<&str> {
+    let pounds = repr.len() - repr.trim_start_matches('#').len();
+    if repr.len() >= pounds + 1 + 1 + pounds
+        && repr[pounds..].starts_with('"')
+        && repr.trim_end_matches('#').len() + pounds == repr.len()
+        && repr[..repr.len() - pounds].ends_with('"')
+    {
+        Some(&repr[pounds + 1..repr.len() - pounds - 1])
+    } else {
+        None
+    }
+}
+
 /// Public implementation details for the `TokenStream` type, such as iterators.
 pub mod token_stream {
     use crate::marker::{ProcMacroAutoTraits, MARKER};

src/rustc_literal_escaper.rs

@@ -1,3 +1,6 @@
+// Vendored from rustc-literal-escaper v0.0.5
+// https://github.com/rust-lang/literal-escaper/tree/v0.0.5
+
 //! Utilities for validating (raw) string, char, and byte literals and
 //! turning escape sequences into the values they represent.
 
@@ -6,9 +9,6 @@ use std::num::NonZero;
 use std::ops::Range;
 use std::str::Chars;
 
-#[cfg(test)]
-mod tests;
-
 /// Errors and warnings that can occur during string, char, and byte unescaping.
 ///
 /// Mostly relating to malformed escape sequences, but also a few other problems.