Implement char extension in Rust tools

bril-rs/Cargo.toml

@@ -25,6 +25,7 @@ ssa = []
 speculate = []
 position = []
 import = []
+char = []
 
 [[example]]
 name = "bril2txt"
@@ -33,10 +34,10 @@ path = "examples/bril2txt.rs"
 # However this currently does not work as expected and is being hashed out in https://github.com/rust-lang/rfcs/pull/3020 and https://github.com/rust-lang/rfcs/pull/2887
 # Until a solution is reached, I'm using `required-features` so that these features must be passed by flag. This is less ergonomic at the moment, however the user will get a nicer error that they need a feature flag instead of an Result::unwrap() error.
 # Note: See dev-dependencies for a hack to not need the user to pass that feature flag.
-required-features = ["memory", "float", "ssa", "speculate", "position", "import"]
+required-features = ["memory", "float", "ssa", "speculate", "position", "import", "char"]
 
 [dev-dependencies]
 # trick to enable all features in test
 # This is actually really hacky because it is used in all tests/examples/benchmarks but since we currently only have one example this works for enabling the following feature flags for our users.
 # If the above rfcs every get resolved, then dev-dependencies will no longer be needed.
-bril-rs = { path = ".", features = ["memory", "float", "ssa", "speculate", "position", "import"] }
+bril-rs = { path = ".", features = ["memory", "float", "ssa", "speculate", "position", "import", "char"] }

bril-rs/bril2json/Cargo.toml

@@ -25,4 +25,4 @@ lalrpop = "0.20"
 [dependencies.bril-rs]
 version = "0.1.0"
 path = "../../bril-rs"
-features = ["ssa", "memory", "float", "speculate", "position", "import"]
+features = ["ssa", "memory", "float", "speculate", "position", "import", "char"]

bril-rs/bril2json/src/bril_grammar.lalrpop

@@ -20,7 +20,7 @@
 
 use std::str::FromStr;
 use std::path::PathBuf;
-use crate::{Lines, ParsingArgs};
+use crate::{Lines, ParsingArgs, escape_control_chars};
 use bril_rs::{AbstractProgram, AbstractFunction, AbstractArgument, AbstractCode, AbstractInstruction, ConstOps, AbstractType, Literal, Import, ImportedFunction};
 
 grammar(lines : &Lines);
@@ -33,6 +33,7 @@ match {
     r"(\+|-)?[0-9]+" => INT_TOKEN, // int
     r"(\+|-)?(((([0-9]+\.?[0-9]*)|(\.[0-9]+))(E|e)(\+|-)?[0-9]+)|(([0-9]+\.[0-9]*)|(\.[0-9]+)))" => FLOAT_TOKEN, // https://stackoverflow.com/questions/12643009/regular-expression-for-floating-point-numbers
     r"(_|%|[[:alpha:]])(_|%|\.|[[:alnum:]])*" => IDENT_TOKEN,
+    r"('.')|('\\[0abtnvfr]')" => CHAR_TOKEN,
     r#""[^"]*""# => STRING_TOKEN,
     _
 }
@@ -189,6 +190,7 @@ Literal: Literal = {
     <n: Num> => Literal::Int(n),
     <b: Bool> => Literal::Bool(b),
     <f: Float> => Literal::Float(f),
+    <c: Char> => Literal::Char(c),
 }
 
 Num: i64 = <s:INT_TOKEN> => i64::from_str(s).unwrap();
@@ -199,6 +201,8 @@ Bool: bool = {
 
 Float: f64 = <f:FLOAT_TOKEN> => f64::from_str(f).unwrap();
 
+Char: char = <c:CHAR_TOKEN> => {let c = c.trim_matches('\''); escape_control_chars(c).unwrap()};
+
 // https://lalrpop.github.io/lalrpop/tutorial/006_macros.html
 Comma<T>: Vec<T> = { // (1)
     <mut v:(<T> ",")*> <e:T?> => match e { // (2)

bril-rs/bril2json/src/lib.rs

@@ -12,6 +12,23 @@ use std::fs::File;
 
 use bril_rs::{AbstractProgram, ColRow, Position};
 
+/// A helper function for processing the accepted Bril characters from their text representation
+#[must_use]
+pub fn escape_control_chars(s: &str) -> Option<char> {
+    match s {
+        "\\0" => Some('\u{0000}'),
+        "\\a" => Some('\u{0007}'),
+        "\\b" => Some('\u{0008}'),
+        "\\t" => Some('\u{0009}'),
+        "\\n" => Some('\u{000A}'),
+        "\\v" => Some('\u{000B}'),
+        "\\f" => Some('\u{000C}'),
+        "\\r" => Some('\u{000D}'),
+        s if s.chars().count() == 1 => s.chars().next(),
+        _ => None,
+    }
+}
+
 #[doc(hidden)]
 #[derive(Clone)]
 pub struct Lines {

bril-rs/brild/src/lib.rs

@@ -2,6 +2,8 @@
 #![warn(missing_docs)]
 #![doc = include_str!("../README.md")]
 #![allow(clippy::module_name_repetitions)]
+// todo until multiple versions of some really far down dependency are fixed
+#![allow(clippy::multiple_crate_versions)]
 
 #[doc(hidden)]
 pub mod cli;

bril-rs/brillvm/Makefile

@@ -2,13 +2,13 @@ TESTS := ../../test/interp/core/*.bril \
          ../../test/interp/float/*.bril \
 		 ../../test/interp/ssa/*.bril \
 		 ../../test/interp/mem/*.bril \
-		 ../../test/interp/mixed/*.bril \
+		 ../../test/interp/mixed/*[^r].bril # A hack to exclude store-char.bril by excluding any file ending in r.bril
 
 # Currently ignoring the Cholesky benchmark because of (probably) a floating point rounding bug.
 BENCHMARKS := ../../benchmarks/core/*.bril \
 			  ../../benchmarks/float/*.bril \
 			  ../../benchmarks/mem/*.bril \
-			  ../../benchmarks/mixed/[!cholesky]*.bril
+			  ../../benchmarks/mixed/[^c]*.bril # A hack to exclude cholesky.bril by excluding any file starting in c.
 
 clean:
 	cargo clean

bril-rs/brillvm/src/lib.rs

@@ -2,6 +2,8 @@
 #![allow(clippy::too_many_lines)]
 #![allow(clippy::needless_for_each)]
 #![doc = include_str!("../README.md")]
+// When you run with --all-targets, you also get --target=all which includes --target=redox. This pulls in redox_sys via a chain of deps through inkwell-parking_lot which uses an older version of bitflags 1.3.2. Given backwards compatibility, it's going to be a very long time, if ever, that this gets updated(because of msrv changes).
+#![allow(clippy::multiple_crate_versions)]
 
 #[doc(hidden)]
 pub mod cli;

bril-rs/src/conversion.rs

@@ -245,6 +245,20 @@ impl TryFrom<AbstractInstruction> for Instruction {
                     "fle" => ValueOps::Fle,
                     #[cfg(feature = "float")]
                     "fge" => ValueOps::Fge,
+                    #[cfg(feature = "char")]
+                    "ceq" => ValueOps::Ceq,
+                    #[cfg(feature = "char")]
+                    "clt" => ValueOps::Clt,
+                    #[cfg(feature = "char")]
+                    "cgt" => ValueOps::Cgt,
+                    #[cfg(feature = "char")]
+                    "cle" => ValueOps::Cle,
+                    #[cfg(feature = "char")]
+                    "cge" => ValueOps::Cge,
+                    #[cfg(feature = "char")]
+                    "char2int" => ValueOps::Char2int,
+                    #[cfg(feature = "char")]
+                    "int2char" => ValueOps::Int2char,
                     #[cfg(feature = "memory")]
                     "alloc" => ValueOps::Alloc,
                     #[cfg(feature = "memory")]
@@ -313,6 +327,8 @@ impl TryFrom<AbstractType> for Type {
             AbstractType::Primitive(t) if t == "bool" => Self::Bool,
             #[cfg(feature = "float")]
             AbstractType::Primitive(t) if t == "float" => Self::Float,
+            #[cfg(feature = "char")]
+            AbstractType::Primitive(t) if t == "char" => Self::Char,
             AbstractType::Primitive(t) => return Err(ConversionError::InvalidPrimitive(t)),
             #[cfg(feature = "memory")]
             AbstractType::Parameterized(t, ty) if t == "ptr" => {

bril-rs/src/program.rs

@@ -451,6 +451,27 @@ pub enum ValueOps {
     /// <https://capra.cs.cornell.edu/bril/lang/float.html#operations>
     #[cfg(feature = "float")]
     Fge,
+    /// <https://capra.cs.cornell.edu/bril/lang/char.html#operations>
+    #[cfg(feature = "char")]
+    Ceq,
+    /// <https://capra.cs.cornell.edu/bril/lang/char.html#operations>
+    #[cfg(feature = "char")]
+    Clt,
+    /// <https://capra.cs.cornell.edu/bril/lang/char.html#operations>
+    #[cfg(feature = "char")]
+    Cgt,
+    /// <https://capra.cs.cornell.edu/bril/lang/char.html#operations>
+    #[cfg(feature = "char")]
+    Cle,
+    /// <https://capra.cs.cornell.edu/bril/lang/char.html#operations>
+    #[cfg(feature = "char")]
+    Cge,
+    /// <https://capra.cs.cornell.edu/bril/lang/char.html#operations>
+    #[cfg(feature = "char")]
+    Char2int,
+    /// <https://capra.cs.cornell.edu/bril/lang/char.html#operations>
+    #[cfg(feature = "char")]
+    Int2char,
     /// <https://capra.cs.cornell.edu/bril/lang/memory.html#operations>
     #[cfg(feature = "memory")]
     Alloc,
@@ -499,6 +520,20 @@ impl Display for ValueOps {
             Self::Fle => write!(f, "fle"),
             #[cfg(feature = "float")]
             Self::Fge => write!(f, "fge"),
+            #[cfg(feature = "char")]
+            Self::Ceq => write!(f, "ceq"),
+            #[cfg(feature = "char")]
+            Self::Clt => write!(f, "clt"),
+            #[cfg(feature = "char")]
+            Self::Cgt => write!(f, "cgt"),
+            #[cfg(feature = "char")]
+            Self::Cle => write!(f, "cle"),
+            #[cfg(feature = "char")]
+            Self::Cge => write!(f, "cge"),
+            #[cfg(feature = "char")]
+            Self::Char2int => write!(f, "char2int"),
+            #[cfg(feature = "char")]
+            Self::Int2char => write!(f, "int2char"),
             #[cfg(feature = "memory")]
             Self::Alloc => write!(f, "alloc"),
             #[cfg(feature = "memory")]
@@ -520,6 +555,9 @@ pub enum Type {
     /// <https://capra.cs.cornell.edu/bril/lang/float.html#types>
     #[cfg(feature = "float")]
     Float,
+    /// <https://capra.cs.cornell.edu/bril/lang/char.html#types>
+    #[cfg(feature = "char")]
+    Char,
     /// <https://capra.cs.cornell.edu/bril/lang/memory.html#types>
     #[cfg(feature = "memory")]
     #[serde(rename = "ptr")]
@@ -533,6 +571,8 @@ impl Display for Type {
             Self::Bool => write!(f, "bool"),
             #[cfg(feature = "float")]
             Self::Float => write!(f, "float"),
+            #[cfg(feature = "char")]
+            Self::Char => write!(f, "char"),
             #[cfg(feature = "memory")]
             Self::Pointer(tpe) => write!(f, "ptr<{tpe}>"),
         }
@@ -551,6 +591,9 @@ pub enum Literal {
     /// Floating Points
     #[cfg(feature = "float")]
     Float(f64),
+    /// UTF-16 Characters
+    #[cfg(feature = "char")]
+    Char(char),
 }
 
 impl Display for Literal {
@@ -560,10 +603,27 @@ impl Display for Literal {
             Self::Bool(b) => write!(f, "{b}"),
             #[cfg(feature = "float")]
             Self::Float(x) => write!(f, "{x}"),
+            #[cfg(feature = "char")]
+            Self::Char(c) => write!(f, "\'{}\'", escape_char(*c)),
         }
     }
 }
 
+#[cfg(feature = "char")]
+fn escape_char(c: char) -> String {
+    match c {
+        '\u{0000}' => "\\0".to_string(),
+        '\u{0007}' => "\\a".to_string(),
+        '\u{0008}' => "\\b".to_string(),
+        '\u{0009}' => "\\t".to_string(),
+        '\u{000A}' => "\\n".to_string(),
+        '\u{000B}' => "\\v".to_string(),
+        '\u{000C}' => "\\f".to_string(),
+        '\u{000D}' => "\\r".to_string(),
+        c => c.to_string(),
+    }
+}
+
 impl Literal {
     /// A helper function to get the type of literal values
     #[must_use]
@@ -573,6 +633,8 @@ impl Literal {
             Self::Bool(_) => Type::Bool,
             #[cfg(feature = "float")]
             Self::Float(_) => Type::Float,
+            #[cfg(feature = "char")]
+            Self::Char(_) => Type::Char,
         }
     }
 }

brilift/Makefile

@@ -9,7 +9,7 @@ endif
 
 # Brilift only supports core Bril for now, so we select those tests &
 # benchmarks.
-TESTS := ../test/interp/core/*.bril ../test/interp/float/*.bril ../test/interp/mem/*.bril ../test/interp/mixed/*.bril
+TESTS := ../test/interp/core/*.bril ../test/interp/float/*.bril ../test/interp/mem/*.bril ../test/interp/mixed/*[^r].bril # A hack to exclude store-char.bril by excluding any file ending in r.bril
 BENCHMARKS := ../benchmarks/core/*.bril ../benchmarks/float/*.bril ../benchmarks/mem/*.bril ../benchmarks/mixed/*.bril
 
 CFLAGS := $(if $(TARGET),-target $(TARGET))

brilift/run.sh

@@ -10,7 +10,7 @@ set -e
 #
 
 OS=$(uname -m)
-if [[ "$OS" == 'arm64' ]]; then
+if [ "$OS" = 'arm64' ]; then
     export TARGET=x86_64-unknown-darwin-macho
 fi
 

brilirs/Cargo.toml

@@ -13,20 +13,20 @@ keywords = ["compiler", "bril", "interpreter", "data-structures", "language"]
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 [build-dependencies]
-clap         = { version = "4.2", features = ["derive"] }
-clap_complete= { version = "4.2", optional = true }
+clap         = { version = "4.3", features = ["derive"] }
+clap_complete= { version = "4.3", optional = true }
 
 [dependencies]
 thiserror    = "1.0"
-clap         = { version = "4.2", features = ["derive"] }
+clap         = { version = "4.3", features = ["derive"] }
 fxhash       = "0.2"
 mimalloc     = "0.1"
 itoa         = "1.0"
 
 [dependencies.bril-rs]
 version      = "0.1.0"
 path         = "../bril-rs"
-features     = ["ssa", "memory", "float", "speculate"]
+features     = ["ssa", "memory", "float", "speculate", "char"]
 
 [dependencies.bril2json]
 version      = "0.1.0"
@@ -39,4 +39,4 @@ lto = true
 panic = "abort"
 
 [features]
-completions = ["clap_complete"]
+completions = ["clap_complete"]

brilirs/Makefile

@@ -2,6 +2,7 @@ TESTS := ../test/check/*.bril \
 ../test/interp*/core*/*.bril \
 ../test/interp*/float/*.bril \
 ../test/interp*/mem*/*.bril \
+../test/interp*/char*/*.bril \
 ../test/interp*/mixed/*.bril \
 ../test/interp*/ssa*/*.bril \
 

brilirs/src/check.rs

@@ -210,6 +210,55 @@ fn type_check_instruction<'a>(
       check_asmt_type(&Type::Bool, op_type)?;
       update_env(env, dest, op_type)
     }
+    Instruction::Value {
+      op: ValueOps::Ceq | ValueOps::Cge | ValueOps::Clt | ValueOps::Cgt | ValueOps::Cle,
+      args,
+      dest,
+      funcs,
+      labels,
+      pos: _,
+      op_type,
+    } => {
+      check_num_args(2, args)?;
+      check_num_funcs(0, funcs)?;
+      check_num_labels(0, labels)?;
+      check_asmt_type(&Type::Char, get_type(env, 0, args)?)?;
+      check_asmt_type(&Type::Char, get_type(env, 1, args)?)?;
+      check_asmt_type(&Type::Bool, op_type)?;
+      update_env(env, dest, op_type)
+    }
+    Instruction::Value {
+      op: ValueOps::Char2int,
+      args,
+      dest,
+      funcs,
+      labels,
+      pos: _,
+      op_type,
+    } => {
+      check_num_args(1, args)?;
+      check_num_funcs(0, funcs)?;
+      check_num_labels(0, labels)?;
+      check_asmt_type(&Type::Char, get_type(env, 0, args)?)?;
+      check_asmt_type(&Type::Int, op_type)?;
+      update_env(env, dest, op_type)
+    }
+    Instruction::Value {
+      op: ValueOps::Int2char,
+      args,
+      dest,
+      funcs,
+      labels,
+      pos: _,
+      op_type,
+    } => {
+      check_num_args(1, args)?;
+      check_num_funcs(0, funcs)?;
+      check_num_labels(0, labels)?;
+      check_asmt_type(&Type::Int, get_type(env, 0, args)?)?;
+      check_asmt_type(&Type::Char, op_type)?;
+      update_env(env, dest, op_type)
+    }
     Instruction::Value {
       op: ValueOps::Call,
       dest,

brilirs/src/error.rs

@@ -21,6 +21,8 @@ pub enum InterpError {
   NoMainFunction,
   #[error("phi node has unequal numbers of labels and args")]
   UnequalPhiNode,
+  #[error("char must have one character")]
+  NotOneChar,
   #[error("multiple functions of the same name found")]
   DuplicateFunction,
   #[error("Expected empty return for `{0}`, found value")]
@@ -53,6 +55,8 @@ pub enum InterpError {
   BadAsmtType(bril_rs::Type, bril_rs::Type), // (expected, actual). For when the LHS type of an instruction is bad
   #[error("There has been an io error: `{0:?}`")]
   IoError(#[from] std::io::Error),
+  #[error("value ${0} cannot be converted to char")]
+  ToCharError(i64),
   #[error("You probably shouldn't see this error, this is here to handle conversions between InterpError and PositionalError")]
   PositionalInterpErrorConversion(#[from] PositionalInterpError),
 }