feat: created stdarch-gen-wasm to create spec file for wasm32

intrinsics

Author: madhav-madhusoodanan

Cargo.lock

@@ -0,0 +1,14 @@
+[package]
+name = "stdarch-gen-wasm"
+version = "0.1.0"
+authors = ["Madhav Madhusoodanan <[email protected]>"]
+edition = "2024"
+
+[dependencies]
+tree-sitter = "0.24"
+tree-sitter-rust = "0.23"
+tree-sitter-c = "0.23"
+clap = { version = "4.5.45", features = ["derive"] }
+
+[build-dependencies]
+cc="*"

crates/stdarch-gen-wasm/Cargo.toml

@@ -0,0 +1,13 @@
+use clap::Parser;
+
+#[derive(Parser, Debug)]
+#[command(version, about, long_about = None)]
+pub struct Args {
+    /// The Rust source file path argument
+    #[arg(short, long)]
+    pub c: String,
+
+    /// The Rust source file path argument
+    #[arg(short, long)]
+    pub rust: String,
+}

crates/stdarch-gen-wasm/src/cli.rs

@@ -0,0 +1,126 @@
+mod cli;
+mod matcher;
+mod structs;
+mod utils;
+
+use std::fs;
+
+use clap::Parser as ClapParser;
+use cli::Args;
+use tree_sitter::{Parser, Tree};
+
+use crate::matcher::match_intrinsic_definitions;
+use crate::structs::{CIntrinsic, RustIntrinsic};
+use crate::utils::leaf_nodes_from_grammar_name;
+
+/// Read the Rust source code and returns its AST
+fn process_rust_code(source: &String) -> (String, Tree) {
+    let mut parser = Parser::new();
+    parser
+        .set_language(&tree_sitter_rust::LANGUAGE.into())
+        .expect("Error loading Rust grammar");
+    let source_code = fs::read_to_string(source);
+
+    if let Err(ref err) = source_code {
+        eprintln!("Rust parsing error: {}", err);
+        panic!()
+    }
+    let source_code_string = source_code.unwrap();
+    let tree = parser.parse(source_code_string.clone(), None).unwrap();
+    (source_code_string, tree)
+}
+
+/// Reads the C source code and returns its AST
+fn process_c_code(source: &String) -> (String, Tree) {
+    let mut parser = Parser::new();
+    parser
+        .set_language(&tree_sitter_c::LANGUAGE.into())
+        .expect("Error loading Rust grammar");
+    let source_code = fs::read_to_string(source);
+
+    if let Err(ref err) = source_code {
+        eprintln!("C parsing error: {}", err);
+        panic!()
+    }
+    let source_code_string = source_code.unwrap();
+    let tree = parser.parse(source_code_string.clone(), None).unwrap();
+    (source_code_string, tree)
+}
+
+/// Creates an entry in the spec sheet that corresponds to a specific intrinsic
+fn generate_spec(c_intrinsic: &CIntrinsic, rust_intrinsic: &RustIntrinsic) -> String {
+    format!(
+        "/// {}
+c-intrinsic-name = {}
+c-arguments = {}
+c-arguments-data-types = {}
+c-return-type = {}
+rust-intrinsic-name = {}
+rust-arguments = {}
+rust-arguments-data-types = {}
+rust-const-generic-arguments = {}
+rust-const-generic-arguments-data-types = {}
+rust-return-type = {}",
+        rust_intrinsic.intrinsic,
+        c_intrinsic.intrinsic,
+        c_intrinsic.arg_names.join(", "),
+        c_intrinsic.arg_types.join(", "),
+        c_intrinsic.return_type.unwrap_or(""),
+        rust_intrinsic.intrinsic,
+        rust_intrinsic.arg_names.join(", "),
+        rust_intrinsic.arg_types.join(", "),
+        rust_intrinsic.generic_arg_names.join(", "),
+        rust_intrinsic.generic_arg_types.join(", "),
+        rust_intrinsic.return_type.unwrap_or(""),
+    )
+}
+
+/// Create the spec sheet.
+/// 
+/// Fields that would be present in the spec sheet:
+/// 1. c-intrinsic-name
+/// 2. c-arguments
+/// 3. c-arguments-data-types
+/// 4. c-return-type
+/// 5. rust-intrinsic-name
+/// 6. rust-arguments
+/// 7. rust-arguments-data-types
+/// 8. rust-const-generic-arguments
+/// 9. rust-const-generic-arguments-data-types
+/// 10. rust-return-type
+fn main() {
+    // Read the file-paths from CLI arguments
+    // obtain the tree of tokens from the code
+    let args = Args::parse();
+    let (c_source, c_tree) = process_c_code(&args.c);
+    let (rust_source, rust_tree) = process_rust_code(&args.rust);
+
+    let preproc_node = c_tree.root_node();
+
+    let c_intrinsics = leaf_nodes_from_grammar_name(preproc_node, "function_definition")
+        .iter()
+        .map(|&node| CIntrinsic::new(node, &c_source))
+        .collect::<Vec<_>>();
+
+    let mut rust_cursor = rust_tree.root_node().walk();
+    let rust_intrinsics = rust_tree
+        .root_node()
+        .children(&mut rust_cursor)
+        .filter(|node| node.grammar_name() == "function_item")
+        .map(|node| RustIntrinsic::new(node, &rust_source))
+        .collect::<Vec<_>>();
+
+    let matching_intrinsics = match_intrinsic_definitions(&c_intrinsics, &rust_intrinsics);
+    println!(
+        "// This code is automatically generated. DO NOT MODIFY.
+// Number of matched intrinsics: {}\n",
+        matching_intrinsics.len()
+    );
+
+    let spec_details = matching_intrinsics
+        .iter()
+        .map(|&(c_intrinsic, rust_intrinsic)| generate_spec(c_intrinsic, rust_intrinsic))
+        .collect::<Vec<_>>()
+        .join("\n\n");
+    println!("{}", spec_details);
+}

crates/stdarch-gen-wasm/src/main.rs

@@ -0,0 +1,47 @@
+use std::collections::HashSet;
+
+use crate::structs::{CIntrinsic, RustIntrinsic};
+
+/// Matches the set of intrinsics in Rust to their C counterpart.
+/// 
+/// This function assumes that the list of Rust definitions
+/// will be a subset of the list of definitions in C.
+pub fn match_intrinsic_definitions<'a>(
+    c_definitions: &'a Vec<CIntrinsic>,
+    rust_definitions: &'a Vec<RustIntrinsic>,
+) -> Vec<(&'a CIntrinsic<'a>, &'a RustIntrinsic<'a>)> {
+    // This function assumes that the list of Rust definitions
+    // will be a subset of the list of definitions in C
+
+    let mut matched_definitions: Vec<(&'a CIntrinsic, &'a RustIntrinsic)> = Vec::new();
+    matched_definitions.reserve(rust_definitions.len());
+
+    for rust_definition in rust_definitions.iter() {
+        let c_definition = c_definitions
+            .iter()
+            .find(|&c_def| match_intrinsic_definition(c_def.intrinsic, rust_definition.intrinsic));
+        if let Some(c_def) = c_definition {
+            matched_definitions.push((c_def, rust_definition));
+        }
+    }
+
+    matched_definitions
+}
+
+/// checks if the function name of the intrinsic in Rust
+/// matches that of the intrinsic in C.
+fn match_intrinsic_definition(c_definition: &str, rust_definition: &str) -> bool {
+    // Most intrinsics in C are of the format: `wasm_v128_load`.
+    // Its Rust counterpart is named `v128_load`.
+    // 
+    // Another one is `wasm_i8x16_const_splat`, and its Rust counterpart is `i8x16_splat`.
+    // 
+    // The pattern that is observed is that, each keyword "chunk" that constructs 
+    // the intrinsic name in Rust will also be used to construct the intrinsic name in C.
+    // These names are constructed by joining the chunks with an underscore (_).
+
+    let c_definition_split: HashSet<_> = c_definition.split('_').collect();
+    rust_definition
+        .split('_')
+        .all(|keyword| c_definition_split.contains(keyword))
+}