feat(xtask): check wasm exports

.github/workflows/wasm_exports.yml

@@ -0,0 +1,36 @@
+name: Check WASM Exports
+
+on:
+  pull_request:
+    paths:
+      - lib/include/tree_sitter/api.h
+      - lib/binding_web/**
+  push:
+    branches: [master]
+    paths:
+      - lib/include/tree_sitter/api.h
+      - lib/binding_rust/bindings.rs
+      - lib/CMakeLists.txt
+
+jobs:
+  check-wasm-exports:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Set up stable Rust toolchain
+        uses: actions-rust-lang/setup-rust-toolchain@v1
+        with:
+          toolchain: stable
+
+      - name: Install wasm-objdump
+        run: sudo apt-get update -y && sudo apt-get install -y wabt
+
+      - name: Build C library (make)
+        run: make -j CFLAGS="$CFLAGS"
+        env:
+          CFLAGS: -g -Werror -Wall -Wextra -Wshadow -Wpedantic -Werror=incompatible-pointer-types
+
+      - name: Check WASM exports
+        run: cargo xtask check-wasm-exports

xtask/src/check_wasm_exports.rs

@@ -0,0 +1,127 @@
+use std::{
+    collections::HashSet,
+    io::BufRead,
+    process::{Command, Stdio},
+};
+
+use anyhow::{anyhow, Result};
+
+use crate::{bail_on_err, build_wasm::run_wasm, BuildWasm};
+
+const EXCLUDES: [&str; 28] = [
+    // Unneeded because the JS side has its own way of implementing it
+    "ts_node_child_by_field_name",
+    "ts_node_edit",
+    // Precomputed and stored in the JS side
+    "ts_node_type",
+    "ts_node_grammar_type",
+    "ts_node_eq",
+    "ts_tree_cursor_current_field_name",
+    "ts_lookahead_iterator_current_symbol_name",
+    // Deprecated
+    "ts_node_child_containing_descendant",
+    // Not used in wasm
+    "ts_init",
+    "ts_set_allocator",
+    "ts_parser_set_cancellation_flag",
+    "ts_parser_cancellation_flag",
+    "ts_parser_print_dot_graphs",
+    "ts_tree_print_dot_graph",
+    "ts_parser_set_wasm_store",
+    "ts_parser_take_wasm_store",
+    "ts_parser_language",
+    "ts_node_language",
+    "ts_tree_language",
+    "ts_lookahead_iterator_language",
+    "ts_parser_logger",
+    "ts_parser_parse_string",
+    "ts_parser_parse_string_encoding",
+    // Query cursor is not managed by user in web bindings
+    "ts_query_cursor_delete",
+    "ts_query_cursor_timeout_micros",
+    "ts_query_cursor_match_limit",
+    "ts_query_cursor_remove_match",
+    "ts_query_cursor_timeout_micros",
+];
+
+pub fn run() -> Result<()> {
+    // Build the wasm module with debug symbols for wasm-objdump
+    run_wasm(&BuildWasm {
+        debug: true,
+        verbose: false,
+        docker: false,
+    })?;
+
+    let mut wasm_exports = include_str!("../../lib/binding_web/exports.txt")
+        .lines()
+        .map(|s| s.replace("_wasm", "").replace("byte", "index"))
+        // remove leading and trailing quotes, trailing comma
+        .map(|s| s[1..s.len() - 2].to_string())
+        .collect::<HashSet<_>>();
+
+    // Run wasm-objdump to see symbols used internally in binding.c but not exposed in any way.
+    let wasm_objdump = Command::new("wasm-objdump")
+        .args([
+            "--details",
+            "lib/binding_web/tree-sitter.wasm",
+            "--section",
+            "Name",
+        ])
+        .output()
+        .expect("Failed to run wasm-objdump");
+    bail_on_err(&wasm_objdump, "Failed to run wasm-objdump")?;
+
+    wasm_exports.extend(
+        wasm_objdump
+            .stdout
+            .lines()
+            .map_while(Result::ok)
+            .skip_while(|line| !line.contains("- func"))
+            .filter_map(|line| {
+                if line.contains("func") {
+                    if let Some(function) = line.split_whitespace().nth(2).map(String::from) {
+                        let trimmed = function.trim_start_matches('<').trim_end_matches('>');
+                        if trimmed.starts_with("ts") && !trimmed.contains("__") {
+                            return Some(trimmed.to_string());
+                        }
+                    }
+                }
+                None
+            }),
+    );
+
+    let nm_child = Command::new("nm")
+        .arg("-W")
+        .arg("-U")
+        .arg("libtree-sitter.so")
+        .stdout(Stdio::piped())
+        .output()
+        .expect("Failed to run nm");
+    bail_on_err(&nm_child, "Failed to run nm")?;
+    let export_reader = nm_child
+        .stdout
+        .lines()
+        .map_while(Result::ok)
+        .filter(|line| line.contains(" T "));
+
+    let exports = export_reader
+        .filter_map(|line| line.split_whitespace().nth(2).map(String::from))
+        .filter(|symbol| !EXCLUDES.contains(&symbol.as_str()))
+        .collect::<HashSet<_>>();
+
+    let mut missing = exports
+        .iter()
+        .filter(|&symbol| !wasm_exports.contains(symbol))
+        .map(|symbol| symbol.as_str())
+        .collect::<Vec<_>>();
+    missing.sort_unstable();
+
+    if !missing.is_empty() {
+        Err(anyhow!(format!(
+            "Unmatched wasm exports:\n{}",
+            missing.join("\n")
+        )))?;
+    }
+
+    Ok(())
+}

xtask/src/main.rs

@@ -1,6 +1,7 @@
 mod benchmark;
 mod build_wasm;
 mod bump;
+mod check_wasm_exports;
 mod clippy;
 mod fetch;
 mod generate;
@@ -28,6 +29,8 @@ enum Commands {
     BuildWasmStdlib,
     /// Bumps the version of the workspace.
     BumpVersion(BumpVersion),
+    /// Checks that WASM exports are synced.
+    CheckWasmExports,
     /// Runs `cargo clippy`.
     Clippy(Clippy),
     /// Fetches emscripten.
@@ -204,6 +207,7 @@ fn run() -> Result<()> {
         Commands::BuildWasm(build_wasm_options) => build_wasm::run_wasm(&build_wasm_options)?,
         Commands::BuildWasmStdlib => build_wasm::run_wasm_stdlib()?,
         Commands::BumpVersion(bump_options) => bump::run(bump_options)?,
+        Commands::CheckWasmExports => check_wasm_exports::run()?,
         Commands::Clippy(clippy_options) => clippy::run(&clippy_options)?,
         Commands::FetchEmscripten => fetch::run_emscripten()?,
         Commands::FetchFixtures => fetch::run_fixtures()?,