treeshake imports based on guest content

crates/spidermonkey-embedding-splicer/src/bindgen.rs

@@ -124,7 +124,13 @@ pub struct Componentization {
     pub resource_imports: Vec<(String, String, u32)>,
 }
 
-pub fn componentize_bindgen(resolve: &Resolve, id: WorldId, name: &str) -> Componentization {
+pub fn componentize_bindgen(
+    resolve: &Resolve,
+    id: WorldId,
+    name: &str,
+    guest_imports: &Vec<String>,
+    guest_exports: &Vec<String>,
+) -> Componentization {
     let mut bindgen = JsBindgen {
         src: Source::default(),
         esm_bindgen: EsmBindgen::default(),
@@ -147,9 +153,9 @@ pub fn componentize_bindgen(resolve: &Resolve, id: WorldId, name: &str) -> Compo
         .local_names
         .exclude_globals(Intrinsic::get_global_names());
 
-    bindgen.imports_bindgen();
+    bindgen.imports_bindgen(&guest_imports);
 
-    bindgen.exports_bindgen();
+    bindgen.exports_bindgen(&guest_exports);
     bindgen.esm_bindgen.populate_export_aliases();
 
     // consolidate import specifiers and generate wrappers
@@ -342,6 +348,7 @@ pub fn componentize_bindgen(resolve: &Resolve, id: WorldId, name: &str) -> Compo
         "$source_mod",
         &mut bindgen.local_names,
         name,
+        &guest_exports,
     );
 
     let js_intrinsics = render_intrinsics(&mut bindgen.all_intrinsics, false, true);
@@ -363,9 +370,28 @@ impl JsBindgen<'_> {
         return intrinsic.name().to_string();
     }
 
-    fn exports_bindgen(&mut self) {
+    fn exports_bindgen(&mut self, guest_exports: &Vec<String>) {
         for (key, export) in &self.resolve.worlds[self.world].exports {
             let name = self.resolve.name_world_key(key);
+            // TODO: figure out how to go from "run" -> wasi:cli/[email protected] and
+            // "incomingHandler" -> wasi:http/[email protected] and in
+            // general go from the sugared up names to explicit name. This is
+            // just to make sure some random export does not mess up the exports
+            // that a components to need to export
+            if name == "wasi:http/[email protected]"
+                && !guest_exports.contains(&"incomingHandler".to_string())
+                && !guest_exports.contains(&"wasi:http/[email protected]".to_string())
+            {
+                continue;
+            }
+
+            if name == "wasi:cli/[email protected]"
+                && !guest_exports.contains(&"run".to_string())
+                && !guest_exports.contains(&"wasi:cli/[email protected]".to_string())
+            {
+                continue;
+            }
+
             match export {
                 WorldItem::Function(func) => {
                     let local_name = self.local_names.create_once(&func.name).to_string();
@@ -451,9 +477,12 @@ impl JsBindgen<'_> {
         }
     }
 
-    fn imports_bindgen(&mut self) {
+    fn imports_bindgen(&mut self, guest_imports: &Vec<String>) {
         for (key, impt) in &self.resolve.worlds[self.world].imports {
             let import_name = self.resolve.name_world_key(key);
+            if !guest_imports.contains(&import_name) {
+                continue;
+            }
             match &impt {
                 WorldItem::Function(f) => {
                     self.import_bindgen(import_name, f, false, None);
@@ -1001,6 +1030,7 @@ impl EsmBindgen {
         imports_object: &str,
         _local_names: &mut LocalNames,
         source_name: &str,
+        guest_exports: &Vec<String>,
     ) {
         // TODO: bring back these validations of imports
         // including using the flattened bindings
@@ -1043,6 +1073,11 @@ impl EsmBindgen {
                 ");
         }
         for (export_name, binding) in &self.exports {
+            if export_name == "wasi:http/[email protected]"
+                && !guest_exports.contains(&"incomingHandler".to_string())
+            {
+                continue;
+            }
             match binding {
                 Binding::Interface(bindings) => {
                     uwrite!(output, "const ");

crates/spidermonkey-embedding-splicer/src/lib.rs

@@ -1,6 +1,9 @@
 use anyhow::{bail, Context, Result};
 use bindgen::BindingItem;
-use std::path::{Path, PathBuf};
+use std::{
+    path::{Path, PathBuf},
+    vec,
+};
 
 mod bindgen;
 mod splice;
@@ -110,6 +113,8 @@ impl Guest for SpidermonkeyEmbeddingSplicerComponent {
         wit_source: Option<String>,
         wit_path: Option<String>,
         world_name: Option<String>,
+        mut guest_imports: Vec<String>,
+        guest_exports: Vec<String>,
         debug: bool,
     ) -> Result<SpliceResult, String> {
         let source_name = source_name.unwrap_or("source.js".to_string());
@@ -131,7 +136,6 @@ impl Guest for SpidermonkeyEmbeddingSplicerComponent {
             .map_err(|e| e.to_string())?;
 
         let mut wasm_bytes = wit_component::dummy_module(&resolve, world);
-        let componentized = bindgen::componentize_bindgen(&resolve, world, &source_name);
 
         // merge the engine world with the target world, retaining the engine producers
         let producers = if let Ok((
@@ -144,6 +148,11 @@ impl Guest for SpidermonkeyEmbeddingSplicerComponent {
             },
         )) = decode(&engine)
         {
+            // merge the imports from the engine with the imports from the guest content.
+            for (k, _) in &engine_resolve.worlds[engine_world].imports {
+                guest_imports.push(engine_resolve.name_world_key(k));
+            }
+
             // we disable the engine run and incoming handler as we recreate these exports
             // when needed, so remove these from the world before initiating the merge
             let maybe_run = engine_resolve.worlds[engine_world]
@@ -187,6 +196,14 @@ impl Guest for SpidermonkeyEmbeddingSplicerComponent {
             None
         };
 
+        let componentized = bindgen::componentize_bindgen(
+            &resolve,
+            world,
+            &source_name,
+            &guest_imports,
+            &guest_exports,
+        );
+
         let encoded = wit_component::metadata::encode(
             &resolve,
             world,
@@ -327,7 +344,6 @@ impl Guest for SpidermonkeyEmbeddingSplicerComponent {
             ));
         }
 
-        // println!("{:?}", &imports);
         // println!("{:?}", &componentized.imports);
         // println!("{:?}", &exports);
         let mut wasm =

crates/spidermonkey-embedding-splicer/src/splice.rs

@@ -38,20 +38,30 @@ pub fn splice(
     let mut module = config.parse(&engine)?;
 
     // since StarlingMonkey implements CLI Run and incoming handler,
-    // we override these in ComponentizeJS, removing them from the
-    // core function exports
-    if let Ok(run) = module.exports.get_func("wasi:cli/[email protected]#run") {
-        let expt = module.exports.get_exported_func(run).unwrap();
-        module.exports.delete(expt.id());
-        module.funcs.delete(run);
+    // we override them only if the guest content exports those functions
+    if exports
+        .iter()
+        .any(|(name, _)| name == "wasi:cli/[email protected]#run")
+    {
+        if let Ok(run) = module.exports.get_func("wasi:cli/[email protected]#run") {
+            let expt = module.exports.get_exported_func(run).unwrap();
+            module.exports.delete(expt.id());
+            module.funcs.delete(run);
+        }
     }
-    if let Ok(serve) = module
-        .exports
-        .get_func("wasi:http/[email protected]#handle")
+
+    if exports
+        .iter()
+        .any(|(name, _)| name == "wasi:http/[email protected]#handle")
     {
-        let expt = module.exports.get_exported_func(serve).unwrap();
-        module.exports.delete(expt.id());
-        module.funcs.delete(serve);
+        if let Ok(serve) = module
+            .exports
+            .get_func("wasi:http/[email protected]#handle")
+        {
+            let expt = module.exports.get_exported_func(serve).unwrap();
+            module.exports.delete(expt.id());
+            module.funcs.delete(serve);
+        }
     }
 
     // we reencode the WASI world component data, so strip it out from the

crates/spidermonkey-embedding-splicer/wit/spidermonkey-embedding-splicer.wit

@@ -33,5 +33,5 @@ world spidermonkey-embedding-splicer {
 
   export stub-wasi: func(engine: list<u8>, features: list<features>, wit-world: option<string>, wit-path: option<string>, world-name: option<string>) -> result<list<u8>, string>;
 
-  export splice-bindings: func(source-name: option<string>, spidermonkey-engine: list<u8>, wit-world: option<string>, wit-path: option<string>, world-name: option<string>, debug: bool) -> result<splice-result, string>;
+  export splice-bindings: func(source-name: option<string>, spidermonkey-engine: list<u8>, wit-world: option<string>, wit-path: option<string>, world-name: option<string>, guest-imports: list<string>, guest-exports: list<string>, debug: bool) -> result<splice-result, string>;
 }

src/componentize.js

@@ -47,12 +47,33 @@ export async function componentize(jsSource, witWorld, opts) {
     enableFeatures = [],
   } = opts || {};
 
+  await lexerInit;
+  let jsImports = [];
+  let jsExports = [];
+  try {
+    [jsImports, jsExports] = parse(jsSource);
+  } catch {
+    // ignore parser errors - will show up as engine parse errors shortly
+  }
+
+  let guestImports = []
+  jsImports.map(k => {
+    guestImports.push(k.n)
+  })
+
+  let guestExports = []
+  jsExports.map(k => {
+    guestExports.push(k.n)
+  })
+
   let { wasm, jsBindings, importWrappers, exports, imports } = spliceBindings(
     sourceName,
     await readFile(engine),
     witWorld,
     maybeWindowsPath(witPath),
     worldName,
+    guestImports,
+    guestExports,
     false
   );
 
@@ -103,13 +124,6 @@ export async function componentize(jsSource, witWorld, opts) {
   await writeFile(input, Buffer.from(wasm));
 
   // rewrite the JS source import specifiers to reference import wrappers
-  await lexerInit;
-  let jsImports = [];
-  try {
-    [jsImports] = parse(jsSource);
-  } catch {
-    // ignore parser errors - will show up as engine parse errors shortly
-  }
   let source = '',
     curIdx = 0;
   for (const jsImpt of jsImports) {