Overhaul how type information gets to the CLI

This commit is a complete overhaul of how the `#[wasm_bindgen]` macro
communicates type information to the CLI tool, and it's done in a somewhat...
unconventional fashion.

Today we've got a problem where the generated JS needs to understand the types
of each function exported or imported. This understanding is what enables it to
generate the appropriate JS wrappers and such. We want to, however, be quite
flexible and extensible in types that are supported across the boundary, which
means that internally we rely on the trait system to resolve what's what.

Communicating the type information historically was done by creating a four byte
"descriptor" and using associated type projections to communicate that to the
CLI tool. Unfortunately four bytes isn't a lot of space to cram information like
arguments to a generic function, tuple types, etc. In general this just wasn't
flexible enough and the way custom references were treated was also already a
bit of a hack.

This commit takes a radical step of creating a **descriptor function** for each
function imported/exported. The really crazy part is that the `wasm-bindgen` CLI
tool now embeds a wasm interpreter and executes these functions when the CLI
tool is invoked. By allowing arbitrary functions to get executed it's now *much*
easier to inform `wasm-bindgen` about complicated structures of types. Rest
assured though that all these descriptor functions are automatically unexported
and gc'd away, so this should not have any impact on binary sizes

A new internal trait, `WasmDescribe`, is added to represent a description of all
types, sort of like a serialization of the structure of a type that
`wasm-bindgen` can understand. This works by calling a special exported function
with a `u32` value a bunch of times. This means that when we run a descriptor we
effectively get a `Vec<u32>` in the `wasm-bindgen` CLI tool. This list of
integers can then be parsed into a rich `enum` for the JS generation to work
with.

This commit currently only retains feature parity with the previous
implementation. I hope to soon solve issues like #123, #104, and #111 with this
support.

crates/cli-support/src/lib.rs

@@ -2,8 +2,10 @@ extern crate parity_wasm;
 extern crate wasm_bindgen_shared as shared;
 extern crate serde_json;
 extern crate wasm_gc;
+extern crate wasmi;
 
 use std::collections::BTreeSet;
+use std::fmt;
 use std::fs::File;
 use std::io::Write;
 use std::path::{Path, PathBuf};
@@ -11,6 +13,7 @@ use std::path::{Path, PathBuf};
 use parity_wasm::elements::*;
 
 mod js;
+mod descriptor;
 pub mod wasm2es6js;
 
 pub struct Bindgen {
@@ -90,11 +93,26 @@ impl Bindgen {
             None => panic!("must have a path input for now"),
         };
         let stem = input.file_stem().unwrap().to_str().unwrap();
-        let mut module = parity_wasm::deserialize_file(input).map_err(|e| {
-            Error(format!("{:?}", e))
-        })?;
+        let mut module = parity_wasm::deserialize_file(input)?;
         let programs = extract_programs(&mut module);
 
+        // Here we're actually instantiating the module we've parsed above for
+        // execution. Why, you might be asking, are we executing wasm code? A
+        // good question!
+        //
+        // Transmitting information from `#[wasm_bindgen]` here to the CLI tool
+        // is pretty tricky. Specifically information about the types involved
+        // with a function signature (especially generic ones) can be hefty to
+        // translate over. As a result, the macro emits a bunch of shims which,
+        // when executed, will describe to us what the types look like.
+        //
+        // This means that whenever we encounter an import or export we'll
+        // execute a shim function which informs us about its type so we can
+        // then generate the appropriate bindings.
+        let instance = wasmi::Module::from_parity_wasm_module(module.clone())?;
+        let instance = wasmi::ModuleInstance::new(&imodule, &MyResolver)?;
+        let instance = instance.not_started_instance();
+
         let (js, ts) = {
             let mut cx = js::Context {
                 globals: String::new(),
@@ -103,16 +121,20 @@ impl Bindgen {
                 typescript: format!("/* tslint:disable */\n"),
                 exposed_globals: Default::default(),
                 required_internal_exports: Default::default(),
-                custom_type_names: Default::default(),
                 imported_names: Default::default(),
                 exported_classes: Default::default(),
                 config: &self,
                 module: &mut module,
                 function_table_needed: false,
+                run_descriptor: &|name| {
+                    let mut v = MyExternals(Vec::new());
+                    let ret = imodulei
+                        .invoke_export(name, &[], &mut v)
+                        .expect("failed to run export");
+                    assert!(ret.is_none());
+                    v.0
+                },
             };
-            for program in programs.iter() {
-                cx.add_custom_type_names(program);
-            }
             for program in programs.iter() {
                 js::SubContext {
                     program,
@@ -233,3 +255,102 @@ to open an issue at https://github.com/alexcrichton/wasm-bindgen/issues!
     });
     return ret
 }
+
+struct MyResolver;
+
+impl wasmi::ImportResolver for MyResolver {
+    fn resolve_func(
+        &self,
+        module_name: &str,
+        field_name: &str,
+        signature: &wasmi::Signature
+    ) -> Result<wasmi::FuncRef, wasmi::Error> {
+        // Route our special "describe" export to 1 and everything else to 0.
+        // That way whenever the function 1 is invoked we know what to do and
+        // when 0 is invoked (by accident) we'll trap and produce an error.
+        let idx = (module_name == "__wbindgen_placeholder__" &&
+            field_name == "__wbindgen_describe") as usize;
+        Ok(wasmi::FuncInstance::alloc_host(signature.clone(), idx))
+    }
+
+    fn resolve_global(
+        &self,
+        _module_name: &str,
+        _field_name: &str,
+        descriptor: &wasmi::GlobalDescriptor
+    ) -> Result<wasmi::GlobalRef, wasmi::Error> {
+        // dummy implementation to ensure instantiation succeeds
+        let val = match descriptor.value_type() {
+            wasmi::ValueType::I32 => wasmi::RuntimeValue::I32(0),
+            wasmi::ValueType::I64 => wasmi::RuntimeValue::I64(0),
+            wasmi::ValueType::F32 => wasmi::RuntimeValue::F32(0.0),
+            wasmi::ValueType::F64 => wasmi::RuntimeValue::F64(0.0),
+        };
+        Ok(wasmi::GlobalInstance::alloc(val, descriptor.is_mutable()))
+    }
+
+    fn resolve_memory(
+        &self,
+        _module_name: &str,
+        _field_name: &str,
+        descriptor: &wasmi::MemoryDescriptor,
+    ) -> Result<wasmi::MemoryRef, wasmi::Error> {
+        // dummy implementation to ensure instantiation succeeds
+        use wasmi::memory_units::Pages;
+        let initial = Pages(descriptor.initial() as usize);
+        let maximum = descriptor.maximum().map(|i| Pages(i as usize));
+        wasmi::MemoryInstance::alloc(initial, maximum)
+    }
+
+    fn resolve_table(
+        &self,
+        _module_name: &str,
+        _field_name: &str,
+        descriptor: &wasmi::TableDescriptor
+    ) -> Result<wasmi::TableRef, wasmi::Error> {
+        // dummy implementation to ensure instantiation succeeds
+        let initial = descriptor.initial();
+        let maximum = descriptor.maximum();
+        wasmi::TableInstance::alloc(initial, maximum)
+    }
+}
+
+struct MyExternals(Vec<u32>);
+#[derive(Debug)]
+struct MyError(String);
+
+impl wasmi::Externals for MyExternals {
+    fn invoke_index(
+        &mut self,
+        index: usize,
+        args: wasmi::RuntimeArgs
+    ) -> Result<Option<wasmi::RuntimeValue>, wasmi::Trap> {
+        macro_rules! bail {
+            ($($t:tt)*) => ({
+                let s = MyError(format!($($t)*));
+                return Err(wasmi::Trap::new(wasmi::TrapKind::Host(Box::new(s))))
+            })
+        }
+        // We only recognize one function here which was mapped to the index 1
+        // by the resolver above.
+        if index != 1 {
+            bail!("only __wbindgen_describe can be run at this time")
+        }
+        if args.len() != 1 {
+            bail!("must have exactly one argument");
+        }
+        match args.nth_value_checked(0)? {
+            wasmi::RuntimeValue::I32(i) => self.0.push(i as u32),
+            _ => bail!("expected one argument of i32 type"),
+        }
+        Ok(None)
+    }
+}
+
+impl wasmi::HostError for MyError {}
+
+impl fmt::Display for MyError {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        self.0.fmt(f)
+    }
+}