Create initial scaffolding for JS WASM assembler

README.md

@@ -15,7 +15,7 @@ emulated in JavaScript.
 
 You'll need:
 
-- [WABT](https://github.com/WebAssembly/wabt)
+- [WABT](https://github.com/WebAssembly/wabt) (not for long mwahaha)
 - [Guile](https://www.gnu.org/software/guile/) (or bring your own HTTP
   server -- see note below)
 

asm.js

@@ -0,0 +1,258 @@
+const HEADER = [
+    0x00, 0x61, 0x73, 0x6d,
+    0x01, 0x00, 0x00, 0x00
+];
+
+const LINE_END = "\n"
+
+class Tokenizer {
+	constructor() {
+		this.delims = new Set([" ", "\r", "\n", "\t"]);
+		this.skips = new Set([" ", "\r", "\t"]);
+        this.comment_start = ";"
+		this.buffer = [];
+        this.comment = false;
+	}
+
+	skip() {
+		const idx = this.buffer.findIndex((cp) => !this.skips.has(cp));
+        this.buffer = idx == -1 ? [] : this.buffer.slice(idx);
+    }
+
+	next() {
+        this.skip();
+        if (this.buffer[0] == LINE_END)
+            return this.buffer.shift();
+
+		const idx = this.buffer.findIndex((cp) => this.delims.has(cp));
+        if (idx != -1) {
+            const token = this.buffer.slice(0, idx).join("");
+            this.buffer = this.buffer.slice(idx);
+            return token;
+        }
+	}
+
+    *handle(src) {
+        this.buffer.push(...src);
+        let token;
+        while (token = this.next()) {
+            if (token == this.comment_start)
+                this.comment = true;
+            else if (this.comment && token == LINE_END)
+                this.comment = false;
+            else if (!this.comment)
+                yield token;
+        }
+	}
+}
+
+const State = Object.freeze({
+    TOP:    0,
+    EXPORT: 1,
+    FUNC:   2,
+    RESULT: 3,
+});
+
+const Action = Object.freeze({
+    APPEND: 0,
+    EXPORT: 1,
+    FUNC:   2,
+    RESULT: 3,
+});
+
+const types = {
+    "func": 0x60,
+    "i32":  0x7f,
+};
+
+const opcodes = {
+    "end":       0x0b,
+    "i32.const": 0x41,
+    "i32.mul":   0x6c,
+};
+
+class Parser {
+	constructor() {
+		this.tokens = [];
+		this.tokenizer = new Tokenizer();
+        this.state = State.TOP;
+        this.directives = {
+            ".export": State.EXPORT,
+            ".func":   State.FUNC,
+            ".result": State.RESULT,
+        };
+        this.handlers = {
+            [State.TOP]:    (token) => this.token_top(token),
+            [State.EXPORT]: (token) => this.token_export(token),
+            [State.FUNC]:   (token) => this.token_func(token),
+            [State.RESULT]: (token) => this.token_result(token),
+        };
+
+        this.results = [];
+	}
+
+    translate_code(token) {
+        return opcodes[token] ?? parseInt(token);
+    }
+
+    translate_type(token) {
+        return types[token];
+    }
+
+    token_top(token) {
+        if (token == LINE_END)
+            return;
+        let state;
+        if (state = this.directives[token]) {
+            this.state = state;
+            return;
+        }
+        const code = this.translate_code(token);
+        return { type: Action.APPEND, code };
+    }
+
+    token_export(token) {
+        this.state = State.TOP;
+        return { type: Action.EXPORT, name: token };
+    }
+
+    token_func(token) {
+        this.state = State.TOP;
+        return { type: Action.FUNC, name: token };
+    }
+
+    token_result(token) {
+        if (token == LINE_END) {
+            const action = { type: Action.RESULT, types: this.results };
+            this.state = State.TOP;
+            this.results = [];
+            return action;
+        } else {
+            this.results.push(this.translate_type(token));
+        }
+    }
+
+	*handle(src) {
+        let action;
+		for (const token of this.tokenizer.handle(src)) {
+            if (action = this.handlers[this.state](token))
+                yield action;
+        }
+	}
+}
+
+const Section = Object.freeze({
+    TYPE:   0x01,
+    FUNC:   0x03,
+    EXPORT: 0x07,
+    CODE:   0x0a
+});
+
+export class Assembler {
+	constructor() {
+        this.encoder = new TextEncoder("utf-8");
+		this.decoder = new TextDecoder("utf-8");
+		this.parser = new Parser();
+        this.handlers = {
+            [Action.APPEND]: (action) => this.action_append(action),
+            [Action.EXPORT]: (action) => this.action_export(action),
+            [Action.FUNC]:   (action) => this.action_func(action),
+            [Action.RESULT]: (action) => this.action_result(action),
+        };
+
+        this.exports = [];
+        this.funcs = {}
+	}
+
+    action_append(action) {
+        this.funcs[this.current_func].body.push(action.code);
+    }
+
+    action_export(action) {
+        const index = Object.keys(this.funcs).indexOf(action.name);
+        this.exports[action.name] = { kind: 0x00, index };
+    }
+
+    action_func(action) {
+        this.funcs[action.name] = {
+            params: {},
+            results: [],
+            locals: {},
+            body: [],
+        }
+        this.current_func = action.name;
+    }
+
+    action_result(action) {
+        this.funcs[this.current_func].results.push(...action.types);
+    }
+
+	push(chunk) {
+		const text = this.decoder.decode(chunk, { stream: true });
+		for (const action of this.parser.handle(text))
+            this.handlers[action.type](action);
+	}
+
+    wasm_section_type() {
+        const funcs = Object.values(this.funcs);
+        const contents = funcs.map(({ params, results }) => {
+            const param_types = Object.values(params);
+            return [
+                types["func"],
+                param_types.length,
+                ...param_types,
+                results.length,
+                ...results,
+            ];
+        });
+        return [ contents.length ].concat(...contents);
+    }
+
+    wasm_section_func() {
+        const func_count = Object.entries(this.funcs).length;
+        return [ func_count, ...Array(func_count).keys() ];
+    }
+
+    wasm_section_export() {
+        const exports = Object.entries(this.exports);
+        const contents = exports.map(([ name, { kind, index }]) => {
+            const name_utf8 = this.encoder.encode(name);
+            return [
+                name_utf8.length,
+                ...name_utf8,
+                kind,
+                index,
+            ];
+        });
+        return [ exports.length ].concat(...contents);
+    }
+
+    wasm_section_code() {
+        const funcs = Object.values(this.funcs);
+        const contents = funcs.map(({ body, locals }) => {
+            const local_count = Object.entries(locals).length;
+            return [
+                body.length + 2,
+                local_count,
+                ...body,
+                opcodes["end"]
+            ]
+        });
+        return [ contents.length ].concat(...contents);
+    }
+
+	wasm() {
+        const template = [
+            [ Section.TYPE,   () => this.wasm_section_type() ],
+            [ Section.FUNC,   () => this.wasm_section_func() ],
+            [ Section.EXPORT, () => this.wasm_section_export() ],
+            [ Section.CODE,   () => this.wasm_section_code() ],
+        ];
+        const sections = template.map(([ code, generator ]) => {
+            const body = generator();
+            return [ code, body.length, ...body ];
+        });
+
+		return new Uint8Array(HEADER.concat(...sections));
+	}
+}