Archive Rust bootstrap — El compiler is now self-hosting

This commit is contained in:
Will Anderson
2026-04-29 22:21:31 -05:00
parent 9a0747aa13
commit 4f3543b068
139 changed files with 8980 additions and 1778 deletions
-8
View File
@@ -1,8 +0,0 @@
[package]
name = "el-compiler"
version = "0.1.0"
description = "el self-hosting compiler — lexer, parser, codegen"
edition = "2026"
[build]
entry = "src/compiler.el"
+9
View File
@@ -0,0 +1,9 @@
package "el-compiler" {
version "0.1.0"
description "el self-hosting compiler — lexer, parser, codegen"
edition "2026"
}
build {
entry "src/compiler.el"
}
+308 -309
View File
@@ -7,6 +7,11 @@
// See the el-compiler/src/bytecode.rs for the canonical enum.
//
// Entry point: fn codegen(stmts: [Map<String, Any>], source: String) -> String
//
// Performance: instruction accumulation uses the VM-native global buffer
// (native_instr_push / native_instr_len / native_instr_patch / native_instr_all)
// instead of passing a growing list through function arguments. This avoids O(n²)
// cloning and keeps compilation time linear in the number of instructions.
// JSON helpers
@@ -59,265 +64,237 @@ fn json_bool(b: Bool) -> String {
// Codegen state
//
// el has no mutable globals. We thread a "ctx" map through all codegen
// functions. The context holds:
// "instrs" -> [String] JSON strings for each emitted instruction
// "patches" -> [Map<String,Any>] pending forward-jump patches
// each patch: { "idx": Int, "kind": String }
// (kind is "JumpIfNot" / "Jump" we store the instruction index and
// patch it at the end of the construct)
fn ctx_new() -> Map<String, Any> {
{ "instrs": native_list_empty(), "patches": native_list_empty() }
}
fn ctx_emit(ctx: Map<String, Any>, instr_json: String) -> Map<String, Any> {
let instrs = ctx["instrs"]
let instrs = native_list_append(instrs, instr_json)
{ "instrs": instrs, "patches": ctx["patches"] }
}
fn ctx_len(ctx: Map<String, Any>) -> Int {
let instrs = ctx["instrs"]
native_list_len(instrs)
}
// Patch a previously-emitted placeholder instruction at index idx.
// For Jump/JumpIf/JumpIfNot, we need to replace the entry in instrs.
// We rebuild the list with the patched value at position idx.
fn ctx_patch(ctx: Map<String, Any>, idx: Int, dest: Int) -> Map<String, Any> {
// offset = dest - (idx + 1)
let offset = dest - (idx + 1)
let instrs = ctx["instrs"]
let total = native_list_len(instrs)
let new_instrs: [String] = native_list_empty()
let i = 0
while i < total {
let instr: String = native_list_get(instrs, i)
if i == idx {
// Replace: determine kind from original instruction string
if native_string_contains(instr, "JumpIfNot") {
let new_instrs = native_list_append(new_instrs, "{\"JumpIfNot\":" + json_int(offset) + "}")
} else {
if native_string_contains(instr, "JumpIf") {
let new_instrs = native_list_append(new_instrs, "{\"JumpIf\":" + json_int(offset) + "}")
} else {
let new_instrs = native_list_append(new_instrs, "{\"Jump\":" + json_int(offset) + "}")
}
}
} else {
let new_instrs = native_list_append(new_instrs, instr)
}
let i = i + 1
}
{ "instrs": new_instrs, "patches": ctx["patches"] }
}
// The instruction accumulator is stored in the VM's global instr_buf.
// native_instr_push(s) append JSON instruction string
// native_instr_len() current instruction count
// native_instr_get(i) get instruction at index i
// native_instr_patch(i,s) replace instruction at index i
// native_instr_all() return full list as [String]
// Instruction emitters
fn emit_halt(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Halt\"")
fn emit_halt() -> Void {
native_instr_push("\"Halt\"")
}
fn emit_nop(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Nop\"")
fn emit_pop() -> Void {
native_instr_push("\"Pop\"")
}
fn emit_pop(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Pop\"")
fn emit_return() -> Void {
native_instr_push("\"Return\"")
}
fn emit_return(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Return\"")
fn emit_add() -> Void {
native_instr_push("\"Add\"")
}
fn emit_add(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Add\"")
fn emit_sub() -> Void {
native_instr_push("\"Sub\"")
}
fn emit_sub(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Sub\"")
fn emit_mul() -> Void {
native_instr_push("\"Mul\"")
}
fn emit_mul(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Mul\"")
fn emit_div() -> Void {
native_instr_push("\"Div\"")
}
fn emit_div(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Div\"")
fn emit_eq() -> Void {
native_instr_push("\"Eq\"")
}
fn emit_eq(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Eq\"")
fn emit_not_eq() -> Void {
native_instr_push("\"NotEq\"")
}
fn emit_not_eq(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"NotEq\"")
fn emit_lt() -> Void {
native_instr_push("\"Lt\"")
}
fn emit_lt(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Lt\"")
fn emit_gt() -> Void {
native_instr_push("\"Gt\"")
}
fn emit_gt(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Gt\"")
fn emit_lt_eq() -> Void {
native_instr_push("\"LtEq\"")
}
fn emit_lt_eq(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"LtEq\"")
fn emit_gt_eq() -> Void {
native_instr_push("\"GtEq\"")
}
fn emit_gt_eq(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"GtEq\"")
fn emit_and() -> Void {
native_instr_push("\"And\"")
}
fn emit_and(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"And\"")
fn emit_or() -> Void {
native_instr_push("\"Or\"")
}
fn emit_or(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Or\"")
fn emit_not() -> Void {
native_instr_push("\"Not\"")
}
fn emit_not(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"Not\"")
fn emit_get_index() -> Void {
native_instr_push("\"GetIndex\"")
}
fn emit_get_index(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "\"GetIndex\"")
fn emit_push_nil() -> Void {
native_instr_push("{\"Push\":\"Nil\"}")
}
fn emit_push_nil(ctx: Map<String, Any>) -> Map<String, Any> {
ctx_emit(ctx, "{\"Push\":\"Nil\"}")
fn emit_push_int(n: Int) -> Void {
native_instr_push("{\"Push\":{\"Int\":" + json_int(n) + "}}")
}
fn emit_push_int(ctx: Map<String, Any>, n: Int) -> Map<String, Any> {
ctx_emit(ctx, "{\"Push\":{\"Int\":" + json_int(n) + "}}")
fn emit_push_bool(b: Bool) -> Void {
native_instr_push("{\"Push\":{\"Bool\":" + json_bool(b) + "}}")
}
fn emit_push_bool(ctx: Map<String, Any>, b: Bool) -> Map<String, Any> {
ctx_emit(ctx, "{\"Push\":{\"Bool\":" + json_bool(b) + "}}")
fn emit_push_str(s: String) -> Void {
native_instr_push("{\"Push\":{\"Str\":" + json_str(s) + "}}")
}
fn emit_push_str(ctx: Map<String, Any>, s: String) -> Map<String, Any> {
ctx_emit(ctx, "{\"Push\":{\"Str\":" + json_str(s) + "}}")
fn emit_load(name: String) -> Void {
native_instr_push("{\"LoadLocal\":" + json_str(name) + "}")
}
fn emit_load(ctx: Map<String, Any>, name: String) -> Map<String, Any> {
ctx_emit(ctx, "{\"LoadLocal\":" + json_str(name) + "}")
fn emit_store(name: String) -> Void {
native_instr_push("{\"StoreLocal\":" + json_str(name) + "}")
}
fn emit_store(ctx: Map<String, Any>, name: String) -> Map<String, Any> {
ctx_emit(ctx, "{\"StoreLocal\":" + json_str(name) + "}")
fn emit_call(name: String, arity: Int) -> Void {
native_instr_push("{\"Call\":{\"name\":" + json_str(name) + ",\"arity\":" + json_int(arity) + "}}")
}
fn emit_call(ctx: Map<String, Any>, name: String, arity: Int) -> Map<String, Any> {
ctx_emit(ctx, "{\"Call\":{\"name\":" + json_str(name) + ",\"arity\":" + json_int(arity) + "}}")
fn emit_get_field(field: String) -> Void {
native_instr_push("{\"GetField\":" + json_str(field) + "}")
}
fn emit_get_field(ctx: Map<String, Any>, field: String) -> Map<String, Any> {
ctx_emit(ctx, "{\"GetField\":" + json_str(field) + "}")
fn emit_build_map(n: Int) -> Void {
native_instr_push("{\"BuildMap\":" + json_int(n) + "}")
}
fn emit_build_map(ctx: Map<String, Any>, n: Int) -> Map<String, Any> {
ctx_emit(ctx, "{\"BuildMap\":" + json_int(n) + "}")
fn emit_build_list(n: Int) -> Void {
native_instr_push("{\"BuildList\":" + json_int(n) + "}")
}
fn emit_build_list(ctx: Map<String, Any>, n: Int) -> Map<String, Any> {
ctx_emit(ctx, "{\"BuildList\":" + json_int(n) + "}")
// Emit a placeholder Jump and return its index (for later patching).
fn emit_jump_placeholder() -> Int {
let idx: Int = native_instr_len()
native_instr_push("{\"Jump\":0}")
idx
}
// Emit a placeholder Jump and return its index (for later patching)
fn emit_jump_placeholder(ctx: Map<String, Any>) -> Map<String, Any> {
let idx = ctx_len(ctx)
let ctx = ctx_emit(ctx, "{\"Jump\":0}")
{ "ctx": ctx, "idx": idx }
fn emit_jump_if_not_placeholder() -> Int {
let idx: Int = native_instr_len()
native_instr_push("{\"JumpIfNot\":0}")
idx
}
fn emit_jump_if_not_placeholder(ctx: Map<String, Any>) -> Map<String, Any> {
let idx = ctx_len(ctx)
let ctx = ctx_emit(ctx, "{\"JumpIfNot\":0}")
{ "ctx": ctx, "idx": idx }
fn emit_jump_to(dest: Int) -> Void {
let here: Int = native_instr_len()
let offset: Int = dest - (here + 1)
native_instr_push("{\"Jump\":" + json_int(offset) + "}")
}
fn emit_jump_to(ctx: Map<String, Any>, dest: Int) -> Map<String, Any> {
let here = ctx_len(ctx)
let offset = dest - (here + 1)
ctx_emit(ctx, "{\"Jump\":" + json_int(offset) + "}")
// Patch a previously-emitted placeholder jump instruction.
fn patch_instr(idx: Int, dest: Int) -> Void {
let offset: Int = dest - (idx + 1)
let original: String = native_instr_get(idx)
if native_string_contains(original, "JumpIfNot") {
native_instr_patch(idx, "{\"JumpIfNot\":" + json_int(offset) + "}")
} else {
if native_string_contains(original, "JumpIf") {
native_instr_patch(idx, "{\"JumpIf\":" + json_int(offset) + "}")
} else {
native_instr_patch(idx, "{\"Jump\":" + json_int(offset) + "}")
}
}
}
// Expression codegen
fn cg_expr(ctx: Map<String, Any>, expr: Map<String, Any>) -> Map<String, Any> {
fn cg_expr(expr: Map<String, Any>) -> Void {
let kind: String = expr["expr"]
if kind == "Int" {
let v: String = expr["value"]
let n: Int = native_str_to_int(v)
return emit_push_int(ctx, n)
emit_push_int(n)
return
}
if kind == "Float" {
let v: String = expr["value"]
// Push as string for now VM handles float parsing from Str
return emit_push_str(ctx, v)
emit_push_str(v)
return
}
if kind == "Str" {
let v: String = expr["value"]
return emit_push_str(ctx, v)
emit_push_str(v)
return
}
if kind == "Bool" {
let v: String = expr["value"]
if v == "true" {
return emit_push_bool(ctx, true)
emit_push_bool(true)
} else {
emit_push_bool(false)
}
return emit_push_bool(ctx, false)
return
}
if kind == "Nil" {
return emit_push_nil(ctx)
emit_push_nil()
return
}
if kind == "Ident" {
let name: String = expr["name"]
return emit_load(ctx, name)
emit_load(name)
return
}
if kind == "Not" {
let inner = expr["inner"]
let ctx = cg_expr(ctx, inner)
return emit_not(ctx)
cg_expr(inner)
emit_not()
return
}
if kind == "Neg" {
// unary minus: push 0, push inner, sub
let ctx = emit_push_int(ctx, 0)
emit_push_int(0)
let inner = expr["inner"]
let ctx = cg_expr(ctx, inner)
return emit_sub(ctx)
cg_expr(inner)
emit_sub()
return
}
if kind == "BinOp" {
let op: String = expr["op"]
let left = expr["left"]
let right = expr["right"]
let ctx = cg_expr(ctx, left)
let ctx = cg_expr(ctx, right)
if op == "Plus" { return emit_add(ctx) }
if op == "Minus" { return emit_sub(ctx) }
if op == "Star" { return emit_mul(ctx) }
if op == "Slash" { return emit_div(ctx) }
if op == "EqEq" { return emit_eq(ctx) }
if op == "NotEq" { return emit_not_eq(ctx) }
if op == "Lt" { return emit_lt(ctx) }
if op == "Gt" { return emit_gt(ctx) }
if op == "LtEq" { return emit_lt_eq(ctx) }
if op == "GtEq" { return emit_gt_eq(ctx) }
if op == "And" { return emit_and(ctx) }
if op == "Or" { return emit_or(ctx) }
return ctx
cg_expr(left)
cg_expr(right)
if op == "Plus" { emit_add() }
if op == "Minus" { emit_sub() }
if op == "Star" { emit_mul() }
if op == "Slash" { emit_div() }
if op == "EqEq" { emit_eq() }
if op == "NotEq" { emit_not_eq() }
if op == "Lt" { emit_lt() }
if op == "Gt" { emit_gt() }
if op == "LtEq" { emit_lt_eq() }
if op == "GtEq" { emit_gt_eq() }
if op == "And" { emit_and() }
if op == "Or" { emit_or() }
return
}
if kind == "Call" {
@@ -328,37 +305,42 @@ fn cg_expr(ctx: Map<String, Any>, expr: Map<String, Any>) -> Map<String, Any> {
let i = 0
while i < arity {
let arg = native_list_get(args, i)
let ctx = cg_expr(ctx, arg)
cg_expr(arg)
let i = i + 1
}
// get function name from func expr
let func_kind: String = func["expr"]
if func_kind == "Ident" {
let fn_name: String = func["name"]
return emit_call(ctx, fn_name, arity)
emit_call(fn_name, arity)
return
}
if func_kind == "Field" {
let obj = func["object"]
let field: String = func["field"]
let ctx = cg_expr(ctx, obj)
return emit_call(ctx, field, arity + 1)
cg_expr(obj)
emit_call(field, arity + 1)
return
}
return emit_call(ctx, "__dynamic__", arity)
emit_call("__dynamic__", arity)
return
}
if kind == "Field" {
let obj = expr["object"]
let field: String = expr["field"]
let ctx = cg_expr(ctx, obj)
return emit_get_field(ctx, field)
cg_expr(obj)
emit_get_field(field)
return
}
if kind == "Index" {
let obj = expr["object"]
let idx = expr["index"]
let ctx = cg_expr(ctx, obj)
let ctx = cg_expr(ctx, idx)
return emit_get_index(ctx)
cg_expr(obj)
cg_expr(idx)
emit_get_index()
return
}
if kind == "Array" {
@@ -367,10 +349,11 @@ fn cg_expr(ctx: Map<String, Any>, expr: Map<String, Any>) -> Map<String, Any> {
let i = 0
while i < n {
let elem = native_list_get(elems, i)
let ctx = cg_expr(ctx, elem)
cg_expr(elem)
let i = i + 1
}
return emit_build_list(ctx, n)
emit_build_list(n)
return
}
if kind == "Map" {
@@ -381,11 +364,12 @@ fn cg_expr(ctx: Map<String, Any>, expr: Map<String, Any>) -> Map<String, Any> {
let pair = native_list_get(pairs, i)
let key: String = pair["key"]
let val = pair["value"]
let ctx = emit_push_str(ctx, key)
let ctx = cg_expr(ctx, val)
emit_push_str(key)
cg_expr(val)
let i = i + 1
}
return emit_build_map(ctx, n)
emit_build_map(n)
return
}
if kind == "If" {
@@ -394,41 +378,36 @@ fn cg_expr(ctx: Map<String, Any>, expr: Map<String, Any>) -> Map<String, Any> {
let else_stmts = expr["else"]
let has_else: Bool = expr["has_else"]
// cond
let ctx = cg_expr(ctx, cond)
cg_expr(cond)
// JumpIfNot placeholder
let r = emit_jump_if_not_placeholder(ctx)
let ctx = r["ctx"]
let jump_false_idx: Int = r["idx"]
let jump_false_idx: Int = emit_jump_if_not_placeholder()
// then body
let ctx = cg_stmts(ctx, then_stmts)
cg_stmts(then_stmts)
if has_else {
// jump over else
let r2 = emit_jump_placeholder(ctx)
let ctx = r2["ctx"]
let jump_end_idx: Int = r2["idx"]
let jump_end_idx: Int = emit_jump_placeholder()
// patch jump_false to here
let else_start = ctx_len(ctx)
let ctx = ctx_patch(ctx, jump_false_idx, else_start)
let else_start: Int = native_instr_len()
patch_instr(jump_false_idx, else_start)
// else body
let ctx = cg_stmts(ctx, else_stmts)
cg_stmts(else_stmts)
// patch jump_end to here
let after_else = ctx_len(ctx)
let ctx = ctx_patch(ctx, jump_end_idx, after_else)
return ctx
let after_else: Int = native_instr_len()
patch_instr(jump_end_idx, after_else)
} else {
let after_then = ctx_len(ctx)
let ctx = ctx_patch(ctx, jump_false_idx, after_then)
return ctx
let after_then: Int = native_instr_len()
patch_instr(jump_false_idx, after_then)
}
return
}
if kind == "Match" {
let subject = expr["subject"]
let arms = expr["arms"]
let n_arms: Int = native_list_len(arms)
let ctx = cg_expr(ctx, subject)
cg_expr(subject)
// store subject in temp var
let ctx = emit_store(ctx, "__match_subj__")
emit_store("__match_subj__")
let end_jump_idxs: [Int] = native_list_empty()
let i = 0
while i < n_arms {
@@ -438,155 +417,187 @@ fn cg_expr(ctx: Map<String, Any>, expr: Map<String, Any>) -> Map<String, Any> {
let pat_kind: String = pattern["pattern"]
if pat_kind == "Wildcard" {
// always matches just emit body
let ctx = cg_expr(ctx, body)
let r = emit_jump_placeholder(ctx)
let ctx = r["ctx"]
let jidx: Int = r["idx"]
cg_expr(body)
let jidx: Int = emit_jump_placeholder()
let end_jump_idxs = native_list_append(end_jump_idxs, jidx)
} else {
if pat_kind == "Binding" {
let bind_name: String = pattern["name"]
let ctx = emit_load(ctx, "__match_subj__")
let ctx = emit_store(ctx, bind_name)
let ctx = cg_expr(ctx, body)
let r = emit_jump_placeholder(ctx)
let ctx = r["ctx"]
let jidx: Int = r["idx"]
emit_load("__match_subj__")
emit_store(bind_name)
cg_expr(body)
let jidx: Int = emit_jump_placeholder()
let end_jump_idxs = native_list_append(end_jump_idxs, jidx)
} else {
// literal pattern: compare subject to literal
let ctx = emit_load(ctx, "__match_subj__")
emit_load("__match_subj__")
if pat_kind == "LitInt" {
let v: String = pattern["value"]
let n: Int = native_str_to_int(v)
let ctx = emit_push_int(ctx, n)
emit_push_int(n)
} else {
if pat_kind == "LitStr" {
let v: String = pattern["value"]
let ctx = emit_push_str(ctx, v)
emit_push_str(v)
} else {
if pat_kind == "LitBool" {
let v: String = pattern["value"]
if v == "true" {
let ctx = emit_push_bool(ctx, true)
emit_push_bool(true)
} else {
let ctx = emit_push_bool(ctx, false)
emit_push_bool(false)
}
} else {
let ctx = emit_push_nil(ctx)
emit_push_nil()
}
}
}
let ctx = emit_eq(ctx)
let r = emit_jump_if_not_placeholder(ctx)
let ctx = r["ctx"]
let no_match_idx: Int = r["idx"]
let ctx = cg_expr(ctx, body)
let r2 = emit_jump_placeholder(ctx)
let ctx = r2["ctx"]
let jidx: Int = r2["idx"]
emit_eq()
let no_match_idx: Int = emit_jump_if_not_placeholder()
cg_expr(body)
let jidx: Int = emit_jump_placeholder()
let end_jump_idxs = native_list_append(end_jump_idxs, jidx)
let next_arm = ctx_len(ctx)
let ctx = ctx_patch(ctx, no_match_idx, next_arm)
let next_arm: Int = native_instr_len()
patch_instr(no_match_idx, next_arm)
}
}
let i = i + 1
}
// default: push nil
let ctx = emit_push_nil(ctx)
let end_pos = ctx_len(ctx)
emit_push_nil()
let end_pos: Int = native_instr_len()
// patch all end jumps
let n_end: Int = native_list_len(end_jump_idxs)
let j = 0
while j < n_end {
let jidx: Int = native_list_get(end_jump_idxs, j)
let ctx = ctx_patch(ctx, jidx, end_pos)
patch_instr(jidx, end_pos)
let j = j + 1
}
return ctx
return
}
if kind == "For" {
// for item in list { body }
// Implementation:
// __for_list__ = list
// __for_len__ = len(list)
// __for_i__ = 0
// loop_start:
// if __for_i__ >= __for_len__ goto done
// item = __for_list__[__for_i__]
// body
// __for_i__ = __for_i__ + 1
// goto loop_start
// done:
let item: String = expr["item"]
let list_expr = expr["list"]
let body = expr["body"]
// emit list, store it
let ctx = cg_expr(ctx, list_expr)
let ctx = emit_store(ctx, "__for_list__")
cg_expr(list_expr)
emit_store("__for_list__")
// compute length
let ctx = emit_load(ctx, "__for_list__")
let ctx = emit_call(ctx, "native_list_len", 1)
let ctx = emit_store(ctx, "__for_len__")
emit_load("__for_list__")
emit_call("native_list_len", 1)
emit_store("__for_len__")
// init counter
let ctx = emit_push_int(ctx, 0)
let ctx = emit_store(ctx, "__for_i__")
emit_push_int(0)
emit_store("__for_i__")
// loop start
let loop_start = ctx_len(ctx)
let loop_start: Int = native_instr_len()
// condition: __for_i__ < __for_len__
let ctx = emit_load(ctx, "__for_i__")
let ctx = emit_load(ctx, "__for_len__")
let ctx = emit_lt(ctx)
let r = emit_jump_if_not_placeholder(ctx)
let ctx = r["ctx"]
let to_done_idx: Int = r["idx"]
emit_load("__for_i__")
emit_load("__for_len__")
emit_lt()
let to_done_idx: Int = emit_jump_if_not_placeholder()
// get current element
let ctx = emit_load(ctx, "__for_list__")
let ctx = emit_load(ctx, "__for_i__")
let ctx = emit_get_index(ctx)
let ctx = emit_store(ctx, item)
emit_load("__for_list__")
emit_load("__for_i__")
emit_get_index()
emit_store(item)
// body
let ctx = cg_stmts(ctx, body)
cg_stmts(body)
// increment counter
let ctx = emit_load(ctx, "__for_i__")
let ctx = emit_push_int(ctx, 1)
let ctx = emit_add(ctx)
let ctx = emit_store(ctx, "__for_i__")
emit_load("__for_i__")
emit_push_int(1)
emit_add()
emit_store("__for_i__")
// jump back
let ctx = emit_jump_to(ctx, loop_start)
emit_jump_to(loop_start)
// patch done
let done_pos = ctx_len(ctx)
let ctx = ctx_patch(ctx, to_done_idx, done_pos)
return ctx
let done_pos: Int = native_instr_len()
patch_instr(to_done_idx, done_pos)
return
}
if kind == "Try" {
// Just emit the inner expression error propagation handled at runtime
// Just emit the inner expression
let inner = expr["inner"]
return cg_expr(ctx, inner)
cg_expr(inner)
return
}
// Fallback
emit_push_nil(ctx)
emit_push_nil()
}
// Statement codegen
fn cg_stmt(ctx: Map<String, Any>, stmt: Map<String, Any>) -> Map<String, Any> {
// cg_stmt_tail emit a statement in tail position (last stmt of fn body).
// For Expr statements, the result is left on the stack (not popped).
// For all other statement kinds, we delegate to cg_stmt and push Nil.
fn cg_stmt_tail(stmt: Map<String, Any>) -> Void {
let kind: String = stmt["stmt"]
if kind == "Expr" {
let val = stmt["value"]
// Emit the expression leave result on stack (tail position).
cg_expr(val)
return
}
if kind == "Return" {
// Explicit return same as cg_stmt
let val = stmt["value"]
cg_expr(val)
emit_return()
return
}
// All other statement kinds (Let, FnDef, While, etc.) don't produce
// a natural return value fall through to cg_stmt then push Nil.
cg_stmt(stmt)
emit_push_nil()
}
// cg_stmts_body emit function body statements.
// All statements except the last use cg_stmt (pops results).
// The last statement uses cg_stmt_tail (leaves value on stack).
fn cg_stmts_body(stmts: [Map<String, Any>]) -> Void {
let n: Int = native_list_len(stmts)
if n == 0 {
// Empty body push Nil as return value.
emit_push_nil()
return
}
let i = 0
while i < n {
let stmt = native_list_get(stmts, i)
let is_last: Bool = (i == n - 1)
if is_last {
cg_stmt_tail(stmt)
} else {
cg_stmt(stmt)
}
let i = i + 1
}
}
fn cg_stmt(stmt: Map<String, Any>) -> Void {
let kind: String = stmt["stmt"]
if kind == "Let" {
let name: String = stmt["name"]
let val = stmt["value"]
let ctx = cg_expr(ctx, val)
return emit_store(ctx, name)
cg_expr(val)
emit_store(name)
return
}
if kind == "Return" {
let val = stmt["value"]
let ctx = cg_expr(ctx, val)
return emit_return(ctx)
cg_expr(val)
emit_return()
return
}
if kind == "FnDef" {
@@ -595,45 +606,40 @@ fn cg_stmt(ctx: Map<String, Any>, stmt: Map<String, Any>) -> Map<String, Any> {
let body = stmt["body"]
let n_params: Int = native_list_len(params)
// emit a Jump to skip over the function body
let r = emit_jump_placeholder(ctx)
let ctx = r["ctx"]
let skip_jump_idx: Int = r["idx"]
let skip_jump_idx: Int = emit_jump_placeholder()
// function body entry: store params in reverse order (caller pushes LR, so pop RL)
let pi = n_params - 1
while pi >= 0 {
let param = native_list_get(params, pi)
let pname: String = param["name"]
let ctx = emit_store(ctx, pname)
emit_store(pname)
let pi = pi - 1
}
// emit body statements
let ctx = cg_stmts(ctx, body)
// implicit nil return
let ctx = emit_push_nil(ctx)
let ctx = emit_return(ctx)
// emit body statements using tail-aware emit (last stmt leaves value on stack)
cg_stmts_body(body)
// return the top-of-stack value
emit_return()
// patch skip jump
let after_fn = ctx_len(ctx)
let ctx = ctx_patch(ctx, skip_jump_idx, after_fn)
let after_fn: Int = native_instr_len()
patch_instr(skip_jump_idx, after_fn)
// register function entry point
let entry_ip = skip_jump_idx + 1
let ctx = emit_push_int(ctx, entry_ip)
let ctx = emit_store(ctx, "__fn_" + fn_name)
return ctx
let entry_ip: Int = skip_jump_idx + 1
emit_push_int(entry_ip)
emit_store("__fn_" + fn_name)
return
}
if kind == "While" {
let cond = stmt["cond"]
let body = stmt["body"]
let loop_start = ctx_len(ctx)
let ctx = cg_expr(ctx, cond)
let r = emit_jump_if_not_placeholder(ctx)
let ctx = r["ctx"]
let to_done_idx: Int = r["idx"]
let ctx = cg_stmts(ctx, body)
let ctx = emit_jump_to(ctx, loop_start)
let done_pos = ctx_len(ctx)
let ctx = ctx_patch(ctx, to_done_idx, done_pos)
return ctx
let loop_start: Int = native_instr_len()
cg_expr(cond)
let to_done_idx: Int = emit_jump_if_not_placeholder()
cg_stmts(body)
emit_jump_to(loop_start)
let done_pos: Int = native_instr_len()
patch_instr(to_done_idx, done_pos)
return
}
if kind == "For" {
@@ -642,53 +648,46 @@ fn cg_stmt(ctx: Map<String, Any>, stmt: Map<String, Any>) -> Map<String, Any> {
let list_expr = stmt["list"]
let body = stmt["body"]
let for_expr = { "expr": "For", "item": item, "list": list_expr, "body": body }
return cg_expr(ctx, for_expr)
cg_expr(for_expr)
return
}
if kind == "Expr" {
let val = stmt["value"]
let val_kind: String = val["expr"]
let ctx = cg_expr(ctx, val)
// Discard result unless it's a control-flow expression
if val_kind == "If" {
return ctx
}
cg_expr(val)
// Discard result unless it's a control-flow expression that doesn't push a value
if val_kind == "For" {
return ctx
return
}
if val_kind == "Match" {
return ctx
}
return emit_pop(ctx)
emit_pop()
return
}
if kind == "TypeDef" {
// compile-time only; no runtime code
return ctx
return
}
if kind == "EnumDef" {
// compile-time only; no runtime code
return ctx
return
}
if kind == "Import" {
// handled at a higher level; skip
return ctx
return
}
ctx
}
fn cg_stmts(ctx: Map<String, Any>, stmts: [Map<String, Any>]) -> Map<String, Any> {
fn cg_stmts(stmts: [Map<String, Any>]) -> Void {
let n: Int = native_list_len(stmts)
let i = 0
while i < n {
let stmt = native_list_get(stmts, i)
let ctx = cg_stmt(ctx, stmt)
cg_stmt(stmt)
let i = i + 1
}
ctx
}
// JSON serialisation
@@ -711,9 +710,9 @@ fn instrs_to_json(instrs: [String]) -> String {
// Entry point
fn codegen(stmts: [Map<String, Any>], source: String) -> String {
let ctx = ctx_new()
let ctx = cg_stmts(ctx, stmts)
let ctx = emit_halt(ctx)
let instrs = ctx["instrs"]
native_instr_reset()
cg_stmts(stmts)
emit_halt()
let instrs: [String] = native_instr_all()
instrs_to_json(instrs)
}
+27
View File
@@ -17,3 +17,30 @@ fn compile(source: String) -> String {
let stmts: [Map<String, Any>] = parse(tokens)
codegen(stmts, source)
}
// main CLI entry point for self-hosted compilation.
//
// Called by: elvm el-compiler.elc <source.el> <output.elc>
//
// Reads pre-resolved El source from args()[0], compiles it to JSON bytecode,
// and writes the result to args()[1]. The output is raw JSON (accepted by
// both elvm and el exec without an ELVM container header).
fn main() -> Void {
let argv: [String] = args()
let argc: Int = native_list_len(argv)
if argc < 2 {
println("el-compiler: usage: elvm el-compiler.elc <source.el> <output.elc>")
exit(1)
}
let src_path: String = native_list_get(argv, 0)
let out_path: String = native_list_get(argv, 1)
let source: String = fs_read(src_path)
let bytecode_json: String = compile(source)
let ok: Bool = fs_write(out_path, bytecode_json)
if ok {
exit(0)
} else {
println("el-compiler: failed to write output")
exit(1)
}
}
+26 -21
View File
@@ -6,6 +6,9 @@
// "value" -> String (the raw text of the token)
//
// Entry point: fn lex(source: String) -> [Map<String, Any>]
//
// Uses global char_buf via native_chars_init / native_char_at / native_char_len
// to avoid O(N²) cloning of the chars list.
// Character helpers
@@ -141,9 +144,11 @@ fn keyword_kind(word: String) -> String {
}
// Scan helpers
// All scan helpers use the global char_buf (native_char_at / native_char_len).
// No chars parameter avoids O(N²) cloning.
// scan_digits advance i while chars[i] is a digit, return { "text": ..., "pos": i }
fn scan_digits(chars: [String], start: Int, total: Int) -> Map<String, Any> {
// scan_digits advance i while char_buf[i] is a digit, return { "text": ..., "pos": i }
fn scan_digits(start: Int, total: Int) -> Map<String, Any> {
let i = start
let text = ""
let running = true
@@ -151,7 +156,7 @@ fn scan_digits(chars: [String], start: Int, total: Int) -> Map<String, Any> {
if i >= total {
let running = false
} else {
let ch = native_list_get(chars, i)
let ch = native_char_at(i)
if is_digit(ch) {
let text = text + ch
let i = i + 1
@@ -163,8 +168,8 @@ fn scan_digits(chars: [String], start: Int, total: Int) -> Map<String, Any> {
{ "text": text, "pos": i }
}
// scan_ident advance i while chars[i] is alphanumeric or underscore
fn scan_ident(chars: [String], start: Int, total: Int) -> Map<String, Any> {
// scan_ident advance i while char_buf[i] is alphanumeric or underscore
fn scan_ident(start: Int, total: Int) -> Map<String, Any> {
let i = start
let text = ""
let running = true
@@ -172,7 +177,7 @@ fn scan_ident(chars: [String], start: Int, total: Int) -> Map<String, Any> {
if i >= total {
let running = false
} else {
let ch = native_list_get(chars, i)
let ch = native_char_at(i)
if is_alnum_or_underscore(ch) {
let text = text + ch
let i = i + 1
@@ -186,7 +191,7 @@ fn scan_ident(chars: [String], start: Int, total: Int) -> Map<String, Any> {
// scan_string scan a quoted string literal, handling \" escapes.
// Starts AFTER the opening quote. Returns { "text": content, "pos": i_after_close }
fn scan_string(chars: [String], start: Int, total: Int) -> Map<String, Any> {
fn scan_string(start: Int, total: Int) -> Map<String, Any> {
let i = start
let text = ""
let closed = false
@@ -195,12 +200,12 @@ fn scan_string(chars: [String], start: Int, total: Int) -> Map<String, Any> {
if i >= total {
let running = false
} else {
let ch = native_list_get(chars, i)
let ch = native_char_at(i)
if ch == "\\" {
// escape: peek next char
let next_i = i + 1
if next_i < total {
let next_ch = native_list_get(chars, next_i)
let next_ch = native_char_at(next_i)
if next_ch == "\"" {
let text = text + "\""
let i = next_i + 1
@@ -244,13 +249,13 @@ fn scan_string(chars: [String], start: Int, total: Int) -> Map<String, Any> {
// Main lexer
fn lex(source: String) -> [Map<String, Any>] {
let chars: [String] = native_string_chars(source)
let total: Int = native_list_len(chars)
native_chars_init(source)
let total: Int = native_char_len()
let tokens: [Map<String, Any>] = native_list_empty()
let i: Int = 0
while i < total {
let ch: String = native_list_get(chars, i)
let ch: String = native_char_at(i)
// Skip whitespace
if is_whitespace(ch) {
@@ -260,7 +265,7 @@ fn lex(source: String) -> [Map<String, Any>] {
if ch == "/" {
let next_i = i + 1
if next_i < total {
let next_ch: String = native_list_get(chars, next_i)
let next_ch: String = native_char_at(next_i)
if next_ch == "/" {
// skip to end of line
let i = i + 2
@@ -269,7 +274,7 @@ fn lex(source: String) -> [Map<String, Any>] {
if i >= total {
let running2 = false
} else {
let lch: String = native_list_get(chars, i)
let lch: String = native_char_at(i)
if lch == "\n" {
let running2 = false
} else {
@@ -288,7 +293,7 @@ fn lex(source: String) -> [Map<String, Any>] {
} else {
// String literal
if ch == "\"" {
let result = scan_string(chars, i + 1, total)
let result = scan_string(i + 1, total)
let str_text: String = result["text"]
let new_pos: Int = result["pos"]
let tokens = native_list_append(tokens, make_tok("Str", str_text))
@@ -296,18 +301,18 @@ fn lex(source: String) -> [Map<String, Any>] {
} else {
// Number literal
if is_digit(ch) {
let result = scan_digits(chars, i, total)
let result = scan_digits(i, total)
let num_text: String = result["text"]
let new_pos: Int = result["pos"]
// check for float (dot followed by digit)
if new_pos < total {
let dot_ch: String = native_list_get(chars, new_pos)
let dot_ch: String = native_char_at(new_pos)
if dot_ch == "." {
let after_dot = new_pos + 1
if after_dot < total {
let after_dot_ch: String = native_list_get(chars, after_dot)
let after_dot_ch: String = native_char_at(after_dot)
if is_digit(after_dot_ch) {
let frac_result = scan_digits(chars, after_dot, total)
let frac_result = scan_digits(after_dot, total)
let frac_text: String = frac_result["text"]
let frac_pos: Int = frac_result["pos"]
let tokens = native_list_append(tokens, make_tok("Float", num_text + "." + frac_text))
@@ -331,7 +336,7 @@ fn lex(source: String) -> [Map<String, Any>] {
} else {
// Identifier or keyword
if is_alpha(ch) || ch == "_" {
let result = scan_ident(chars, i, total)
let result = scan_ident(i, total)
let word: String = result["text"]
let new_pos: Int = result["pos"]
let kw = keyword_kind(word)
@@ -346,7 +351,7 @@ fn lex(source: String) -> [Map<String, Any>] {
let peek_i = i + 1
let peek_ch = ""
if peek_i < total {
let peek_ch = native_list_get(chars, peek_i)
let peek_ch = native_char_at(peek_i)
}
if ch == "=" {
+146 -140
View File
@@ -6,26 +6,30 @@
// The cursor (integer position into the token list) is threaded through every
// parse function. Functions return { "node": <map>, "pos": <int> }.
//
// The token list is stored in the VM's global token buffer via
// native_tokens_init / native_token_at / native_token_len. This avoids
// O(n²) cloning that occurs when passing the list as a function argument.
//
// Entry point: fn parse(tokens: [Map<String, Any>]) -> [Map<String, Any>]
// Token access helpers
fn tok_at(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
native_list_get(tokens, pos)
fn tok_at(pos: Int) -> Map<String, Any> {
native_token_at(pos)
}
fn tok_kind(tokens: [Map<String, Any>], pos: Int) -> String {
let t = tok_at(tokens, pos)
fn tok_kind(pos: Int) -> String {
let t = native_token_at(pos)
t["kind"]
}
fn tok_value(tokens: [Map<String, Any>], pos: Int) -> String {
let t = tok_at(tokens, pos)
fn tok_value(pos: Int) -> String {
let t = native_token_at(pos)
t["value"]
}
fn expect(tokens: [Map<String, Any>], pos: Int, kind: String) -> Int {
let k = tok_kind(tokens, pos)
fn expect(pos: Int, kind: String) -> Int {
let k = tok_kind(pos)
if k == kind {
return pos + 1
}
@@ -43,26 +47,26 @@ fn make_result(node: Map<String, Any>, pos: Int) -> Map<String, Any> {
// Skips over a type annotation, returning the new position.
// Types can be: Ident, [Type], Map<K,V>, Type?, Type<Type,...>
fn skip_type(tokens: [Map<String, Any>], pos: Int) -> Int {
let k = tok_kind(tokens, pos)
fn skip_type(pos: Int) -> Int {
let k = tok_kind(pos)
// Array type: [Type]
if k == "LBracket" {
let p = pos + 1
let p = skip_type(tokens, p)
let p = expect(tokens, p, "RBracket")
let p = skip_type(p)
let p = expect(p, "RBracket")
return p
}
// Named type (possibly generic)
if k == "Ident" {
let p = pos + 1
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "Lt" {
// Generic params: skip until matching >
let p = p + 1
let depth = 1
let running = true
while running {
let kk = tok_kind(tokens, p)
let kk = tok_kind(p)
if kk == "Eof" {
let running = false
} else {
@@ -82,7 +86,7 @@ fn skip_type(tokens: [Map<String, Any>], pos: Int) -> Int {
}
}
}
let k3 = tok_kind(tokens, p)
let k3 = tok_kind(p)
if k3 == "QuestionMark" {
let p = p + 1
}
@@ -100,12 +104,12 @@ fn skip_type(tokens: [Map<String, Any>], pos: Int) -> Int {
// Parameter list
// Parses (name: Type, name: Type, ...) returns { "params": [...], "pos": ... }
fn parse_params(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let p = expect(tokens, pos, "LParen")
fn parse_params(pos: Int) -> Map<String, Any> {
let p = expect(pos, "LParen")
let params: [Map<String, Any>] = native_list_empty()
let running = true
while running {
let k = tok_kind(tokens, p)
let k = tok_kind(p)
if k == "RParen" {
let running = false
} else {
@@ -113,28 +117,28 @@ fn parse_params(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let running = false
} else {
// param name
let pname = tok_value(tokens, p)
let pname = tok_value(p)
let p = p + 1
let p = expect(tokens, p, "Colon")
let p = skip_type(tokens, p)
let p = expect(p, "Colon")
let p = skip_type(p)
let param = { "name": pname }
let params = native_list_append(params, param)
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "Comma" {
let p = p + 1
}
}
}
}
let p = expect(tokens, p, "RParen")
let p = expect(p, "RParen")
{ "params": params, "pos": p }
}
// Expression parsing
fn parse_primary(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let k = tok_kind(tokens, pos)
let v = tok_value(tokens, pos)
fn parse_primary(pos: Int) -> Map<String, Any> {
let k = tok_kind(pos)
let v = tok_value(pos)
// Integer literal
if k == "Int" {
@@ -163,10 +167,10 @@ fn parse_primary(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
// Grouped expression
if k == "LParen" {
let r = parse_expr(tokens, pos + 1)
let r = parse_expr(pos + 1)
let node = r["node"]
let p = r["pos"]
let p = expect(tokens, p, "RParen")
let p = expect(p, "RParen")
return make_result(node, p)
}
@@ -176,25 +180,25 @@ fn parse_primary(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let elems: [Map<String, Any>] = native_list_empty()
let running = true
while running {
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "RBracket" {
let running = false
} else {
if k2 == "Eof" {
let running = false
} else {
let r = parse_expr(tokens, p)
let r = parse_expr(p)
let elem = r["node"]
let p = r["pos"]
let elems = native_list_append(elems, elem)
let k3 = tok_kind(tokens, p)
let k3 = tok_kind(p)
if k3 == "Comma" {
let p = p + 1
}
}
}
}
let p = expect(tokens, p, "RBracket")
let p = expect(p, "RBracket")
return make_result({ "expr": "Array", "elems": elems }, p)
}
@@ -204,7 +208,7 @@ fn parse_primary(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let pairs: [Map<String, Any>] = native_list_empty()
let running = true
while running {
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "RBrace" {
let running = false
} else {
@@ -212,46 +216,46 @@ fn parse_primary(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let running = false
} else {
// key: Str token
let key = tok_value(tokens, p)
let key = tok_value(p)
let p = p + 1
let p = expect(tokens, p, "Colon")
let r = parse_expr(tokens, p)
let p = expect(p, "Colon")
let r = parse_expr(p)
let val_node = r["node"]
let p = r["pos"]
let pair = { "key": key, "value": val_node }
let pairs = native_list_append(pairs, pair)
let k3 = tok_kind(tokens, p)
let k3 = tok_kind(p)
if k3 == "Comma" {
let p = p + 1
}
}
}
}
let p = expect(tokens, p, "RBrace")
let p = expect(p, "RBrace")
return make_result({ "expr": "Map", "pairs": pairs }, p)
}
// if expression
if k == "If" {
let r = parse_if(tokens, pos)
let r = parse_if(pos)
return r
}
// match expression
if k == "Match" {
let r = parse_match(tokens, pos)
let r = parse_match(pos)
return r
}
// for expression (used as statement)
if k == "For" {
let r = parse_for_expr(tokens, pos)
let r = parse_for_expr(pos)
return r
}
// Unary not
if k == "Not" {
let r = parse_primary(tokens, pos + 1)
let r = parse_primary(pos + 1)
let inner = r["node"]
let p = r["pos"]
return make_result({ "expr": "Not", "inner": inner }, p)
@@ -259,7 +263,7 @@ fn parse_primary(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
// Unary minus
if k == "Minus" {
let r = parse_primary(tokens, pos + 1)
let r = parse_primary(pos + 1)
let inner = r["node"]
let p = r["pos"]
return make_result({ "expr": "Neg", "inner": inner }, p)
@@ -269,29 +273,29 @@ fn parse_primary(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
make_result({ "expr": "Nil" }, pos + 1)
}
fn parse_if(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let p = expect(tokens, pos, "If")
let r = parse_expr(tokens, p)
fn parse_if(pos: Int) -> Map<String, Any> {
let p = expect(pos, "If")
let r = parse_expr(p)
let cond = r["node"]
let p = r["pos"]
let r2 = parse_block(tokens, p)
let r2 = parse_block(p)
let then_stmts = r2["stmts"]
let p = r2["pos"]
let has_else = false
let else_stmts: [Map<String, Any>] = native_list_empty()
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "Else" {
let p = p + 1
let k3 = tok_kind(tokens, p)
let k3 = tok_kind(p)
if k3 == "If" {
// else-if chain: parse as nested if
let r3 = parse_if(tokens, p)
let r3 = parse_if(p)
let nested = r3["node"]
let p = r3["pos"]
let else_stmts = native_list_append(else_stmts, { "stmt": "Expr", "value": nested })
let has_else = true
} else {
let r3 = parse_block(tokens, p)
let r3 = parse_block(p)
let else_stmts = r3["stmts"]
let p = r3["pos"]
let has_else = true
@@ -300,16 +304,16 @@ fn parse_if(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
make_result({ "expr": "If", "cond": cond, "then": then_stmts, "else": else_stmts, "has_else": has_else }, p)
}
fn parse_match(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let p = expect(tokens, pos, "Match")
let r = parse_expr(tokens, p)
fn parse_match(pos: Int) -> Map<String, Any> {
let p = expect(pos, "Match")
let r = parse_expr(p)
let subject = r["node"]
let p = r["pos"]
let p = expect(tokens, p, "LBrace")
let p = expect(p, "LBrace")
let arms: [Map<String, Any>] = native_list_empty()
let running = true
while running {
let k = tok_kind(tokens, p)
let k = tok_kind(p)
if k == "RBrace" {
let running = false
} else {
@@ -317,131 +321,131 @@ fn parse_match(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let running = false
} else {
// parse pattern => body
let r2 = parse_pattern(tokens, p)
let r2 = parse_pattern(p)
let pattern = r2["node"]
let p = r2["pos"]
let p = expect(tokens, p, "FatArrow")
let r3 = parse_expr(tokens, p)
let p = expect(p, "FatArrow")
let r3 = parse_expr(p)
let body = r3["node"]
let p = r3["pos"]
let arm = { "pattern": pattern, "body": body }
let arms = native_list_append(arms, arm)
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "Comma" {
let p = p + 1
}
}
}
}
let p = expect(tokens, p, "RBrace")
let p = expect(p, "RBrace")
make_result({ "expr": "Match", "subject": subject, "arms": arms }, p)
}
fn parse_pattern(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let k = tok_kind(tokens, pos)
fn parse_pattern(pos: Int) -> Map<String, Any> {
let k = tok_kind(pos)
if k == "Ident" {
let v = tok_value(tokens, pos)
let v = tok_value(pos)
if v == "_" {
return make_result({ "pattern": "Wildcard" }, pos + 1)
}
return make_result({ "pattern": "Binding", "name": v }, pos + 1)
}
if k == "Int" {
return make_result({ "pattern": "LitInt", "value": tok_value(tokens, pos) }, pos + 1)
return make_result({ "pattern": "LitInt", "value": tok_value(pos) }, pos + 1)
}
if k == "Str" {
return make_result({ "pattern": "LitStr", "value": tok_value(tokens, pos) }, pos + 1)
return make_result({ "pattern": "LitStr", "value": tok_value(pos) }, pos + 1)
}
if k == "Bool" {
return make_result({ "pattern": "LitBool", "value": tok_value(tokens, pos) }, pos + 1)
return make_result({ "pattern": "LitBool", "value": tok_value(pos) }, pos + 1)
}
// Wildcard _
make_result({ "pattern": "Wildcard" }, pos + 1)
}
fn parse_for_expr(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let p = expect(tokens, pos, "For")
let item_name = tok_value(tokens, p)
fn parse_for_expr(pos: Int) -> Map<String, Any> {
let p = expect(pos, "For")
let item_name = tok_value(p)
let p = p + 1
let p = expect(tokens, p, "In")
let r = parse_expr(tokens, p)
let p = expect(p, "In")
let r = parse_expr(p)
let list_expr = r["node"]
let p = r["pos"]
let r2 = parse_block(tokens, p)
let r2 = parse_block(p)
let body = r2["stmts"]
let p = r2["pos"]
make_result({ "expr": "For", "item": item_name, "list": list_expr, "body": body }, p)
}
fn parse_block(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let p = expect(tokens, pos, "LBrace")
fn parse_block(pos: Int) -> Map<String, Any> {
let p = expect(pos, "LBrace")
let stmts: [Map<String, Any>] = native_list_empty()
let running = true
while running {
let k = tok_kind(tokens, p)
let k = tok_kind(p)
if k == "RBrace" {
let running = false
} else {
if k == "Eof" {
let running = false
} else {
let r = parse_stmt(tokens, p)
let r = parse_stmt(p)
let stmt = r["node"]
let p = r["pos"]
let stmts = native_list_append(stmts, stmt)
}
}
}
let p = expect(tokens, p, "RBrace")
let p = expect(p, "RBrace")
{ "stmts": stmts, "pos": p }
}
// Postfix expressions (calls, field access, index)
fn parse_postfix(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let r = parse_primary(tokens, pos)
fn parse_postfix(pos: Int) -> Map<String, Any> {
let r = parse_primary(pos)
let node = r["node"]
let p = r["pos"]
let running = true
while running {
let k = tok_kind(tokens, p)
let k = tok_kind(p)
if k == "LParen" {
// function call
let p = p + 1
let args: [Map<String, Any>] = native_list_empty()
let run2 = true
while run2 {
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "RParen" {
let run2 = false
} else {
if k2 == "Eof" {
let run2 = false
} else {
let r2 = parse_expr(tokens, p)
let r2 = parse_expr(p)
let arg = r2["node"]
let p = r2["pos"]
let args = native_list_append(args, arg)
let k3 = tok_kind(tokens, p)
let k3 = tok_kind(p)
if k3 == "Comma" {
let p = p + 1
}
}
}
}
let p = expect(tokens, p, "RParen")
let p = expect(p, "RParen")
let node = { "expr": "Call", "func": node, "args": args }
} else {
if k == "Dot" {
let field = tok_value(tokens, p + 1)
let field = tok_value(p + 1)
let p = p + 2
let node = { "expr": "Field", "object": node, "field": field }
} else {
if k == "LBracket" {
let r2 = parse_expr(tokens, p + 1)
let r2 = parse_expr(p + 1)
let idx = r2["node"]
let p = r2["pos"]
let p = expect(tokens, p, "RBracket")
let p = expect(p, "RBracket")
let node = { "expr": "Index", "object": node, "index": idx }
} else {
if k == "QuestionMark" {
@@ -491,18 +495,18 @@ fn is_binop(kind: String) -> Bool {
false
}
fn parse_binop(tokens: [Map<String, Any>], pos: Int, min_prec: Int) -> Map<String, Any> {
let r = parse_postfix(tokens, pos)
fn parse_binop(pos: Int, min_prec: Int) -> Map<String, Any> {
let r = parse_postfix(pos)
let left = r["node"]
let p = r["pos"]
let running = true
while running {
let k = tok_kind(tokens, p)
let k = tok_kind(p)
let prec = op_precedence(k)
if is_binop(k) {
if prec >= min_prec {
let op = k
let r2 = parse_binop(tokens, p + 1, prec + 1)
let r2 = parse_binop(p + 1, prec + 1)
let right = r2["node"]
let p = r2["pos"]
let left = { "expr": "BinOp", "op": op, "left": left, "right": right }
@@ -516,28 +520,28 @@ fn parse_binop(tokens: [Map<String, Any>], pos: Int, min_prec: Int) -> Map<Strin
make_result(left, p)
}
fn parse_expr(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
parse_binop(tokens, pos, 1)
fn parse_expr(pos: Int) -> Map<String, Any> {
parse_binop(pos, 1)
}
// Statement parsing
fn parse_stmt(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let k = tok_kind(tokens, pos)
fn parse_stmt(pos: Int) -> Map<String, Any> {
let k = tok_kind(pos)
// let binding
if k == "Let" {
let p = pos + 1
let name = tok_value(tokens, p)
let name = tok_value(p)
let p = p + 1
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
// optional type annotation: name: Type
if k2 == "Colon" {
let p = p + 1
let p = skip_type(tokens, p)
let p = skip_type(p)
}
let p = expect(tokens, p, "Eq")
let r = parse_expr(tokens, p)
let p = expect(p, "Eq")
let r = parse_expr(p)
let val = r["node"]
let p = r["pos"]
return make_result({ "stmt": "Let", "name": name, "value": val }, p)
@@ -546,14 +550,14 @@ fn parse_stmt(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
// return statement
if k == "Return" {
let p = pos + 1
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "RBrace" {
return make_result({ "stmt": "Return", "value": { "expr": "Nil" } }, p)
}
if k2 == "Eof" {
return make_result({ "stmt": "Return", "value": { "expr": "Nil" } }, p)
}
let r = parse_expr(tokens, p)
let r = parse_expr(p)
let val = r["node"]
let p = r["pos"]
return make_result({ "stmt": "Return", "value": val }, p)
@@ -562,18 +566,18 @@ fn parse_stmt(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
// fn definition
if k == "Fn" {
let p = pos + 1
let name = tok_value(tokens, p)
let name = tok_value(p)
let p = p + 1
let r = parse_params(tokens, p)
let r = parse_params(p)
let params = r["params"]
let p = r["pos"]
// return type annotation: -> Type
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "Arrow" {
let p = p + 1
let p = skip_type(tokens, p)
let p = skip_type(p)
}
let r2 = parse_block(tokens, p)
let r2 = parse_block(p)
let body = r2["stmts"]
let p = r2["pos"]
return make_result({ "stmt": "FnDef", "name": name, "params": params, "body": body }, p)
@@ -582,69 +586,69 @@ fn parse_stmt(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
// type definition
if k == "Type" {
let p = pos + 1
let name = tok_value(tokens, p)
let name = tok_value(p)
let p = p + 1
let p = expect(tokens, p, "LBrace")
let p = expect(p, "LBrace")
let fields: [Map<String, Any>] = native_list_empty()
let running = true
while running {
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "RBrace" {
let running = false
} else {
if k2 == "Eof" {
let running = false
} else {
let fname = tok_value(tokens, p)
let fname = tok_value(p)
let p = p + 1
let p = expect(tokens, p, "Colon")
let p = skip_type(tokens, p)
let p = expect(p, "Colon")
let p = skip_type(p)
let fields = native_list_append(fields, { "name": fname })
let k3 = tok_kind(tokens, p)
let k3 = tok_kind(p)
if k3 == "Comma" {
let p = p + 1
}
}
}
}
let p = expect(tokens, p, "RBrace")
let p = expect(p, "RBrace")
return make_result({ "stmt": "TypeDef", "name": name, "fields": fields }, p)
}
// enum definition
if k == "Enum" {
let p = pos + 1
let name = tok_value(tokens, p)
let name = tok_value(p)
let p = p + 1
let p = expect(tokens, p, "LBrace")
let p = expect(p, "LBrace")
let variants: [Map<String, Any>] = native_list_empty()
let running = true
while running {
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "RBrace" {
let running = false
} else {
if k2 == "Eof" {
let running = false
} else {
let vname = tok_value(tokens, p)
let vname = tok_value(p)
let p = p + 1
let variants = native_list_append(variants, { "name": vname })
let k3 = tok_kind(tokens, p)
let k3 = tok_kind(p)
if k3 == "Comma" {
let p = p + 1
}
}
}
}
let p = expect(tokens, p, "RBrace")
let p = expect(p, "RBrace")
return make_result({ "stmt": "EnumDef", "name": name, "variants": variants }, p)
}
// import statement
if k == "Import" {
let p = pos + 1
let path = tok_value(tokens, p)
let path = tok_value(p)
let p = p + 1
return make_result({ "stmt": "Import", "path": path }, p)
}
@@ -652,20 +656,20 @@ fn parse_stmt(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
// from ... import { ... }
if k == "From" {
let p = pos + 1
let module_name = tok_value(tokens, p)
let module_name = tok_value(p)
let p = p + 1
// skip "import" keyword
let k2 = tok_kind(tokens, p)
let k2 = tok_kind(p)
if k2 == "Import" {
let p = p + 1
}
// skip { Name, ... }
let k3 = tok_kind(tokens, p)
let k3 = tok_kind(p)
if k3 == "LBrace" {
let p = p + 1
let running = true
while running {
let k4 = tok_kind(tokens, p)
let k4 = tok_kind(p)
if k4 == "RBrace" {
let running = false
} else {
@@ -673,14 +677,14 @@ fn parse_stmt(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let running = false
} else {
let p = p + 1
let k5 = tok_kind(tokens, p)
let k5 = tok_kind(p)
if k5 == "Comma" {
let p = p + 1
}
}
}
}
let p = expect(tokens, p, "RBrace")
let p = expect(p, "RBrace")
}
return make_result({ "stmt": "Import", "path": module_name }, p)
}
@@ -688,10 +692,10 @@ fn parse_stmt(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
// while loop
if k == "While" {
let p = pos + 1
let r = parse_expr(tokens, p)
let r = parse_expr(p)
let cond = r["node"]
let p = r["pos"]
let r2 = parse_block(tokens, p)
let r2 = parse_block(p)
let body = r2["stmts"]
let p = r2["pos"]
return make_result({ "stmt": "While", "cond": cond, "body": body }, p)
@@ -700,13 +704,13 @@ fn parse_stmt(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
// for loop
if k == "For" {
let p = pos + 1
let item_name = tok_value(tokens, p)
let item_name = tok_value(p)
let p = p + 1
let p = expect(tokens, p, "In")
let r = parse_expr(tokens, p)
let p = expect(p, "In")
let r = parse_expr(p)
let list_expr = r["node"]
let p = r["pos"]
let r2 = parse_block(tokens, p)
let r2 = parse_block(p)
let body = r2["stmts"]
let p = r2["pos"]
return make_result({ "stmt": "For", "item": item_name, "list": list_expr, "body": body }, p)
@@ -717,11 +721,11 @@ fn parse_stmt(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
let p = pos + 1
// skip decorator name
let p = p + 1
return parse_stmt(tokens, p)
return parse_stmt(p)
}
// bare expression or if/match statement
let r = parse_expr(tokens, pos)
let r = parse_expr(pos)
let val = r["node"]
let p = r["pos"]
make_result({ "stmt": "Expr", "value": val }, p)
@@ -730,7 +734,9 @@ fn parse_stmt(tokens: [Map<String, Any>], pos: Int) -> Map<String, Any> {
// Top-level parse
fn parse(tokens: [Map<String, Any>]) -> [Map<String, Any>] {
let total: Int = native_list_len(tokens)
// Store tokens in global buffer to avoid O(n²) cloning on every recursive call.
native_tokens_init(tokens)
let total: Int = native_token_len()
let stmts: [Map<String, Any>] = native_list_empty()
let pos: Int = 0
let running = true
@@ -738,11 +744,11 @@ fn parse(tokens: [Map<String, Any>]) -> [Map<String, Any>] {
if pos >= total {
let running = false
} else {
let k = tok_kind(tokens, pos)
let k = tok_kind(pos)
if k == "Eof" {
let running = false
} else {
let r = parse_stmt(tokens, pos)
let r = parse_stmt(pos)
let stmt = r["node"]
let new_pos: Int = r["pos"]
let stmts = native_list_append(stmts, stmt)