ludic/selfhost/parse.ludic

# parse.ludic — recursive-descent parser: the token slice -> an AST.
# Core grammar: properties, vars/lets, consts, fns, entry; the ECS/ui/match
# constructs live in parse_game.ludic. Uses the global `toks` and a cursor `pi`.

var pi: int = 0
var prog: []Node                 # the top-level declarations
var g_game_name: ptr             # the `game`/`module` name

fn cur() -> Tok { return toks[pi] }
fn pk(o: int) -> Tok { return toks[pi + o] }
fn is_op(v: ptr) -> bool { let t = toks[pi];  return t.kind == TK_OP and (t.text == v) }
fn is_id(v: ptr) -> bool { let t = toks[pi];  return t.kind == TK_ID and (t.text == v) }
fn is_kw(v: ptr) -> bool { return is_id(v) }

fn perr(msg: ptr) -> void {
  let e = file_stderr()
  file_write(e, "ludicc(self): parse error: ", 27)
  file_write(e, msg, len(msg))
  file_write(e, "\n", 1)
  os_exit(1)
}

fn eat_op(v: ptr) -> void { if not is_op(v) { perr(v) }; pi = pi + 1 }
fn eat_id() -> ptr {
  let t = toks[pi]
  if t.kind != TK_ID { perr("expected identifier") }
  pi = pi + 1
  return t.text
}
fn skipnl() -> void { while toks[pi].kind == TK_NL { pi = pi + 1 } }

# a type: `[]T` slice, or a plain name (int/ptr/str/bool/struct)
fn ptype() -> ptr {
  if is_op("[") {
    pi = pi + 1
    eat_op("]")
    let el = ptype()
    let out = bytes(len(el) + 3)
    out[0] = 91; out[1] = 93              # "[]"
    var i = 0
    while el[i] != 0 { out[2 + i] = el[i]; i = i + 1 }
    out[2 + i] = 0
    return out
  }
  return eat_id()
}

# ---- expressions -----------------------------------------------------------
fn expr() -> Node { return p_or() }

fn args_call(call: Node) -> void {
  eat_op("("); skipnl()
  while not is_op(")") { push(call.kids, expr()); skipnl(); if is_op(",") { pi = pi + 1; skipnl() } }
  eat_op(")")
}

# ---- string interpolation --------------------------------------------------
# `text {expr} text` desugars to a `+` chain of string literals and `str(expr)`
# holes, so it reuses the string-concat operator and needs no new runtime.
fn interp_lit(buf: ptr, len: int) -> Node { let n = node(E_STR); n.s = buf[0..0 + len]; return n }
fn interp_add(acc: Node, part: Node) -> Node {
  if acc == null { return part }
  return mkbin("+", acc, part)
}
fn interp_str(e: Node) -> Node {                  # wrap a hole in str(...)
  let c = node(E_CALL); let id = node(E_ID); id.s = "str"; c.a = id; push(c.kids, e); return c
}
fn parse_hole(inner: ptr) -> Node {               # re-lex+parse an embedded expression
  let saved_toks = toks;  let saved_pi = pi
  lex(inner);  pi = 0;  skipnl()
  let e = expr()
  toks = saved_toks;  pi = saved_pi
  return e
}
fn parse_interp(raw: ptr) -> Node {
  let n = len(raw)
  var acc: Node = null
  let lit = bytes(n + 1)
  var lj = 0
  var i = 0
  while i < n {
    let c = raw[i]
    if c == 123 {                                 # '{'
      if raw[i + 1] == 123 { lit[lj] = 123; lj = lj + 1; i = i + 2; continue }   # {{ -> {
      if lj > 0 { acc = interp_add(acc, interp_lit(lit, lj));  lj = 0 }
      i = i + 1
      let hs = i
      var depth = 1
      while i < n and depth > 0 {
        let d = raw[i]
        if d == 123 { depth = depth + 1 }
        else { if d == 125 { depth = depth - 1;  if depth == 0 { break } } }
        i = i + 1
      }
      acc = interp_add(acc, interp_str(parse_hole(raw[hs..i])))
      i = i + 1                                   # skip the closing '}'
    } else {
      if c == 125 and raw[i + 1] == 125 { lit[lj] = 125; lj = lj + 1; i = i + 2; continue }  # }} -> }
      if c == 92 {                                # backslash escape in literal text
        let e = raw[i + 1];  var r = e
        if e == 110 { r = 10 }
        if e == 116 { r = 9 }
        lit[lj] = r; lj = lj + 1; i = i + 2; continue
      }
      lit[lj] = c; lj = lj + 1; i = i + 1
    }
  }
  if lj > 0 or (acc == null) { acc = interp_add(acc, interp_lit(lit, lj)) }
  return acc
}

fn p_primary() -> Node {
  let t = toks[pi]
  if t.kind == TK_INTERP { pi = pi + 1; return parse_interp(t.text) }
  if t.kind == TK_INT { let n = node(E_INT); n.ival = t.ival; pi = pi + 1; return n }
  if t.kind == TK_FLOAT { let n = node(E_FLOAT); n.ival = t.ival; pi = pi + 1; return n }
  if t.kind == TK_STR { let n = node(E_STR); n.s = t.text; pi = pi + 1; return n }
  if t.kind == TK_ID {
    if (t.text == "true")  { let n = node(E_BOOL); n.ival = 1; pi = pi + 1; return n }
    if (t.text == "false") { let n = node(E_BOOL); n.ival = 0; pi = pi + 1; return n }
    if (t.text == "null")  { pi = pi + 1; return node(E_NULL) }
    if (t.text == "new")   { pi = pi + 1;  let n = node(E_NEW); n.s = ptype(); return n }
    let n = node(E_ID); n.s = t.text; pi = pi + 1; return n
  }
  if is_op("(") { pi = pi + 1; skipnl(); let e = expr(); skipnl(); eat_op(")"); return e }
  perr("expected expression")
  return node(E_INT)
}

fn p_postfix() -> Node {
  var e = p_primary()
  while true {
    if is_op(".") { pi = pi + 1; let m = node(E_MEMBER); m.a = e; m.s = eat_id(); e = m }
    else { if is_op("[") { pi = pi + 1; let lo = expr()
        if is_op("..") { pi = pi + 1; let sl = node(E_SLICE); sl.a = e; sl.b = lo; sl.c = expr(); eat_op("]"); e = sl }   # s[a..b] substring
        else { let ix = node(E_INDEX); ix.a = e; ix.b = lo; eat_op("]"); e = ix } }
      else { if is_op("(") { let c = node(E_CALL); c.a = e; args_call(c); e = c } else { break } } }
  }
  return e
}

fn p_unary() -> Node {
  if is_op("-") { pi = pi + 1; let n = node(E_UN); n.s = "-"; n.a = p_unary(); return n }
  if is_op("~") { pi = pi + 1; let n = node(E_UN); n.s = "~"; n.a = p_unary(); return n }   # bitwise not
  if is_id("not") { pi = pi + 1; let n = node(E_UN); n.s = "not"; n.a = p_unary(); return n }
  return p_postfix()
}

fn mkbin(op: ptr, l: Node, r: Node) -> Node { let b = node(E_BIN); b.s = op; b.a = l; b.b = r; return b }

# precedence (Go-style, so `flags & MASK == 0` needs no parens): shifts and `&`
# bind like `*`; `|` and `^` bind like `+`; both tighter than comparison.
fn p_mul() -> Node {
  var l = p_unary()
  while is_op("*") or is_op("/") or is_op("%") or is_op("<<") or is_op(">>") or is_op("&") {
    let op = toks[pi].text; pi = pi + 1; l = mkbin(op, l, p_unary()) }
  return l
}
fn p_add() -> Node {
  var l = p_mul()
  while is_op("+") or is_op("-") or is_op("|") or is_op("^") {
    let op = toks[pi].text; pi = pi + 1; l = mkbin(op, l, p_mul()) }
  return l
}
fn p_cmp() -> Node {
  var l = p_add()
  while is_op("<") or is_op("<=") or is_op(">") or is_op(">=") or is_op("==") or is_op("!=") {
    let op = toks[pi].text; pi = pi + 1; l = mkbin(op, l, p_add())
  }
  return l
}
fn p_and() -> Node {
  var l = p_cmp()
  while is_id("and") { pi = pi + 1; l = mkbin("and", l, p_cmp()) }
  return l
}
fn p_or() -> Node {
  var l = p_and()
  while is_id("or") { pi = pi + 1; l = mkbin("or", l, p_and()) }
  return l
}

# a record literal `{ field: value, ... }` — used by spawn component inits.
# Rule A: a named part uses `:` (`=` is assignment/binding only).
fn record() -> Node {
  eat_op("{")
  let r = node(E_REC)
  while true { skipnl(); if is_op("}") { break }
    let fi = node(E_FINIT); fi.s = eat_id(); eat_op(":"); fi.a = expr(); push(r.kids, fi)
    if is_op(",") { pi = pi + 1 } }
  eat_op("}"); return r
}

# ---- statements ------------------------------------------------------------
fn block() -> Node {
  skipnl(); eat_op("{")
  let b = node(N_BLOCK)
  while true { skipnl(); if is_op("}") { break }
    push(b.kids, stmt())
    # Rule B: statements are separated by a newline or ';' (both lex to TK_NL).
    # After a statement the next token must be that separator or the block's end —
    # two statements may not sit adjacent with only spaces between them.
    var sep = toks[pi].kind == TK_NL
    if is_op("}") { sep = true }
    if not sep { perr("expected newline or ';' between statements") }
  }
  eat_op("}")
  return b
}

fn stmt() -> Node {
  let t = toks[pi]
  if t.kind == TK_ID {
    if (t.text == "let") or (t.text == "var") {
      var mut = 0; if (t.text == "var") { mut = 1 }     # let = immutable, var = mutable
      pi = pi + 1; let n = node(S_LET); n.ival = mut; n.line = toks[pi].line; n.s = eat_id()
      if is_op(":") { pi = pi + 1; n.ty = ptype() }
      if is_op("=") { pi = pi + 1; n.a = expr() }
      return n
    }
    if (t.text == "return") {
      pi = pi + 1; let n = node(S_RETURN)
      if toks[pi].kind != TK_NL and not is_op("}") { n.a = expr() }
      return n
    }
    if (t.text == "if") {
      pi = pi + 1; let n = node(S_IF); n.a = expr(); n.b = block()
      let save = pi; skipnl()                 # peek past newlines for a trailing `else`
      if is_id("else") { pi = pi + 1; skipnl()
        if is_id("if") { n.c = stmt() } else { n.c = block() } }
      else { pi = save }                      # no else: keep the separator for block()'s Rule-B check
      return n
    }
    if (t.text == "while") { pi = pi + 1; let n = node(S_WHILE); n.a = expr(); n.b = block(); return n }
    if (t.text == "for") {
      pi = pi + 1
      if is_op("(") { return parse_query_for() }
      let n = node(S_FOR); n.s = eat_id()
      let inkw = eat_id()                       # 'in'
      n.a = expr(); eat_op(".."); n.b = expr(); n.c = block()
      return n
    }
    if (t.text == "spawn") { return parse_spawn() }
    if (t.text == "machine") {
      pi = pi + 1; let n = node(S_MACHINE); n.a = expr(); skipnl(); eat_op("{")
      var sidx = 0                              # states auto-number by declaration order
      while true { skipnl(); if is_op("}") { break }
        let stkw = eat_id()                     # 'state'
        let s = node(S_STATE); s.s = eat_id()
        if is_op("=") { pi = pi + 1; s.b = expr() }              # explicit value (still allowed)
        else { let iv = node(E_INT); iv.ival = sidx; s.b = iv }  # else its ordinal
        skipnl(); s.a = block()
        push(n.kids, s); sidx = sidx + 1 }
      eat_op("}"); return n
    }
    if (t.text == "become") { pi = pi + 1; let n = node(S_BECOME); n.s = eat_id(); return n }
    if (t.text == "despawn") { pi = pi + 1; let n = node(S_DESPAWN); n.a = expr(); return n }
    if (t.text == "enable") or (t.text == "disable") {
      var en = 0; if (t.text == "enable") { en = 1 }
      pi = pi + 1; let n = node(S_TOGGLE); n.ival = en; n.s = eat_id()     # `enable P on e` / `disable Model` / `disable Handler`
      if is_id("on") { pi = pi + 1; n.a = expr() }                        # property on an entity
      return n
    }
    if (t.text == "break") { pi = pi + 1; return node(S_BREAK) }
    if (t.text == "continue") { pi = pi + 1; return node(S_CONTINUE) }
    if (t.text == "match") {
      pi = pi + 1; let n = node(S_MATCH); n.a = expr(); skipnl(); eat_op("{")
      while true {
        skipnl(); if is_op("}") { break }
        let arm = node(S_MARM)
        while true { push(arm.kids, expr()); if is_op(",") { pi = pi + 1; skipnl(); continue }; break }
        eat_op("=>"); skipnl()
        if is_op("{") { arm.a = block() } else { let b = node(N_BLOCK); push(b.kids, stmt()); arm.a = b }
        push(n.kids, arm)
      }
      eat_op("}"); return n
    }
  }
  let e = expr()
  if is_op("=") or is_op("+=") or is_op("-=") or is_op("*=") or is_op("/=") {
    let n = node(S_ASSIGN); n.line = toks[pi].line; n.s = toks[pi].text; pi = pi + 1; n.a = e; n.b = expr(); return n
  }
  let n = node(S_EXPR); n.a = e; return n
}

# ---- declarations ----------------------------------------------------------
fn parse_var() -> Node {
  pi = pi + 1; let n = node(N_VAR); n.s = eat_id(); eat_op(":"); n.ty = ptype()
  if is_op("=") { pi = pi + 1; n.a = expr() }
  return n
}
fn parse_const() -> Node {
  pi = pi + 1; let n = node(N_CONST); n.s = eat_id(); eat_op(":"); n.ty = ptype(); eat_op("="); n.a = expr()
  return n
}
fn parse_fn() -> Node {
  pi = pi + 1; let n = node(N_FN); n.s = eat_id(); eat_op("(")
  while not is_op(")") {
    let p = node(N_PARAM); p.s = eat_id(); eat_op(":"); p.ty = ptype(); push(n.kids, p)
    if is_op(",") { pi = pi + 1 }
  }
  eat_op(")")
  n.ty = "void"
  if is_op("->") { pi = pi + 1; n.ty = ptype() }
  n.a = block()
  return n
}
fn parse_main() -> Node { pi = pi + 1; let n = node(N_MAIN); n.a = block(); return n }

# directory part of a path, including the trailing '/', or "" if none
fn dir_of(path: ptr) -> ptr {
  var last = 0 - 1
  var i = 0
  while path[i] != 0 { if path[i] == 47 { last = i };  i = i + 1 }
  if last < 0 { return "" }
  return path[0..0 + (last + 1)]
}
fn path_join(dir: ptr, rel: ptr) -> ptr {
  if rel[0] == 47 { return rel }               # absolute
  return (dir + rel)
}

var loaded_paths: []ptr
var cur_dir: ptr

fn already_loaded(full: ptr) -> bool {
  var i = 0
  while i < len(loaded_paths) { if (loaded_paths[i] == full) { return true };  i = i + 1 }
  return false
}

# parse one top-level declaration (or resolve an import) into `prog`.
# Modifiers are `@annotations` in front of the declaration: `@export`, `@edge`,
# `@pure`, `@deterministic`, … — one channel, not a zoo of prefix keywords.
fn parse_one_decl() -> void {
  var is_export = false
  var qspec: Node = null
  var onspawn_model: ptr = null
  var ondespawn_model: ptr = null
  var onattach_prop: ptr = null
  var onenable_prop: ptr = null
  var ondisable_prop: ptr = null
  var hook_phase: ptr = null            # @OnStart / @OnQuit override the phase
  while is_op("@") {
    pi = pi + 1; let a = eat_id()             # collect a leading @annotation
    if (a == "export") { is_export = true }
    else { if (a == "Queries") { qspec = parse_queries_anno() }               # @Queries(these: [...], on: ...)
    else { if (a == "OnSpawn") { eat_op("("); onspawn_model = eat_id(); eat_op(")") }
    else { if (a == "OnDespawn") { eat_op("("); ondespawn_model = eat_id(); eat_op(")") }
    else { if (a == "OnAttach") { eat_op("("); onattach_prop = eat_id(); eat_op(")") }
    else { if (a == "OnEnable") { eat_op("("); onenable_prop = eat_id(); eat_op(")") }
    else { if (a == "OnDisable") { eat_op("("); ondisable_prop = eat_id(); eat_op(")") }
    else { if (a == "OnStart") { hook_phase = "Start" }                       # boot
    else { if (a == "OnQuit") { hook_phase = "OnQuit" }                       # shutdown
    else { if is_op("(") { var d = 0          # any other @anno(args) — parsed and skipped
        while true { if is_op("(") { d = d + 1 };  if is_op(")") { d = d - 1 };  pi = pi + 1;  if d == 0 { break } } } } } } } } } } } }
    skipnl()
  }
  if is_id("import") { pi = pi + 1
    let t = toks[pi]
    if t.kind != TK_STR { perr("expected \"path\" after import") }
    let rel = t.text;  pi = pi + 1
    do_import(rel)
    return
  }
  if is_id("enum") { push(prog, parse_enum()); return }
  if is_id("property") { push(prog, parse_component()); return }
  if is_id("model") { push(prog, parse_archetype()); return }
  if is_id("handler") {
    let h = parse_system()
    if (onspawn_model != null) { register_onspawn(onspawn_model, h.a); return }       # spawn hook
    if (ondespawn_model != null) { register_ondespawn(ondespawn_model, h.a); return } # despawn hook
    if (onattach_prop != null) { register_onattach(onattach_prop, h.a); return }      # attach hook
    if (onenable_prop != null) { register_onenable(onenable_prop, h.a); return }      # enable hook
    if (ondisable_prop != null) { register_ondisable(ondisable_prop, h.a); return }   # disable hook
    if (hook_phase != null) { h.ty = hook_phase }                                     # @OnStart/@OnQuit
    if (qspec != null) {               # @Queries wraps the body in its S_QUERY
      qspec.a = h.a
      let wrap = node(N_BLOCK); push(wrap.kids, qspec); h.a = wrap
    }
    push(prog, h); return
  }
  if is_id("ui") { push(prog, parse_ui()); return }
  if is_id("var")    { push(prog, parse_var()); return }
  if is_id("const")  { push(prog, parse_const()); return }
  if is_id("fn") { let f = parse_fn(); if is_export { f.ival = 1 };  push(prog, f); return }
  if is_id("extern") { push(prog, parse_extern()); return }
  if is_id("entry")  { push(prog, parse_main()); return }
  perr("expected declaration")
}

# lex and parse an imported fragment into `prog`, saving/restoring lexer state
fn do_import(rel: ptr) -> void {
  let full = path_join(cur_dir, rel)
  if already_loaded(full) { return }
  push(loaded_paths, full)
  let src = read_file(full)
  if (src == null) { perr(`cannot open import {full}`) }
  let saved_toks = toks;  let saved_pi = pi;  let saved_dir = cur_dir
  cur_dir = dir_of(full)
  lex(src)                                   # resets the global token stream
  pi = 0
  skipnl()
  while toks[pi].kind != TK_EOF { parse_one_decl(); skipnl() }
  toks = saved_toks;  pi = saved_pi;  cur_dir = saved_dir
}

# a game (has systems/components) links the Ludic runtime; auto-splice it the
# way the C compiler does. Tools (a `main` block, no ECS) get nothing.
fn maybe_splice_runtime() -> void {
  if not has_ecs() { return }
  let saved = cur_dir
  cur_dir = ""
  do_import("runtime/native/core.ludic")
  cur_dir = saved
}

fn parse_program() -> void {
  prog = new []Node
  g_computed = new []Node
  g_onspawn = new []Node
  g_ondespawn = new []Node
  g_onattach = new []Node
  g_onenable = new []Node
  g_ondisable = new []Node
  loaded_paths = new []ptr
  skipnl()
  g_game_name = "Ludic"
  # imports may precede the program block
  while is_id("import") { pi = pi + 1; let t = toks[pi];  let rel = t.text;  pi = pi + 1;  do_import(rel); skipnl() }
  # @annotations on the program itself (e.g. @Handles(Movement)) — parsed, skipped
  while is_op("@") {
    pi = pi + 1; let a = eat_id()
    if is_op("(") { var d = 0
      while true { if is_op("(") { d = d + 1 };  if is_op(")") { d = d - 1 };  pi = pi + 1;  if d == 0 { break } } }
    skipnl()
  }
  if is_id("program") { pi = pi + 1; g_game_name = eat_id(); skipnl(); eat_op("{") }
  while true {
    skipnl()
    if toks[pi].kind == TK_EOF { break }
    if is_op("}") { break }
    parse_one_decl()
  }
}