2186 lines
60 KiB
Plaintext
2186 lines
60 KiB
Plaintext
var parse = function(tokens, src, filename, tokenizer) {
|
|
var _src_len = length(src)
|
|
|
|
var template_escape_map = {
|
|
n: "\n", t: "\t", r: "\r", "\\": "\\",
|
|
"`": "`", "$": "$", "0": character(0)
|
|
}
|
|
|
|
// ============================================================
|
|
// Parser Cursor
|
|
// ============================================================
|
|
|
|
var cursor = 0
|
|
var tok = null
|
|
var got_lf = false
|
|
var prev_tok = null
|
|
var _control_depth = 0
|
|
var _control_type = null
|
|
var _expecting_body = false
|
|
|
|
var advance = function() {
|
|
var t = null
|
|
var k = null
|
|
prev_tok = tok
|
|
cursor = cursor + 1
|
|
got_lf = false
|
|
while (cursor < length(tokens)) {
|
|
t = tokens[cursor]
|
|
k = t.kind
|
|
if (k == "space" || k == "comment") {
|
|
cursor = cursor + 1
|
|
continue
|
|
}
|
|
if (k == "newline") {
|
|
got_lf = true
|
|
cursor = cursor + 1
|
|
continue
|
|
}
|
|
tok = t
|
|
return null
|
|
}
|
|
tok = tokens[length(tokens) - 1]
|
|
}
|
|
|
|
var peek_ahead = function(n) {
|
|
var c = cursor + 1
|
|
var count = 0
|
|
var t = null
|
|
var k = null
|
|
while (c < length(tokens)) {
|
|
t = tokens[c]
|
|
k = t.kind
|
|
if (k != "space" && k != "comment" && k != "newline") {
|
|
count = count + 1
|
|
if (count == n) return t
|
|
}
|
|
c = c + 1
|
|
}
|
|
return tokens[length(tokens) - 1]
|
|
}
|
|
|
|
var init_cursor = function() {
|
|
cursor = -1
|
|
advance()
|
|
}
|
|
|
|
// ============================================================
|
|
// AST Helpers
|
|
// ============================================================
|
|
|
|
var errors = []
|
|
var error_count = 0
|
|
var fn_counter = 1
|
|
|
|
var ast_node = function(kind, token) {
|
|
return {
|
|
kind: kind,
|
|
at: token.at,
|
|
from_row: token.from_row,
|
|
from_column: token.from_column
|
|
}
|
|
}
|
|
|
|
var ast_node_end = function(node) {
|
|
node.to_row = prev_tok.to_row
|
|
node.to_column = prev_tok.to_column
|
|
return node
|
|
}
|
|
|
|
var parse_error = function(token, msg) {
|
|
if (error_count >= 5) return null
|
|
error_count = error_count + 1
|
|
push(errors, {
|
|
message: msg,
|
|
line: token.from_row + 1,
|
|
column: token.from_column + 1,
|
|
offset: token.at
|
|
})
|
|
}
|
|
|
|
var _keywords = {
|
|
"if": true, in: true, "do": true, go: true,
|
|
"var": true, def: true, "for": true,
|
|
"else": true, "this": true, "null": true, "true": true,
|
|
"false": true, "while": true, "break": true,
|
|
"return": true, "delete": true,
|
|
disrupt: true, "function": true, "continue": true,
|
|
disruption: true
|
|
}
|
|
|
|
var is_keyword = function(kind) {
|
|
return _keywords[kind] == true
|
|
}
|
|
|
|
// ============================================================
|
|
// Expression Parsing
|
|
// ============================================================
|
|
|
|
// Forward declarations via var
|
|
var parse_expr = null
|
|
var parse_assign_expr = null
|
|
var parse_assign = null
|
|
var parse_statement = null
|
|
var parse_block_statements = null
|
|
var parse_function_inner = null
|
|
var parse_arrow_function = null
|
|
|
|
var is_arrow_function = function() {
|
|
if (tok.kind != "(") return false
|
|
var c = cursor + 1
|
|
var depth = 1
|
|
var k = null
|
|
while (c < length(tokens) && depth > 0) {
|
|
k = tokens[c].kind
|
|
if (k == "(") { depth = depth + 1 }
|
|
else if (k == ")") { depth = depth - 1 }
|
|
else if (k == "text" || k == "number") { null }
|
|
c = c + 1
|
|
}
|
|
while (c < length(tokens)) {
|
|
k = tokens[c].kind
|
|
if (k != "space" && k != "newline" && k != "comment") break
|
|
c = c + 1
|
|
}
|
|
if (c >= length(tokens)) return false
|
|
return tokens[c].kind == "=>"
|
|
}
|
|
|
|
var parse_primary = function() {
|
|
var start = tok
|
|
var node = null
|
|
var k = tok.kind
|
|
var list = null
|
|
var pair = null
|
|
var left = null
|
|
var right = null
|
|
var is_ident = false
|
|
var is_kw = false
|
|
var p1 = null
|
|
var elem = null
|
|
var fn_start = null
|
|
var fn = null
|
|
var name_item = null
|
|
var params = null
|
|
var param = null
|
|
var rpos = 0
|
|
var pattern_parts = null
|
|
var flags_parts = null
|
|
var tv = null
|
|
var has_interp = false
|
|
var ti = 0
|
|
var tpl_list = null
|
|
var fmt_parts = null
|
|
var idx = 0
|
|
var tvi = 0
|
|
var tvlen = 0
|
|
var depth = 0
|
|
var expr_parts = null
|
|
var expr_str = null
|
|
var tc = null
|
|
var tq = null
|
|
var esc_ch = null
|
|
var esc_val = null
|
|
var expr_tokens = null
|
|
var sub_ast = null
|
|
var sub_stmt = null
|
|
var sub_expr = null
|
|
var meth_old_cd = 0
|
|
var meth_old_ct = null
|
|
var meth_old_eb = false
|
|
|
|
if (k == "number") {
|
|
node = ast_node("number", start)
|
|
node.value = tok.value
|
|
node.number = tok.number
|
|
advance()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "text") {
|
|
// Check for template interpolation: ${...}
|
|
tv = tok.value
|
|
has_interp = false
|
|
ti = 0
|
|
while (ti < length(tv) - 1) {
|
|
if (tv[ti] == "$" && tv[ti + 1] == "{") {
|
|
if (ti == 0 || tv[ti - 1] != "\\") {
|
|
has_interp = true
|
|
break
|
|
}
|
|
}
|
|
ti = ti + 1
|
|
}
|
|
if (!has_interp || tokenizer == null) {
|
|
node = ast_node("text", start)
|
|
node.value = tok.value
|
|
advance()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
// Template literal with interpolation
|
|
node = ast_node("text literal", start)
|
|
tpl_list = []
|
|
node.list = tpl_list
|
|
fmt_parts = []
|
|
idx = 0
|
|
tvi = 0
|
|
tvlen = length(tv)
|
|
while (tvi < tvlen) {
|
|
if (tv[tvi] == "\\" && tvi + 1 < tvlen) {
|
|
esc_ch = tv[tvi + 1]
|
|
esc_val = template_escape_map[esc_ch]
|
|
if (esc_val != null) { push(fmt_parts, esc_val) }
|
|
else { push(fmt_parts, esc_ch) }
|
|
tvi = tvi + 2
|
|
} else if (tv[tvi] == "$" && tvi + 1 < tvlen && tv[tvi + 1] == "{") {
|
|
tvi = tvi + 2
|
|
depth = 1
|
|
expr_parts = []
|
|
while (tvi < tvlen && depth > 0) {
|
|
tc = tv[tvi]
|
|
if (tc == "{") { depth = depth + 1; push(expr_parts, tc); tvi = tvi + 1 }
|
|
else if (tc == "}") {
|
|
depth = depth - 1
|
|
if (depth > 0) { push(expr_parts, tc) }
|
|
tvi = tvi + 1
|
|
}
|
|
else if (tc == "'" || tc == "\"" || tc == "`") {
|
|
tq = tc
|
|
push(expr_parts, tc)
|
|
tvi = tvi + 1
|
|
while (tvi < tvlen && tv[tvi] != tq) {
|
|
if (tv[tvi] == "\\" && tvi + 1 < tvlen) {
|
|
push(expr_parts, tv[tvi])
|
|
tvi = tvi + 1
|
|
}
|
|
push(expr_parts, tv[tvi])
|
|
tvi = tvi + 1
|
|
}
|
|
if (tvi < tvlen) { push(expr_parts, tv[tvi]); tvi = tvi + 1 }
|
|
} else {
|
|
push(expr_parts, tc)
|
|
tvi = tvi + 1
|
|
}
|
|
}
|
|
expr_str = text(expr_parts)
|
|
expr_tokens = tokenizer(expr_str, "<template>").tokens
|
|
sub_ast = parse(expr_tokens, expr_str, "<template>", tokenizer)
|
|
if (sub_ast != null && sub_ast.statements != null && length(sub_ast.statements) > 0) {
|
|
sub_stmt = sub_ast.statements[0]
|
|
sub_expr = null
|
|
if (sub_stmt.kind == "call") {
|
|
sub_expr = sub_stmt.expression
|
|
} else {
|
|
sub_expr = sub_stmt
|
|
}
|
|
push(tpl_list, sub_expr)
|
|
}
|
|
push(fmt_parts, "{")
|
|
push(fmt_parts, text(idx))
|
|
push(fmt_parts, "}")
|
|
idx = idx + 1
|
|
} else {
|
|
push(fmt_parts, tv[tvi])
|
|
tvi = tvi + 1
|
|
}
|
|
}
|
|
node.value = text(fmt_parts)
|
|
advance()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "name") {
|
|
p1 = peek_ahead(1)
|
|
if (p1.kind == "=>") {
|
|
return parse_arrow_function()
|
|
}
|
|
node = ast_node("name", start)
|
|
node.name = tok.value
|
|
advance()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "null") {
|
|
node = ast_node("null", start)
|
|
advance()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "true") {
|
|
node = ast_node("true", start)
|
|
advance()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "false") {
|
|
node = ast_node("false", start)
|
|
advance()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "this") {
|
|
node = ast_node("this", start)
|
|
advance()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "[") {
|
|
node = ast_node("array", start)
|
|
list = []
|
|
node.list = list
|
|
advance()
|
|
while (tok.kind != "]" && tok.kind != "eof") {
|
|
elem = parse_assign_expr()
|
|
if (elem != null) push(list, elem)
|
|
if (tok.kind == ",") advance()
|
|
else break
|
|
}
|
|
ast_node_end(node)
|
|
if (tok.kind == "]") advance()
|
|
else if (tok.kind == "eof") parse_error(tok, "unterminated array literal, expected ']'")
|
|
return node
|
|
}
|
|
if (k == "{") {
|
|
node = ast_node("record", start)
|
|
list = []
|
|
node.list = list
|
|
advance()
|
|
while (tok.kind != "}" && tok.kind != "eof") {
|
|
pair = {}
|
|
is_ident = (tok.kind == "name")
|
|
is_kw = is_keyword(tok.kind)
|
|
if (is_ident || is_kw || tok.kind == "text" || tok.kind == "number") {
|
|
if (is_kw) {
|
|
left = ast_node("name", tok)
|
|
left.name = tok.kind
|
|
advance()
|
|
ast_node_end(left)
|
|
} else {
|
|
left = parse_primary()
|
|
}
|
|
pair.left = left
|
|
} else if (tok.kind == "[") {
|
|
advance()
|
|
left = parse_assign_expr()
|
|
pair.left = left
|
|
if (tok.kind == "]") advance()
|
|
else parse_error(tok, "expected ']' after computed property")
|
|
} else {
|
|
parse_error(tok, "expected property name in object literal")
|
|
break
|
|
}
|
|
if (tok.kind == ":") {
|
|
advance()
|
|
right = parse_assign_expr()
|
|
pair.right = right
|
|
} else if (tok.kind == "(") {
|
|
fn_start = tok
|
|
fn = ast_node("function", fn_start)
|
|
name_item = pair.left
|
|
if (name_item != null && name_item.name != null) {
|
|
fn.name = name_item.name
|
|
}
|
|
params = []
|
|
fn.list = params
|
|
advance()
|
|
while (tok.kind != ")" && tok.kind != "eof") {
|
|
if (tok.kind == "name") {
|
|
param = ast_node("name", tok)
|
|
param.name = tok.value
|
|
advance()
|
|
ast_node_end(param)
|
|
if (tok.kind == "=" || tok.kind == "|") {
|
|
advance()
|
|
param.expression = parse_assign_expr()
|
|
}
|
|
push(params, param)
|
|
} else {
|
|
parse_error(tok, "expected parameter name")
|
|
break
|
|
}
|
|
if (tok.kind == ",") advance()
|
|
else break
|
|
}
|
|
if (tok.kind == ")") advance()
|
|
else if (tok.kind == "eof") parse_error(tok, "unterminated method parameter list")
|
|
if (length(params) > 4) parse_error(tok, "functions cannot have more than 4 parameters")
|
|
fn.arity = length(params)
|
|
meth_old_cd = _control_depth
|
|
meth_old_ct = _control_type
|
|
meth_old_eb = _expecting_body
|
|
_control_depth = 0
|
|
_control_type = null
|
|
_expecting_body = false
|
|
if (tok.kind == "{") {
|
|
advance()
|
|
fn.statements = parse_block_statements()
|
|
if (tok.kind == "}") advance()
|
|
else if (tok.kind == "eof") parse_error(tok, "unterminated method body")
|
|
} else {
|
|
parse_error(tok, "expected '{' for method body")
|
|
}
|
|
_control_depth = meth_old_cd
|
|
_control_type = meth_old_ct
|
|
_expecting_body = meth_old_eb
|
|
fn.function_nr = fn_counter
|
|
fn_counter = fn_counter + 1
|
|
ast_node_end(fn)
|
|
pair.right = fn
|
|
} else if (is_ident && (tok.kind == "," || tok.kind == "}")) {
|
|
right = ast_node("name", pair.left)
|
|
right.name = pair.left.name
|
|
ast_node_end(right)
|
|
pair.right = right
|
|
} else {
|
|
parse_error(tok, "expected ':' after property name")
|
|
}
|
|
push(list, pair)
|
|
if (tok.kind == ",") advance()
|
|
else if (tok.kind == "{") {
|
|
if (right && right.kind == "(") {
|
|
parse_error(tok, "unexpected '{' after property value; use method shorthand `name(args) { ... }` or `name: function(args) { ... }`")
|
|
} else {
|
|
parse_error(tok, "expected ',' or '}' in object literal")
|
|
}
|
|
break
|
|
} else break
|
|
}
|
|
ast_node_end(node)
|
|
if (tok.kind == "}") advance()
|
|
else if (tok.kind == "eof") parse_error(tok, "unterminated object literal, expected '}'")
|
|
return node
|
|
}
|
|
if (k == "(") {
|
|
if (is_arrow_function()) {
|
|
return parse_arrow_function()
|
|
}
|
|
advance()
|
|
node = parse_expr()
|
|
if (tok.kind == ")") advance()
|
|
else if (tok.kind == "eof") parse_error(tok, "unterminated parenthesized expression, expected ')'")
|
|
else parse_error(tok, "expected ')' after expression")
|
|
return node
|
|
}
|
|
if (k == "function") {
|
|
return parse_function_inner()
|
|
}
|
|
if (k == "/") {
|
|
node = ast_node("regexp", start)
|
|
rpos = tok.at + 1
|
|
pattern_parts = []
|
|
flags_parts = []
|
|
while (rpos < _src_len && src[rpos] != "/") {
|
|
if (src[rpos] == "\\" && rpos + 1 < _src_len) {
|
|
push(pattern_parts, src[rpos])
|
|
push(pattern_parts, src[rpos + 1])
|
|
rpos = rpos + 2
|
|
} else {
|
|
push(pattern_parts, src[rpos])
|
|
rpos = rpos + 1
|
|
}
|
|
}
|
|
if (rpos < _src_len) rpos = rpos + 1
|
|
while (rpos < _src_len && is_letter(src[rpos])) {
|
|
push(flags_parts, src[rpos])
|
|
rpos = rpos + 1
|
|
}
|
|
node.pattern = text(pattern_parts)
|
|
if (length(flags_parts) > 0) node.flags = text(flags_parts)
|
|
// Skip all tokens consumed by the regex re-scan
|
|
while (true) {
|
|
advance()
|
|
if (tok.kind == "eof" || tok.at >= rpos) break
|
|
}
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "eof") {
|
|
parse_error(start, "unexpected end of input")
|
|
} else {
|
|
parse_error(start, "unexpected token where expression expected")
|
|
}
|
|
advance()
|
|
return null
|
|
}
|
|
|
|
var parse_postfix = function() {
|
|
var node = parse_primary()
|
|
var start = null
|
|
var new_node = null
|
|
var index = null
|
|
var arg = null
|
|
var args_list = null
|
|
var one_node = null
|
|
var binop_node = null
|
|
var op = null
|
|
if (node == null) return null
|
|
while (true) {
|
|
start = tok
|
|
if (tok.kind == ".") {
|
|
advance()
|
|
new_node = ast_node(".", start)
|
|
new_node.left = node
|
|
if (tok.kind == "name" || is_keyword(tok.kind)) {
|
|
if (tok.kind == "name") {
|
|
new_node.right = tok.value
|
|
} else {
|
|
new_node.right = tok.kind
|
|
}
|
|
advance()
|
|
} else {
|
|
parse_error(tok, "expected property name after '.'")
|
|
}
|
|
ast_node_end(new_node)
|
|
node = new_node
|
|
} else if (tok.kind == "[") {
|
|
advance()
|
|
new_node = ast_node("[", start)
|
|
new_node.left = node
|
|
if (tok.kind == "]") {
|
|
advance()
|
|
} else {
|
|
index = parse_assign_expr()
|
|
new_node.right = index
|
|
if (tok.kind == "]") advance()
|
|
else parse_error(tok, "expected ']'")
|
|
}
|
|
ast_node_end(new_node)
|
|
node = new_node
|
|
} else if (tok.kind == "(") {
|
|
advance()
|
|
new_node = ast_node("(", start)
|
|
new_node.expression = node
|
|
args_list = []
|
|
new_node.list = args_list
|
|
while (tok.kind != ")" && tok.kind != "eof") {
|
|
arg = parse_assign_expr()
|
|
if (arg != null) push(args_list, arg)
|
|
if (tok.kind == ",") advance()
|
|
else break
|
|
}
|
|
if (tok.kind == ")") advance()
|
|
else parse_error(tok, "unterminated argument list, expected ')'")
|
|
ast_node_end(new_node)
|
|
node = new_node
|
|
} else if (tok.kind == "++") {
|
|
new_node = ast_node("++", start)
|
|
new_node.expression = node
|
|
new_node.postfix = true
|
|
advance()
|
|
ast_node_end(new_node)
|
|
node = new_node
|
|
} else if (tok.kind == "--") {
|
|
new_node = ast_node("--", start)
|
|
new_node.expression = node
|
|
new_node.postfix = true
|
|
advance()
|
|
ast_node_end(new_node)
|
|
node = new_node
|
|
} else {
|
|
break
|
|
}
|
|
}
|
|
return node
|
|
}
|
|
|
|
var parse_unary = function() {
|
|
var start = tok
|
|
var node = null
|
|
var expr = null
|
|
var k = tok.kind
|
|
var operand = null
|
|
var one_node = null
|
|
var binop_node = null
|
|
var op = null
|
|
if (k == "!") {
|
|
advance()
|
|
node = ast_node("!", start)
|
|
node.expression = parse_unary()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "~") {
|
|
advance()
|
|
node = ast_node("~", start)
|
|
node.expression = parse_unary()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "+") {
|
|
advance()
|
|
node = ast_node("+unary", start)
|
|
node.expression = parse_unary()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "-") {
|
|
advance()
|
|
node = ast_node("-unary", start)
|
|
node.expression = parse_unary()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "++" || k == "--") {
|
|
advance()
|
|
operand = parse_unary()
|
|
one_node = ast_node("number", start)
|
|
one_node.number = 1
|
|
one_node.value = "1"
|
|
ast_node_end(one_node)
|
|
op = "+"
|
|
if (k == "--") op = "-"
|
|
binop_node = ast_node(op, start)
|
|
binop_node.left = operand
|
|
binop_node.right = one_node
|
|
ast_node_end(binop_node)
|
|
node = ast_node("assign", start)
|
|
node.left = operand
|
|
node.right = binop_node
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
if (k == "delete") {
|
|
advance()
|
|
node = ast_node("delete", start)
|
|
node.expression = parse_unary()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
return parse_postfix()
|
|
}
|
|
|
|
// Binary operator precedence
|
|
var binop_prec = {
|
|
"**": 14,
|
|
"*": 13, "/": 13, "%": 13,
|
|
"+": 12, "-": 12,
|
|
"<<": 11, ">>": 11, ">>>": 11,
|
|
"<": 10, ">": 10, "<=": 10, ">=": 10, in: 10,
|
|
"==": 9, "!=": 9, "===": 9, "!==": 9,
|
|
"&": 8, "^": 7, "|": 6,
|
|
"&&": 5, "||": 4
|
|
}
|
|
|
|
var parse_binary = function(min_prec) {
|
|
var left_node = parse_unary()
|
|
var start = null
|
|
var op = null
|
|
var prec = null
|
|
var next_prec = 0
|
|
var right_node = null
|
|
var node = null
|
|
if (left_node == null) return null
|
|
while (true) {
|
|
start = tok
|
|
op = tok.kind
|
|
prec = binop_prec[op]
|
|
if (prec == null || prec < min_prec) break
|
|
advance()
|
|
next_prec = prec + 1
|
|
if (prec == 14) next_prec = prec // right-assoc for **
|
|
right_node = parse_binary(next_prec)
|
|
node = ast_node(op, start)
|
|
node.left = left_node
|
|
node.right = right_node
|
|
ast_node_end(node)
|
|
left_node = node
|
|
}
|
|
return left_node
|
|
}
|
|
|
|
var parse_ternary = function() {
|
|
var cond = parse_binary(1)
|
|
var start = null
|
|
var then_expr = null
|
|
var else_expr = null
|
|
var node = null
|
|
if (cond == null) return null
|
|
if (tok.kind == "?") {
|
|
start = tok
|
|
advance()
|
|
then_expr = parse_assign_expr()
|
|
if (tok.kind == ":") advance()
|
|
else parse_error(tok, "expected ':' in ternary expression")
|
|
else_expr = parse_assign_expr()
|
|
node = ast_node("then", start)
|
|
node.expression = cond
|
|
node.then = then_expr
|
|
node.else = else_expr
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
return cond
|
|
}
|
|
|
|
// Assign operators
|
|
var assign_ops = {
|
|
"=": "assign", "+=": "+=", "-=": "-=", "*=": "*=", "/=": "/=", "%=": "%=",
|
|
"<<=": "<<=", ">>=": ">>=", ">>>=": ">>>=",
|
|
"&=": "&=", "^=": "^=", "|=": "|=", "**=": "**=",
|
|
"&&=": "&&=", "||=": "||="
|
|
}
|
|
|
|
var compound_binop = {
|
|
"+=": "+", "-=": "-", "*=": "*", "/=": "/", "%=": "%",
|
|
"<<=": "<<", ">>=": ">>", ">>>=": ">>>",
|
|
"&=": "&", "^=": "^", "|=": "|", "**=": "**",
|
|
"&&=": "&&", "||=": "||"
|
|
}
|
|
|
|
parse_assign = function(unused) {
|
|
var left_node = parse_ternary()
|
|
var start = null
|
|
var kind = null
|
|
var right_node = null
|
|
var node = null
|
|
var left_kind = null
|
|
var right_kind = null
|
|
var binop = null
|
|
var binop_node = null
|
|
if (left_node == null) return null
|
|
start = tok
|
|
kind = assign_ops[tok.kind]
|
|
if (kind == null) return left_node
|
|
|
|
left_kind = left_node.kind
|
|
if (left_kind != "name" && left_kind != "." && left_kind != "[") {
|
|
parse_error(start, "invalid assignment left-hand side")
|
|
}
|
|
|
|
advance()
|
|
right_node = parse_assign()
|
|
|
|
binop = compound_binop[kind]
|
|
if (binop != null) {
|
|
binop_node = ast_node(binop, start)
|
|
binop_node.left = left_node
|
|
binop_node.right = right_node
|
|
ast_node_end(binop_node)
|
|
node = ast_node("assign", start)
|
|
node.left = left_node
|
|
node.right = binop_node
|
|
} else {
|
|
node = ast_node(kind, start)
|
|
node.left = left_node
|
|
node.right = right_node
|
|
if (left_node.kind == "[" && left_node.right == null) node.push = true
|
|
if (right_node != null && right_node.kind == "[" && right_node.right == null) node.pop = true
|
|
}
|
|
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
parse_assign_expr = function(unused) {
|
|
return parse_assign()
|
|
}
|
|
|
|
parse_expr = function(unused) {
|
|
var left_node = parse_assign()
|
|
var start = null
|
|
var right_node = null
|
|
var node = null
|
|
if (left_node == null) return null
|
|
while (tok.kind == ",") {
|
|
start = tok
|
|
advance()
|
|
right_node = parse_assign()
|
|
node = ast_node(",", start)
|
|
node.left = left_node
|
|
node.right = right_node
|
|
ast_node_end(node)
|
|
left_node = node
|
|
}
|
|
return left_node
|
|
}
|
|
|
|
// ============================================================
|
|
// Statement Parsing
|
|
// ============================================================
|
|
|
|
var in_disruption = 0
|
|
|
|
var expect_semi = function() {
|
|
if (tok.kind == ";") { advance(); return null }
|
|
if (tok.kind == "eof" || tok.kind == "}" || got_lf || tok.kind == "else") return null
|
|
parse_error(tok, "expecting ';'")
|
|
}
|
|
|
|
var sync_to_statement = function() {
|
|
var k = null
|
|
while (tok.kind != "eof") {
|
|
k = tok.kind
|
|
if (k == ";") { advance(); return null }
|
|
if (k == "}") return null
|
|
if (k == "var" || k == "def" || k == "if" || k == "while" ||
|
|
k == "for" || k == "return" || k == "disrupt" ||
|
|
k == "function" || k == "break" || k == "continue" || k == "do") return null
|
|
advance()
|
|
}
|
|
}
|
|
|
|
parse_block_statements = function(unused) {
|
|
var stmts = []
|
|
var before = null
|
|
var stmt = null
|
|
while (tok.kind != "}" && tok.kind != "eof") {
|
|
before = cursor
|
|
stmt = parse_statement()
|
|
if (stmt != null) {
|
|
push(stmts, stmt)
|
|
} else if (cursor == before) {
|
|
sync_to_statement()
|
|
}
|
|
}
|
|
return stmts
|
|
}
|
|
|
|
parse_function_inner = function(unused) {
|
|
var start = tok
|
|
var node = ast_node("function", start)
|
|
var params = []
|
|
var stmts = null
|
|
var param = null
|
|
var prev_names = null
|
|
var pname = null
|
|
var old_dis = 0
|
|
var old_cd = _control_depth
|
|
var old_ct = _control_type
|
|
var old_eb = _expecting_body
|
|
|
|
if (in_disruption) {
|
|
parse_error(tok, "cannot define function inside disruption clause")
|
|
}
|
|
|
|
advance() // skip 'function'
|
|
|
|
if (tok.kind == "name") {
|
|
node.name = tok.value
|
|
advance()
|
|
}
|
|
|
|
node.list = params
|
|
if (tok.kind == "(") {
|
|
advance()
|
|
prev_names = []
|
|
while (tok.kind != ")" && tok.kind != "eof") {
|
|
if (tok.kind == "name") {
|
|
param = ast_node("name", tok)
|
|
param.name = tok.value
|
|
pname = tok.value
|
|
if (find(prev_names, pname) != null) parse_error(tok, "duplicate parameter name '" + pname + "'")
|
|
push(prev_names, pname)
|
|
advance()
|
|
ast_node_end(param)
|
|
if (tok.kind == "=" || tok.kind == "|") {
|
|
advance()
|
|
param.expression = parse_assign_expr()
|
|
}
|
|
push(params, param)
|
|
} else {
|
|
parse_error(tok, "expected parameter name")
|
|
break
|
|
}
|
|
if (tok.kind == ",") advance()
|
|
else break
|
|
}
|
|
if (tok.kind == ")") advance()
|
|
else if (tok.kind == "eof") parse_error(tok, "unterminated function parameter list, expected ')'")
|
|
} else {
|
|
parse_error(tok, "expected '(' after function name")
|
|
}
|
|
|
|
if (length(params) > 4) parse_error(tok, "functions cannot have more than 4 parameters")
|
|
node.arity = length(params)
|
|
|
|
_control_depth = 0
|
|
_control_type = null
|
|
_expecting_body = false
|
|
if (tok.kind == "{") {
|
|
advance()
|
|
stmts = parse_block_statements()
|
|
node.statements = stmts
|
|
if (tok.kind == "}") advance()
|
|
else if (tok.kind == "eof") parse_error(tok, "unterminated function body, expected '}'")
|
|
} else {
|
|
parse_error(tok, "expected '{' for function body")
|
|
}
|
|
|
|
if (tok.kind == "disruption") {
|
|
advance()
|
|
if (tok.kind == "{") {
|
|
advance()
|
|
old_dis = in_disruption
|
|
in_disruption = 1
|
|
node.disruption = parse_block_statements()
|
|
in_disruption = old_dis
|
|
if (tok.kind == "}") advance()
|
|
else if (tok.kind == "eof") parse_error(tok, "unterminated disruption clause, expected '}'")
|
|
} else {
|
|
parse_error(tok, "expected '{' after disruption")
|
|
}
|
|
}
|
|
|
|
_control_depth = old_cd
|
|
_control_type = old_ct
|
|
_expecting_body = old_eb
|
|
node.function_nr = fn_counter
|
|
fn_counter = fn_counter + 1
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
parse_arrow_function = function(unused) {
|
|
var start = tok
|
|
var node = ast_node("function", start)
|
|
var params = []
|
|
var param = null
|
|
var stmts = null
|
|
var ret = null
|
|
var expr = null
|
|
var prev_names = null
|
|
var pname = null
|
|
var old_cd = _control_depth
|
|
var old_ct = _control_type
|
|
var old_eb = _expecting_body
|
|
node.arrow = true
|
|
|
|
if (in_disruption) {
|
|
parse_error(tok, "cannot define function inside disruption clause")
|
|
}
|
|
|
|
node.list = params
|
|
|
|
if (tok.kind == "name") {
|
|
param = ast_node("name", tok)
|
|
param.name = tok.value
|
|
advance()
|
|
ast_node_end(param)
|
|
push(params, param)
|
|
} else if (tok.kind == "(") {
|
|
advance()
|
|
prev_names = []
|
|
while (tok.kind != ")" && tok.kind != "eof") {
|
|
if (tok.kind == "name") {
|
|
param = ast_node("name", tok)
|
|
param.name = tok.value
|
|
pname = tok.value
|
|
if (find(prev_names, pname) != null) parse_error(tok, "duplicate parameter name '" + pname + "'")
|
|
push(prev_names, pname)
|
|
advance()
|
|
ast_node_end(param)
|
|
if (tok.kind == "=" || tok.kind == "|") {
|
|
advance()
|
|
param.expression = parse_assign_expr()
|
|
}
|
|
push(params, param)
|
|
} else {
|
|
parse_error(tok, "expected parameter name")
|
|
break
|
|
}
|
|
if (tok.kind == ",") advance()
|
|
else break
|
|
}
|
|
if (tok.kind == ")") advance()
|
|
}
|
|
|
|
if (length(params) > 4) parse_error(tok, "functions cannot have more than 4 parameters")
|
|
node.arity = length(params)
|
|
|
|
if (tok.kind != "=>") {
|
|
parse_error(tok, "expected '=>' in arrow function")
|
|
} else {
|
|
advance()
|
|
}
|
|
|
|
_control_depth = 0
|
|
_control_type = null
|
|
_expecting_body = false
|
|
if (tok.kind == "{") {
|
|
advance()
|
|
stmts = parse_block_statements()
|
|
node.statements = stmts
|
|
if (tok.kind == "}") advance()
|
|
} else {
|
|
stmts = []
|
|
ret = ast_node("return", tok)
|
|
expr = parse_assign_expr()
|
|
ret.expression = expr
|
|
ast_node_end(ret)
|
|
push(stmts, ret)
|
|
node.statements = stmts
|
|
}
|
|
|
|
_control_depth = old_cd
|
|
_control_type = old_ct
|
|
_expecting_body = old_eb
|
|
node.function_nr = fn_counter
|
|
fn_counter = fn_counter + 1
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
parse_statement = function(unused) {
|
|
var start = tok
|
|
var node = null
|
|
var k = tok.kind
|
|
var stmts = null
|
|
var cond = null
|
|
var then_stmts = null
|
|
var else_stmts = null
|
|
var else_ifs = null
|
|
var body = null
|
|
var expr = null
|
|
var init = null
|
|
var test = null
|
|
var update = null
|
|
var left_node = null
|
|
var right_node = null
|
|
var kind_name = null
|
|
var is_def = false
|
|
var decls = null
|
|
var decl_count = 0
|
|
var var_name = null
|
|
var right_kind = null
|
|
var elif = null
|
|
var p1_tok = null
|
|
var labeled_stmt = null
|
|
var depth = 0
|
|
var saved_ct = null
|
|
var saved_cd = 0
|
|
var saved_eb = false
|
|
|
|
if (k == "{") {
|
|
if (!_expecting_body) {
|
|
parse_error(start, "bare block '{ ... }' is not a valid statement; use a function, if, while, or for instead")
|
|
advance()
|
|
depth = 1
|
|
while (tok.kind != "eof" && depth > 0) {
|
|
if (tok.kind == "{") depth = depth + 1
|
|
else if (tok.kind == "}") depth = depth - 1
|
|
if (depth > 0) advance()
|
|
}
|
|
if (tok.kind == "}") advance()
|
|
return null
|
|
}
|
|
_expecting_body = false
|
|
node = ast_node("block", start)
|
|
advance()
|
|
stmts = parse_block_statements()
|
|
node.statements = stmts
|
|
if (tok.kind == "}") advance()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "var" || k == "def") {
|
|
if (_control_depth > 0) {
|
|
parse_error(start, "'" + k + "' declarations must appear at function body level, not inside '" + _control_type + "'; move this declaration before the '" + _control_type + "' statement")
|
|
}
|
|
kind_name = k
|
|
is_def = (k == "def")
|
|
advance()
|
|
if (tok.kind != "name") {
|
|
parse_error(tok, "expected identifier after '" + kind_name + "'")
|
|
return null
|
|
}
|
|
decls = []
|
|
decl_count = 0
|
|
while (tok.kind == "name") {
|
|
node = ast_node(kind_name, start)
|
|
left_node = ast_node("name", tok)
|
|
left_node.name = tok.value
|
|
var_name = tok.value
|
|
advance()
|
|
ast_node_end(left_node)
|
|
node.left = left_node
|
|
if (tok.kind == "=") {
|
|
advance()
|
|
right_node = parse_assign_expr()
|
|
node.right = right_node
|
|
if (right_node != null && right_node.kind == "[" && right_node.right == null) {
|
|
node.pop = true
|
|
}
|
|
} else if (is_def) {
|
|
parse_error(start, "missing initializer for constant '" + var_name + "'")
|
|
} else {
|
|
parse_error(start, "'var' declarations must be initialized; use 'var " + var_name + " = null' if no value is needed")
|
|
}
|
|
ast_node_end(node)
|
|
push(decls, node)
|
|
decl_count = decl_count + 1
|
|
if (tok.kind == ",") advance()
|
|
else break
|
|
}
|
|
expect_semi()
|
|
if (decl_count == 1) {
|
|
return decls[0]
|
|
}
|
|
node = ast_node("var_list", start)
|
|
node.list = decls
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "if") {
|
|
node = ast_node("if", start)
|
|
advance()
|
|
if (tok.kind == "(") advance()
|
|
else parse_error(tok, "expected '(' before condition")
|
|
cond = parse_expr()
|
|
node.expression = cond
|
|
if (tok.kind == ")") advance()
|
|
else parse_error(tok, "expected ')' after if condition")
|
|
then_stmts = []
|
|
node.then = then_stmts
|
|
saved_ct = _control_type
|
|
saved_cd = _control_depth
|
|
_control_type = "if"
|
|
_control_depth = _control_depth + 1
|
|
_expecting_body = true
|
|
body = parse_statement()
|
|
if (body != null) push(then_stmts, body)
|
|
else_ifs = []
|
|
node.list = else_ifs
|
|
if (tok.kind == "else") {
|
|
advance()
|
|
if (tok.kind == "if") {
|
|
_control_depth = saved_cd
|
|
_control_type = saved_ct
|
|
elif = parse_statement()
|
|
if (elif != null) push(else_ifs, elif)
|
|
ast_node_end(node)
|
|
return node
|
|
} else {
|
|
else_stmts = []
|
|
node.else = else_stmts
|
|
_expecting_body = true
|
|
body = parse_statement()
|
|
if (body != null) push(else_stmts, body)
|
|
}
|
|
}
|
|
_control_depth = saved_cd
|
|
_control_type = saved_ct
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "while") {
|
|
node = ast_node("while", start)
|
|
advance()
|
|
if (tok.kind == "(") advance()
|
|
else parse_error(tok, "expected '(' before condition")
|
|
cond = parse_expr()
|
|
node.expression = cond
|
|
if (tok.kind == ")") advance()
|
|
else parse_error(tok, "expected ')' after while condition")
|
|
stmts = []
|
|
node.statements = stmts
|
|
saved_ct = _control_type
|
|
saved_cd = _control_depth
|
|
_control_type = "while"
|
|
_control_depth = _control_depth + 1
|
|
_expecting_body = true
|
|
body = parse_statement()
|
|
if (body != null) push(stmts, body)
|
|
_control_depth = saved_cd
|
|
_control_type = saved_ct
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "do") {
|
|
node = ast_node("do", start)
|
|
advance()
|
|
stmts = []
|
|
node.statements = stmts
|
|
saved_ct = _control_type
|
|
saved_cd = _control_depth
|
|
_control_type = "do"
|
|
_control_depth = _control_depth + 1
|
|
_expecting_body = true
|
|
body = parse_statement()
|
|
if (body != null) push(stmts, body)
|
|
_control_depth = saved_cd
|
|
_control_type = saved_ct
|
|
if (tok.kind == "while") advance()
|
|
else parse_error(tok, "expected 'while' after do body")
|
|
if (tok.kind == "(") advance()
|
|
else parse_error(tok, "expected '(' before condition")
|
|
cond = parse_expr()
|
|
node.expression = cond
|
|
if (tok.kind == ")") advance()
|
|
else parse_error(tok, "expected ')' after do-while condition")
|
|
expect_semi()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "for") {
|
|
node = ast_node("for", start)
|
|
advance()
|
|
if (tok.kind == "(") advance()
|
|
else parse_error(tok, "expected '(' after for")
|
|
if (tok.kind != ";") {
|
|
if (tok.kind == "var" || tok.kind == "def") {
|
|
parse_error(tok, "'" + tok.kind + "' declarations cannot appear in the for initializer; declare variables before the for loop")
|
|
init = parse_statement()
|
|
node.init = init
|
|
} else {
|
|
init = parse_expr()
|
|
node.init = init
|
|
if (tok.kind == ";") advance()
|
|
}
|
|
} else {
|
|
advance()
|
|
}
|
|
if (tok.kind != ";") {
|
|
test = parse_expr()
|
|
node.test = test
|
|
}
|
|
if (tok.kind == ";") advance()
|
|
if (tok.kind != ")") {
|
|
update = parse_expr()
|
|
node.update = update
|
|
}
|
|
if (tok.kind == ")") advance()
|
|
else parse_error(tok, "expected ')' after for clauses")
|
|
stmts = []
|
|
node.statements = stmts
|
|
saved_ct = _control_type
|
|
saved_cd = _control_depth
|
|
_control_type = "for"
|
|
_control_depth = _control_depth + 1
|
|
_expecting_body = true
|
|
body = parse_statement()
|
|
if (body != null) push(stmts, body)
|
|
_control_depth = saved_cd
|
|
_control_type = saved_ct
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "return") {
|
|
node = ast_node("return", start)
|
|
advance()
|
|
if (tok.kind != ";" && tok.kind != "}" && !got_lf) {
|
|
expr = parse_expr()
|
|
node.expression = expr
|
|
}
|
|
expect_semi()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "go") {
|
|
node = ast_node("go", start)
|
|
advance()
|
|
if (tok.kind != ";" && tok.kind != "}" && !got_lf) {
|
|
expr = parse_expr()
|
|
node.expression = expr
|
|
}
|
|
expect_semi()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "disrupt") {
|
|
node = ast_node("disrupt", start)
|
|
advance()
|
|
expect_semi()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "break") {
|
|
node = ast_node("break", start)
|
|
advance()
|
|
if (tok.kind == "name" && !got_lf) {
|
|
node.name = tok.value
|
|
advance()
|
|
}
|
|
expect_semi()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "continue") {
|
|
node = ast_node("continue", start)
|
|
advance()
|
|
if (tok.kind == "name" && !got_lf) {
|
|
node.name = tok.value
|
|
advance()
|
|
}
|
|
expect_semi()
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
|
|
if (k == "function") {
|
|
return parse_function_inner()
|
|
}
|
|
|
|
if (k == ";") {
|
|
advance()
|
|
return null
|
|
}
|
|
|
|
if (k == "name") {
|
|
if (tok.value == "try" || tok.value == "catch" || tok.value == "finally") {
|
|
parse_error(start, "'" + tok.value + "' is not supported; use disrupt/disruption instead")
|
|
sync_to_statement()
|
|
return null
|
|
}
|
|
if (tok.value == "throw") {
|
|
parse_error(start, "'throw' is not supported; use disrupt instead")
|
|
sync_to_statement()
|
|
return null
|
|
}
|
|
if (tok.value == "class") {
|
|
parse_error(start, "'class' is not supported; use meme()/proto() instead")
|
|
sync_to_statement()
|
|
return null
|
|
}
|
|
if (tok.value == "new") {
|
|
parse_error(start, "'new' is not supported; use meme()/proto() instead")
|
|
sync_to_statement()
|
|
return null
|
|
}
|
|
if (tok.value == "switch" || tok.value == "case") {
|
|
parse_error(start, "'" + tok.value + "' is not supported; use if/else instead")
|
|
sync_to_statement()
|
|
return null
|
|
}
|
|
if (tok.value == "let" || tok.value == "const") {
|
|
parse_error(start, "'" + tok.value + "' is not supported; use var/def instead")
|
|
sync_to_statement()
|
|
return null
|
|
}
|
|
p1_tok = peek_ahead(1)
|
|
if (p1_tok.kind == ":") {
|
|
node = ast_node("label", start)
|
|
node.name = tok.value
|
|
advance() // skip identifier
|
|
advance() // skip colon
|
|
labeled_stmt = parse_statement()
|
|
node.statement = labeled_stmt
|
|
ast_node_end(node)
|
|
return node
|
|
}
|
|
}
|
|
|
|
expr = parse_expr()
|
|
if (expr != null) {
|
|
node = ast_node("call", start)
|
|
node.expression = expr
|
|
ast_node_end(node)
|
|
expect_semi()
|
|
return node
|
|
}
|
|
parse_error(start, "unexpected token at start of statement")
|
|
return null
|
|
}
|
|
|
|
// ============================================================
|
|
// Program
|
|
// ============================================================
|
|
|
|
var parse_program = function() {
|
|
var root = {kind: "program", filename: filename}
|
|
var functions = []
|
|
var statements = []
|
|
var before = 0
|
|
var stmt = null
|
|
root.functions = functions
|
|
root.statements = statements
|
|
|
|
while (tok.kind != "eof") {
|
|
before = cursor
|
|
stmt = parse_statement()
|
|
if (stmt != null) {
|
|
if (stmt.kind == "function") {
|
|
push(functions, stmt)
|
|
} else {
|
|
push(statements, stmt)
|
|
}
|
|
} else if (cursor == before) {
|
|
sync_to_statement()
|
|
}
|
|
}
|
|
return root
|
|
}
|
|
|
|
// ============================================================
|
|
// Semantic Analysis
|
|
// ============================================================
|
|
|
|
var sem_errors = []
|
|
var scopes_array = []
|
|
var intrinsics = []
|
|
|
|
var sem_error = function(node, msg) {
|
|
var err = {message: msg}
|
|
if (node.from_row != null) err.line = node.from_row + 1
|
|
if (node.from_column != null) err.column = node.from_column + 1
|
|
push(sem_errors, err)
|
|
}
|
|
|
|
var make_scope = function(parent, fn_nr, opts) {
|
|
return {
|
|
parent: parent,
|
|
vars: [],
|
|
in_loop: opts.in_loop == true,
|
|
function_nr: fn_nr,
|
|
is_function_scope: opts.is_func == true,
|
|
func_node: null,
|
|
has_inner_func: false
|
|
}
|
|
}
|
|
|
|
var sem_add_var = function(scope, name, make_opts) {
|
|
push(scope.vars, {
|
|
name: name,
|
|
is_const: make_opts.is_const == true,
|
|
make: make_opts.make,
|
|
function_nr: make_opts.fn_nr,
|
|
nr_uses: 0,
|
|
closure: 0
|
|
})
|
|
}
|
|
|
|
var sem_lookup_var = function(scope, name) {
|
|
var result = {v: null, level: 0, def_function_nr: -1}
|
|
var cur_fn = scope.function_nr
|
|
var s = scope
|
|
var i = 0
|
|
while (s != null) {
|
|
i = 0
|
|
while (i < length(s.vars)) {
|
|
if (s.vars[i].name == name) {
|
|
result.v = s.vars[i]
|
|
result.def_function_nr = s.vars[i].function_nr
|
|
return result
|
|
}
|
|
i = i + 1
|
|
}
|
|
if (s.parent != null && s.parent.function_nr != cur_fn) {
|
|
result.level = result.level + 1
|
|
cur_fn = s.parent.function_nr
|
|
}
|
|
s = s.parent
|
|
}
|
|
return result
|
|
}
|
|
|
|
var sem_find_var = function(scope, name) {
|
|
var r = sem_lookup_var(scope, name)
|
|
return r.v
|
|
}
|
|
|
|
var sem_in_loop = function(scope) {
|
|
var s = scope
|
|
while (s != null) {
|
|
if (s.in_loop) return true
|
|
s = s.parent
|
|
}
|
|
return false
|
|
}
|
|
|
|
var sem_find_func_scope = function(scope) {
|
|
var s = scope
|
|
while (s != null) {
|
|
if (s.is_function_scope) return s
|
|
s = s.parent
|
|
}
|
|
return null
|
|
}
|
|
|
|
var sem_add_intrinsic = function(name) {
|
|
if (find(intrinsics, name) == null) push(intrinsics, name)
|
|
}
|
|
|
|
var functino_names = {
|
|
"+!": true, "-!": true, "*!": true, "/!": true, "%!": true, "**!": true,
|
|
"<!": true, ">!": true, "<=!": true, ">=!": true, "=!": true, "!=!": true,
|
|
"&!": true, "|!": true, "^!": true, "<<!": true, ">>!": true, ">>>!": true,
|
|
"&&!": true, "||!": true, "~!": true, "[]!": true
|
|
}
|
|
|
|
var is_functino_name = function(name) {
|
|
return functino_names[name] == true
|
|
}
|
|
|
|
var derive_type_tag = function(expr) {
|
|
if (expr == null) return null
|
|
var k = expr.kind
|
|
if (k == "array") return "array"
|
|
if (k == "record") return "record"
|
|
if (k == "function") return "function"
|
|
if (k == "text" || k == "text literal") return "text"
|
|
if (k == "number") {
|
|
if (is_integer(expr.number)) return "integer"
|
|
return "number"
|
|
}
|
|
if (k == "true" || k == "false") return "logical"
|
|
if (k == "null") return "null"
|
|
return null
|
|
}
|
|
|
|
var _assign_kinds = {
|
|
assign: true, "+=": true, "-=": true, "*=": true, "/=": true, "%=": true,
|
|
"<<=": true, ">>=": true, ">>>=": true,
|
|
"&=": true, "^=": true, "|=": true, "**=": true,
|
|
"&&=": true, "||=": true
|
|
}
|
|
|
|
var sem_propagate_vars = function(parent, child) {
|
|
parent.vars = array(parent.vars, child.vars)
|
|
}
|
|
|
|
var sem_build_scope_record = function(scope) {
|
|
var rec = {function_nr: scope.function_nr}
|
|
var slots = 0
|
|
var close_slots = 0
|
|
var i = 0
|
|
var v = null
|
|
while (i < length(scope.vars)) {
|
|
v = scope.vars[i]
|
|
rec[v.name] = {
|
|
make: v.make,
|
|
function_nr: v.function_nr,
|
|
nr_uses: v.nr_uses,
|
|
closure: v.closure == 1,
|
|
level: 0,
|
|
type_tag: v.type_tag
|
|
}
|
|
slots = slots + 1
|
|
if (v.closure) close_slots = close_slots + 1
|
|
i = i + 1
|
|
}
|
|
return {rec: rec, nr_slots: slots, nr_close: close_slots}
|
|
}
|
|
|
|
// Forward declarations
|
|
var sem_check_expr = null
|
|
var sem_check_stmt = null
|
|
|
|
var sem_predeclare_vars = function(scope, stmts) {
|
|
var i = 0
|
|
var stmt = null
|
|
var kind = null
|
|
var name = null
|
|
var item = null
|
|
var ik = null
|
|
var j = 0
|
|
while (i < length(stmts)) {
|
|
stmt = stmts[i]
|
|
kind = stmt.kind
|
|
if (kind == "function") {
|
|
name = stmt.name
|
|
if (name != null && sem_find_var(scope, name) == null) {
|
|
sem_add_var(scope, name, {make: "function", fn_nr: scope.function_nr})
|
|
}
|
|
} else if (kind == "var") {
|
|
name = stmt.left.name
|
|
if (name != null && sem_find_var(scope, name) == null) {
|
|
sem_add_var(scope, name, {make: "var", fn_nr: scope.function_nr})
|
|
}
|
|
} else if (kind == "var_list") {
|
|
j = 0
|
|
while (j < length(stmt.list)) {
|
|
item = stmt.list[j]
|
|
ik = item.kind
|
|
if (ik == "var") {
|
|
name = item.left.name
|
|
if (name != null && sem_find_var(scope, name) == null) {
|
|
sem_add_var(scope, name, {make: "var", fn_nr: scope.function_nr})
|
|
}
|
|
}
|
|
j = j + 1
|
|
}
|
|
}
|
|
i = i + 1
|
|
}
|
|
}
|
|
|
|
var sem_check_assign_target = function(scope, left_node) {
|
|
if (left_node == null) return null
|
|
var kind = left_node.kind
|
|
var name = null
|
|
var v = null
|
|
var r = null
|
|
var obj_expr = null
|
|
|
|
if (kind == "name") {
|
|
name = left_node.name
|
|
if (name == null) return null
|
|
v = sem_find_var(scope, name)
|
|
if (v == null) {
|
|
sem_error(left_node, "cannot assign to unbound variable '" + name + "'")
|
|
} else if (v.is_const) {
|
|
sem_error(left_node, "cannot assign to constant '" + name + "'")
|
|
}
|
|
r = sem_lookup_var(scope, name)
|
|
if (r.v != null) {
|
|
left_node.level = r.level
|
|
left_node.function_nr = r.def_function_nr
|
|
} else {
|
|
left_node.level = -1
|
|
}
|
|
} else if (kind == "." || kind == "[") {
|
|
obj_expr = left_node.left
|
|
sem_check_expr(scope, obj_expr)
|
|
if (kind == "[" && left_node.right != null) {
|
|
sem_check_expr(scope, left_node.right)
|
|
}
|
|
// Type error detection for known-type constant objects
|
|
if (obj_expr != null && obj_expr.kind == "name" && obj_expr.name != null) {
|
|
v = sem_find_var(scope, obj_expr.name)
|
|
if (v != null && v.is_const && v.type_tag != null) {
|
|
if (kind == ".") {
|
|
if (v.type_tag == "array") {
|
|
sem_error(left_node, "cannot set property on array '" + obj_expr.name + "'")
|
|
}
|
|
} else if (kind == "[") {
|
|
if (left_node.right == null) {
|
|
// Push: a[] = val
|
|
if (v.type_tag != "array") {
|
|
sem_error(left_node, "push only works on arrays, not " + v.type_tag + " '" + obj_expr.name + "'")
|
|
}
|
|
} else if (v.type_tag == "array") {
|
|
if (left_node.right.kind == "text") {
|
|
sem_error(left_node, "cannot use text key on array '" + obj_expr.name + "'")
|
|
}
|
|
} else if (v.type_tag == "record") {
|
|
if (left_node.right.kind == "number" && is_integer(left_node.right.number)) {
|
|
sem_error(left_node, "cannot use integer key on record '" + obj_expr.name + "'; use text key")
|
|
}
|
|
}
|
|
}
|
|
} else if (v != null && v.is_const && v.type_tag == null) {
|
|
// Infer type_tag from usage pattern (def only)
|
|
if (kind == ".") {
|
|
v.type_tag = "record"
|
|
} else if (kind == "[") {
|
|
if (left_node.right == null) {
|
|
// Push: a[] = val → array
|
|
v.type_tag = "array"
|
|
} else if (left_node.right.kind == "number" && is_integer(left_node.right.number)) {
|
|
v.type_tag = "array"
|
|
} else if (left_node.right.kind == "text") {
|
|
v.type_tag = "record"
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
sem_check_expr = function(scope, expr) {
|
|
if (expr == null) return null
|
|
var kind = expr.kind
|
|
if (kind == null) return null
|
|
var name = null
|
|
var r = null
|
|
var i = 0
|
|
var operand = null
|
|
var v = null
|
|
var prop = null
|
|
var val = null
|
|
var fn_nr_val = null
|
|
var fn_scope = null
|
|
var pname = null
|
|
var def_val = null
|
|
var sr = null
|
|
var enclosing = null
|
|
|
|
if (_assign_kinds[kind] == true) {
|
|
sem_check_assign_target(scope, expr.left)
|
|
sem_check_expr(scope, expr.right)
|
|
return null
|
|
}
|
|
|
|
if (kind == "++" || kind == "--") {
|
|
operand = expr.expression
|
|
if (operand != null && operand.kind == "name") {
|
|
name = operand.name
|
|
if (name != null) {
|
|
v = sem_find_var(scope, name)
|
|
if (v == null) {
|
|
sem_error(expr, "cannot assign to unbound variable '" + name + "'")
|
|
} else if (v.is_const) {
|
|
sem_error(expr, "cannot assign to constant '" + name + "'")
|
|
}
|
|
r = sem_lookup_var(scope, name)
|
|
if (r.v != null) {
|
|
operand.level = r.level
|
|
operand.function_nr = r.def_function_nr
|
|
} else {
|
|
operand.level = -1
|
|
}
|
|
}
|
|
} else if (operand != null) {
|
|
sem_check_assign_target(scope, operand)
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "this") {
|
|
if (scope.function_nr == 0) {
|
|
sem_error(expr, "'this' cannot be used at the top level of a program")
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "[") {
|
|
sem_check_expr(scope, expr.left)
|
|
sem_check_expr(scope, expr.right)
|
|
if (expr.right != null) {
|
|
if (expr.right.kind == "number" && is_integer(expr.right.number)) {
|
|
expr.access_kind = "index"
|
|
} else if (expr.right.kind == "text") {
|
|
expr.access_kind = "field"
|
|
}
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "," || kind == "+" || kind == "-" || kind == "*" ||
|
|
kind == "/" || kind == "%" || kind == "==" || kind == "!=" ||
|
|
kind == "<" || kind == ">" || kind == "<=" || kind == ">=" ||
|
|
kind == "&&" || kind == "||" || kind == "&" ||
|
|
kind == "|" || kind == "^" || kind == "<<" || kind == ">>" ||
|
|
kind == ">>>" || kind == "**" || kind == "in" ||
|
|
kind == ".") {
|
|
sem_check_expr(scope, expr.left)
|
|
sem_check_expr(scope, expr.right)
|
|
return null
|
|
}
|
|
|
|
if (kind == "then") {
|
|
sem_check_expr(scope, expr.expression)
|
|
sem_check_expr(scope, expr.then)
|
|
sem_check_expr(scope, expr.else)
|
|
return null
|
|
}
|
|
|
|
if (kind == "(") {
|
|
sem_check_expr(scope, expr.expression)
|
|
i = 0
|
|
while (i < length(expr.list)) {
|
|
sem_check_expr(scope, expr.list[i])
|
|
i = i + 1
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "!" || kind == "~" || kind == "delete" ||
|
|
kind == "-unary" || kind == "+unary") {
|
|
sem_check_expr(scope, expr.expression)
|
|
return null
|
|
}
|
|
|
|
if (kind == "array") {
|
|
i = 0
|
|
while (i < length(expr.list)) {
|
|
sem_check_expr(scope, expr.list[i])
|
|
i = i + 1
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "record") {
|
|
i = 0
|
|
while (i < length(expr.list)) {
|
|
prop = expr.list[i]
|
|
val = prop.right
|
|
sem_check_expr(scope, val)
|
|
i = i + 1
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "function") {
|
|
enclosing = sem_find_func_scope(scope)
|
|
if (enclosing != null) enclosing.has_inner_func = true
|
|
fn_nr_val = expr.function_nr
|
|
if (fn_nr_val == null) fn_nr_val = scope.function_nr
|
|
fn_scope = make_scope(scope, fn_nr_val, {is_func: true})
|
|
fn_scope.func_node = expr
|
|
expr.outer = scope.function_nr
|
|
i = 0
|
|
while (i < length(expr.list)) {
|
|
pname = expr.list[i].name
|
|
if (pname != null) sem_add_var(fn_scope, pname, {is_const: true, make: "input", fn_nr: fn_nr_val})
|
|
def_val = expr.list[i].expression
|
|
if (def_val != null) sem_check_expr(fn_scope, def_val)
|
|
i = i + 1
|
|
}
|
|
if (expr.statements != null) {
|
|
sem_predeclare_vars(fn_scope, expr.statements)
|
|
i = 0
|
|
while (i < length(expr.statements)) {
|
|
sem_check_stmt(fn_scope, expr.statements[i])
|
|
i = i + 1
|
|
}
|
|
}
|
|
if (expr.disruption != null) {
|
|
i = 0
|
|
while (i < length(expr.disruption)) {
|
|
sem_check_stmt(fn_scope, expr.disruption[i])
|
|
i = i + 1
|
|
}
|
|
}
|
|
sr = sem_build_scope_record(fn_scope)
|
|
push(scopes_array, sr.rec)
|
|
expr.nr_slots = sr.nr_slots
|
|
expr.nr_close_slots = sr.nr_close
|
|
return null
|
|
}
|
|
|
|
if (kind == "text literal") {
|
|
i = 0
|
|
while (i < length(expr.list)) {
|
|
sem_check_expr(scope, expr.list[i])
|
|
i = i + 1
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "name") {
|
|
name = expr.name
|
|
if (name != null) {
|
|
if (is_functino_name(name)) {
|
|
expr.make = "functino"
|
|
expr.level = -1
|
|
return null
|
|
}
|
|
r = sem_lookup_var(scope, name)
|
|
if (r.v != null) {
|
|
expr.level = r.level
|
|
expr.function_nr = r.def_function_nr
|
|
r.v.nr_uses = r.v.nr_uses + 1
|
|
if (r.level > 0) r.v.closure = 1
|
|
} else {
|
|
expr.level = -1
|
|
expr.intrinsic = true
|
|
sem_add_intrinsic(name)
|
|
}
|
|
}
|
|
return null
|
|
}
|
|
}
|
|
|
|
sem_check_stmt = function(scope, stmt) {
|
|
if (stmt == null) return null
|
|
var kind = stmt.kind
|
|
if (kind == null) return null
|
|
var name = null
|
|
var existing = null
|
|
var i = 0
|
|
var loop_scope = null
|
|
var do_scope = null
|
|
var for_scope = null
|
|
var init_kind = null
|
|
var fn_nr_val = null
|
|
var fn_scope = null
|
|
var pname = null
|
|
var def_val = null
|
|
var sr = null
|
|
var enclosing = null
|
|
var func_scope = null
|
|
var tt = null
|
|
|
|
if (kind == "var_list") {
|
|
i = 0
|
|
while (i < length(stmt.list)) {
|
|
sem_check_stmt(scope, stmt.list[i])
|
|
i = i + 1
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "var") {
|
|
name = stmt.left.name
|
|
if (name != null) {
|
|
existing = sem_find_var(scope, name)
|
|
if (existing != null && existing.is_const) {
|
|
sem_error(stmt.left, "cannot redeclare constant '" + name + "'")
|
|
}
|
|
if (existing == null || existing.function_nr != scope.function_nr) {
|
|
sem_add_var(scope, name, {make: "var", fn_nr: scope.function_nr})
|
|
}
|
|
}
|
|
sem_check_expr(scope, stmt.right)
|
|
return null
|
|
}
|
|
|
|
if (kind == "def") {
|
|
name = stmt.left.name
|
|
if (name != null) {
|
|
existing = sem_find_var(scope, name)
|
|
if (existing != null && existing.is_const) {
|
|
sem_error(stmt.left, "cannot redeclare constant '" + name + "'")
|
|
} else if (existing != null && !existing.is_const && existing.function_nr == scope.function_nr) {
|
|
existing.is_const = 1
|
|
existing.make = "def"
|
|
} else {
|
|
sem_add_var(scope, name, {is_const: true, make: "def", fn_nr: scope.function_nr})
|
|
}
|
|
}
|
|
sem_check_expr(scope, stmt.right)
|
|
if (name != null) {
|
|
tt = derive_type_tag(stmt.right)
|
|
if (tt != null && tt != "null") {
|
|
existing = sem_find_var(scope, name)
|
|
if (existing != null) existing.type_tag = tt
|
|
}
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "call") {
|
|
sem_check_expr(scope, stmt.expression)
|
|
return null
|
|
}
|
|
|
|
if (kind == "if") {
|
|
sem_check_expr(scope, stmt.expression)
|
|
i = 0
|
|
while (i < length(stmt.then)) {
|
|
sem_check_stmt(scope, stmt.then[i])
|
|
i = i + 1
|
|
}
|
|
i = 0
|
|
while (i < length(stmt.list)) {
|
|
sem_check_stmt(scope, stmt.list[i])
|
|
i = i + 1
|
|
}
|
|
if (stmt.else != null) {
|
|
i = 0
|
|
while (i < length(stmt.else)) {
|
|
sem_check_stmt(scope, stmt.else[i])
|
|
i = i + 1
|
|
}
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "while") {
|
|
sem_check_expr(scope, stmt.expression)
|
|
loop_scope = make_scope(scope, scope.function_nr, {in_loop: true})
|
|
i = 0
|
|
while (i < length(stmt.statements)) {
|
|
sem_check_stmt(loop_scope, stmt.statements[i])
|
|
i = i + 1
|
|
}
|
|
sem_propagate_vars(scope, loop_scope)
|
|
return null
|
|
}
|
|
|
|
if (kind == "do") {
|
|
do_scope = make_scope(scope, scope.function_nr, {in_loop: true})
|
|
i = 0
|
|
while (i < length(stmt.statements)) {
|
|
sem_check_stmt(do_scope, stmt.statements[i])
|
|
i = i + 1
|
|
}
|
|
sem_propagate_vars(scope, do_scope)
|
|
sem_check_expr(scope, stmt.expression)
|
|
return null
|
|
}
|
|
|
|
if (kind == "for") {
|
|
for_scope = make_scope(scope, scope.function_nr, {in_loop: true})
|
|
if (stmt.init != null) {
|
|
init_kind = stmt.init.kind
|
|
if (init_kind == "var" || init_kind == "def") {
|
|
sem_check_stmt(for_scope, stmt.init)
|
|
} else {
|
|
sem_check_expr(for_scope, stmt.init)
|
|
}
|
|
}
|
|
sem_check_expr(for_scope, stmt.test)
|
|
sem_check_expr(for_scope, stmt.update)
|
|
i = 0
|
|
while (i < length(stmt.statements)) {
|
|
sem_check_stmt(for_scope, stmt.statements[i])
|
|
i = i + 1
|
|
}
|
|
sem_propagate_vars(scope, for_scope)
|
|
return null
|
|
}
|
|
|
|
if (kind == "go") {
|
|
sem_check_expr(scope, stmt.expression)
|
|
if (stmt.expression == null || stmt.expression.kind != "(") {
|
|
sem_error(stmt, "'go' must be followed by a function call")
|
|
} else {
|
|
func_scope = sem_find_func_scope(scope)
|
|
if (func_scope != null && func_scope.func_node != null) {
|
|
if (func_scope.func_node.disruption != null) {
|
|
sem_error(stmt, "cannot use 'go' in a function with a disruption clause")
|
|
}
|
|
if (func_scope.has_inner_func) {
|
|
sem_error(stmt, "cannot use 'go' in a function that defines inner functions")
|
|
}
|
|
}
|
|
stmt.tail = true
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "return") {
|
|
sem_check_expr(scope, stmt.expression)
|
|
if (stmt.expression != null && stmt.expression.kind == "(") {
|
|
stmt.tail = true
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "disrupt") {
|
|
return null
|
|
}
|
|
|
|
if (kind == "break") {
|
|
if (!sem_in_loop(scope)) {
|
|
sem_error(stmt, "'break' used outside of loop")
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "continue") {
|
|
if (!sem_in_loop(scope)) {
|
|
sem_error(stmt, "'continue' used outside of loop")
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "block") {
|
|
i = 0
|
|
while (i < length(stmt.statements)) {
|
|
sem_check_stmt(scope, stmt.statements[i])
|
|
i = i + 1
|
|
}
|
|
return null
|
|
}
|
|
|
|
if (kind == "label") {
|
|
sem_check_stmt(scope, stmt.statement)
|
|
return null
|
|
}
|
|
|
|
if (kind == "function") {
|
|
enclosing = sem_find_func_scope(scope)
|
|
if (enclosing != null) enclosing.has_inner_func = true
|
|
name = stmt.name
|
|
if (name != null && sem_find_var(scope, name) == null) sem_add_var(scope, name, {make: "function", fn_nr: scope.function_nr})
|
|
fn_nr_val = stmt.function_nr
|
|
if (fn_nr_val == null) fn_nr_val = scope.function_nr
|
|
fn_scope = make_scope(scope, fn_nr_val, {is_func: true})
|
|
fn_scope.func_node = stmt
|
|
stmt.outer = scope.function_nr
|
|
i = 0
|
|
while (i < length(stmt.list)) {
|
|
pname = stmt.list[i].name
|
|
if (pname != null) sem_add_var(fn_scope, pname, {is_const: true, make: "input", fn_nr: fn_nr_val})
|
|
def_val = stmt.list[i].expression
|
|
if (def_val != null) sem_check_expr(fn_scope, def_val)
|
|
i = i + 1
|
|
}
|
|
sem_predeclare_vars(fn_scope, stmt.statements)
|
|
i = 0
|
|
while (i < length(stmt.statements)) {
|
|
sem_check_stmt(fn_scope, stmt.statements[i])
|
|
i = i + 1
|
|
}
|
|
if (stmt.disruption != null) {
|
|
i = 0
|
|
while (i < length(stmt.disruption)) {
|
|
sem_check_stmt(fn_scope, stmt.disruption[i])
|
|
i = i + 1
|
|
}
|
|
}
|
|
sr = sem_build_scope_record(fn_scope)
|
|
push(scopes_array, sr.rec)
|
|
stmt.nr_slots = sr.nr_slots
|
|
stmt.nr_close_slots = sr.nr_close
|
|
return null
|
|
}
|
|
}
|
|
|
|
var semantic_check = function(ast) {
|
|
var global_scope = make_scope(null, 0, {is_func: true})
|
|
var i = 0
|
|
var stmt = null
|
|
var name = null
|
|
var sr = null
|
|
var new_scopes = null
|
|
|
|
i = 0
|
|
while (i < length(ast.functions)) {
|
|
name = ast.functions[i].name
|
|
if (name != null) sem_add_var(global_scope, name, {make: "function", fn_nr: 0})
|
|
i = i + 1
|
|
}
|
|
|
|
i = 0
|
|
while (i < length(ast.statements)) {
|
|
sem_check_stmt(global_scope, ast.statements[i])
|
|
i = i + 1
|
|
}
|
|
|
|
i = 0
|
|
while (i < length(ast.functions)) {
|
|
sem_check_stmt(global_scope, ast.functions[i])
|
|
i = i + 1
|
|
}
|
|
|
|
sr = sem_build_scope_record(global_scope)
|
|
new_scopes = [sr.rec]
|
|
i = 0
|
|
while (i < length(scopes_array)) {
|
|
push(new_scopes, scopes_array[i])
|
|
i = i + 1
|
|
}
|
|
scopes_array = new_scopes
|
|
|
|
ast.scopes = scopes_array
|
|
ast.intrinsics = intrinsics
|
|
if (length(sem_errors) > 0) {
|
|
ast.errors = sem_errors
|
|
}
|
|
}
|
|
|
|
// ============================================================
|
|
// Main
|
|
// ============================================================
|
|
|
|
init_cursor()
|
|
var ast = parse_program()
|
|
|
|
if (error_count == 0) {
|
|
semantic_check(ast)
|
|
}
|
|
|
|
// Merge parse errors
|
|
var _mi = 0
|
|
if (length(errors) > 0) {
|
|
if (ast.errors != null) {
|
|
_mi = 0
|
|
while (_mi < length(errors)) {
|
|
push(ast.errors, errors[_mi])
|
|
_mi = _mi + 1
|
|
}
|
|
} else {
|
|
ast.errors = errors
|
|
}
|
|
}
|
|
|
|
return ast
|
|
}
|
|
|
|
return parse
|