|
| 1 | +// SPDX-License-Identifier: MPL-2.0 |
| 2 | +// Cst.affine — Concrete Syntax Tree for Error-Lang (ported from compiler/src/Cst.res). |
| 3 | +// |
| 4 | +// A CST preserves all source text (whitespace, comments, exact tokens) so source |
| 5 | +// round-trips. Ported to AffineScript's functional style (no mut struct fields / |
| 6 | +// record spread): stateful classify_trivia loops a `let mut` local; lex_with_trivia |
| 7 | +// folds raw tokens carrying the last token in hand (so newline-trailing-trivia and |
| 8 | +// trailing source are attached without array-index mutation). |
| 9 | +// |
| 10 | +// Selective `Types` import: Cst's node kinds (MainBlock/FunctionDecl/LetStmt/...) |
| 11 | +// would collide with Types' Decl/Stmt constructors under a glob import, so only the |
| 12 | +// token/location types (+ the Newline constructor for matching) are imported. |
| 13 | +// |
| 14 | +// Omitted vs Cst.res (documented): |
| 15 | +// - node_at: returning a deepest *subtree* needs a node used both to recurse into |
| 16 | +// and to return — not expressible under affine ownership without a shared/clone |
| 17 | +// type. Auxiliary (IDE cursor lookup); deferred. |
| 18 | +// - run_tests: the test harness (Console-based) is not part of the compiler. |
| 19 | + |
| 20 | +module Cst; |
| 21 | + |
| 22 | +use prelude::*; |
| 23 | +use string::{length, substring, join, char_at}; |
| 24 | +use collections::{flat_map}; |
| 25 | +use Types::{TokenType, Position, Location, Token, Newline}; |
| 26 | +use Lexer::{lex}; |
| 27 | + |
| 28 | +// ---- trivia ---- |
| 29 | + |
| 30 | +enum TriviaKind { TkWhitespace, TkLineComment, TkNewline } |
| 31 | + |
| 32 | +struct Trivia { |
| 33 | + kind: TriviaKind, |
| 34 | + text: String, |
| 35 | + loc: Location |
| 36 | +} |
| 37 | + |
| 38 | +struct CstToken { |
| 39 | + tokenKind: TokenType, |
| 40 | + text: String, |
| 41 | + leadingTrivia: [Trivia], |
| 42 | + trailingTrivia: [Trivia], |
| 43 | + loc: Location |
| 44 | +} |
| 45 | + |
| 46 | +enum CstNodeKind { |
| 47 | + SourceFile, MainBlock, FunctionDecl, StructDecl, LetStmt, IfStmt, WhileStmt, |
| 48 | + ForStmt, ReturnStmt, BreakStmt, ContinueStmt, PrintStmt, GutterBlock, ExprStmt, |
| 49 | + BinaryExpr, UnaryExpr, CallExpr, IndexExpr, MemberExpr, TernaryExpr, LambdaExpr, |
| 50 | + ArrayLitExpr, ParamList, ArgList, ErrorNode |
| 51 | +} |
| 52 | + |
| 53 | +enum CstNode { |
| 54 | + NodeToken(CstToken), |
| 55 | + NodeTree(CstTree) |
| 56 | +} |
| 57 | + |
| 58 | +struct CstTree { |
| 59 | + kind: CstNodeKind, |
| 60 | + children: [CstNode], |
| 61 | + loc: Location |
| 62 | +} |
| 63 | + |
| 64 | +// ---- location helpers ---- |
| 65 | + |
| 66 | +fn mk_pos(line: Int, column: Int, offset: Int) -> Position { |
| 67 | + #{ line: line, column: column, offset: offset } |
| 68 | +} |
| 69 | + |
| 70 | +fn mk_loc(start_off: Int, end_off: Int, file: String) -> Location { |
| 71 | + #{ start: mk_pos(0, 0, start_off), end_: mk_pos(0, 0, end_off), file: file } |
| 72 | +} |
| 73 | + |
| 74 | +fn code_at_str(s: String, i: Int) -> Int { |
| 75 | + match char_at(s, i) { |
| 76 | + Some(c) => char_to_int(c), |
| 77 | + None => -1 |
| 78 | + } |
| 79 | +} |
| 80 | + |
| 81 | +fn is_ws_code(c: Int) -> Bool { c == 32 || c == 9 || c == 13 } |
| 82 | + |
| 83 | +// ---- source reconstruction (round-trip) ---- |
| 84 | + |
| 85 | +fn trivia_text(t: Trivia) -> String { t.text } |
| 86 | + |
| 87 | +pub fn to_source(node: CstNode) -> String { |
| 88 | + match node { |
| 89 | + NodeToken(tok) => { |
| 90 | + let leading = join(map(tok.leadingTrivia, trivia_text), ""); |
| 91 | + let trailing = join(map(tok.trailingTrivia, trivia_text), ""); |
| 92 | + leading ++ tok.text ++ trailing |
| 93 | + }, |
| 94 | + NodeTree(tree) => tree_to_source(tree) |
| 95 | + } |
| 96 | +} |
| 97 | + |
| 98 | +pub fn tree_to_source(tree: CstTree) -> String { |
| 99 | + join(map(tree.children, to_source), "") |
| 100 | +} |
| 101 | + |
| 102 | +// ---- token collection (document order) ---- |
| 103 | + |
| 104 | +fn node_tokens(node: CstNode) -> [CstToken] { |
| 105 | + match node { |
| 106 | + NodeToken(tok) => [tok], |
| 107 | + NodeTree(subtree) => tokens(subtree) |
| 108 | + } |
| 109 | +} |
| 110 | + |
| 111 | +pub fn tokens(tree: CstTree) -> [CstToken] { |
| 112 | + flat_map(node_tokens, tree.children) |
| 113 | +} |
| 114 | + |
| 115 | +// ---- trivia classification (gap of source -> trivia items) ---- |
| 116 | + |
| 117 | +pub fn classify_trivia(gap: String, file: String, base_offset: Int) -> [Trivia] { |
| 118 | + let mut result = []; |
| 119 | + let mut i = 0; |
| 120 | + let glen = length(gap); |
| 121 | + while i < glen { |
| 122 | + let c = code_at_str(gap, i); |
| 123 | + if c == 35 { // '#': line comment to end of line |
| 124 | + let start = i; |
| 125 | + let mut j = i; |
| 126 | + while (j < glen) && (code_at_str(gap, j) != 10) { j = j + 1; } |
| 127 | + let loc = mk_loc(base_offset + start, base_offset + j, file); |
| 128 | + result = result ++ [#{ kind: TkLineComment, text: substring(gap, start, j), loc: loc }]; |
| 129 | + i = j; |
| 130 | + } else if c == 10 { // newline |
| 131 | + let loc = mk_loc(base_offset + i, base_offset + i + 1, file); |
| 132 | + result = result ++ [#{ kind: TkNewline, text: "\n", loc: loc }]; |
| 133 | + i = i + 1; |
| 134 | + } else if is_ws_code(c) { // contiguous whitespace |
| 135 | + let start = i; |
| 136 | + let mut j = i; |
| 137 | + while (j < glen) && is_ws_code(code_at_str(gap, j)) { j = j + 1; } |
| 138 | + let loc = mk_loc(base_offset + start, base_offset + j, file); |
| 139 | + result = result ++ [#{ kind: TkWhitespace, text: substring(gap, start, j), loc: loc }]; |
| 140 | + i = j; |
| 141 | + } else { // anything else (shouldn't occur in gaps) |
| 142 | + let loc = mk_loc(base_offset + i, base_offset + i + 1, file); |
| 143 | + result = result ++ [#{ kind: TkWhitespace, text: substring(gap, i, i + 1), loc: loc }]; |
| 144 | + i = i + 1; |
| 145 | + } |
| 146 | + } |
| 147 | + result |
| 148 | +} |
| 149 | + |
| 150 | +// ---- trivia-aware lexing ---- |
| 151 | + |
| 152 | +struct LexAcc { |
| 153 | + committed: [CstToken], |
| 154 | + last: Option<CstToken>, |
| 155 | + prev_end: Int |
| 156 | +} |
| 157 | + |
| 158 | +fn set_trailing(t: CstToken, trailing: [Trivia]) -> CstToken { |
| 159 | + #{ tokenKind: t.tokenKind, text: t.text, leadingTrivia: t.leadingTrivia, |
| 160 | + trailingTrivia: trailing, loc: t.loc } |
| 161 | +} |
| 162 | + |
| 163 | +pub fn lex_with_trivia(source: String, file: String, run_number: Int) -> [CstToken] { |
| 164 | + let (raw_tokens, _diags) = lex(source, file, run_number); |
| 165 | + |
| 166 | + let final = fold(raw_tokens, #{ committed: [], last: None, prev_end: 0 }, |acc, tok| { |
| 167 | + let tok_start = tok.loc.start.offset; |
| 168 | + let tok_end = tok.loc.end_.offset; |
| 169 | + let leading = if tok_start > acc.prev_end { |
| 170 | + classify_trivia(substring(source, acc.prev_end, tok_start), file, acc.prev_end) |
| 171 | + } else { |
| 172 | + [] |
| 173 | + }; |
| 174 | + match tok.type_ { |
| 175 | + Newline => { |
| 176 | + let nl = #{ kind: TkNewline, text: tok.lexeme, loc: tok.loc }; |
| 177 | + match acc.last { |
| 178 | + None => { |
| 179 | + let synth = #{ tokenKind: tok.type_, text: tok.lexeme, leadingTrivia: leading, |
| 180 | + trailingTrivia: [], loc: tok.loc }; |
| 181 | + #{ committed: acc.committed, last: Some(synth), prev_end: tok_end } |
| 182 | + }, |
| 183 | + Some(lt) => { |
| 184 | + let updated = set_trailing(lt, lt.trailingTrivia ++ leading ++ [nl]); |
| 185 | + #{ committed: acc.committed, last: Some(updated), prev_end: tok_end } |
| 186 | + } |
| 187 | + } |
| 188 | + }, |
| 189 | + _ => { |
| 190 | + let new_committed = match acc.last { |
| 191 | + Some(lt) => acc.committed ++ [lt], |
| 192 | + None => acc.committed |
| 193 | + }; |
| 194 | + let new_tok = #{ tokenKind: tok.type_, text: tok.lexeme, leadingTrivia: leading, |
| 195 | + trailingTrivia: [], loc: tok.loc }; |
| 196 | + #{ committed: new_committed, last: Some(new_tok), prev_end: tok_end } |
| 197 | + } |
| 198 | + } |
| 199 | + }); |
| 200 | + |
| 201 | + // attach trailing source after the last token, then commit the last token |
| 202 | + let total_len = length(source); |
| 203 | + match final.last { |
| 204 | + None => final.committed, |
| 205 | + Some(lt) => { |
| 206 | + let lt2 = if final.prev_end < total_len { |
| 207 | + set_trailing(lt, lt.trailingTrivia ++ classify_trivia(substring(source, final.prev_end, total_len), file, final.prev_end)) |
| 208 | + } else { |
| 209 | + lt |
| 210 | + }; |
| 211 | + final.committed ++ [lt2] |
| 212 | + } |
| 213 | + } |
| 214 | +} |
| 215 | + |
| 216 | +// ---- public API ---- |
| 217 | + |
| 218 | +pub fn parse_to_cst(source: String, file: String, run_number: Int) -> CstTree { |
| 219 | + let cst_tokens = lex_with_trivia(source, file, run_number); |
| 220 | + let children = map(cst_tokens, |tok| NodeToken(tok)); |
| 221 | + let n = len(cst_tokens); |
| 222 | + let loc = if n == 0 { |
| 223 | + mk_loc(0, 0, file) |
| 224 | + } else { |
| 225 | + #{ start: cst_tokens[0].loc.start, end_: cst_tokens[n - 1].loc.end_, file: file } |
| 226 | + }; |
| 227 | + #{ kind: SourceFile, children: children, loc: loc } |
| 228 | +} |
0 commit comments