ast.rs 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393
  1. pub use crate::lexer::{FileRef, Located, Span};
  2. use std::fmt;
  3. pub type StrRef = string_interner::DefaultSymbol;
  4. pub type Name = Located<StrRef>;
  5. pub struct ASTArena {
  6. files: Vec<String>,
  7. strings: string_interner::StringInterner,
  8. exprs: Vec<Expr>,
  9. }
  10. impl Default for ASTArena {
  11. fn default() -> Self {
  12. Self::new()
  13. }
  14. }
  15. impl ASTArena {
  16. pub fn new() -> ASTArena {
  17. ASTArena {
  18. files: Vec::new(),
  19. strings: string_interner::StringInterner::new(),
  20. exprs: vec![Expr::Nil],
  21. }
  22. }
  23. pub fn expr_nil(&self) -> ExprId {
  24. ExprId { idx: 0 }
  25. }
  26. pub fn add_expr(&mut self, e: Expr) -> ExprId {
  27. let idx = self.exprs.len();
  28. self.exprs.push(e);
  29. ExprId { idx }
  30. }
  31. pub fn add_string(&mut self, s: &str) -> string_interner::DefaultSymbol {
  32. self.strings.get_or_intern(s)
  33. }
  34. pub fn show_stmt(&self, stmt: &Stmt, f: &mut fmt::Formatter) -> fmt::Result {
  35. match stmt {
  36. Stmt::Puts(expr) => {
  37. write!(f, "Puts ")?;
  38. self.show_expr(&self[expr.item], f, 0)
  39. }
  40. Stmt::Fix(name) => writeln!(f, "Fix {}", &self[name.item]),
  41. Stmt::Assn(fixed, name, expr) => {
  42. write!(
  43. f,
  44. "Assn {} {} ",
  45. if *fixed { "fixed" } else { "" },
  46. &self[name.item]
  47. )?;
  48. self.show_expr(&self[expr.item], f, 0)
  49. }
  50. Stmt::LitAssn(fixed, name, strs) => {
  51. write!(
  52. f,
  53. "LitAssn {} {}, [ ",
  54. if *fixed { "fixed" } else { "" },
  55. &self[name.item],
  56. )?;
  57. for str in strs.iter() {
  58. let s = &self[str.item];
  59. write!(f, " {} ", s)?;
  60. }
  61. writeln!(f, "]")
  62. }
  63. }
  64. }
  65. pub fn add_file(&mut self, file: String) -> FileRef {
  66. let idx = self.files.len();
  67. self.files.push(file);
  68. FileRef { idx }
  69. }
  70. pub fn get_line(&self, file: FileRef, span: Span) -> String {
  71. let mut line_number = 1;
  72. let mut start_of_line = 0;
  73. let mut end_of_line = None;
  74. let src = &self.files[file.idx];
  75. for (i, ch) in src.char_indices() {
  76. if ch == '\n' {
  77. if i < span.start as usize {
  78. line_number += 1;
  79. start_of_line = i;
  80. }
  81. if i >= span.end as usize && end_of_line.is_none() {
  82. end_of_line = Some(i);
  83. }
  84. }
  85. }
  86. let end_of_line = end_of_line.unwrap_or(src.len());
  87. let mut result = format!("{:3} |", line_number);
  88. result.push_str(&src[start_of_line..end_of_line]);
  89. result.push_str("\n ");
  90. for _ in start_of_line..(span.start as usize) {
  91. result.push(' ');
  92. }
  93. for _ in span.start..span.end {
  94. result.push('^');
  95. }
  96. result
  97. }
  98. fn indent(&self, f: &mut fmt::Formatter, depth: usize) -> fmt::Result {
  99. for _ in 0..depth {
  100. write!(f, " ")?;
  101. }
  102. Ok(())
  103. }
  104. fn show_pat(&self, pat: &Pat, f: &mut fmt::Formatter) -> fmt::Result {
  105. match pat {
  106. Pat::Wildcard => write!(f, "_"),
  107. Pat::Var(n) => write!(f, "{}", &self[n.item]),
  108. Pat::Lit(Literal::Atom(n)) => write!(f, "{}", &self[n.item]),
  109. Pat::Lit(lit) => write!(f, "{:?}", lit),
  110. Pat::Tup(tup) => {
  111. write!(f, "Tup( ")?;
  112. for t in tup {
  113. self.show_pat(t, f)?;
  114. write!(f, " ")?;
  115. }
  116. write!(f, ")")
  117. }
  118. }
  119. }
  120. fn show_expr(&self, expr: &Expr, f: &mut fmt::Formatter, depth: usize) -> fmt::Result {
  121. match expr {
  122. Expr::Nil => writeln!(f, "Nil"),
  123. Expr::Var(v) => writeln!(f, "Var({})", &self[v.item]),
  124. Expr::Lit(Literal::Atom(n)) => writeln!(f, "Lit(Atom({}))", &self[n.item]),
  125. Expr::Lit(lit) => writeln!(f, "{:?}", lit),
  126. Expr::Range(from, to) => {
  127. writeln!(f, "Range(")?;
  128. self.indent(f, depth + 2)?;
  129. self.show_expr(&self[*from], f, depth + 2)?;
  130. self.indent(f, depth + 2)?;
  131. self.show_expr(&self[*to], f, depth + 2)?;
  132. self.indent(f, depth)?;
  133. writeln!(f, ")")
  134. }
  135. Expr::Ap(func, args) => {
  136. writeln!(f, "Ap(")?;
  137. self.indent(f, depth + 2)?;
  138. self.show_expr(&self[*func], f, depth + 2)?;
  139. for arg in args {
  140. self.indent(f, depth + 2)?;
  141. self.show_expr(&self[*arg], f, depth + 2)?;
  142. }
  143. self.indent(f, depth)?;
  144. writeln!(f, ")")
  145. }
  146. Expr::Tup(expr) => {
  147. writeln!(f, "Tup(")?;
  148. for e in expr {
  149. self.indent(f, depth + 2)?;
  150. self.show_expr(&self[*e], f, depth + 2)?;
  151. }
  152. self.indent(f, depth)?;
  153. writeln!(f, ")")
  154. }
  155. Expr::Cat(expr) => {
  156. writeln!(f, "Cat(")?;
  157. for e in expr {
  158. self.indent(f, depth + 2)?;
  159. self.show_expr(&self[*e], f, depth + 2)?;
  160. }
  161. self.indent(f, depth)?;
  162. writeln!(f, ")")
  163. }
  164. Expr::Chc(expr) => {
  165. writeln!(f, "Chc(")?;
  166. for e in expr {
  167. if let Some(s) = e.weight {
  168. self.indent(f, depth + 2)?;
  169. writeln!(f, "{}:", s)?;
  170. self.indent(f, depth + 4)?;
  171. self.show_expr(&self[e.value], f, depth + 4)?;
  172. } else {
  173. self.indent(f, depth + 2)?;
  174. self.show_expr(&self[e.value], f, depth + 2)?;
  175. }
  176. }
  177. self.indent(f, depth)?;
  178. writeln!(f, ")")
  179. }
  180. Expr::Let(fixed, name, expr, body) => {
  181. writeln!(
  182. f,
  183. "Let({}{}",
  184. if *fixed { "fixed " } else { "" },
  185. &self[name.item]
  186. )?;
  187. self.indent(f, depth + 2)?;
  188. self.show_expr(&self[*expr], f, depth + 2)?;
  189. self.indent(f, depth + 2)?;
  190. self.show_expr(&self[*body], f, depth + 2)?;
  191. self.indent(f, depth)?;
  192. writeln!(f, ")")
  193. }
  194. Expr::Fun(cases) => {
  195. writeln!(f, "Fun(")?;
  196. for case in cases {
  197. self.indent(f, depth + 2)?;
  198. for pat in case.pats.iter() {
  199. self.show_pat(pat, f)?;
  200. writeln!(f, ", ")?;
  201. }
  202. writeln!(f, " =>")?;
  203. self.indent(f, depth + 4)?;
  204. self.show_expr(&self[case.expr], f, depth + 4)?;
  205. }
  206. self.indent(f, depth)?;
  207. writeln!(f, ")")
  208. }
  209. Expr::Case(expr, cases) => {
  210. writeln!(f, "Case(")?;
  211. self.indent(f, depth)?;
  212. self.show_expr(&self[*expr], f, depth)?;
  213. for case in cases {
  214. self.indent(f, depth + 2)?;
  215. self.show_pat(&case.pats[0], f)?;
  216. writeln!(f, " =>")?;
  217. self.indent(f, depth + 4)?;
  218. self.show_expr(&self[case.expr], f, depth + 4)?;
  219. }
  220. self.indent(f, depth)?;
  221. writeln!(f, ")")
  222. }
  223. }
  224. }
  225. }
  226. impl std::ops::Index<string_interner::DefaultSymbol> for ASTArena {
  227. type Output = str;
  228. fn index(&self, sf: string_interner::DefaultSymbol) -> &str {
  229. self.strings.resolve(sf).unwrap()
  230. }
  231. }
  232. impl std::ops::Index<ExprRef> for ASTArena {
  233. type Output = Expr;
  234. fn index(&self, rf: ExprRef) -> &Self::Output {
  235. &self.exprs[rf.item.idx]
  236. }
  237. }
  238. impl std::ops::Index<ExprId> for ASTArena {
  239. type Output = Expr;
  240. fn index(&self, rf: ExprId) -> &Self::Output {
  241. &self.exprs[rf.idx]
  242. }
  243. }
  244. /// A `Printable` struct is a bundle of another value and an
  245. /// `ASTArena`, which allows us to fetch the various indices and
  246. /// dereference the interned strings.
  247. pub struct Printable<'a, T> {
  248. arena: &'a ASTArena,
  249. value: &'a T,
  250. }
  251. impl<'a> std::fmt::Debug for Printable<'a, Stmt> {
  252. fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
  253. self.arena.show_stmt(self.value, f)
  254. }
  255. }
  256. impl<'a> std::fmt::Debug for Printable<'a, Expr> {
  257. fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
  258. self.arena.show_expr(self.value, f, 0)
  259. }
  260. }
  261. /// A top-level Matzo statement
  262. #[derive(Debug, Clone)]
  263. pub enum Stmt {
  264. /// evaluate and print the value of an expression
  265. Puts(ExprRef),
  266. /// replace a named item with the forced version of that item
  267. Fix(Name),
  268. /// assign a value to a name which may or may not be forced
  269. Assn(bool, Name, ExprRef),
  270. /// assign one of a set of strings to a name, which may or may not
  271. /// be forced
  272. LitAssn(bool, Name, Vec<Name>),
  273. }
  274. impl Stmt {
  275. pub fn show<'a>(&'a self, ast: &'a ASTArena) -> Printable<'a, Stmt> {
  276. Printable {
  277. arena: ast,
  278. value: self,
  279. }
  280. }
  281. }
  282. /// A Matzo expression
  283. #[derive(Debug, Clone)]
  284. pub enum Expr {
  285. Var(Name),
  286. Cat(Vec<ExprRef>),
  287. Chc(Vec<Choice>),
  288. Lit(Literal),
  289. Ap(ExprRef, Vec<ExprRef>),
  290. Tup(Vec<ExprRef>),
  291. Let(bool, Name, ExprRef, ExprRef),
  292. Fun(Vec<Case>),
  293. Range(ExprRef, ExprRef),
  294. Case(ExprRef, Vec<Case>),
  295. Nil,
  296. }
  297. pub type ExprRef = Located<ExprId>;
  298. #[derive(Debug, Copy, Clone, PartialEq, Eq)]
  299. pub struct ExprId {
  300. idx: usize,
  301. }
  302. impl ExprId {
  303. pub fn nil(&self) -> bool {
  304. self.idx == 0
  305. }
  306. }
  307. /// A single case in an anonymous function or `case` statement
  308. #[derive(Debug, Clone)]
  309. pub struct Case {
  310. pub pats: Vec<Pat>,
  311. pub expr: ExprRef,
  312. }
  313. /// A pattern, e.g. in an anonymous function or `case` statement
  314. #[derive(Debug, Clone)]
  315. pub enum Pat {
  316. Var(Name),
  317. Wildcard,
  318. Lit(Literal),
  319. Tup(Vec<Pat>),
  320. }
  321. /// A single element in a choice, with an optional weight (which
  322. /// defaults to 1) and a value
  323. #[derive(Debug, Clone)]
  324. pub struct Choice {
  325. pub weight: Option<i64>,
  326. pub value: ExprRef,
  327. }
  328. impl Choice {
  329. /// Fetch a weight from a `Choice`, defaulting to 1
  330. pub fn weight(&self) -> i64 {
  331. self.weight.unwrap_or(1)
  332. }
  333. }
  334. /// An atomic literal: a string, a number, or an atom
  335. #[derive(Debug, Clone)]
  336. pub enum Literal {
  337. Str(String),
  338. Atom(Name),
  339. Num(i64),
  340. }
  341. impl PartialEq for Literal {
  342. fn eq(&self, other: &Literal) -> bool {
  343. match (self, other) {
  344. (Literal::Str(s1), Literal::Str(s2)) => s1 == s2,
  345. (Literal::Atom(a1), Literal::Atom(a2)) => a1.item == a2.item,
  346. (Literal::Num(n1), Literal::Num(n2)) => n1 == n2,
  347. (_, _) => false,
  348. }
  349. }
  350. }