1use std::path::PathBuf;23use anyhow::Result;4use ast::{lower, AstSrc};5use itertools::Itertools;6use kinds::{KindsSrc, TokenKind};7use proc_macro2::{Punct, Spacing, TokenStream};8use quote::{format_ident, quote};9use ungrammar::Grammar;10use util::{ensure_file_contents, reformat, to_pascal_case, to_upper_snake_case};1112mod ast;13mod kinds;14mod util;1516enum SpecialName {17 Literal,18 Meta,19 Error,20}21fn classify_special(name: &str) -> Option<(SpecialName, &str)> {22 let name = name.strip_suffix('!')?;23 Some(if let Some(name) = name.strip_prefix("LIT_") {24 (SpecialName::Literal, name)25 } else if let Some(name) = name.strip_prefix("META_") {26 (SpecialName::Meta, name)27 } else if let Some(name) = name.strip_prefix("ERROR_") {28 (SpecialName::Error, name)29 } else {30 return None;31 })32}3334pub fn generate_ungrammar() -> Result<()> {35 let grammar: Grammar = include_str!(concat!(36 env!("CARGO_MANIFEST_DIR"),37 "/../crates/jrsonnet-rowan-parser/jsonnet.ungram"38 ))39 .parse()?;4041 let mut kinds = kinds::jsonnet_kinds();42 let ast = lower(&kinds, &grammar);4344 for token in grammar.tokens() {45 let token = &grammar[token];46 let token = &token.name.clone();47 if !kinds.is_token(token) {48 if let Some((special, name)) = classify_special(token) {49 match special {50 SpecialName::Literal => panic!("literal is not defined: {name}"),51 SpecialName::Meta => {52 eprintln!("implicit meta: {}", name);53 kinds.define_token(TokenKind::Meta {54 grammar_name: token.to_owned(),55 name: format!("META_{}", name),56 })57 }58 SpecialName::Error => {59 eprintln!("implicit error: {}", name);60 kinds.define_token(TokenKind::Error {61 grammar_name: token.to_owned(),62 name: format!("ERROR_{}", name),63 regex: None,64 priority: None,65 is_lexer_error: true,66 })67 }68 };69 continue;70 };71 let name = to_upper_snake_case(token);72 eprintln!("implicit kw: {}", token);73 kinds.define_token(TokenKind::Keyword {74 code: token.to_owned(),75 name: format!("{name}_KW"),76 });77 }78 }79 for node in &ast.nodes {80 let name = to_upper_snake_case(&node.name);81 kinds.define_node(&name);82 }83 for enum_ in &ast.enums {84 let name = to_upper_snake_case(&enum_.name);85 kinds.define_node(&name);86 }87 for token_enum in &ast.token_enums {88 let name = to_upper_snake_case(&token_enum.name);89 kinds.define_node(&name);90 }9192 let syntax_kinds = generate_syntax_kinds(&kinds, &ast)?;9394 let nodes = generate_nodes(&kinds, &ast)?;95 ensure_file_contents(96 &PathBuf::from(concat!(97 env!("CARGO_MANIFEST_DIR"),98 "/../crates/jrsonnet-rowan-parser/src/generated/syntax_kinds.rs",99 )),100 &syntax_kinds,101 )?;102 ensure_file_contents(103 &PathBuf::from(concat!(104 env!("CARGO_MANIFEST_DIR"),105 "/../crates/jrsonnet-rowan-parser/src/generated/nodes.rs",106 )),107 &nodes,108 )?;109 Ok(())110}111112fn generate_syntax_kinds(kinds: &KindsSrc, grammar: &AstSrc) -> Result<String> {113 let t_macros = kinds.tokens().filter_map(TokenKind::expand_t_macros);114 let token_kinds = kinds.tokens().map(TokenKind::expand_kind);115116 let keywords = kinds117 .tokens()118 .filter(|k| matches!(k, TokenKind::Keyword { .. }))119 .map(TokenKind::name)120 .map(|n| format_ident!("{n}"));121122 let nodes = kinds123 .nodes124 .iter()125 .map(|name| format_ident!("{}", name))126 .collect::<Vec<_>>();127128 let enums = grammar129 .enums130 .iter()131 .map(|e| format_ident!("{}", to_upper_snake_case(&e.name)))132 .chain(133 grammar134 .token_enums135 .iter()136 .map(|e| format_ident!("{}", to_upper_snake_case(&e.name))),137 );138139 let ast = quote! {140 #![allow(bad_style, missing_docs, unreachable_pub, clippy::manual_non_exhaustive, clippy::match_like_matches_macro)]141 use logos::Logos;142143 144 #[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash, Debug, Logos)]145 #[repr(u16)]146 pub enum SyntaxKind {147 #[doc(hidden)]148 TOMBSTONE,149 #[doc(hidden)]150 EOF,151 #(#token_kinds,)*152 #[error]153 ERROR,154 #(#nodes,)*155 #[doc(hidden)]156 __LAST,157 }158 use self::SyntaxKind::*;159160 impl SyntaxKind {161 pub fn is_keyword(self) -> bool {162 match self {163 #(#keywords)|* => true,164 _ => false,165 }166 }167 pub fn is_enum(self) -> bool {168 match self {169 #(#enums)|* => true,170 _ => false,171 }172 }173174 pub fn from_raw(r: u16) -> Self {175 assert!(r < Self::__LAST as u16);176 unsafe { std::mem::transmute(r) }177 }178 pub fn into_raw(self) -> u16 {179 self as u16180 }181 }182183 #[macro_export]184 macro_rules! T {#(#t_macros);*}185 pub use T;186 };187188 reformat(&ast.to_string())189}190191fn generate_nodes(kinds: &KindsSrc, grammar: &AstSrc) -> Result<String> {192 let (node_defs, node_boilerplate_impls): (Vec<_>, Vec<_>) = grammar193 .nodes194 .iter()195 .map(|node| {196 let name = format_ident!("{}", node.name);197 let kind = format_ident!("{}", to_upper_snake_case(&node.name));198 let traits = node.traits.iter().map(|trait_name| {199 let trait_name = format_ident!("{}", trait_name);200 quote!(impl ast::#trait_name for #name {})201 });202203 let methods = node.fields.iter().map(|field| {204 let method_name = field.method_name(kinds);205 let ty = field.ty();206207 if field.is_many() {208 quote! {209 pub fn #method_name(&self) -> AstChildren<#ty> {210 support::children(&self.syntax)211 }212 }213 } else if let Some(token_kind) = field.token_kind(kinds) {214 quote! {215 pub fn #method_name(&self) -> Option<#ty> {216 support::token(&self.syntax, #token_kind)217 }218 }219 } else if field.is_token_enum(grammar) {220 quote! {221 pub fn #method_name(&self) -> Option<#ty> {222 support::token_child(&self.syntax)223 }224 }225 } else {226 quote! {227 pub fn #method_name(&self) -> Option<#ty> {228 support::child(&self.syntax)229 }230 }231 }232 });233 (234 quote! {235 #[pretty_doc_comment_placeholder_workaround]236 #[derive(Debug, Clone, PartialEq, Eq, Hash)]237 pub struct #name {238 pub(crate) syntax: SyntaxNode,239 }240241 #(#traits)*242243 impl #name {244 #(#methods)*245 }246 },247 quote! {248 impl AstNode for #name {249 fn can_cast(kind: SyntaxKind) -> bool {250 kind == #kind251 }252 fn cast(syntax: SyntaxNode) -> Option<Self> {253 if Self::can_cast(syntax.kind()) { Some(Self { syntax }) } else { None }254 }255 fn syntax(&self) -> &SyntaxNode { &self.syntax }256 }257 },258 )259 })260 .unzip();261262 let (enum_defs, enum_boilerplate_impls): (Vec<_>, Vec<_>) = grammar263 .enums264 .iter()265 .map(|en| {266 let variants: Vec<_> = en267 .variants268 .iter()269 .map(|var| format_ident!("{}", var))270 .collect();271 let name = format_ident!("{}", en.name);272 let kinds: Vec<_> = variants273 .iter()274 .map(|name| format_ident!("{}", to_upper_snake_case(&name.to_string())))275 .collect();276 let traits = en.traits.iter().map(|trait_name| {277 let trait_name = format_ident!("{}", trait_name);278 quote!(impl ast::#trait_name for #name {})279 });280281 let ast_node = quote! {282 impl AstNode for #name {283 fn can_cast(kind: SyntaxKind) -> bool {284 match kind {285 #(#kinds)|* => true,286 _ => false,287 }288 }289 fn cast(syntax: SyntaxNode) -> Option<Self> {290 let res = match syntax.kind() {291 #(292 #kinds => #name::#variants(#variants { syntax }),293 )*294 _ => return None,295 };296 Some(res)297 }298 fn syntax(&self) -> &SyntaxNode {299 match self {300 #(301 #name::#variants(it) => &it.syntax,302 )*303 }304 }305 }306 };307308 (309 quote! {310 #[pretty_doc_comment_placeholder_workaround]311 #[derive(Debug, Clone, PartialEq, Eq, Hash)]312 pub enum #name {313 #(#variants(#variants),)*314 }315316 #(#traits)*317 },318 quote! {319 #(320 impl From<#variants> for #name {321 fn from(node: #variants) -> #name {322 #name::#variants(node)323 }324 }325 )*326 #ast_node327 },328 )329 })330 .unzip();331332 let (token_enum_defs, token_enum_boilerplate_impls): (Vec<_>, Vec<_>) = grammar333 .token_enums334 .iter()335 .map(|en| {336 let variants: Vec<_> = en337 .variants338 .iter()339 .map(|token| {340 format_ident!(341 "{}",342 to_pascal_case(kinds.token(token).expect("token exists").name())343 )344 })345 .collect();346 let name = format_ident!("{}", en.name);347 let kind_name = format_ident!("{}Kind", en.name);348 let kinds: Vec<_> = variants349 .iter()350 .map(|name| format_ident!("{}", to_upper_snake_case(&name.to_string())))351 .collect();352353 let ast_node = quote! {354 impl AstToken for #name {355 fn can_cast(kind: SyntaxKind) -> bool {356 #kind_name::can_cast(kind)357 }358 fn cast(syntax: SyntaxToken) -> Option<Self> {359 let kind = #kind_name::cast(syntax.kind())?;360 Some(#name { syntax, kind })361 }362 fn syntax(&self) -> &SyntaxToken {363 &self.syntax364 }365 }366367 impl #kind_name {368 fn can_cast(kind: SyntaxKind) -> bool {369 match kind {370 #(#kinds)|* => true,371 _ => false,372 }373 }374 pub fn cast(kind: SyntaxKind) -> Option<Self> {375 let res = match kind {376 #(#kinds => Self::#variants,)*377 _ => return None,378 };379 Some(res)380 }381 }382 };383384 (385 quote! {386 #[pretty_doc_comment_placeholder_workaround]387 #[derive(Debug, Clone, PartialEq, Eq, Hash)]388 pub struct #name { syntax: SyntaxToken, kind: #kind_name }389390 #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]391 pub enum #kind_name {392 #(#variants,)*393 }394 },395 quote! {396 #ast_node397398 impl #name {399 pub fn kind(&self) -> #kind_name {400 self.kind401 }402 }403404 impl std::fmt::Display for #name {405 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {406 std::fmt::Display::fmt(self.syntax(), f)407 }408 }409 },410 )411 })412 .unzip();413414 let (any_node_defs, any_node_boilerplate_impls): (Vec<_>, Vec<_>) = grammar415 .nodes416 .iter()417 .flat_map(|node| node.traits.iter().map(move |t| (t, node)))418 .into_group_map()419 .into_iter()420 .sorted_by_key(|(k, _)| *k)421 .map(|(trait_name, nodes)| {422 let name = format_ident!("Any{}", trait_name);423 let trait_name = format_ident!("{}", trait_name);424 let kinds: Vec<_> = nodes425 .iter()426 .map(|name| format_ident!("{}", to_upper_snake_case(&name.name.to_string())))427 .collect();428429 (430 quote! {431 #[pretty_doc_comment_placeholder_workaround]432 #[derive(Debug, Clone, PartialEq, Eq, Hash)]433 pub struct #name {434 pub(crate) syntax: SyntaxNode,435 }436 impl ast::#trait_name for #name {}437 },438 quote! {439 impl #name {440 #[inline]441 pub fn new<T: ast::#trait_name>(node: T) -> #name {442 #name {443 syntax: node.syntax().clone()444 }445 }446 }447 impl AstNode for #name {448 fn can_cast(kind: SyntaxKind) -> bool {449 match kind {450 #(#kinds)|* => true,451 _ => false,452 }453 }454 fn cast(syntax: SyntaxNode) -> Option<Self> {455 Self::can_cast(syntax.kind()).then(|| #name { syntax })456 }457 fn syntax(&self) -> &SyntaxNode {458 &self.syntax459 }460 }461 },462 )463 })464 .unzip();465466 let enum_names = grammar.enums.iter().map(|it| &it.name);467 let node_names = grammar.nodes.iter().map(|it| &it.name);468469 let display_impls = enum_names470 .chain(node_names.clone())471 .map(|it| format_ident!("{}", it))472 .map(|name| {473 quote! {474 impl std::fmt::Display for #name {475 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {476 std::fmt::Display::fmt(self.syntax(), f)477 }478 }479 }480 });481482 let ast = quote! {483 #![allow(non_snake_case, clippy::match_like_matches_macro)]484485 use crate::{486 SyntaxNode, SyntaxToken, SyntaxKind::{self, *},487 ast::{AstNode, AstToken, AstChildren, support},488 T,489 };490491 #(#node_defs)*492 #(#enum_defs)*493 #(#token_enum_defs)*494 #(#any_node_defs)*495 #(#node_boilerplate_impls)*496 #(#enum_boilerplate_impls)*497 #(#token_enum_boilerplate_impls)*498 #(#any_node_boilerplate_impls)*499 #(#display_impls)*500 };501502 let ast = ast.to_string().replace("T ! [", "T![");503504 let mut res = String::with_capacity(ast.len() * 2);505506 let mut docs = grammar507 .nodes508 .iter()509 .map(|it| &it.doc)510 .chain(grammar.enums.iter().map(|it| &it.doc));511512 for chunk in ast.split("# [pretty_doc_comment_placeholder_workaround] ") {513 res.push_str(chunk);514 if let Some(doc) = docs.next() {515 write_doc_comment(doc, &mut res);516 }517 }518519 let res = reformat(&res)?;520 Ok(res.replace("#[derive", "\n#[derive"))521}522523fn write_doc_comment(contents: &[String], dest: &mut String) {524 use std::fmt::Write;525 for line in contents {526 writeln!(dest, "///{}", line).unwrap();527 }528}529530pub fn escape_token_macro(token: &str) -> TokenStream {531 if "{}[]()$".contains(token) {532 let c = token.chars().next().unwrap();533 quote! { #c }534 } else if token.contains('$') {535 quote! { #token }536 } else {537 let cs = token.chars().map(|c| Punct::new(c, Spacing::Joint));538 quote! { #(#cs)* }539 }540}