Merge pull request #631 from avityuk/issue-630-parse-performance

Avoid repeated parser table decode and cut parse setup overhead (issue 630)

Author: ratmice

lrpar/src/lib/ctbuilder.rs

@@ -1093,7 +1093,7 @@ where
         let run_parser = match self.yacckind.unwrap() {
             YaccKind::Original(YaccOriginalActionKind::GenericParseTree) => {
                 quote! {
-                    ::lrpar::RTParserBuilder::new(&grm, &stable)
+                    ::lrpar::RTParserBuilder::new(grm, stable)
                         .recoverer(#recoverer)
                         .parse_map(
                             lexer,
@@ -1104,7 +1104,7 @@ where
             }
             YaccKind::Original(YaccOriginalActionKind::NoAction) => {
                 quote! {
-                    ::lrpar::RTParserBuilder::new(&grm, &stable)
+                    ::lrpar::RTParserBuilder::new(grm, stable)
                         .recoverer(#recoverer)
                         .parse_map(lexer, &|_| (), &|_, _| ()).1
                 }
@@ -1145,7 +1145,7 @@ where
                                     #action_fn_parse_param_ty
                             ) -> #actionskind #type_generics
                         > = ::std::vec![#(&#wrappers,)*];
-                    match ::lrpar::RTParserBuilder::new(&grm, &stable)
+                    match ::lrpar::RTParserBuilder::new(grm, stable)
                         .recoverer(#recoverer)
                         .parse_actions(lexer, &actions, #action_fn_parse_param) {
                             (Some(#actionskind::#action_ident(x)), y) => (Some(x), y),
@@ -1205,14 +1205,24 @@ where
             const __GRM_DATA: &[u8] = &[#(#grm_data,)*];
             const __STABLE_DATA: &[u8] = &[#(#stable_data,)*];
 
+            fn __lrpar_parser_data() -> &'static ::lrpar::ParserData<#storaget> {
+                static DATA: ::std::sync::OnceLock<::lrpar::ParserData<#storaget>>
+                    = ::std::sync::OnceLock::new();
+                DATA.get_or_init(
+                    || ::lrpar::ctbuilder::_reconstitute(__GRM_DATA, __STABLE_DATA)
+                )
+            }
+
             #[allow(dead_code)]
             pub fn parse #generics (
                  lexer: &'lexer dyn ::lrpar::NonStreamingLexer<'input, #lexertypest>,
                  #parse_fn_parse_param
             ) -> #parse_fn_return_ty
             #where_clause
             {
-                let (grm, stable) = ::lrpar::ctbuilder::_reconstitute(__GRM_DATA, __STABLE_DATA);
+                let __data = __lrpar_parser_data();
+                let grm = __data.grm();
+                let stable = __data.stable();
                 #run_parser
             }
         })
@@ -1601,16 +1611,34 @@ where
     }
 }
 
+/// Bundles `YaccGrammar` + `StateTable` so that generated parsers can hold
+/// them in a `OnceLock` without naming `lrtable` directly.
+#[doc(hidden)]
+pub struct ParserData<StorageT: Eq + Hash> {
+    grm: YaccGrammar<StorageT>,
+    stable: StateTable<StorageT>,
+}
+
+impl<StorageT: Eq + Hash> ParserData<StorageT> {
+    pub fn grm(&self) -> &YaccGrammar<StorageT> {
+        &self.grm
+    }
+
+    pub fn stable(&self) -> &StateTable<StorageT> {
+        &self.stable
+    }
+}
+
 /// This function is called by generated files; it exists so that generated files don't require a
 /// direct dependency on bincode.
 #[doc(hidden)]
-pub fn _reconstitute<StorageT: Decode<()> + Hash + PrimInt + Unsigned + 'static>(
+pub fn _reconstitute<StorageT: Decode<()> + Eq + Hash + PrimInt + Unsigned + 'static>(
     grm_buf: &[u8],
     stable_buf: &[u8],
-) -> (YaccGrammar<StorageT>, StateTable<StorageT>) {
+) -> ParserData<StorageT> {
     let (grm, _) = decode_from_slice(grm_buf, bincode::config::standard()).unwrap();
     let (stable, _) = decode_from_slice(stable_buf, bincode::config::standard()).unwrap();
-    (grm, stable)
+    ParserData { grm, stable }
 }
 
 /// An interface to the result of [CTParserBuilder::build()].

lrpar/src/lib/mod.rs

@@ -206,7 +206,7 @@ pub mod parser;
 pub mod test_utils;
 
 pub use crate::{
-    ctbuilder::{CTParser, CTParserBuilder, RustEdition, Visibility},
+    ctbuilder::{CTParser, CTParserBuilder, ParserData, RustEdition, Visibility},
     lex_api::{LexError, Lexeme, Lexer, LexerTypes, NonStreamingLexer},
     parser::{LexParseError, ParseError, ParseRepair, RTParserBuilder, RecoveryKind},
 };

lrpar/src/lib/parser.rs

@@ -112,7 +112,7 @@ pub(super) struct Parser<
 {
     rcvry_kind: RecoveryKind,
     pub(super) grm: &'a YaccGrammar<StorageT>,
-    pub(super) token_cost: Box<TokenCostFn<'a, StorageT>>,
+    pub(super) token_cost: TokenCostFn<'a, StorageT>,
     pub(super) stable: &'a StateTable<StorageT>,
     lexer: &'b dyn NonStreamingLexer<'input, LexerTypesT>,
     // In the long term, we should remove the `lexemes` field entirely, as the `NonStreamingLexer` API is
@@ -176,7 +176,7 @@ where
         let psr = Parser {
             rcvry_kind,
             grm,
-            token_cost: Box::new(token_cost),
+            token_cost,
             stable,
             lexer,
             lexemes,
@@ -274,7 +274,7 @@ where
         let psr = Parser {
             rcvry_kind,
             grm,
-            token_cost: Box::new(token_cost),
+            token_cost,
             stable,
             lexer,
             lexemes,
@@ -970,13 +970,10 @@ where
         fterm: &dyn Fn(LexerTypesT::LexemeT) -> Node,
         fnonterm: &dyn Fn(RIdx<StorageT>, Vec<Node>) -> Node,
     ) -> (Option<Node>, Vec<LexParseError<StorageT, LexerTypesT>>) {
-        let mut lexemes = vec![];
-        for e in lexer.iter().collect::<Vec<_>>() {
-            match e {
-                Ok(l) => lexemes.push(l),
-                Err(e) => return (None, vec![e.into()]),
-            }
-        }
+        let lexemes = match lexer.iter().collect() {
+            Ok(lexemes) => lexemes,
+            Err(e) => return (None, vec![e.into()]),
+        };
         Parser::<
             StorageT,
             LexerTypesT,
@@ -1019,13 +1016,10 @@ where
         actions: &'a [ActionFn<'a, 'b, 'input, StorageT, LexerTypesT, ActionT, ParamT>],
         param: ParamT,
     ) -> (Option<ActionT>, Vec<LexParseError<StorageT, LexerTypesT>>) {
-        let mut lexemes = vec![];
-        for e in lexer.iter().collect::<Vec<_>>() {
-            match e {
-                Ok(l) => lexemes.push(l),
-                Err(e) => return (None, vec![e.into()]),
-            }
-        }
+        let lexemes = match lexer.iter().collect() {
+            Ok(lexemes) => lexemes,
+            Err(e) => return (None, vec![e.into()]),
+        };
         Parser::parse_actions(
             self.recoverer,
             self.grm,