Allow combining #[cache] with arguments

This is a naïve first attempt at enabling caching with arguments,
supporting both values and ownable references. References are less
efficient because it is extremely difficult to satisfy the
constraint `K: Borrow<Q>` for `HashMap` with a tuple. Most
recommendations seem to be to either do some contortions with trait
objects, or to use the hashbrown crate and implement
`hashbrown::Equivalent`, but I have not been able to successfully
think through how to do this in a way which works with arbitrary
tuples of foreign types. Still, a half-measure which can be
improved later is hopefully better than nothing.

Author: csnover

peg-macros/ast.rs

@@ -49,6 +49,16 @@ pub struct Rule {
     pub no_eof: bool,
 }
 
+impl Rule {
+    pub fn cacheable(&self) -> bool {
+        self.ty_params.is_none()
+            && self
+                .params
+                .iter()
+                .all(|param| matches!(param.ty, RuleParamTy::Rust(..)))
+    }
+}
+
 #[derive(Debug)]
 pub struct RuleParam {
     pub name: Ident,

peg-macros/translate.rs

@@ -84,10 +84,10 @@ pub(crate) fn compile_grammar(grammar: &Grammar) -> TokenStream {
                     continue;
                 }
 
-                if rule.cache.is_some() && !(rule.params.is_empty() && rule.ty_params.is_none()) {
+                if rule.cache.is_some() && !rule.cacheable() {
                     items.push(report_error(
-                            rule.name.span(),
-                            "rules with generics or parameters cannot use #[cache] or #[cache_left_rec]".to_string(),
+                        rule.name.span(),
+                        "rules with generics cannot use #[cache] or #[cache_left_rec]".to_string(),
                     ));
                     continue;
                 }
@@ -158,11 +158,29 @@ fn make_parse_state(grammar: &Grammar) -> TokenStream {
     let mut cache_fields_def: Vec<TokenStream> = Vec::new();
     let mut cache_fields: Vec<Ident> = Vec::new();
     for rule in grammar.iter_rules() {
-        if rule.cache.is_some() && rule.params.is_empty() && rule.ty_params.is_none() {
+        if rule.cache.is_some() && rule.cacheable() {
             let name = format_ident!("{}_cache", rule.name);
             let ret_ty = rule.ret_type.clone().unwrap_or_else(|| quote!(()));
+
+            // This could be written more simply as `(usize, #(, #param_ty)*))`,
+            // but this generates unnecessary brackets when `rule.params` is
+            // empty, and new releases of clippy have a bad habit of suddenly
+            // triggering on code generated by proc-macros when `quote_spanned!`
+            // is used because it thinks that the generated code was handwritten
+            let key = if rule.params.is_empty() {
+                quote_spanned!(span=> usize)
+            } else {
+                let param_ty = rule.params.iter().map(|param| {
+                    let RuleParamTy::Rust(ty) = &param.ty else {
+                        unreachable!()
+                    };
+                    quote_spanned!(span=> <::peg::chomp_ref!(#ty) as ::std::borrow::ToOwned>::Owned)
+                });
+                quote_spanned!(span=> (usize, #(#param_ty),*))
+            };
+
             cache_fields_def.push(
-                quote_spanned! { span =>  #name: ::std::collections::HashMap<usize, ::peg::RuleResult<#ret_ty>> },
+                quote_spanned!(span=> #name: ::std::collections::HashMap<#key, ::peg::RuleResult<#ret_ty>>),
             );
             cache_fields.push(name);
         }
@@ -274,28 +292,38 @@ fn compile_rule(context: &Context, rule: &Rule) -> TokenStream {
                 quote!()
             };
 
+            let param = rule.params.iter().map(|param| {
+                let name = &param.name;
+                quote_spanned!(span=> #name.to_owned())
+            });
+            let key = if rule.params.is_empty() {
+                quote_spanned!(span=> __pos)
+            } else {
+                quote_spanned!(span=> (__pos, #(#param),*))
+            };
+
             match cache_type {
                 Cache::Simple => quote_spanned! { span =>
-                    if let Some(entry) = __state.#cache_field.get(&__pos) {
+                    if let Some(entry) = __state.#cache_field.get(&#key) {
                         #cache_trace
                         return entry.clone();
                     }
 
                     let __rule_result = #wrapped_body;
-                    __state.#cache_field.insert(__pos, __rule_result.clone());
+                    __state.#cache_field.insert(#key, __rule_result.clone());
                     __rule_result
                 },
                 Cache::Recursive =>
                 // `#[cache_left_rec] support for recursive rules using the technique described here:
                 // <https://medium.com/@gvanrossum_83706/left-recursive-peg-grammars-65dab3c580e1>
                 {
                     quote_spanned! { span =>
-                        if let Some(entry) = __state.#cache_field.get(&__pos) {
+                        if let Some(entry) = __state.#cache_field.get(&#key) {
                             #cache_trace
                             return entry.clone();
                         }
 
-                        __state.#cache_field.insert(__pos, ::peg::RuleResult::Failed);
+                        __state.#cache_field.insert(#key, ::peg::RuleResult::Failed);
                         let mut __last_result = ::peg::RuleResult::Failed;
                         loop {
                             let __current_result = { #wrapped_body };
@@ -305,7 +333,7 @@ fn compile_rule(context: &Context, rule: &Rule) -> TokenStream {
                                     match __last_result {
                                         ::peg::RuleResult::Matched(__last_endpos, _) if __current_endpos <= __last_endpos => break,
                                         _ => {
-                                            __state.#cache_field.insert(__pos, __current_result.clone());
+                                            __state.#cache_field.insert(#key, __current_result.clone());
                                             __last_result = __current_result;
                                         },
                                     }

peg-runtime/lib.rs

@@ -71,3 +71,19 @@ pub fn call_custom_closure<I, T>(
 ) -> RuleResult<T> {
     f(input, pos)
 }
+
+// this is used to convert references to ownable types for cache keys, as a
+// cleaner alternative to filtering the token tree
+#[doc(hidden)]
+#[macro_export]
+macro_rules! chomp_ref {
+    (& $lt:lifetime $ty:ty) => {
+        $ty
+    };
+    (& $ty:ty) => {
+        $ty
+    };
+    ($ty:ty) => {
+        $ty
+    };
+}

src/lib.rs

@@ -312,7 +312,7 @@
 //!
 //! ## Caching and left recursion
 //!
-//! A `rule` without parameters can be prefixed with `#[cache]` if it is likely
+//! A `rule` without any generics can be prefixed with `#[cache]` if it is likely
 //! to be checked repeatedly in the same position. This memoizes the rule result
 //! as a function of input position, in the style of a [packrat
 //! parser][wp-peg-packrat].
@@ -334,6 +334,11 @@
 //! The `precedence!{}` syntax is another way to handle nested operators and avoid
 //! repeatedly matching an expression rule.
 //!
+//! Currently, rules that receive arguments can only be cached if their types
+//! are `Clone + Hash`. Explicit lifetimes in references are not currently
+//! supported. References are converted to values when they are inserted to the
+//! cache by cloning.
+//!
 //! ## Tracing
 //!
 //! If you pass the `peg/trace` feature to Cargo when building your project, a

tests/compile-fail/cache_with_args.rs

@@ -1,11 +1,8 @@
 extern crate peg;
 
-peg::parser!(grammar foo() for str { 
-    #[cache]
-    rule foo(x: u32) = "foo" //~ ERROR
-
+peg::parser!(grammar foo() for str {
     #[cache]
     rule ltarg<'a>() -> &'a str = { "" } //~ ERROR
 });
 
-fn main() {}
+fn main() {}

tests/compile-fail/cache_with_args.stderr

@@ -1,11 +1,5 @@
-error: rules with generics or parameters cannot use #[cache] or #[cache_left_rec]
- --> $DIR/cache_with_args.rs:5:10
+error: rules with generics cannot use #[cache] or #[cache_left_rec]
+ --> tests/compile-fail/cache_with_args.rs:5:10
   |
-5 |     rule foo(x: u32) = "foo" //~ ERROR
-  |          ^^^
-
-error: rules with generics or parameters cannot use #[cache] or #[cache_left_rec]
- --> $DIR/cache_with_args.rs:8:10
-  |
-8 |     rule ltarg<'a>() -> &'a str = { "" } //~ ERROR
+5 |     rule ltarg<'a>() -> &'a str = { "" } //~ ERROR
   |          ^^^^^

tests/pass/cache_with_args.rs

@@ -0,0 +1,58 @@
+peg::parser!(grammar foo() for str {
+    pub rule main()
+    = yepnope(true)
+      yepnope(false)
+    / yepnope(true)
+      yepnope(true)
+      yepnope(false)
+
+    #[cache]
+    rule yepnope(yep: bool)
+    = &assert(yep, "yep") "yep"
+    / !assert(yep, "yep") "nope"
+
+    pub rule main_ref()
+    = yepnope_ref(&true)
+      yepnope_ref(&false)
+    / yepnope_ref(&true)
+      yepnope_ref(&true)
+      yepnope_ref(&false)
+
+    #[cache]
+    rule yepnope_ref(yep: &bool)
+    = &assert(*yep, "yep") "yep"
+    / !assert(*yep, "yep") "nope"
+
+    pub rule main_ref_lifetime()
+    = yepnope_ref(&true)
+      yepnope_ref(&false)
+    / yepnope_ref(&true)
+      yepnope_ref(&true)
+      yepnope_ref(&false)
+
+    #[cache]
+    rule yepnope_ref_lifetime(yep: &'input bool)
+    = &assert(*yep, "yep") "yep"
+    / !assert(*yep, "yep") "nope"
+
+    rule assert(v: bool, msg: &'static str)
+    = {? if v { Ok(()) } else { Err(msg) } }
+});
+
+#[test]
+fn main() {
+    foo::main("yepnope").unwrap();
+    foo::main("nopeyep").unwrap_err();
+    foo::main("yepyepnope").unwrap();
+    foo::main("nopeyepnope").unwrap_err();
+
+    foo::main_ref("yepnope").unwrap();
+    foo::main_ref("nopeyep").unwrap_err();
+    foo::main_ref("yepyepnope").unwrap();
+    foo::main_ref("nopeyepnope").unwrap_err();
+
+    foo::main_ref_lifetime("yepnope").unwrap();
+    foo::main_ref_lifetime("nopeyep").unwrap_err();
+    foo::main_ref_lifetime("yepyepnope").unwrap();
+    foo::main_ref_lifetime("nopeyepnope").unwrap_err();
+}

tests/pass/mod.rs

@@ -7,6 +7,7 @@ mod arithmetic_with_left_recursion;
 mod assembly_ast_dyn_type_param_bounds;
 mod borrow_from_input;
 mod bytes;
+mod cache_with_args;
 mod conditional_block;
 mod crate_import;
 mod custom_expr;