Slightly cursed visitor stuff

Author: CraftSpider

examples/zero-copy.rs

@@ -1,4 +1,5 @@
 use chumsky::zero_copy::prelude::*;
+use chumsky::zero_copy::problems::Problems;
 
 #[derive(PartialEq, Debug)]
 enum Token<'a> {
@@ -27,6 +28,17 @@ fn parser<'a>() -> impl Parser<'a, str, [(SimpleSpan<usize>, Token<'a>); 6]> {
         .collect()
 }
 
+fn bad_parser() -> impl for<'a> Parser<'a, str, Vec<()>, extra::Err<Rich<str>>> + Problems {
+    just("").or(just("b"))
+        .or_not()
+        .separated_by(just(""))
+        .ignored()
+        .then(just("").ignored())
+        .ignored()
+        .repeated()
+        .collect::<_>()
+}
+
 fn main() {
     assert_eq!(
         parser()
@@ -41,4 +53,9 @@ fn main() {
             ((31..37).into(), Token::Ident("tokens")),
         ]),
     );
+
+    let p = bad_parser();
+    for p in p.find_problems() {
+        println!("{p}");
+    }
 }

src/zero_copy/mod.rs

@@ -44,12 +44,14 @@ pub mod error;
 pub mod extra;
 pub mod input;
 pub mod primitive;
+pub mod problems;
 pub mod recovery;
 pub mod recursive;
 #[cfg(feature = "regex")]
 pub mod regex;
 pub mod span;
 pub mod text;
+pub(crate) mod visit;
 
 /// Commonly used functions, traits and types.
 ///

src/zero_copy/primitive.rs

@@ -147,7 +147,7 @@ impl<T> Default for JustCfg<T> {
 
 /// See [`just`].
 pub struct Just<T, I: ?Sized, E = EmptyErr> {
-    seq: T,
+    pub(crate) seq: T,
     phantom: PhantomData<(E, I)>,
 }
 

src/zero_copy/problems.rs

@@ -0,0 +1,90 @@
+//! TODO
+
+use crate::zero_copy::visit::{ParserInfo, ParserVisitor, Visitable};
+
+#[derive(Clone, PartialEq)]
+enum CheckState {
+    None,
+    Check(String),
+    Defer(String),
+}
+
+struct ProblemVisitor {
+    problems: Vec<String>,
+
+    first: bool,
+    check_nonempty: CheckState,
+}
+
+impl ProblemVisitor {
+    fn new() -> Self {
+        ProblemVisitor {
+            problems: Vec::new(),
+
+            first: true,
+            check_nonempty: CheckState::None,
+        }
+    }
+
+    fn into_problems(self) -> Vec<String> {
+        self.problems
+    }
+}
+
+impl ParserVisitor for ProblemVisitor {
+    fn visit<P>(&mut self, info: &ParserInfo<'_, P>)
+    where
+        P: ?Sized + Visitable,
+    {
+        if self.first {
+            self.first = false;
+            if info.size_hint.lower() == 0 {
+                self.problems.push(format!(
+                    "The top-level `{}` parser can potentially consume no input, meaning that it cannot fail.\nThis is probably a bug.\nConsider adding `.then_ignore(end())` to the parser.", info.name
+                ))
+            }
+        }
+
+        let state = core::mem::replace(&mut self.check_nonempty, CheckState::None);
+        if let CheckState::Check(name) = state {
+            if info.size_hint.lower() == 0 {
+                self.problems.push(format!(
+                    "`{}` parser has an inner parser that can potentially consume no input, meaning that it cannot fail.\nThis is probably a bug, because it could lead to an infinite loop.\nConsider using `.at_least(1)` to force the inner parser to consume some input.",
+                    name,
+                ))
+            }
+            self.check_nonempty = CheckState::Defer(name);
+        }
+
+        match info.name {
+            "repeated" | "separated_by" => {
+                let old = core::mem::replace(
+                    &mut self.check_nonempty,
+                    CheckState::Check(info.name.to_string())
+                );
+                info.visit(self);
+                self.check_nonempty = old;
+            }
+            _ => info.visit(self),
+        }
+
+        let state = core::mem::replace(&mut self.check_nonempty, CheckState::None);
+        if let CheckState::Defer(name) = state {
+            self.check_nonempty = CheckState::Check(name);
+        }
+    }
+}
+
+/// TODO
+pub trait Problems: Visitable {
+    /// TODO
+    fn find_problems(&self) -> Vec<String>;
+}
+
+impl<P: Visitable> Problems for P {
+    fn find_problems(&self) -> Vec<String> {
+        let mut visitor = ProblemVisitor::new();
+        self.visit(&mut visitor);
+        visitor.into_problems()
+    }
+}