From df99d78a64df159d947e2619a10573868727bc42 Mon Sep 17 00:00:00 2001
From: Chris Mounce <christophermounce@gmail.com>
Date: Wed, 28 May 2025 00:13:03 -0700
Subject: [PATCH] Add special matchers, lookahead operators

---
 peg_macro/src/lib.rs | 66 ++++++++++++++++++++++++++++++++++++++++++--
 src/peg/peg.rs       | 36 ++++++++++++++++++++++--
 2 files changed, 97 insertions(+), 5 deletions(-)

diff --git a/peg_macro/src/lib.rs b/peg_macro/src/lib.rs
index 493386f..d2ea7fc 100644
--- a/peg_macro/src/lib.rs
+++ b/peg_macro/src/lib.rs
@@ -1,5 +1,6 @@
 use std::ops::RangeInclusive;
 
+use kw::{ANY, EOI};
 use proc_macro::TokenStream;
 use quote::quote;
 use syn::{
@@ -20,16 +21,25 @@ struct Rule {
 
 #[derive(Debug)]
 enum Term {
+    AnyChar,
     Choice(Vec<Term>),
+    EOI,
     Literal(String),
+    NegLookahead(Box<Term>),
     Optional(Box<Term>),
     Plus(Box<Term>),
+    PosLookahead(Box<Term>),
     Range(RangeInclusive<char>),
     Rule(Ident),
     Sequence(Vec<Term>),
     Star(Box<Term>),
 }
 
+mod kw {
+    syn::custom_keyword!(ANY);
+    syn::custom_keyword!(EOI);
+}
+
 impl Parse for Grammar {
     fn parse(input: ParseStream) -> syn::Result<Self> {
         Ok(Grammar {
@@ -59,7 +69,13 @@ impl Parse for Term {
         fn parse_atom(input: ParseStream) -> syn::Result<Term> {
             let look = input.lookahead1();
             if look.peek(Ident) {
-                input.parse().map(Term::Rule)
+                if input.parse::<ANY>().is_ok() {
+                    Ok(Term::AnyChar)
+                } else if input.parse::<EOI>().is_ok() {
+                    Ok(Term::EOI)
+                } else {
+                    input.parse().map(Term::Rule)
+                }
             } else if look.peek(LitStr) {
                 input.parse::<LitStr>().map(|x| Term::Literal(x.value()))
             } else if look.peek(LitChar) {
@@ -89,10 +105,20 @@ impl Parse for Term {
             Ok(result)
         }
 
+        fn parse_lookahead(input: ParseStream) -> syn::Result<Term> {
+            if input.parse::<Token![!]>().is_ok() {
+                parse_repeat(input).map(|x| Term::NegLookahead(x.into()))
+            } else if input.parse::<Token![&]>().is_ok() {
+                parse_repeat(input).map(|x| Term::PosLookahead(x.into()))
+            } else {
+                parse_repeat(input)
+            }
+        }
+
         fn parse_sequence(input: ParseStream) -> syn::Result<Term> {
-            let mut terms = vec![parse_repeat(input)?];
+            let mut terms = vec![parse_lookahead(input)?];
             while !input.is_empty() && !input.peek(Token![/]) && !input.peek(Token![;]) {
-                terms.push(parse_repeat(input)?);
+                terms.push(parse_lookahead(input)?);
             }
             if terms.len() == 1 {
                 Ok(terms.pop().unwrap())
@@ -121,6 +147,12 @@ impl Parse for Term {
 impl Term {
     fn generate_code(&self) -> proc_macro2::TokenStream {
         match self {
+            Term::AnyChar => quote! {
+                p.any()
+            },
+            Term::EOI => quote! {
+                p.eoi()
+            },
             Term::Rule(ident) => quote! {
                 #ident(p)
             },
@@ -186,6 +218,34 @@ impl Term {
                     p.range(#lo..=#hi)
                 }
             }
+            Term::NegLookahead(term) => {
+                let code = term.generate_code();
+                quote! {
+                    {
+                        let save = p.save();
+                        if #code {
+                            p.restore(save);
+                            false
+                        } else {
+                            true
+                        }
+                    }
+                }
+            }
+            Term::PosLookahead(term) => {
+                let code = term.generate_code();
+                quote! {
+                    {
+                        let save = p.save();
+                        if #code {
+                            p.restore(save);
+                            true
+                        } else {
+                            false
+                        }
+                    }
+                }
+            }
         }
     }
 }
diff --git a/src/peg/peg.rs b/src/peg/peg.rs
index 1716457..ce8be35 100644
--- a/src/peg/peg.rs
+++ b/src/peg/peg.rs
@@ -45,6 +45,19 @@ impl ParseState {
         }
         false
     }
+
+    pub fn any(&mut self) -> bool {
+        if let Some(c) = self.input[self.offset..].chars().next() {
+            self.offset += c.len_utf8();
+            true
+        } else {
+            false
+        }
+    }
+
+    pub fn eoi(&self) -> bool {
+        self.offset >= self.input.len()
+    }
 }
 
 #[cfg(test)]
@@ -54,7 +67,7 @@ mod tests {
     use super::*;
 
     grammar! {
-        fake_csv = line "\n" line;
+        fake_csv = line "\n" line EOI;
         line = item "," item;
         item = "foo" / "bar";
 
@@ -69,11 +82,15 @@ mod tests {
 
         words = "(" (plain_word ("," plain_word)*)? ")";
         nested_choice = "(" ("a" / "b") ("c" / "d") ")";
+
+        dq = "\"";
+        quoted = dq ("\\" ANY / !dq ANY)* dq;
     }
 
     fn parse<T: Fn(&mut ParseState) -> bool>(rule: T, s: &str) -> bool {
+        println!("About to parse: {}", s);
         let mut p = ParseState::new(s);
-        rule(&mut p)
+        rule(&mut p) && p.eoi()
     }
 
     #[test]
@@ -82,6 +99,8 @@ mod tests {
         assert!(fake_csv(&mut p));
         let mut p = ParseState::new("foo,foo\nfoo;foo");
         assert!(!fake_csv(&mut p));
+        let mut p = ParseState::new("foo,foo\nfoo,foo\njunk at end");
+        assert!(!fake_csv(&mut p));
     }
 
     #[test]
@@ -131,4 +150,17 @@ mod tests {
         assert!(!parse(nested_choice, "(c)"));
         assert!(!parse(nested_choice, "(d)"));
     }
+
+    #[test]
+    fn test_quoted() {
+        assert!(parse(quoted, r#""""#));
+        assert!(parse(quoted, r#""foo""#));
+        assert!(parse(quoted, r#""\"""#));
+        assert!(parse(quoted, r#""foo \"bar\" baz""#));
+        assert!(parse(quoted, r#""C:\\>""#));
+
+        assert!(!parse(quoted, r#""no end"#));
+        assert!(!parse(quoted, r#""foo " bar""#));
+        assert!(!parse(quoted, r#""false end \""#));
+    }
 }