From 2ad770734906916e1b018620da1de776e1eafa59 Mon Sep 17 00:00:00 2001
From: Greg Shuflin <greg@everydayimshuflin.com>
Date: Sat, 27 Jan 2024 02:20:58 -0800
Subject: [PATCH] Repeat test

---
 src/combinators.rs | 142 ++++++++++++++++++++++++++++++++++++++++++---
 src/lib.rs         |  10 ++++
 2 files changed, 144 insertions(+), 8 deletions(-)

diff --git a/src/combinators.rs b/src/combinators.rs
index 4ffe931..f4913f6 100644
--- a/src/combinators.rs
+++ b/src/combinators.rs
@@ -2,19 +2,21 @@ use std::marker::PhantomData;
 
 use crate::{ParseResult, Parser};
 
-pub fn repeated<P, I, O, E>(parser: P) -> impl Parser<I, Vec<O>, E>
+pub fn repeated<P, I, O, E>(parser: P) -> Repeated<P, I, O, E>
 where
     P: Parser<I, O, E>,
 {
     Repeated::new(parser)
 }
 
-struct Repeated<P, I, O, E>
+pub struct Repeated<P, I, O, E>
 where
     P: Parser<I, O, E>,
 {
     inner_parser: P,
     phantom: PhantomData<(I, O, E)>,
+    at_least: Option<u32>,
+    at_most: Option<u32>,
 }
 
 impl<P, I, O, E> Repeated<P, I, O, E>
@@ -25,6 +27,31 @@ where
         Self {
             inner_parser,
             phantom: PhantomData,
+            at_least: None,
+            at_most: None,
+        }
+    }
+    pub fn at_least(self, at_least: u32) -> Self {
+        Self {
+            at_least: Some(at_least),
+            ..self
+        }
+    }
+    pub fn at_most(self, at_most: u32) -> Self {
+        Self {
+            at_most: Some(at_most),
+            ..self
+        }
+    }
+    pub fn separated_by<D>(self, delimiter: D) -> SeparatedBy<D, P, I, O, E>
+    where
+        D: Parser<I, (), E>,
+        E: Default,
+    {
+        SeparatedBy {
+            inner_repeated: self,
+            delimiter,
+            allow_trailing: false,
         }
     }
 }
@@ -32,24 +59,123 @@ where
 impl<P, I, O, E> Parser<I, Vec<O>, E> for Repeated<P, I, O, E>
 where
     P: Parser<I, O, E>,
+    E: Default,
 {
-    fn parse(&self, input: I) -> ParseResult<I, Vec<O>, E> {
-        let mut acc = input;
+    fn parse(&self, mut input: I) -> ParseResult<I, Vec<O>, E> {
+        let at_least = self.at_least.unwrap_or(0);
+        let at_most = self.at_most.unwrap_or(u32::MAX);
+
         let mut results = vec![];
+        let mut count = 0;
+
+        if at_most == 0 {
+            return Ok((vec![], input));
+        }
 
         loop {
-            match self.inner_parser.parse(acc) {
+            match self.inner_parser.parse(input) {
                 Ok((item, rest)) => {
                     results.push(item);
-                    acc = rest;
+                    input = rest;
+                    count += 1;
+                    if count >= at_most {
+                        break;
+                    }
                 }
                 Err((_err, rest)) => {
-                    acc = rest;
+                    input = rest;
                     break;
                 }
             }
         }
-        Ok((results, acc))
+
+        if count < at_least {
+            return Err((Default::default(), input));
+        }
+        Ok((results, input))
+    }
+}
+
+pub struct SeparatedBy<D, P, I, O, E>
+where
+    D: Parser<I, (), E>,
+    P: Parser<I, O, E>,
+    E: Default,
+{
+    inner_repeated: Repeated<P, I, O, E>,
+    delimiter: D,
+    allow_trailing: bool,
+}
+
+impl<D, P, I, O, E> Parser<I, Vec<O>, E> for SeparatedBy<D, P, I, O, E>
+where
+    D: Parser<I, (), E>,
+    P: Parser<I, O, E>,
+    E: Default,
+{
+    fn parse(&self, mut input: I) -> ParseResult<I, Vec<O>, E> {
+        let at_least = self.inner_repeated.at_least.unwrap_or(0);
+        let at_most = self.inner_repeated.at_most.unwrap_or(u32::MAX);
+        let inner = &self.inner_repeated.inner_parser;
+        let delimiter = &self.delimiter;
+
+        if at_most == 0 {
+            return Ok((vec![], input));
+        }
+
+        let mut results = Vec::new();
+        let mut count: u32 = 0;
+
+        match inner.parse(input) {
+            Ok((item, rest)) => {
+                results.push(item);
+                input = rest;
+            }
+            Err((err, rest)) => {
+                if at_least > 0 {
+                    return Err((err, rest));
+                } else {
+                    return Ok((vec![], rest));
+                }
+            }
+        }
+
+        loop {
+            match delimiter.parse(input) {
+                Ok(((), rest)) => {
+                    input = rest;
+                    count += 1;
+                }
+                Err((_err, rest)) => {
+                    input = rest;
+                    break;
+                }
+            }
+            match inner.parse(input) {
+                Ok((item, rest)) => {
+                    input = rest;
+                    results.push(item);
+                }
+                Err((err, rest)) => {
+                    if self.allow_trailing {
+                        input = rest;
+                        break;
+                    } else {
+                        return Err((err, rest));
+                    }
+                }
+            }
+            if count >= at_most {
+                break;
+            }
+        }
+
+        if count < at_least {
+            //return Err(??, rest) <- need to handle errors better
+            unimplemented!();
+        }
+
+        Ok((results, input))
     }
 }
 
diff --git a/src/lib.rs b/src/lib.rs
index 30ca740..f0d0478 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -87,4 +87,14 @@ mod tests {
         let output2 = parser.parse("alpha   beta").unwrap();
         assert_eq!(output2.0 .0, Some("alpha"));
     }
+
+    #[test]
+    fn test_repeated() {
+        let spaces = repeated(literal_char(' ')).at_least(1);
+        let bongo = literal("bongo");
+        let parser = repeated(bongo).separated_by(map(spaces, |_| ()));
+        let output = parser.parse("bongo bongo     bongo           bongo");
+        let output = output.unwrap();
+        assert_eq!(output.0, vec!["bongo", "bongo", "bongo", "bongo"]);
+    }
 }