From edf799904e34b8349b568fbf8783fa3c7207147c Mon Sep 17 00:00:00 2001
From: Artemy Egorov <artegoser@gmail.com>
Date: Wed, 7 Aug 2024 22:41:55 +0300
Subject: [PATCH] feat: symbols, text, number lexer

---
 src/daleth/lexer/mod.rs                 | 68 +++++++++++++++++++++++++
 src/daleth/{lexer.rs => lexer/types.rs} |  8 ++-
 2 files changed, 75 insertions(+), 1 deletion(-)
 create mode 100644 src/daleth/lexer/mod.rs
 rename src/daleth/{lexer.rs => lexer/types.rs} (87%)

diff --git a/src/daleth/lexer/mod.rs b/src/daleth/lexer/mod.rs
new file mode 100644
index 0000000..5960d08
--- /dev/null
+++ b/src/daleth/lexer/mod.rs
@@ -0,0 +1,68 @@
+use chumsky::prelude::*;
+use types::{Span, Spanned, Token};
+pub mod types;
+
+pub fn lexer<'src>(
+) -> impl Parser<'src, &'src str, Vec<Spanned<Token<'src>>>, extra::Err<Rich<'src, char, Span>>> {
+    let symbol = choice((
+        just("(").to(Token::LParen).labelled("LParen"),
+        just(")").to(Token::RParen).labelled("RParen"),
+        just("{").to(Token::LAngle).labelled("LAngle"),
+        just("}").to(Token::RAngle).labelled("RAngle"),
+        just("[").to(Token::LSquare).labelled("LSquare"),
+        just("]").to(Token::RSquare).labelled("RSquare"),
+        just(":").to(Token::Colon).labelled("Colon"),
+    ))
+    .labelled("symbol");
+
+    let number = text::int(10)
+        .from_str()
+        .unwrapped()
+        .map(Token::Number)
+        .labelled("number");
+
+    let textual = {
+        let escape = just('\\')
+            .ignore_then(choice((
+                just("\\`").to('`'.to_owned()),
+                just("\\]").to(']'.to_owned()),
+            )))
+            .labelled("escape sequence");
+
+        let text = none_of("]\n")
+            .or(escape.clone())
+            .repeated()
+            .to_slice()
+            .map(Token::Text);
+
+        let multiline_text = none_of("`").or(escape).repeated();
+
+        let mltext = multiline_text
+            .clone()
+            .delimited_by(just('`'), just('`'))
+            .to_slice()
+            .map(Token::MLText)
+            .labelled("multiline text");
+
+        let mlmstext = multiline_text
+            .delimited_by(just("`#"), just('`'))
+            .to_slice()
+            .map(Token::RMLText)
+            .labelled("raw multiline text");
+
+        choice((mltext, mlmstext, text))
+    };
+
+    let comment = just("#")
+        .then(none_of("\n").repeated())
+        .to_slice()
+        .map(Token::Comment);
+
+    let token = choice((symbol, number, textual, comment));
+
+    token
+        .map_with(|t, e| (t, e.span()))
+        .padded()
+        .repeated()
+        .collect()
+}
diff --git a/src/daleth/lexer.rs b/src/daleth/lexer/types.rs
similarity index 87%
rename from src/daleth/lexer.rs
rename to src/daleth/lexer/types.rs
index 936e0dc..9e2a903 100644
--- a/src/daleth/lexer.rs
+++ b/src/daleth/lexer/types.rs
@@ -1,10 +1,11 @@
 use chumsky::prelude::*;
 
 pub type Span = SimpleSpan<usize>;
+pub type Spanned<T> = (T, Span);
 
 #[derive(Clone, Debug, PartialEq)]
 pub enum Token<'src> {
-    // Brackets
+    // Symbols
     /// (
     LParen,
     /// )
@@ -17,6 +18,8 @@ pub enum Token<'src> {
     LSquare,
     /// ]
     RSquare,
+    /// :
+    Colon,
 
     // Values
     Number(u8),
@@ -28,6 +31,9 @@ pub enum Token<'src> {
     /// Raw Multi line text
     RMLText(&'src str),
 
+    /// Special
+    Comment(&'src str),
+
     // Tags
     El,
     H,