From d72e13c7c8535ddd9406af70f189bb62313deeff Mon Sep 17 00:00:00 2001
From: KtorZ <matthias.benkort@gmail.com>
Date: Sat, 18 Feb 2023 10:55:39 +0100
Subject: [PATCH] Emit parse error when finding a ByteArray literal instead of
 String literal.

---
 crates/aiken-lang/src/parser.rs       | 64 +++++++++++++++++++++++++++
 crates/aiken-lang/src/parser/error.rs | 37 ++++++++++++++++
 2 files changed, 101 insertions(+)

diff --git a/crates/aiken-lang/src/parser.rs b/crates/aiken-lang/src/parser.rs
index df6a143c..1e14b1e0 100644
--- a/crates/aiken-lang/src/parser.rs
+++ b/crates/aiken-lang/src/parser.rs
@@ -533,12 +533,70 @@ pub fn anon_fn_param_parser() -> impl Parser<Token, ast::UntypedArg, Error = Par
     })
 }
 
+fn unexpected_bytearray_literal(
+    token: Token,
+    expr: &expr::UntypedExpr,
+) -> Option<(ParseError, String)> {
+    match expr {
+        expr::UntypedExpr::ByteArray {
+            bytes,
+            preferred_format: ByteArrayFormatPreference::Utf8String,
+            location,
+            ..
+        } => {
+            let literal = String::from_utf8(bytes.clone()).unwrap();
+
+            Some((
+                ParseError::unexpected_bytearray_literal(*location, token, literal.clone()),
+                literal,
+            ))
+        }
+        _ => None,
+    }
+}
+
+fn validate_error_todo(
+    token: Token,
+    reason: Option<expr::UntypedExpr>,
+    emit: &mut dyn FnMut(ParseError),
+) -> Option<expr::UntypedExpr> {
+    if let Some(reason) = reason {
+        match unexpected_bytearray_literal(token, &reason) {
+            None => Some(reason),
+            Some((err, value)) => {
+                emit(err);
+                Some(expr::UntypedExpr::String {
+                    location: reason.location(),
+                    value,
+                })
+            }
+        }
+    } else {
+        reason
+    }
+}
+
 pub fn expr_seq_parser() -> impl Parser<Token, expr::UntypedExpr, Error = ParseError> {
     recursive(|r| {
         choice((
             just(Token::Trace)
                 .ignore_then(expr_parser(r.clone()))
                 .then(r.clone())
+                .validate(|(text, then), _span, emit| {
+                    match unexpected_bytearray_literal(Token::Trace, &text) {
+                        None => (text, then),
+                        Some((err, value)) => {
+                            emit(err);
+                            (
+                                expr::UntypedExpr::String {
+                                    location: text.location(),
+                                    value,
+                                },
+                                then,
+                            )
+                        }
+                    }
+                })
                 .map_with_span(|(text, then_), span| expr::UntypedExpr::Trace {
                     kind: TraceKind::Trace,
                     location: span,
@@ -547,9 +605,11 @@ pub fn expr_seq_parser() -> impl Parser<Token, expr::UntypedExpr, Error = ParseE
                 }),
             just(Token::ErrorTerm)
                 .ignore_then(expr_parser(r.clone()).or_not())
+                .validate(|reason, _span, emit| validate_error_todo(Token::ErrorTerm, reason, emit))
                 .map_with_span(|reason, span| expr::UntypedExpr::error(span, reason)),
             just(Token::Todo)
                 .ignore_then(expr_parser(r.clone()).or_not())
+                .validate(|reason, _span, emit| validate_error_todo(Token::Todo, reason, emit))
                 .map_with_span(|reason, span| expr::UntypedExpr::todo(span, reason)),
             expr_parser(r.clone())
                 .then(r.repeated())
@@ -924,6 +984,7 @@ pub fn expr_parser(
                             .then_ignore(one_of(Token::RArrow).not().rewind())
                             .or_not(),
                     )
+                    .validate(|reason, _span, emit| validate_error_todo(Token::Todo, reason, emit))
                     .map_with_span(|reason, span| expr::UntypedExpr::todo(span, reason)),
                 just(Token::ErrorTerm)
                     .ignore_then(
@@ -931,6 +992,9 @@ pub fn expr_parser(
                             .then_ignore(just(Token::RArrow).not().rewind())
                             .or_not(),
                     )
+                    .validate(|reason, _span, emit| {
+                        validate_error_todo(Token::ErrorTerm, reason, emit)
+                    })
                     .map_with_span(|reason, span| expr::UntypedExpr::error(span, reason)),
             )))
             .map_with_span(
diff --git a/crates/aiken-lang/src/parser/error.rs b/crates/aiken-lang/src/parser/error.rs
index 4b2d1055..7893244b 100644
--- a/crates/aiken-lang/src/parser/error.rs
+++ b/crates/aiken-lang/src/parser/error.rs
@@ -55,6 +55,16 @@ impl ParseError {
             label: None,
         }
     }
+
+    pub fn unexpected_bytearray_literal(span: Span, keyword: Token, literal: String) -> Self {
+        Self {
+            kind: ErrorKind::UnexpectedByteArrayLiteral(keyword, literal),
+            span,
+            while_parsing: None,
+            expected: HashSet::new(),
+            label: Some("use @"),
+        }
+    }
 }
 
 impl PartialEq for ParseError {
@@ -102,15 +112,18 @@ impl<T: Into<Pattern>> chumsky::Error<T> for ParseError {
 pub enum ErrorKind {
     #[error("I arrived at the end of the file unexpectedly.")]
     UnexpectedEnd,
+
     #[error("{0}")]
     #[diagnostic(help("{}", .0.help().unwrap_or_else(|| Box::new(""))))]
     Unexpected(Pattern),
+
     #[error("I discovered an invalid tuple index.")]
     #[diagnostic()]
     InvalidTupleIndex {
         #[help]
         hint: Option<String>,
     },
+
     #[error("I tripped over a malformed base16-encoded string literal.")]
     #[diagnostic(help("{}", formatdoc! {
         r#"You can declare literal bytearrays from base16-encoded (a.k.a. hexadecimal) string literals.
@@ -123,6 +136,7 @@ pub enum ErrorKind {
         "#
     , "pub const".bright_blue(), "=".yellow(), "\"f4c9f9c4252d86702c2f4c2e49e6648c7cffe3c8f2b6b7d779788f50\"".bright_purple()}))]
     MalformedBase16StringLiteral,
+
     #[error("I failed to understand a when clause guard.")]
     #[diagnostic(url("https://aiken-lang.org/language-tour/control-flow#checking-equality-and-ordering-in-patterns"))]
     #[diagnostic(help("{}", formatdoc! {
@@ -145,6 +159,29 @@ pub enum ErrorKind {
         , bad = "✖️".red()
     }))]
     InvalidWhenClause,
+
+    #[error(
+        "I noticed a {} literal where I would expect a {}.",
+        "ByteArray".bold().bright_blue(),
+        "String".bold().bright_blue(),
+    )]
+    #[diagnostic(url("https://aiken-lang.org/language-tour/primitive-types#string"))]
+    #[diagnostic(help("{}", formatdoc! {
+        r#"The keyword {keyword} accepts a {type_String} as argument. A {type_String} literal is denoted by {syntax}. Simple double-quotes are interpreted as {type_ByteArray} of UTF-8 encoded bytes. Juggling between {type_ByteArray} and {type_String} can be a little confusing.
+
+           In this instance, you probably meant to write the following:
+
+           ╰─▶ {keyword} {symbol_at}{symbol_doublequotes}{literal}{symbol_doublequotes}
+        "#,
+        type_String = "String".bold().bright_blue(),
+        type_ByteArray = "ByteArray".bold().bright_blue(),
+        symbol_at = "@".purple(),
+        symbol_doublequotes = "\"".purple(),
+        syntax = format!("{}{}", "@".purple(), "\"...\"".purple()),
+        keyword = format!("{}", .0).replace('"', "").bold(),
+        literal = .1.purple(),
+    }))]
+    UnexpectedByteArrayLiteral(Token, String),
 }
 
 #[derive(Debug, PartialEq, Eq, Hash, Diagnostic, thiserror::Error)]