test: Add test for query which blocks on chumsky (#1982)

* test: Add test for query which blocks on chumsky

Based on #1978.

It will block tests, probably until timeout

* Add a very minimal test

prql-compiler/src/parser/lexer.rs

@@ -77,8 +77,10 @@ pub fn lexer() -> impl Parser<char, Vec<(Token, std::ops::Range<usize>)>, Error
         literal,
         keyword,
         ident,
-    ))
-    .recover_with(skip_then_retry_until([]).skip_start());
+    ));
+    // TODO: Add this back when https://github.com/zesterer/chumsky/issues/301
+    // is fixed.
+    // .recover_with(skip_then_retry_until([]).skip_start());
 
     let comment = just('#').then(none_of('\n').repeated());
     let comments = comment

prql-compiler/src/parser/mod.rs

@@ -54,7 +54,13 @@ pub fn parse(source: &str) -> Result<Vec<Stmt>> {
 }
 
 fn convert_lexer_error(source: &str, e: Cheap<char>) -> Error {
-    let found = source[e.span()].to_string();
+    // TODO: is there a neater way of taking a span? We want to take it based on
+    // the chars, not the bytes, so can't just index into the str.
+    let found = source
+        .chars()
+        .skip(e.span().start)
+        .take(e.span().end() - e.span().start)
+        .collect();
     let span = common::into_span(e.span());
 
     Error::new(Reason::Unexpected { found }).with_span(span)
@@ -164,7 +170,7 @@ mod test {
 
     use super::*;
     use anyhow::anyhow;
-    use insta::assert_yaml_snapshot;
+    use insta::{assert_debug_snapshot, assert_yaml_snapshot};
 
     fn parse_expr(source: &str) -> Result<Expr, Vec<anyhow::Error>> {
         let tokens = Parser::parse(&lexer::lexer(), source).map_err(|errs| {
@@ -2206,4 +2212,33 @@ join s=salaries [==id]
         Param: 2_any_text
         "###);
     }
+
+    #[test]
+    fn test_error_unicode_string() {
+        // Test various unicode strings successfully parse errors. We were
+        // getting loops in the lexer before.
+        parse("s’ ").unwrap_err();
+        parse("s’").unwrap_err();
+        parse(" s’").unwrap_err();
+        parse(" ’ s").unwrap_err();
+        parse("’s").unwrap_err();
+        parse("👍 s’").unwrap_err();
+
+        let source = "Mississippi has four S’s and four I’s.";
+        assert_debug_snapshot!(parse(source).unwrap_err(), @r###"
+        Errors(
+            [
+                Error {
+                    span: Some(
+                        span-chars-22-23,
+                    ),
+                    reason: Unexpected {
+                        found: "’",
+                    },
+                    help: None,
+                },
+            ],
+        )
+        "###);
+    }
 }

prql-compiler/src/test.rs

@@ -2939,6 +2939,16 @@ fn test_errors() {
        ·           ╰─── `take` expected int or range, but found 1.8
     ───╯
     "###);
+
+    assert_display_snapshot!(compile("Mississippi has four S’s and four I’s.").unwrap_err(), @r###"
+    Error:
+       ╭─[:1:23]
+       │
+     1 │ Mississippi has four S’s and four I’s.
+       ·                       ┬
+       ·                       ╰── unexpected ’
+    ───╯
+    "###);
 }
 
 #[test]