Reapply "For # to start a comment, then it either need to be the first chara…" (#14606)

This reverts commit d94b344342.

crates/nu-parser/src/lex.rs

@@ -51,7 +51,7 @@ impl BlockKind {
 }
 
 // A baseline token is terminated if it's not nested inside of a paired
-// delimiter and the next character is one of: `|`, `;`, `#` or any
+// delimiter and the next character is one of: `|`, `;` or any
 // whitespace.
 fn is_item_terminator(
     block_level: &[BlockKind],
@@ -115,6 +115,7 @@ pub fn lex_item(
     //   character (whitespace, `|`, `;` or `#`) is encountered, the baseline
     //   token is done.
     // - Otherwise, accumulate the character into the current baseline token.
+    let mut previous_char = None;
     while let Some(c) = input.get(*curr_offset) {
         let c = *c;
 
@@ -147,11 +148,9 @@ pub fn lex_item(
                 // Also need to check to make sure we aren't escaped
                 quote_start = None;
             }
-        } else if c == b'#' {
-            if is_item_terminator(&block_level, c, additional_whitespace, special_tokens) {
-                break;
-            }
-            in_comment = true;
+        } else if c == b'#' && !in_comment {
+            // To start a comment, It either need to be the first character of the token or prefixed with space.
+            in_comment = previous_char.map(|pc| pc == b' ').unwrap_or(true);
         } else if c == b'\n' || c == b'\r' {
             in_comment = false;
             if is_item_terminator(&block_level, c, additional_whitespace, special_tokens) {
@@ -254,6 +253,7 @@ pub fn lex_item(
         }
 
         *curr_offset += 1;
+        previous_char = Some(c);
     }
 
     let span = Span::new(span_offset + token_start, span_offset + *curr_offset);

crates/nu-parser/tests/test_lex.rs

@@ -159,6 +159,29 @@ fn lex_comment() {
     );
 }
 
+#[test]
+fn lex_not_comment_needs_space_in_front_of_hashtag() {
+    let file = b"1..10 | each {echo test#testing }";
+
+    let output = lex(file, 0, &[], &[], false);
+
+    assert!(output.1.is_none());
+}
+
+#[test]
+fn lex_comment_with_space_in_front_of_hashtag() {
+    let file = b"1..10 | each {echo test #testing }";
+
+    let output = lex(file, 0, &[], &[], false);
+
+    assert!(output.1.is_some());
+    assert!(matches!(
+        output.1.unwrap(),
+        ParseError::UnexpectedEof(missing_token, span) if missing_token == "}"
+            && span == Span::new(33, 34)
+    ));
+}
+
 #[test]
 fn lex_is_incomplete() {
     let file = b"let x = 300 | ;";

tests/repl/test_parser.rs

@@ -169,6 +169,41 @@ fn comment_skipping_in_pipeline_3() -> TestResult {
     )
 }
 
+#[test]
+fn still_string_if_hashtag_is_middle_of_string() -> TestResult {
+    run_test(r#"echo test#testing"#, "test#testing")
+}
+
+#[test]
+fn non_comment_hashtag_in_comment_does_not_stop_comment() -> TestResult {
+    run_test(r#"# command_bar_text: { fg: '#C4C9C6' },"#, "")
+}
+
+#[test]
+fn non_comment_hashtag_in_comment_does_not_stop_comment_in_block() -> TestResult {
+    run_test(
+        r#"{
+        explore: {
+            # command_bar_text: { fg: '#C4C9C6' },
+        }
+    } | get explore | is-empty"#,
+        "true",
+    )
+}
+
+#[test]
+fn still_string_if_hashtag_is_middle_of_string_inside_each() -> TestResult {
+    run_test(
+        r#"1..1 | each {echo test#testing } | get 0"#,
+        "test#testing",
+    )
+}
+
+#[test]
+fn still_string_if_hashtag_is_middle_of_string_inside_each_also_with_dot() -> TestResult {
+    run_test(r#"1..1 | each {echo '.#testing' } | get 0"#, ".#testing")
+}
+
 #[test]
 fn bad_var_name() -> TestResult {
     fail_test(r#"let $"foo bar" = 4"#, "can't contain")
@@ -282,6 +317,11 @@ fn raw_string_with_equals() -> TestResult {
     )
 }
 
+#[test]
+fn raw_string_with_hashtag() -> TestResult {
+    run_test(r#"r##' one # two '##"#, "one # two")
+}
+
 #[test]
 fn list_quotes_with_equals() -> TestResult {
     run_test(