Merge pull request #374 from Flakebi/fix-escape

Fix multiple escape sequences

Cargo.lock

@@ -1499,9 +1499,9 @@ checksum = "a7f741b240f1a48843f9b8e0444fb55fb2a4ff67293b50a9179dfd5ea67f8d41"
 
 [[package]]
 name = "trycmd"
-version = "0.4.0"
+version = "0.5.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0b15571a9c85f2dc93e93907813b8b82583a60c1ee738ef5fa5123f4c96863b5"
+checksum = "2d034538089e906ac14df42c19aae52a55aa1014102d6895d9748080e043cc48"
 dependencies = [
  "concolor-control",
  "difflib",

Cargo.toml

@@ -94,7 +94,7 @@ maplit = "1.0"
 
 [dev-dependencies]
 assert_fs = "1.0"
-trycmd = "0.4"
+trycmd = "0.5.1"
 criterion = "0.3"
 
 [profile.dev]

crates/typos/src/tokens.rs

@@ -190,8 +190,8 @@ mod parser {
             terminated(base64_literal, sep1),
             terminated(email_literal, sep1),
             terminated(url_literal, sep1),
-            terminated(c_escape, sep1),
-            terminated(printf, sep1),
+            c_escape,
+            printf,
             sep1,
         )))(input)
     }
@@ -410,6 +410,10 @@ mod parser {
         <T as nom::InputTakeAtPosition>::Item: AsChar + Copy,
         <T as nom::InputIter>::Item: AsChar + Copy,
     {
+        // We don't know whether the string we are parsing is a literal string (no escaping) or
+        // regular string that does escaping. The escaped letter might be part of a word, or it
+        // might not be. Rather than guess and be wrong part of the time and correct people's words
+        // incorrectly, we opt for just not evaluating it at all.
         preceded(take_while1(is_escape), take_while(is_xid_continue))(input)
     }
 
@@ -1103,6 +1107,36 @@ mod test {
         assert_eq!(expected, actual);
     }
 
+    #[test]
+    fn tokenize_double_escape() {
+        let parser = TokenizerBuilder::new().build();
+
+        let input = "Hello \\n\\n World";
+        let expected: Vec<Identifier> = vec![
+            Identifier::new_unchecked("Hello", Case::None, 0),
+            Identifier::new_unchecked("World", Case::None, 11),
+        ];
+        let actual: Vec<_> = parser.parse_bytes(input.as_bytes()).collect();
+        assert_eq!(expected, actual);
+        let actual: Vec<_> = parser.parse_str(input).collect();
+        assert_eq!(expected, actual);
+    }
+
+    #[test]
+    fn tokenize_ignore_escape() {
+        let parser = TokenizerBuilder::new().build();
+
+        let input = "Hello \\nanana\\nanana World";
+        let expected: Vec<Identifier> = vec![
+            Identifier::new_unchecked("Hello", Case::None, 0),
+            Identifier::new_unchecked("World", Case::None, 21),
+        ];
+        let actual: Vec<_> = parser.parse_bytes(input.as_bytes()).collect();
+        assert_eq!(expected, actual);
+        let actual: Vec<_> = parser.parse_str(input).collect();
+        assert_eq!(expected, actual);
+    }
+
     #[test]
     fn tokenize_printf() {
         let parser = TokenizerBuilder::new().build();

tests/cmd/double-escaped.stdin

@@ -0,0 +1,2 @@
+\n\n
+Destory

tests/cmd/double-escaped.stdout

@@ -0,0 +1,2 @@
+\n\n
+Destroy

tests/cmd/double-escaped.toml

@@ -0,0 +1,3 @@
+bin.name = "typos"
+args = "--write-changes -"
+status.code = 0