From e72d617f4b2a4e55bda379185e6dcf07ca97b985 Mon Sep 17 00:00:00 2001
From: Dhruv Manilawala <dhruvmanila@gmail.com>
Date: Sun, 1 Oct 2023 07:37:59 +0530
Subject: [PATCH] Remove escaped mac/windows eol from AST string value (#7724)

## Summary

This PR fixes the bug where the value of a string node type includes the
escaped mac/windows newline character.

Note that the token value still includes them, it's only removed when
parsing the string content.

## Test Plan

Add new test cases for the string node type to check that the escapes
aren't being included in the string value.

fixes: #7723
---
 ..._tests__string_parser_escaped_mac_eol.snap | 23 ++++++++++++
 ...tests__string_parser_escaped_unix_eol.snap | 23 ++++++++++++
 ...ts__string_parser_escaped_windows_eol.snap | 23 ++++++++++++
 crates/ruff_python_parser/src/string.rs       | 35 ++++++++++++++++++-
 4 files changed, 103 insertions(+), 1 deletion(-)
 create mode 100644 crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_mac_eol.snap
 create mode 100644 crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_unix_eol.snap
 create mode 100644 crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_windows_eol.snap

diff --git a/crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_mac_eol.snap b/crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_mac_eol.snap
new file mode 100644
index 0000000000000..451aae8765f97
--- /dev/null
+++ b/crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_mac_eol.snap
@@ -0,0 +1,23 @@
+---
+source: crates/ruff_python_parser/src/string.rs
+expression: parse_ast
+---
+[
+    Expr(
+        StmtExpr {
+            range: 0..18,
+            value: Constant(
+                ExprConstant {
+                    range: 0..18,
+                    value: Str(
+                        StringConstant {
+                            value: "text more text",
+                            unicode: false,
+                            implicit_concatenated: false,
+                        },
+                    ),
+                },
+            ),
+        },
+    ),
+]
diff --git a/crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_unix_eol.snap b/crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_unix_eol.snap
new file mode 100644
index 0000000000000..451aae8765f97
--- /dev/null
+++ b/crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_unix_eol.snap
@@ -0,0 +1,23 @@
+---
+source: crates/ruff_python_parser/src/string.rs
+expression: parse_ast
+---
+[
+    Expr(
+        StmtExpr {
+            range: 0..18,
+            value: Constant(
+                ExprConstant {
+                    range: 0..18,
+                    value: Str(
+                        StringConstant {
+                            value: "text more text",
+                            unicode: false,
+                            implicit_concatenated: false,
+                        },
+                    ),
+                },
+            ),
+        },
+    ),
+]
diff --git a/crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_windows_eol.snap b/crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_windows_eol.snap
new file mode 100644
index 0000000000000..9b839986a3f55
--- /dev/null
+++ b/crates/ruff_python_parser/src/snapshots/ruff_python_parser__string__tests__string_parser_escaped_windows_eol.snap
@@ -0,0 +1,23 @@
+---
+source: crates/ruff_python_parser/src/string.rs
+expression: parse_ast
+---
+[
+    Expr(
+        StmtExpr {
+            range: 0..19,
+            value: Constant(
+                ExprConstant {
+                    range: 0..19,
+                    value: Str(
+                        StringConstant {
+                            value: "text more text",
+                            unicode: false,
+                            implicit_concatenated: false,
+                        },
+                    ),
+                },
+            ),
+        },
+    ),
+]
diff --git a/crates/ruff_python_parser/src/string.rs b/crates/ruff_python_parser/src/string.rs
index 0b2ecd28a84c2..81cbe2f9a1ce3 100644
--- a/crates/ruff_python_parser/src/string.rs
+++ b/crates/ruff_python_parser/src/string.rs
@@ -178,6 +178,12 @@ impl<'a> StringParser<'a> {
                     'N' if !self.kind.is_any_bytes() => self.parse_unicode_name()?,
                     // Special cases where the escape sequence is not a single character
                     '\n' => return Ok(String::new()),
+                    '\r' => {
+                        if self.peek() == Some('\n') {
+                            self.next_char();
+                        }
+                        return Ok(String::new());
+                    }
                     c => {
                         if self.kind.is_any_bytes() && !c.is_ascii() {
                             return Err(LexicalError {
@@ -558,10 +564,37 @@ impl From<FStringError> for crate::parser::LalrpopError<TextSize, Tok, LexicalEr
 mod tests {
     use crate::lexer::LexicalErrorType;
     use crate::parser::parse_suite;
-    use crate::ParseErrorType;
+    use crate::{ParseErrorType, Suite};
 
     use super::*;
 
+    const WINDOWS_EOL: &str = "\r\n";
+    const MAC_EOL: &str = "\r";
+    const UNIX_EOL: &str = "\n";
+
+    fn string_parser_escaped_eol(eol: &str) -> Suite {
+        let source = format!(r"'text \{eol}more text'");
+        parse_suite(&source, "<test>").unwrap()
+    }
+
+    #[test]
+    fn test_string_parser_escaped_unix_eol() {
+        let parse_ast = string_parser_escaped_eol(UNIX_EOL);
+        insta::assert_debug_snapshot!(parse_ast);
+    }
+
+    #[test]
+    fn test_string_parser_escaped_mac_eol() {
+        let parse_ast = string_parser_escaped_eol(MAC_EOL);
+        insta::assert_debug_snapshot!(parse_ast);
+    }
+
+    #[test]
+    fn test_string_parser_escaped_windows_eol() {
+        let parse_ast = string_parser_escaped_eol(WINDOWS_EOL);
+        insta::assert_debug_snapshot!(parse_ast);
+    }
+
     #[test]
     fn test_parse_fstring() {
         let source = r#"f"{a}{ b }{{foo}}""#;