[`ISC001`] fix panic when string literals are unclosed (#21034)

Co-authored-by: Micha Reiser <micha@reiser.io>

crates/ruff_linter/resources/test/fixtures/flake8_implicit_str_concat/ISC_syntax_error_2.py

@@ -0,0 +1,7 @@
+# Regression test for https://github.com/astral-sh/ruff/issues/21023
+'' '
+"" ""
+'' '' '
+"" "" "
+f"" f"
+f"" f"" f"

crates/ruff_linter/src/rules/flake8_implicit_str_concat/mod.rs

@@ -23,6 +23,14 @@ mod tests {
         Rule::MultiLineImplicitStringConcatenation,
         Path::new("ISC_syntax_error.py")
     )]
+    #[test_case(
+        Rule::SingleLineImplicitStringConcatenation,
+        Path::new("ISC_syntax_error_2.py")
+    )]
+    #[test_case(
+        Rule::MultiLineImplicitStringConcatenation,
+        Path::new("ISC_syntax_error_2.py")
+    )]
     #[test_case(Rule::ExplicitStringConcatenation, Path::new("ISC.py"))]
     fn rules(rule_code: Rule, path: &Path) -> Result<()> {
         let snapshot = format!("{}_{}", rule_code.noqa_code(), path.to_string_lossy());

crates/ruff_linter/src/rules/flake8_implicit_str_concat/rules/implicit.rs

@@ -1,13 +1,12 @@
 use std::borrow::Cow;
 
 use itertools::Itertools;
-
 use ruff_macros::{ViolationMetadata, derive_message_formats};
-use ruff_python_ast::str::{leading_quote, trailing_quote};
+use ruff_python_ast::StringFlags;
 use ruff_python_index::Indexer;
-use ruff_python_parser::{TokenKind, Tokens};
+use ruff_python_parser::{Token, TokenKind, Tokens};
 use ruff_source_file::LineRanges;
-use ruff_text_size::{Ranged, TextRange};
+use ruff_text_size::{Ranged, TextLen, TextRange};
 
 use crate::Locator;
 use crate::checkers::ast::LintContext;
@@ -169,7 +168,8 @@ pub(crate) fn implicit(
                 SingleLineImplicitStringConcatenation,
                 TextRange::new(a_range.start(), b_range.end()),
             ) {
-                if let Some(fix) = concatenate_strings(a_range, b_range, locator) {
+                if let Some(fix) = concatenate_strings(a_token, b_token, a_range, b_range, locator)
+                {
                     diagnostic.set_fix(fix);
                 }
             }
@@ -177,38 +177,55 @@ pub(crate) fn implicit(
     }
 }
 
-fn concatenate_strings(a_range: TextRange, b_range: TextRange, locator: &Locator) -> Option<Fix> {
+/// Concatenates two strings
-    let a_text = locator.slice(a_range);
+///
-    let b_text = locator.slice(b_range);
+/// The `a_string_range` and `b_string_range` are the range of the entire string,
-
+/// not just of the string token itself (important for interpolated strings where
-    let a_leading_quote = leading_quote(a_text)?;
+/// the start token doesn't span the entire token).
-    let b_leading_quote = leading_quote(b_text)?;
+fn concatenate_strings(
-
+    a_token: &Token,
-    // Require, for now, that the leading quotes are the same.
+    b_token: &Token,
-    if a_leading_quote != b_leading_quote {
+    a_string_range: TextRange,
+    b_string_range: TextRange,
+    locator: &Locator,
+) -> Option<Fix> {
+    if a_token.string_flags()?.is_unclosed() || b_token.string_flags()?.is_unclosed() {
         return None;
     }
 
-    let a_trailing_quote = trailing_quote(a_text)?;
+    let a_string_flags = a_token.string_flags()?;
-    let b_trailing_quote = trailing_quote(b_text)?;
+    let b_string_flags = b_token.string_flags()?;
 
-    // Require, for now, that the trailing quotes are the same.
+    let a_prefix = a_string_flags.prefix();
-    if a_trailing_quote != b_trailing_quote {
+    let b_prefix = b_string_flags.prefix();
+
+    // Require, for now, that the strings have the same prefix,
+    // quote style, and number of quotes
+    if a_prefix != b_prefix
+        || a_string_flags.quote_style() != b_string_flags.quote_style()
+        || a_string_flags.is_triple_quoted() != b_string_flags.is_triple_quoted()
+    {
         return None;
     }
 
+    let a_text = locator.slice(a_string_range);
+    let b_text = locator.slice(b_string_range);
+
+    let quotes = a_string_flags.quote_str();
+
+    let opener_len = a_string_flags.opener_len();
+    let closer_len = a_string_flags.closer_len();
+
     let mut a_body =
-        Cow::Borrowed(&a_text[a_leading_quote.len()..a_text.len() - a_trailing_quote.len()]);
+        Cow::Borrowed(&a_text[TextRange::new(opener_len, a_text.text_len() - closer_len)]);
-    let b_body = &b_text[b_leading_quote.len()..b_text.len() - b_trailing_quote.len()];
+    let b_body = &b_text[TextRange::new(opener_len, b_text.text_len() - closer_len)];
 
-    if a_leading_quote.find(['r', 'R']).is_none()
+    if !a_string_flags.is_raw_string() && matches!(b_body.bytes().next(), Some(b'0'..=b'7')) {
-        && matches!(b_body.bytes().next(), Some(b'0'..=b'7'))
-    {
         normalize_ending_octal(&mut a_body);
     }
 
-    let concatenation = format!("{a_leading_quote}{a_body}{b_body}{a_trailing_quote}");
+    let concatenation = format!("{a_prefix}{quotes}{a_body}{b_body}{quotes}");
-    let range = TextRange::new(a_range.start(), b_range.end());
+    let range = TextRange::new(a_string_range.start(), b_string_range.end());
 
     Some(Fix::safe_edit(Edit::range_replacement(
         concatenation,

crates/ruff_linter/src/rules/flake8_implicit_str_concat/snapshots/ruff_linter__rules__flake8_implicit_str_concat__tests__ISC001_ISC_syntax_error_2.py.snap

@@ -0,0 +1,134 @@
+---
+source: crates/ruff_linter/src/rules/flake8_implicit_str_concat/mod.rs
+---
+ISC001 Implicitly concatenated string literals on one line
+ --> ISC_syntax_error_2.py:2:1
+  |
+1 | # Regression test for https://github.com/astral-sh/ruff/issues/21023
+2 | '' '
+  | ^^^^
+3 | "" ""
+4 | '' '' '
+  |
+help: Combine string literals
+
+invalid-syntax: missing closing quote in string literal
+ --> ISC_syntax_error_2.py:2:4
+  |
+1 | # Regression test for https://github.com/astral-sh/ruff/issues/21023
+2 | '' '
+  |    ^
+3 | "" ""
+4 | '' '' '
+  |
+
+ISC001 Implicitly concatenated string literals on one line
+ --> ISC_syntax_error_2.py:3:1
+  |
+1 | # Regression test for https://github.com/astral-sh/ruff/issues/21023
+2 | '' '
+3 | "" ""
+  | ^^^^^
+4 | '' '' '
+5 | "" "" "
+  |
+help: Combine string literals
+
+ISC001 Implicitly concatenated string literals on one line
+ --> ISC_syntax_error_2.py:4:1
+  |
+2 | '' '
+3 | "" ""
+4 | '' '' '
+  | ^^^^^
+5 | "" "" "
+6 | f"" f"
+  |
+help: Combine string literals
+
+ISC001 Implicitly concatenated string literals on one line
+ --> ISC_syntax_error_2.py:4:4
+  |
+2 | '' '
+3 | "" ""
+4 | '' '' '
+  |    ^^^^
+5 | "" "" "
+6 | f"" f"
+  |
+help: Combine string literals
+
+invalid-syntax: missing closing quote in string literal
+ --> ISC_syntax_error_2.py:4:7
+  |
+2 | '' '
+3 | "" ""
+4 | '' '' '
+  |       ^
+5 | "" "" "
+6 | f"" f"
+  |
+
+ISC001 Implicitly concatenated string literals on one line
+ --> ISC_syntax_error_2.py:5:1
+  |
+3 | "" ""
+4 | '' '' '
+5 | "" "" "
+  | ^^^^^
+6 | f"" f"
+7 | f"" f"" f"
+  |
+help: Combine string literals
+
+ISC001 Implicitly concatenated string literals on one line
+ --> ISC_syntax_error_2.py:5:4
+  |
+3 | "" ""
+4 | '' '' '
+5 | "" "" "
+  |    ^^^^
+6 | f"" f"
+7 | f"" f"" f"
+  |
+help: Combine string literals
+
+invalid-syntax: missing closing quote in string literal
+ --> ISC_syntax_error_2.py:5:7
+  |
+3 | "" ""
+4 | '' '' '
+5 | "" "" "
+  |       ^
+6 | f"" f"
+7 | f"" f"" f"
+  |
+
+invalid-syntax: f-string: unterminated string
+ --> ISC_syntax_error_2.py:6:7
+  |
+4 | '' '' '
+5 | "" "" "
+6 | f"" f"
+  |       ^
+7 | f"" f"" f"
+  |
+
+ISC001 Implicitly concatenated string literals on one line
+ --> ISC_syntax_error_2.py:7:1
+  |
+5 | "" "" "
+6 | f"" f"
+7 | f"" f"" f"
+  | ^^^^^^^
+  |
+help: Combine string literals
+
+invalid-syntax: f-string: unterminated string
+ --> ISC_syntax_error_2.py:7:11
+  |
+5 | "" "" "
+6 | f"" f"
+7 | f"" f"" f"
+  |           ^
+  |

crates/ruff_linter/src/rules/flake8_implicit_str_concat/snapshots/ruff_linter__rules__flake8_implicit_str_concat__tests__ISC002_ISC_syntax_error_2.py.snap

@@ -0,0 +1,53 @@
+---
+source: crates/ruff_linter/src/rules/flake8_implicit_str_concat/mod.rs
+---
+invalid-syntax: missing closing quote in string literal
+ --> ISC_syntax_error_2.py:2:4
+  |
+1 | # Regression test for https://github.com/astral-sh/ruff/issues/21023
+2 | '' '
+  |    ^
+3 | "" ""
+4 | '' '' '
+  |
+
+invalid-syntax: missing closing quote in string literal
+ --> ISC_syntax_error_2.py:4:7
+  |
+2 | '' '
+3 | "" ""
+4 | '' '' '
+  |       ^
+5 | "" "" "
+6 | f"" f"
+  |
+
+invalid-syntax: missing closing quote in string literal
+ --> ISC_syntax_error_2.py:5:7
+  |
+3 | "" ""
+4 | '' '' '
+5 | "" "" "
+  |       ^
+6 | f"" f"
+7 | f"" f"" f"
+  |
+
+invalid-syntax: f-string: unterminated string
+ --> ISC_syntax_error_2.py:6:7
+  |
+4 | '' '' '
+5 | "" "" "
+6 | f"" f"
+  |       ^
+7 | f"" f"" f"
+  |
+
+invalid-syntax: f-string: unterminated string
+ --> ISC_syntax_error_2.py:7:11
+  |
+5 | "" "" "
+6 | f"" f"
+7 | f"" f"" f"
+  |           ^
+  |