[flake8-implicit-str-concat] Normalize octals before merging concat…

…enated strings in `single-line-implicit-string-concatenation` (`ISC001`) (#13118) Co-authored-by: Alex Waygood <[email protected]>
astral-sh · Aug 27, 2024 · 483748c · 483748c
1 parent eb3dc37
commit 483748c
Show file tree

Hide file tree

Showing 5 changed files with 456 additions and 3 deletions.
diff --git a/crates/ruff_linter/resources/test/fixtures/flake8_implicit_str_concat/ISC.py b/crates/ruff_linter/resources/test/fixtures/flake8_implicit_str_concat/ISC.py
@@ -79,3 +79,15 @@
     + f"second"} d"
 _ = f"a {f"first {f"middle"}"
     + f"second"} d"
+
+# See https://github.com/astral-sh/ruff/issues/12936
+_ = "\12""0" # fix should be "\0120"
+_ = "\\12""0" # fix should be "\\120"
+_ = "\\\12""0" # fix should be "\\\0120"
+_ = "\12 0""0" # fix should be "\12 00"
+_ = r"\12"r"0" # fix should be r"\120"
+_ = "\12 and more""0" # fix should be "\12 and more0"
+_ = "\8""0" # fix should be "\80"
+_ = "\12""8" # fix should be "\128"
+_ = "\12""foo" # fix should be "\12foo"
+_ = "\12" ""  # fix should be "\12"
diff --git a/crates/ruff_linter/src/rules/flake8_implicit_str_concat/rules/implicit.rs b/crates/ruff_linter/src/rules/flake8_implicit_str_concat/rules/implicit.rs
@@ -1,3 +1,5 @@
+use std::borrow::Cow;
+
 use itertools::Itertools;
 
 use ruff_diagnostics::{Diagnostic, Edit, Fix, FixAvailability, Violation};
@@ -172,9 +174,16 @@ fn concatenate_strings(a_range: TextRange, b_range: TextRange, locator: &Locator
         return None;
     }
 
-    let a_body = &a_text[a_leading_quote.len()..a_text.len() - a_trailing_quote.len()];
+    let mut a_body =
+        Cow::Borrowed(&a_text[a_leading_quote.len()..a_text.len() - a_trailing_quote.len()]);
     let b_body = &b_text[b_leading_quote.len()..b_text.len() - b_trailing_quote.len()];
 
+    if a_leading_quote.find(['r', 'R']).is_none()
+        && 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 range = TextRange::new(a_range.start(), b_range.end());
 
@@ -183,3 +192,39 @@ fn concatenate_strings(a_range: TextRange, b_range: TextRange, locator: &Locator
         range,
     )))
 }
+
+/// Pads an octal at the end of the string
+/// to three digits, if necessary.
+fn normalize_ending_octal(text: &mut Cow<'_, str>) {
+    // Early return for short strings
+    if text.len() < 2 {
+        return;
+    }
+
+    let mut rev_bytes = text.bytes().rev();
+    if let Some(last_byte @ b'0'..=b'7') = rev_bytes.next() {
+        // "\y" -> "\00y"
+        if has_odd_consecutive_backslashes(&mut rev_bytes.clone()) {
+            let prefix = &text[..text.len() - 2];
+            *text = Cow::Owned(format!("{prefix}\\00{}", last_byte as char));
+        }
+        // "\xy" -> "\0xy"
+        else if let Some(penultimate_byte @ b'0'..=b'7') = rev_bytes.next() {
+            if has_odd_consecutive_backslashes(&mut rev_bytes.clone()) {
+                let prefix = &text[..text.len() - 3];
+                *text = Cow::Owned(format!(
+                    "{prefix}\\0{}{}",
+                    penultimate_byte as char, last_byte as char
+                ));
+            }
+        }
+    }
+}
+
+fn has_odd_consecutive_backslashes(mut itr: impl Iterator<Item = u8>) -> bool {
+    let mut odd_backslashes = false;
+    while let Some(b'\\') = itr.next() {
+        odd_backslashes = !odd_backslashes;
+    }
+    odd_backslashes
+}
diff --git a/...oncat/snapshots/ruff_linter__rules__flake8_implicit_str_concat__tests__ISC001_ISC.py.snap b/...oncat/snapshots/ruff_linter__rules__flake8_implicit_str_concat__tests__ISC001_ISC.py.snap
@@ -296,4 +296,202 @@ ISC.py:73:20: ISC001 [*] Implicitly concatenated string literals on one line
 75 75 |     f"def"} g"
 76 76 | 
 
+ISC.py:84:5: ISC001 [*] Implicitly concatenated string literals on one line
+   |
+83 | # See https://github.com/astral-sh/ruff/issues/12936
+84 | _ = "\12""0" # fix should be "\0120"
+   |     ^^^^^^^^ ISC001
+85 | _ = "\\12""0" # fix should be "\\120"
+86 | _ = "\\\12""0" # fix should be "\\\0120"
+   |
+   = help: Combine string literals
+
+ℹ Safe fix
+81 81 |     + f"second"} d"
+82 82 | 
+83 83 | # See https://github.com/astral-sh/ruff/issues/12936
+84    |-_ = "\12""0" # fix should be "\0120"
+   84 |+_ = "\0120" # fix should be "\0120"
+85 85 | _ = "\\12""0" # fix should be "\\120"
+86 86 | _ = "\\\12""0" # fix should be "\\\0120"
+87 87 | _ = "\12 0""0" # fix should be "\12 00"
+
+ISC.py:85:5: ISC001 [*] Implicitly concatenated string literals on one line
+   |
+83 | # See https://github.com/astral-sh/ruff/issues/12936
+84 | _ = "\12""0" # fix should be "\0120"
+85 | _ = "\\12""0" # fix should be "\\120"
+   |     ^^^^^^^^^ ISC001
+86 | _ = "\\\12""0" # fix should be "\\\0120"
+87 | _ = "\12 0""0" # fix should be "\12 00"
+   |
+   = help: Combine string literals
+
+ℹ Safe fix
+82 82 | 
+83 83 | # See https://github.com/astral-sh/ruff/issues/12936
+84 84 | _ = "\12""0" # fix should be "\0120"
+85    |-_ = "\\12""0" # fix should be "\\120"
+   85 |+_ = "\\120" # fix should be "\\120"
+86 86 | _ = "\\\12""0" # fix should be "\\\0120"
+87 87 | _ = "\12 0""0" # fix should be "\12 00"
+88 88 | _ = r"\12"r"0" # fix should be r"\120"
+
+ISC.py:86:5: ISC001 [*] Implicitly concatenated string literals on one line
+   |
+84 | _ = "\12""0" # fix should be "\0120"
+85 | _ = "\\12""0" # fix should be "\\120"
+86 | _ = "\\\12""0" # fix should be "\\\0120"
+   |     ^^^^^^^^^^ ISC001
+87 | _ = "\12 0""0" # fix should be "\12 00"
+88 | _ = r"\12"r"0" # fix should be r"\120"
+   |
+   = help: Combine string literals
 
+ℹ Safe fix
+83 83 | # See https://github.com/astral-sh/ruff/issues/12936
+84 84 | _ = "\12""0" # fix should be "\0120"
+85 85 | _ = "\\12""0" # fix should be "\\120"
+86    |-_ = "\\\12""0" # fix should be "\\\0120"
+   86 |+_ = "\\\0120" # fix should be "\\\0120"
+87 87 | _ = "\12 0""0" # fix should be "\12 00"
+88 88 | _ = r"\12"r"0" # fix should be r"\120"
+89 89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+
+ISC.py:87:5: ISC001 [*] Implicitly concatenated string literals on one line
+   |
+85 | _ = "\\12""0" # fix should be "\\120"
+86 | _ = "\\\12""0" # fix should be "\\\0120"
+87 | _ = "\12 0""0" # fix should be "\12 00"
+   |     ^^^^^^^^^^ ISC001
+88 | _ = r"\12"r"0" # fix should be r"\120"
+89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+   |
+   = help: Combine string literals
+
+ℹ Safe fix
+84 84 | _ = "\12""0" # fix should be "\0120"
+85 85 | _ = "\\12""0" # fix should be "\\120"
+86 86 | _ = "\\\12""0" # fix should be "\\\0120"
+87    |-_ = "\12 0""0" # fix should be "\12 00"
+   87 |+_ = "\12 00" # fix should be "\12 00"
+88 88 | _ = r"\12"r"0" # fix should be r"\120"
+89 89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+90 90 | _ = "\8""0" # fix should be "\80"
+
+ISC.py:88:5: ISC001 [*] Implicitly concatenated string literals on one line
+   |
+86 | _ = "\\\12""0" # fix should be "\\\0120"
+87 | _ = "\12 0""0" # fix should be "\12 00"
+88 | _ = r"\12"r"0" # fix should be r"\120"
+   |     ^^^^^^^^^^ ISC001
+89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+90 | _ = "\8""0" # fix should be "\80"
+   |
+   = help: Combine string literals
+
+ℹ Safe fix
+85 85 | _ = "\\12""0" # fix should be "\\120"
+86 86 | _ = "\\\12""0" # fix should be "\\\0120"
+87 87 | _ = "\12 0""0" # fix should be "\12 00"
+88    |-_ = r"\12"r"0" # fix should be r"\120"
+   88 |+_ = r"\120" # fix should be r"\120"
+89 89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+90 90 | _ = "\8""0" # fix should be "\80"
+91 91 | _ = "\12""8" # fix should be "\128"
+
+ISC.py:89:5: ISC001 [*] Implicitly concatenated string literals on one line
+   |
+87 | _ = "\12 0""0" # fix should be "\12 00"
+88 | _ = r"\12"r"0" # fix should be r"\120"
+89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+   |     ^^^^^^^^^^^^^^^^^ ISC001
+90 | _ = "\8""0" # fix should be "\80"
+91 | _ = "\12""8" # fix should be "\128"
+   |
+   = help: Combine string literals
+
+ℹ Safe fix
+86 86 | _ = "\\\12""0" # fix should be "\\\0120"
+87 87 | _ = "\12 0""0" # fix should be "\12 00"
+88 88 | _ = r"\12"r"0" # fix should be r"\120"
+89    |-_ = "\12 and more""0" # fix should be "\12 and more0"
+   89 |+_ = "\12 and more0" # fix should be "\12 and more0"
+90 90 | _ = "\8""0" # fix should be "\80"
+91 91 | _ = "\12""8" # fix should be "\128"
+92 92 | _ = "\12""foo" # fix should be "\12foo"
+
+ISC.py:90:5: ISC001 [*] Implicitly concatenated string literals on one line
+   |
+88 | _ = r"\12"r"0" # fix should be r"\120"
+89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+90 | _ = "\8""0" # fix should be "\80"
+   |     ^^^^^^^ ISC001
+91 | _ = "\12""8" # fix should be "\128"
+92 | _ = "\12""foo" # fix should be "\12foo"
+   |
+   = help: Combine string literals
+
+ℹ Safe fix
+87 87 | _ = "\12 0""0" # fix should be "\12 00"
+88 88 | _ = r"\12"r"0" # fix should be r"\120"
+89 89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+90    |-_ = "\8""0" # fix should be "\80"
+   90 |+_ = "\80" # fix should be "\80"
+91 91 | _ = "\12""8" # fix should be "\128"
+92 92 | _ = "\12""foo" # fix should be "\12foo"
+93 93 | _ = "\12" ""  # fix should be "\12"
+
+ISC.py:91:5: ISC001 [*] Implicitly concatenated string literals on one line
+   |
+89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+90 | _ = "\8""0" # fix should be "\80"
+91 | _ = "\12""8" # fix should be "\128"
+   |     ^^^^^^^^ ISC001
+92 | _ = "\12""foo" # fix should be "\12foo"
+93 | _ = "\12" ""  # fix should be "\12"
+   |
+   = help: Combine string literals
+
+ℹ Safe fix
+88 88 | _ = r"\12"r"0" # fix should be r"\120"
+89 89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+90 90 | _ = "\8""0" # fix should be "\80"
+91    |-_ = "\12""8" # fix should be "\128"
+   91 |+_ = "\128" # fix should be "\128"
+92 92 | _ = "\12""foo" # fix should be "\12foo"
+93 93 | _ = "\12" ""  # fix should be "\12"
+
+ISC.py:92:5: ISC001 [*] Implicitly concatenated string literals on one line
+   |
+90 | _ = "\8""0" # fix should be "\80"
+91 | _ = "\12""8" # fix should be "\128"
+92 | _ = "\12""foo" # fix should be "\12foo"
+   |     ^^^^^^^^^^ ISC001
+93 | _ = "\12" ""  # fix should be "\12"
+   |
+   = help: Combine string literals
+
+ℹ Safe fix
+89 89 | _ = "\12 and more""0" # fix should be "\12 and more0"
+90 90 | _ = "\8""0" # fix should be "\80"
+91 91 | _ = "\12""8" # fix should be "\128"
+92    |-_ = "\12""foo" # fix should be "\12foo"
+   92 |+_ = "\12foo" # fix should be "\12foo"
+93 93 | _ = "\12" ""  # fix should be "\12"
+
+ISC.py:93:5: ISC001 [*] Implicitly concatenated string literals on one line
+   |
+91 | _ = "\12""8" # fix should be "\128"
+92 | _ = "\12""foo" # fix should be "\12foo"
+93 | _ = "\12" ""  # fix should be "\12"
+   |     ^^^^^^^^ ISC001
+   |
+   = help: Combine string literals
+
+ℹ Safe fix
+90 90 | _ = "\8""0" # fix should be "\80"
+91 91 | _ = "\12""8" # fix should be "\128"
+92 92 | _ = "\12""foo" # fix should be "\12foo"
+93    |-_ = "\12" ""  # fix should be "\12"
+   93 |+_ = "\12"  # fix should be "\12"
diff --git a/...oncat/snapshots/ruff_linter__rules__flake8_implicit_str_concat__tests__ISC003_ISC.py.snap b/...oncat/snapshots/ruff_linter__rules__flake8_implicit_str_concat__tests__ISC003_ISC.py.snap
@@ -50,6 +50,6 @@ ISC.py:80:10: ISC003 Explicitly concatenated string should be implicitly concate
    |  __________^
 81 | |     + f"second"} d"
    | |_______________^ ISC003
+82 |   
+83 |   # See https://github.com/astral-sh/ruff/issues/12936
    |
-
-