Use shared leading_quote implementation in ruff_python_formatter (#3396)

Cargo.lock

@@ -2154,6 +2154,7 @@ dependencies = [
  "itertools",
  "once_cell",
  "ruff_formatter",
+ "ruff_python_ast",
  "ruff_python_stdlib",
  "ruff_rustpython",
  "ruff_testing_macros",

crates/ruff/src/checkers/ast/mod.rs

@@ -27,7 +27,7 @@ use ruff_python_ast::types::{
 use ruff_python_ast::typing::{match_annotated_subscript, Callable, SubscriptKind};
 use ruff_python_ast::visitor::{walk_excepthandler, walk_pattern, Visitor};
 use ruff_python_ast::{
-    branch_detection, cast, helpers, operations, strings, typing, visibility, visitor,
+    branch_detection, cast, helpers, operations, str, typing, visibility, visitor,
 };
 use ruff_python_stdlib::builtins::{BUILTINS, MAGIC_GLOBALS};
 use ruff_python_stdlib::path::is_python_stub_file;
@@ -5292,7 +5292,7 @@ impl<'a> Checker<'a> {
                     Location::new(expr.location.row(), expr.location.column()),
                 ));
 
-                let body = strings::raw_contents(contents);
+                let body = str::raw_contents(contents);
                 let docstring = Docstring {
                     kind: definition.kind,
                     expr,

crates/ruff/src/rules/pydocstyle/rules/ends_with_period.rs

@@ -1,7 +1,7 @@
 use strum::IntoEnumIterator;
 
 use ruff_macros::{derive_message_formats, violation};
-use ruff_python_ast::strings::leading_quote;
+use ruff_python_ast::str::leading_quote;
 use ruff_python_ast::types::Range;
 
 use crate::checkers::ast::Checker;

crates/ruff/src/rules/pydocstyle/rules/ends_with_punctuation.rs

@@ -1,7 +1,7 @@
 use strum::IntoEnumIterator;
 
 use ruff_macros::{derive_message_formats, violation};
-use ruff_python_ast::strings::leading_quote;
+use ruff_python_ast::str::leading_quote;
 use ruff_python_ast::types::Range;
 
 use crate::checkers::ast::Checker;

crates/ruff/src/rules/pydocstyle/rules/multi_line_summary_start.rs

@@ -1,8 +1,7 @@
 use ruff_macros::{derive_message_formats, violation};
-use ruff_python_ast::strings::leading_quote;
+use ruff_python_ast::str::{is_triple_quote, leading_quote};
 use ruff_python_ast::types::Range;
 use ruff_python_ast::whitespace::LinesWithTrailingNewline;
-use ruff_python_stdlib::str::TRIPLE_QUOTE_PREFIXES;
 
 use crate::checkers::ast::Checker;
 use crate::docstrings::definition::{DefinitionKind, Docstring};
@@ -54,7 +53,7 @@ pub fn multi_line_summary_start(checker: &mut Checker, docstring: &Docstring) {
     {
         return;
     };
-    if TRIPLE_QUOTE_PREFIXES.contains(&first_line) {
+    if is_triple_quote(first_line) {
         if checker
             .settings
             .rules

crates/ruff/src/rules/pydocstyle/rules/no_surrounding_whitespace.rs

@@ -1,5 +1,5 @@
 use ruff_macros::{derive_message_formats, violation};
-use ruff_python_ast::strings::leading_quote;
+use ruff_python_ast::str::leading_quote;
 use ruff_python_ast::types::Range;
 use ruff_python_ast::whitespace::LinesWithTrailingNewline;
 

crates/ruff/src/rules/pydocstyle/rules/one_liner.rs

@@ -1,5 +1,5 @@
 use ruff_macros::{derive_message_formats, violation};
-use ruff_python_ast::strings::{leading_quote, trailing_quote};
+use ruff_python_ast::str::{leading_quote, trailing_quote};
 use ruff_python_ast::types::Range;
 use ruff_python_ast::whitespace::LinesWithTrailingNewline;
 

crates/ruff/src/rules/pyflakes/fixes.rs

@@ -4,7 +4,7 @@ use rustpython_parser::ast::{Excepthandler, Expr};
 use rustpython_parser::{lexer, Mode, Tok};
 
 use ruff_python_ast::source_code::{Locator, Stylist};
-use ruff_python_ast::strings::raw_contents;
+use ruff_python_ast::str::raw_contents;
 use ruff_python_ast::types::Range;
 
 use crate::cst::matchers::{match_expr, match_module};

crates/ruff/src/rules/pylint/rules/bad_string_format_type.rs

@@ -6,7 +6,7 @@ use rustpython_parser::ast::{Constant, Expr, ExprKind, Location, Operator};
 use rustpython_parser::{lexer, Mode, Tok};
 
 use ruff_macros::{derive_message_formats, violation};
-use ruff_python_ast::strings::{leading_quote, trailing_quote};
+use ruff_python_ast::str::{leading_quote, trailing_quote};
 use ruff_python_ast::types::Range;
 
 use crate::checkers::ast::Checker;

crates/ruff/src/rules/pyupgrade/rules/f_strings.rs

@@ -6,7 +6,7 @@ use rustpython_parser::ast::{Constant, Expr, ExprKind, KeywordData};
 use rustpython_parser::{lexer, Mode, Tok};
 
 use ruff_macros::{derive_message_formats, violation};
-use ruff_python_ast::strings::{leading_quote, trailing_quote};
+use ruff_python_ast::str::{leading_quote, trailing_quote};
 use ruff_python_ast::types::Range;
 
 use crate::checkers::ast::Checker;

crates/ruff/src/rules/pyupgrade/rules/printf_string_formatting.rs

@@ -7,7 +7,7 @@ use rustpython_parser::ast::{Constant, Expr, ExprKind, Location};
 use rustpython_parser::{lexer, Mode, Tok};
 
 use ruff_macros::{derive_message_formats, violation};
-use ruff_python_ast::strings::{leading_quote, trailing_quote};
+use ruff_python_ast::str::{leading_quote, trailing_quote};
 use ruff_python_ast::types::Range;
 use ruff_python_ast::whitespace::indentation;
 use ruff_python_stdlib::identifiers::is_identifier;

crates/ruff_python_ast/src/lib.rs

@@ -9,7 +9,7 @@ pub mod logging;
 pub mod operations;
 pub mod relocate;
 pub mod source_code;
-pub mod strings;
+pub mod str;
 pub mod types;
 pub mod typing;
 pub mod visibility;

crates/ruff_python_ast/src/source_code/stylist.rs

@@ -10,7 +10,7 @@ use rustpython_parser::{lexer, Mode, Tok};
 use crate::source_code::Locator;
 use ruff_rustpython::vendor;
 
-use crate::strings::leading_quote;
+use crate::str::leading_quote;
 use crate::types::Range;
 
 pub struct Stylist<'a> {

crates/ruff_python_ast/src/str.rs

@@ -0,0 +1,97 @@
+/// See: <https://docs.python.org/3/reference/lexical_analysis.html#string-and-bytes-literals>
+const TRIPLE_QUOTE_STR_PREFIXES: &[&str] = &[
+    "u\"\"\"", "u'''", "r\"\"\"", "r'''", "U\"\"\"", "U'''", "R\"\"\"", "R'''", "\"\"\"", "'''",
+];
+const SINGLE_QUOTE_STR_PREFIXES: &[&str] = &[
+    "u\"", "u'", "r\"", "r'", "U\"", "U'", "R\"", "R'", "\"", "'",
+];
+pub const TRIPLE_QUOTE_BYTE_PREFIXES: &[&str] = &[
+    "br'''", "rb'''", "bR'''", "Rb'''", "Br'''", "rB'''", "RB'''", "BR'''", "b'''", "br\"\"\"",
+    "rb\"\"\"", "bR\"\"\"", "Rb\"\"\"", "Br\"\"\"", "rB\"\"\"", "RB\"\"\"", "BR\"\"\"", "b\"\"\"",
+    "B\"\"\"",
+];
+pub const SINGLE_QUOTE_BYTE_PREFIXES: &[&str] = &[
+    "br'", "rb'", "bR'", "Rb'", "Br'", "rB'", "RB'", "BR'", "b'", "br\"", "rb\"", "bR\"", "Rb\"",
+    "Br\"", "rB\"", "RB\"", "BR\"", "b\"", "B\"",
+];
+const TRIPLE_QUOTE_SUFFIXES: &[&str] = &["\"\"\"", "'''"];
+const SINGLE_QUOTE_SUFFIXES: &[&str] = &["\"", "'"];
+
+/// Strip the leading and trailing quotes from a docstring.
+pub fn raw_contents(contents: &str) -> &str {
+    for pattern in TRIPLE_QUOTE_STR_PREFIXES
+        .iter()
+        .chain(TRIPLE_QUOTE_BYTE_PREFIXES)
+    {
+        if contents.starts_with(pattern) {
+            return &contents[pattern.len()..contents.len() - 3];
+        }
+    }
+    for pattern in SINGLE_QUOTE_STR_PREFIXES
+        .iter()
+        .chain(SINGLE_QUOTE_BYTE_PREFIXES)
+    {
+        if contents.starts_with(pattern) {
+            return &contents[pattern.len()..contents.len() - 1];
+        }
+    }
+    unreachable!("Expected docstring to start with a valid triple- or single-quote prefix")
+}
+
+/// Return the leading quote for a string or byte literal (e.g., `"""`).
+pub fn leading_quote(content: &str) -> Option<&str> {
+    if let Some(first_line) = content.lines().next() {
+        for pattern in TRIPLE_QUOTE_STR_PREFIXES
+            .iter()
+            .chain(TRIPLE_QUOTE_BYTE_PREFIXES)
+            .chain(SINGLE_QUOTE_STR_PREFIXES)
+            .chain(SINGLE_QUOTE_BYTE_PREFIXES)
+        {
+            if first_line.starts_with(pattern) {
+                return Some(pattern);
+            }
+        }
+    }
+    None
+}
+
+/// Return the trailing quote string for a string or byte literal (e.g., `"""`).
+pub fn trailing_quote(content: &str) -> Option<&&str> {
+    TRIPLE_QUOTE_SUFFIXES
+        .iter()
+        .chain(SINGLE_QUOTE_SUFFIXES)
+        .find(|&pattern| content.ends_with(pattern))
+}
+
+/// Return `true` if the string is a triple-quote string or byte prefix.
+pub fn is_triple_quote(content: &str) -> bool {
+    TRIPLE_QUOTE_STR_PREFIXES.contains(&content) || TRIPLE_QUOTE_BYTE_PREFIXES.contains(&content)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{
+        SINGLE_QUOTE_BYTE_PREFIXES, SINGLE_QUOTE_STR_PREFIXES, TRIPLE_QUOTE_BYTE_PREFIXES,
+        TRIPLE_QUOTE_STR_PREFIXES,
+    };
+
+    #[test]
+    fn test_prefixes() {
+        let prefixes = TRIPLE_QUOTE_STR_PREFIXES
+            .iter()
+            .chain(TRIPLE_QUOTE_BYTE_PREFIXES)
+            .chain(SINGLE_QUOTE_STR_PREFIXES)
+            .chain(SINGLE_QUOTE_BYTE_PREFIXES)
+            .collect::<Vec<_>>();
+        for (i, prefix_i) in prefixes.iter().enumerate() {
+            for (j, prefix_j) in prefixes.iter().enumerate() {
+                if i > j {
+                    assert!(
+                        !prefix_i.starts_with(*prefix_j),
+                        "Prefixes are not unique: {prefix_i} starts with {prefix_j}",
+                    );
+                }
+            }
+        }
+    }
+}

crates/ruff_python_ast/src/strings.rs

@@ -1,38 +0,0 @@
-use ruff_python_stdlib::str::{
-    SINGLE_QUOTE_PREFIXES, SINGLE_QUOTE_SUFFIXES, TRIPLE_QUOTE_PREFIXES, TRIPLE_QUOTE_SUFFIXES,
-};
-
-/// Strip the leading and trailing quotes from a docstring.
-pub fn raw_contents(contents: &str) -> &str {
-    for pattern in TRIPLE_QUOTE_PREFIXES {
-        if contents.starts_with(pattern) {
-            return &contents[pattern.len()..contents.len() - 3];
-        }
-    }
-    for pattern in SINGLE_QUOTE_PREFIXES {
-        if contents.starts_with(pattern) {
-            return &contents[pattern.len()..contents.len() - 1];
-        }
-    }
-    unreachable!("Expected docstring to start with a valid triple- or single-quote prefix")
-}
-
-/// Return the leading quote string for a docstring (e.g., `"""`).
-pub fn leading_quote(content: &str) -> Option<&str> {
-    if let Some(first_line) = content.lines().next() {
-        for pattern in TRIPLE_QUOTE_PREFIXES.iter().chain(SINGLE_QUOTE_PREFIXES) {
-            if first_line.starts_with(pattern) {
-                return Some(pattern);
-            }
-        }
-    }
-    None
-}
-
-/// Return the trailing quote string for a docstring (e.g., `"""`).
-pub fn trailing_quote(content: &str) -> Option<&&str> {
-    TRIPLE_QUOTE_SUFFIXES
-        .iter()
-        .chain(SINGLE_QUOTE_SUFFIXES)
-        .find(|&pattern| content.ends_with(pattern))
-}

crates/ruff_python_formatter/Cargo.toml

@@ -7,6 +7,7 @@ rust-version = { workspace = true }
 
 [dependencies]
 ruff_formatter = { path = "../ruff_formatter" }
+ruff_python_ast = { path = "../ruff_python_ast" }
 ruff_python_stdlib = { path = "../ruff_python_stdlib" }
 ruff_rustpython = { path = "../ruff_rustpython" }
 ruff_text_size = { path = "../ruff_text_size" }

crates/ruff_python_formatter/src/core/helpers.rs

@@ -3,31 +3,6 @@ use rustpython_parser::ast::Location;
 use crate::core::locator::Locator;
 use crate::core::types::Range;
 
-/// Return the leading quote for a string or byte literal (e.g., `"""`).
-pub fn leading_quote(content: &str) -> Option<&str> {
-    if let Some(first_line) = content.lines().next() {
-        for pattern in ruff_python_stdlib::str::TRIPLE_QUOTE_PREFIXES
-            .iter()
-            .chain(ruff_python_stdlib::bytes::TRIPLE_QUOTE_PREFIXES)
-            .chain(ruff_python_stdlib::str::SINGLE_QUOTE_PREFIXES)
-            .chain(ruff_python_stdlib::bytes::SINGLE_QUOTE_PREFIXES)
-        {
-            if first_line.starts_with(pattern) {
-                return Some(pattern);
-            }
-        }
-    }
-    None
-}
-
-/// Return the trailing quote string for a string or byte literal (e.g., `"""`).
-pub fn trailing_quote(content: &str) -> Option<&&str> {
-    ruff_python_stdlib::str::TRIPLE_QUOTE_SUFFIXES
-        .iter()
-        .chain(ruff_python_stdlib::str::SINGLE_QUOTE_SUFFIXES)
-        .find(|&pattern| content.ends_with(pattern))
-}
-
 /// Return `true` if the given string is a radix literal (e.g., `0b101`).
 pub fn is_radix_literal(content: &str) -> bool {
     content.starts_with("0b")
@@ -155,30 +130,3 @@ pub fn is_elif(orelse: &[rustpython_parser::ast::Stmt], locator: &Locator) -> bo
     }
     false
 }
-
-#[cfg(test)]
-mod tests {
-    #[test]
-    fn test_prefixes() {
-        let prefixes = ruff_python_stdlib::str::TRIPLE_QUOTE_PREFIXES
-            .iter()
-            .chain(ruff_python_stdlib::bytes::TRIPLE_QUOTE_PREFIXES)
-            .chain(ruff_python_stdlib::str::SINGLE_QUOTE_PREFIXES)
-            .chain(ruff_python_stdlib::bytes::SINGLE_QUOTE_PREFIXES)
-            .collect::<Vec<_>>();
-        for i in 1..prefixes.len() {
-            for j in 0..i - 1 {
-                if i != j {
-                    if prefixes[i].starts_with(prefixes[j]) {
-                        assert!(
-                            !prefixes[i].starts_with(prefixes[j]),
-                            "Prefixes are not unique: {} starts with {}",
-                            prefixes[i],
-                            prefixes[j]
-                        );
-                    }
-                }
-            }
-        }
-    }
-}

crates/ruff_python_formatter/src/format/strings.rs

@@ -2,10 +2,10 @@ use rustpython_parser::{Mode, Tok};
 
 use ruff_formatter::prelude::*;
 use ruff_formatter::{write, Format};
+use ruff_python_ast::str::{leading_quote, trailing_quote};
 use ruff_text_size::TextSize;
 
 use crate::context::ASTFormatContext;
-use crate::core::helpers::{leading_quote, trailing_quote};
 use crate::core::types::Range;
 use crate::cst::Expr;
 

crates/ruff_python_stdlib/src/bytes.rs

@@ -1,10 +0,0 @@
-/// See: <https://docs.python.org/3/reference/lexical_analysis.html#string-and-bytes-literals>
-pub const TRIPLE_QUOTE_PREFIXES: &[&str] = &[
-    "br'''", "rb'''", "bR'''", "Rb'''", "Br'''", "rB'''", "RB'''", "BR'''", "b'''", "br\"\"\"",
-    "rb\"\"\"", "bR\"\"\"", "Rb\"\"\"", "Br\"\"\"", "rB\"\"\"", "RB\"\"\"", "BR\"\"\"", "b\"\"\"",
-    "B\"\"\"",
-];
-pub const SINGLE_QUOTE_PREFIXES: &[&str] = &[
-    "br'", "rb'", "bR'", "Rb'", "Br'", "rB'", "RB'", "BR'", "b'", "br\"", "rb\"", "bR\"", "Rb\"",
-    "Br\"", "rB\"", "RB\"", "BR\"", "b\"", "B\"",
-];

crates/ruff_python_stdlib/src/lib.rs

@@ -1,5 +1,4 @@
 pub mod builtins;
-pub mod bytes;
 pub mod future;
 pub mod identifiers;
 pub mod keyword;

crates/ruff_python_stdlib/src/str.rs

@@ -1,6 +1,3 @@
-use once_cell::sync::Lazy;
-use regex::Regex;
-
 /// See: <https://docs.python.org/3/reference/lexical_analysis.html#string-and-bytes-literals>
 pub const TRIPLE_QUOTE_PREFIXES: &[&str] = &[
     "u\"\"\"", "u'''", "r\"\"\"", "r'''", "U\"\"\"", "U'''", "R\"\"\"", "R'''", "\"\"\"", "'''",
@@ -11,9 +8,6 @@ pub const SINGLE_QUOTE_PREFIXES: &[&str] = &[
 pub const TRIPLE_QUOTE_SUFFIXES: &[&str] = &["\"\"\"", "'''"];
 pub const SINGLE_QUOTE_SUFFIXES: &[&str] = &["\"", "'"];
 
-pub static STRING_QUOTE_PREFIX_REGEX: Lazy<Regex> =
-    Lazy::new(|| Regex::new(r#"^(?i)[urb]*['"](?P<raw>.*)['"]$"#).unwrap());
-
 pub fn is_lower(s: &str) -> bool {
     let mut cased = false;
     for c in s.chars() {