Use Cow in printf rewrite rule (#7986)

Small thing that bothered me when looking into the regex update.

crates/ruff_linter/src/rules/pyupgrade/helpers.rs

@@ -1,22 +1,21 @@
 use once_cell::sync::Lazy;
 use regex::{Captures, Regex};
+use std::borrow::Cow;
 
 static CURLY_BRACES: Lazy<Regex> = Lazy::new(|| Regex::new(r"(\\N\{[^}]+})|([{}])").unwrap());
 
-pub(super) fn curly_escape(text: &str) -> String {
+pub(super) fn curly_escape(text: &str) -> Cow<'_, str> {
     // Match all curly braces. This will include named unicode escapes (like
     // \N{SNOWMAN}), which we _don't_ want to escape, so take care to preserve them.
-    CURLY_BRACES
+    CURLY_BRACES.replace_all(text, |caps: &Captures| {
-        .replace_all(text, |caps: &Captures| {
+        if let Some(match_) = caps.get(1) {
-            if let Some(match_) = caps.get(1) {
+            match_.as_str().to_string()
-                match_.as_str().to_string()
+        } else {
+            if &caps[2] == "{" {
+                "{{".to_string()
             } else {
-                if &caps[2] == "{" {
+                "}}".to_string()
-                    "{{".to_string()
-                } else {
-                    "}}".to_string()
-                }
             }
-        })
+        }
-        .to_string()
+    })
 }

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

@@ -1,3 +1,4 @@
+use std::borrow::Cow;
 use std::str::FromStr;
 
 use ruff_diagnostics::{Diagnostic, Edit, Fix, FixAvailability, Violation};
@@ -105,7 +106,7 @@ fn simplify_conversion_flag(flags: CConversionFlags) -> String {
 }
 
 /// Convert a [`PercentFormat`] struct into a `String`.
-fn handle_part(part: &CFormatPart<String>) -> String {
+fn handle_part(part: &CFormatPart<String>) -> Cow<'_, str> {
     match part {
         CFormatPart::Literal(item) => curly_escape(item),
         CFormatPart::Spec(spec) => {
@@ -114,7 +115,7 @@ fn handle_part(part: &CFormatPart<String>) -> String {
             // TODO(charlie): What case is this?
             if spec.format_char == '%' {
                 format_string.push('%');
-                return format_string;
+                return Cow::Owned(format_string);
             }
 
             format_string.push('{');
@@ -171,26 +172,25 @@ fn handle_part(part: &CFormatPart<String>) -> String {
                 format_string.push(spec.format_char);
             }
             format_string.push('}');
-            format_string
+            Cow::Owned(format_string)
         }
     }
 }
 
 /// Convert a [`CFormatString`] into a `String`.
 fn percent_to_format(format_string: &CFormatString) -> String {
-    let mut contents = String::new();
+    format_string
-    for (.., format_part) in format_string.iter() {
+        .iter()
-        contents.push_str(&handle_part(format_part));
+        .map(|(_, part)| handle_part(part))
-    }
+        .collect()
-    contents
 }
 
 /// If a tuple has one argument, remove the comma; otherwise, return it as-is.
-fn clean_params_tuple(right: &Expr, locator: &Locator) -> String {
+fn clean_params_tuple<'a>(right: &Expr, locator: &Locator<'a>) -> Cow<'a, str> {
-    let mut contents = locator.slice(right).to_string();
     if let Expr::Tuple(ast::ExprTuple { elts, .. }) = &right {
         if elts.len() == 1 {
             if !locator.contains_line_break(right.range()) {
+                let mut contents = locator.slice(right).to_string();
                 for (i, character) in contents.chars().rev().enumerate() {
                     if character == ',' {
                         let correct_index = contents.len() - i - 1;
@@ -198,10 +198,12 @@ fn clean_params_tuple(right: &Expr, locator: &Locator) -> String {
                         break;
                     }
                 }
+                return Cow::Owned(contents);
             }
         }
     }
-    contents
+
+    Cow::Borrowed(locator.slice(right))
 }
 
 /// Converts a dictionary to a function call while preserving as much styling as
@@ -419,16 +421,16 @@ pub(crate) fn printf_string_formatting(checker: &mut Checker, expr: &Expr, right
     // Parse the parameters.
     let params_string = match right {
         Expr::Constant(_) | Expr::FString(_) => {
-            format!("({})", checker.locator().slice(right))
+            Cow::Owned(format!("({})", checker.locator().slice(right)))
         }
         Expr::Name(_) | Expr::Attribute(_) | Expr::Subscript(_) | Expr::Call(_) => {
             if num_keyword_arguments > 0 {
                 // If we have _any_ named fields, assume the right-hand side is a mapping.
-                format!("(**{})", checker.locator().slice(right))
+                Cow::Owned(format!("(**{})", checker.locator().slice(right)))
             } else if num_positional_arguments > 1 {
                 // If we have multiple fields, but no named fields, assume the right-hand side is a
                 // tuple.
-                format!("(*{})", checker.locator().slice(right))
+                Cow::Owned(format!("(*{})", checker.locator().slice(right)))
             } else {
                 // Otherwise, if we have a single field, we can't make any assumptions about the
                 // right-hand side. It _could_ be a tuple, but it could also be a single value,
@@ -444,13 +446,12 @@ pub(crate) fn printf_string_formatting(checker: &mut Checker, expr: &Expr, right
         }
         Expr::Tuple(_) => clean_params_tuple(right, checker.locator()),
         Expr::Dict(_) => {
-            if let Some(params_string) =
+            let Some(params_string) =
                 clean_params_dictionary(right, checker.locator(), checker.stylist())
-            {
+            else {
-                params_string
-            } else {
                 return;
-            }
+            };
+            Cow::Owned(params_string)
         }
         _ => return,
     };