Implement "native literals" check from pyupgrade (#1350)

README.md

@@ -629,6 +629,7 @@ For more, see [pyupgrade](https://pypi.org/project/pyupgrade/3.2.0/) on PyPI.
 | UP015 | RedundantOpenModes | Unnecessary open mode parameters | 🛠 |
 | UP016 | RemoveSixCompat | Unnecessary `six` compatibility usage | 🛠 |
 | UP017 | DatetimeTimezoneUTC | Use `datetime.UTC` alias | 🛠 |
+| UP018 | NativeLiterals | Unnecessary call to `str` and `bytes` | 🛠 |
 
 ### pep8-naming (N)
 
@@ -1232,7 +1233,7 @@ natively, including:
 - [`pep8-naming`](https://pypi.org/project/pep8-naming/)
 - [`pydocstyle`](https://pypi.org/project/pydocstyle/)
 - [`pygrep-hooks`](https://github.com/pre-commit/pygrep-hooks) (3/10)
-- [`pyupgrade`](https://pypi.org/project/pyupgrade/) (17/33)
+- [`pyupgrade`](https://pypi.org/project/pyupgrade/) (18/33)
 - [`yesqa`](https://github.com/asottile/yesqa)
 
 Note that, in some cases, Ruff uses different error code prefixes than would be found in the

resources/test/fixtures/pyupgrade/UP018.py

@@ -0,0 +1,25 @@
+# These remain unchanged
+str(1)
+str(*a)
+str("foo", *a)
+str(**k)
+str("foo", **k)
+str("foo", encoding="UTF-8")
+str("foo"
+    "bar")
+bytes("foo", encoding="UTF-8")
+bytes(*a)
+bytes("foo", *a)
+bytes("foo", **a)
+bytes(b"foo"
+      b"bar")
+
+# These become string or byte literals
+str()
+str("foo")
+str("""
+foo""")
+bytes()
+bytes(b"foo")
+bytes(b"""
+foo""")

src/checkers/ast.rs

@@ -1644,6 +1644,9 @@ where
                 if self.settings.enabled.contains(&CheckCode::UP016) {
                     pyupgrade::plugins::remove_six_compat(self, expr);
                 }
+                if self.settings.enabled.contains(&CheckCode::UP018) {
+                    pyupgrade::plugins::native_literals(self, expr, func, args, keywords);
+                }
 
                 // flake8-super
                 if self.settings.enabled.contains(&CheckCode::UP008) {

src/checks.rs

@@ -226,6 +226,7 @@ pub enum CheckCode {
     UP015,
     UP016,
     UP017,
+    UP018,
     // pydocstyle
     D100,
     D101,
@@ -829,6 +830,7 @@ pub enum CheckKind {
     RedundantOpenModes,
     RemoveSixCompat,
     DatetimeTimezoneUTC,
+    NativeLiterals,
     // pydocstyle
     BlankLineAfterLastSection(String),
     BlankLineAfterSection(String),
@@ -1203,6 +1205,7 @@ impl CheckCode {
             CheckCode::UP015 => CheckKind::RedundantOpenModes,
             CheckCode::UP016 => CheckKind::RemoveSixCompat,
             CheckCode::UP017 => CheckKind::DatetimeTimezoneUTC,
+            CheckCode::UP018 => CheckKind::NativeLiterals,
             // pydocstyle
             CheckCode::D100 => CheckKind::PublicModule,
             CheckCode::D101 => CheckKind::PublicClass,
@@ -1621,6 +1624,7 @@ impl CheckCode {
             CheckCode::UP015 => CheckCategory::Pyupgrade,
             CheckCode::UP016 => CheckCategory::Pyupgrade,
             CheckCode::UP017 => CheckCategory::Pyupgrade,
+            CheckCode::UP018 => CheckCategory::Pyupgrade,
             CheckCode::W292 => CheckCategory::Pycodestyle,
             CheckCode::W605 => CheckCategory::Pycodestyle,
             CheckCode::YTT101 => CheckCategory::Flake82020,
@@ -1832,6 +1836,7 @@ impl CheckKind {
             CheckKind::RedundantOpenModes => &CheckCode::UP015,
             CheckKind::RemoveSixCompat => &CheckCode::UP016,
             CheckKind::DatetimeTimezoneUTC => &CheckCode::UP017,
+            CheckKind::NativeLiterals => &CheckCode::UP018,
             // pydocstyle
             CheckKind::BlankLineAfterLastSection(_) => &CheckCode::D413,
             CheckKind::BlankLineAfterSection(_) => &CheckCode::D410,
@@ -2555,6 +2560,7 @@ impl CheckKind {
             CheckKind::RedundantOpenModes => "Unnecessary open mode parameters".to_string(),
             CheckKind::RemoveSixCompat => "Unnecessary `six` compatibility usage".to_string(),
             CheckKind::DatetimeTimezoneUTC => "Use `datetime.UTC` alias".to_string(),
+            CheckKind::NativeLiterals => "Unnecessary call to `str` and `bytes`".to_string(),
             CheckKind::ConvertTypedDictFunctionalToClass(name) => {
                 format!("Convert `{name}` from `TypedDict` functional to class syntax")
             }
@@ -2956,6 +2962,7 @@ impl CheckKind {
                 | CheckKind::ConvertNamedTupleFunctionalToClass(..)
                 | CheckKind::ConvertTypedDictFunctionalToClass(..)
                 | CheckKind::DashedUnderlineAfterSection(..)
+                | CheckKind::DatetimeTimezoneUTC
                 | CheckKind::DeprecatedUnittestAlias(..)
                 | CheckKind::DoNotAssertFalse
                 | CheckKind::DoNotAssignLambda
@@ -2969,6 +2976,7 @@ impl CheckKind {
                 | CheckKind::KeyInDict(..)
                 | CheckKind::MisplacedComparisonConstant(..)
                 | CheckKind::MissingReturnTypeSpecialMethod(..)
+                | CheckKind::NativeLiterals
                 | CheckKind::NewLineAfterLastParagraph
                 | CheckKind::NewLineAfterSectionName(..)
                 | CheckKind::NoBlankLineAfterFunction(..)
@@ -2991,7 +2999,6 @@ impl CheckKind {
                 | CheckKind::RedundantOpenModes
                 | CheckKind::RedundantTupleInExceptionHandler(..)
                 | CheckKind::RemoveSixCompat
-                | CheckKind::DatetimeTimezoneUTC
                 | CheckKind::SectionNameEndsInColon(..)
                 | CheckKind::SectionNotOverIndented(..)
                 | CheckKind::SectionUnderlineAfterName(..)

src/checks_gen.rs

@@ -513,6 +513,7 @@ pub enum CheckCodePrefix {
     UP015,
     UP016,
     UP017,
+    UP018,
     W,
     W2,
     W29,
@@ -2088,6 +2089,7 @@ impl CheckCodePrefix {
                     CheckCode::UP015,
                     CheckCode::UP016,
                     CheckCode::UP017,
+                    CheckCode::UP018,
                 ]
             }
             CheckCodePrefix::U0 => {
@@ -2114,6 +2116,7 @@ impl CheckCodePrefix {
                     CheckCode::UP015,
                     CheckCode::UP016,
                     CheckCode::UP017,
+                    CheckCode::UP018,
                 ]
             }
             CheckCodePrefix::U00 => {
@@ -2222,6 +2225,7 @@ impl CheckCodePrefix {
                     CheckCode::UP015,
                     CheckCode::UP016,
                     CheckCode::UP017,
+                    CheckCode::UP018,
                 ]
             }
             CheckCodePrefix::U010 => {
@@ -2313,6 +2317,7 @@ impl CheckCodePrefix {
                 CheckCode::UP015,
                 CheckCode::UP016,
                 CheckCode::UP017,
+                CheckCode::UP018,
             ],
             CheckCodePrefix::UP0 => vec![
                 CheckCode::UP001,
@@ -2331,6 +2336,7 @@ impl CheckCodePrefix {
                 CheckCode::UP015,
                 CheckCode::UP016,
                 CheckCode::UP017,
+                CheckCode::UP018,
             ],
             CheckCodePrefix::UP00 => vec![
                 CheckCode::UP001,
@@ -2359,6 +2365,7 @@ impl CheckCodePrefix {
                 CheckCode::UP015,
                 CheckCode::UP016,
                 CheckCode::UP017,
+                CheckCode::UP018,
             ],
             CheckCodePrefix::UP010 => vec![CheckCode::UP010],
             CheckCodePrefix::UP011 => vec![CheckCode::UP011],
@@ -2368,6 +2375,7 @@ impl CheckCodePrefix {
             CheckCodePrefix::UP015 => vec![CheckCode::UP015],
             CheckCodePrefix::UP016 => vec![CheckCode::UP016],
             CheckCodePrefix::UP017 => vec![CheckCode::UP017],
+            CheckCodePrefix::UP018 => vec![CheckCode::UP018],
             CheckCodePrefix::W => vec![CheckCode::W292, CheckCode::W605],
             CheckCodePrefix::W2 => vec![CheckCode::W292],
             CheckCodePrefix::W29 => vec![CheckCode::W292],
@@ -2923,6 +2931,7 @@ impl CheckCodePrefix {
             CheckCodePrefix::UP015 => SuffixLength::Three,
             CheckCodePrefix::UP016 => SuffixLength::Three,
             CheckCodePrefix::UP017 => SuffixLength::Three,
+            CheckCodePrefix::UP018 => SuffixLength::Three,
             CheckCodePrefix::W => SuffixLength::Zero,
             CheckCodePrefix::W2 => SuffixLength::One,
             CheckCodePrefix::W29 => SuffixLength::Two,

src/pyupgrade/mod.rs

@@ -37,6 +37,7 @@ mod tests {
     #[test_case(CheckCode::UP014, Path::new("UP014.py"); "UP014")]
     #[test_case(CheckCode::UP015, Path::new("UP015.py"); "UP015")]
     #[test_case(CheckCode::UP016, Path::new("UP016.py"); "UP016")]
+    #[test_case(CheckCode::UP018, Path::new("UP018.py"); "UP018")]
     fn checks(check_code: CheckCode, path: &Path) -> Result<()> {
         let snapshot = format!("{}_{}", check_code.as_ref(), path.to_string_lossy());
         let mut checks = test_path(

src/pyupgrade/plugins/mod.rs

@@ -2,6 +2,7 @@ pub use convert_named_tuple_functional_to_class::convert_named_tuple_functional_
 pub use convert_typed_dict_functional_to_class::convert_typed_dict_functional_to_class;
 pub use datetime_utc_alias::datetime_utc_alias;
 pub use deprecated_unittest_alias::deprecated_unittest_alias;
+pub use native_literals::native_literals;
 pub use redundant_open_modes::redundant_open_modes;
 pub use remove_six_compat::remove_six_compat;
 pub use super_call_with_parameters::super_call_with_parameters;
@@ -18,6 +19,7 @@ mod convert_named_tuple_functional_to_class;
 mod convert_typed_dict_functional_to_class;
 mod datetime_utc_alias;
 mod deprecated_unittest_alias;
+mod native_literals;
 mod redundant_open_modes;
 mod remove_six_compat;
 mod super_call_with_parameters;

src/pyupgrade/plugins/native_literals.rs

@@ -0,0 +1,73 @@
+use rustpython_ast::{Constant, Expr, ExprKind, Keyword};
+use rustpython_parser::lexer;
+use rustpython_parser::lexer::Tok;
+
+use crate::ast::types::Range;
+use crate::autofix::Fix;
+use crate::checkers::ast::Checker;
+use crate::checks::{Check, CheckCode, CheckKind};
+
+/// UP018
+pub fn native_literals(
+    checker: &mut Checker,
+    expr: &Expr,
+    func: &Expr,
+    args: &[Expr],
+    keywords: &[Keyword],
+) {
+    let ExprKind::Name { id, .. } = &func.node else { return; };
+
+    if (id == "str" || id == "bytes")
+        && keywords.is_empty()
+        && args.len() <= 1
+        && checker.is_builtin(id)
+    {
+        let Some(arg) = args.get(0) else {
+            let mut check = Check::new(CheckKind::NativeLiterals, Range::from_located(expr));
+            if checker.patch(&CheckCode::UP018) {
+                check.amend(Fix::replacement(
+                    format!("{}\"\"", if id == "bytes" { "b" } else { "" }),
+                    expr.location,
+                    expr.end_location.unwrap(),
+                ));
+            }
+            checker.add_check(check);
+            return;
+        };
+
+        if !matches!(
+            &arg.node,
+            ExprKind::Constant {
+                value: Constant::Str(_) | Constant::Bytes(_),
+                ..
+            }
+        ) {
+            return;
+        }
+
+        // rust-python merges adjacent string/bytes literals into one node, but we can't
+        // safely remove the outer call in this situation. We're following pyupgrade
+        // here and skip.
+        let arg_code = checker
+            .locator
+            .slice_source_code_range(&Range::from_located(arg));
+        if lexer::make_tokenizer(&arg_code)
+            .flatten()
+            .filter(|(_, tok, _)| matches!(tok, Tok::String { .. } | Tok::Bytes { .. }))
+            .count()
+            > 1
+        {
+            return;
+        }
+
+        let mut check = Check::new(CheckKind::NativeLiterals, Range::from_located(expr));
+        if checker.patch(&CheckCode::UP018) {
+            check.amend(Fix::replacement(
+                arg_code.to_string(),
+                expr.location,
+                expr.end_location.unwrap(),
+            ));
+        }
+        checker.add_check(check);
+    }
+}

src/pyupgrade/snapshots/ruff__pyupgrade__tests__UP018_UP018.py.snap

@@ -0,0 +1,95 @@
+---
+source: src/pyupgrade/mod.rs
+expression: checks
+---
+- kind: NativeLiterals
+  location:
+    row: 18
+    column: 0
+  end_location:
+    row: 18
+    column: 5
+  fix:
+    content: "\"\""
+    location:
+      row: 18
+      column: 0
+    end_location:
+      row: 18
+      column: 5
+- kind: NativeLiterals
+  location:
+    row: 19
+    column: 0
+  end_location:
+    row: 19
+    column: 10
+  fix:
+    content: "\"foo\""
+    location:
+      row: 19
+      column: 0
+    end_location:
+      row: 19
+      column: 10
+- kind: NativeLiterals
+  location:
+    row: 20
+    column: 0
+  end_location:
+    row: 21
+    column: 7
+  fix:
+    content: "\"\"\"\nfoo\"\"\""
+    location:
+      row: 20
+      column: 0
+    end_location:
+      row: 21
+      column: 7
+- kind: NativeLiterals
+  location:
+    row: 22
+    column: 0
+  end_location:
+    row: 22
+    column: 7
+  fix:
+    content: "b\"\""
+    location:
+      row: 22
+      column: 0
+    end_location:
+      row: 22
+      column: 7
+- kind: NativeLiterals
+  location:
+    row: 23
+    column: 0
+  end_location:
+    row: 23
+    column: 13
+  fix:
+    content: "b\"foo\""
+    location:
+      row: 23
+      column: 0
+    end_location:
+      row: 23
+      column: 13
+- kind: NativeLiterals
+  location:
+    row: 24
+    column: 0
+  end_location:
+    row: 25
+    column: 7
+  fix:
+    content: "b\"\"\"\nfoo\"\"\""
+    location:
+      row: 24
+      column: 0
+    end_location:
+      row: 25
+      column: 7
+