Add support for unions to our Python builtins type system

crates/ruff/resources/test/fixtures/pylint/bad_string_format_type.py

@@ -13,6 +13,7 @@
 "%(key)d" % {"key": []}
 print("%d" % ("%s" % ("nested",),))
 "%d" % ((1, 2, 3),)
+"%d" % (1 if x > 0 else [])
 
 # False negatives
 WORD = "abc"
@@ -55,3 +56,4 @@
 "%d" % (len(foo),)
 '(%r, %r, %r, %r)' % (hostname, address, username, '$PASSWORD')
 '%r' % ({'server_school_roles': server_school_roles, 'is_school_multiserver_domain': is_school_multiserver_domain}, )
+"%d" % (1 if x > 0 else 2)

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

@@ -9,7 +9,7 @@ use rustc_hash::FxHashMap;
 use ruff_diagnostics::{Diagnostic, Violation};
 use ruff_macros::{derive_message_formats, violation};
 use ruff_python_ast::str::{leading_quote, trailing_quote};
-use ruff_python_semantic::analyze::type_inference::PythonType;
+use ruff_python_semantic::analyze::type_inference::{NumberLike, PythonType, ResolvedPythonType};
 
 use crate::checkers::ast::Checker;
 
@@ -59,14 +59,16 @@ impl FormatType {
             | PythonType::Set
             | PythonType::Tuple
             | PythonType::Generator
-            | PythonType::Complex
-            | PythonType::Bool
             | PythonType::Ellipsis
             | PythonType::None => matches!(
                 self,
                 FormatType::Unknown | FormatType::String | FormatType::Repr
             ),
-            PythonType::Integer => matches!(
+            PythonType::Number(NumberLike::Complex | NumberLike::Bool) => matches!(
+                self,
+                FormatType::Unknown | FormatType::String | FormatType::Repr
+            ),
+            PythonType::Number(NumberLike::Integer) => matches!(
                 self,
                 FormatType::Unknown
                     | FormatType::String
@@ -75,15 +77,14 @@ impl FormatType {
                     | FormatType::Float
                     | FormatType::Number
             ),
-            PythonType::Float => matches!(
+            PythonType::Number(NumberLike::Float) => matches!(
                 self,
                 FormatType::Unknown
                     | FormatType::String
                     | FormatType::Repr
                     | FormatType::Float
                     | FormatType::Number
             ),
-            PythonType::Unknown => true,
         }
     }
 }
@@ -118,16 +119,22 @@ fn collect_specs(formats: &[CFormatStrOrBytes<String>]) -> Vec<&CFormatSpec> {
 
 /// Return `true` if the format string is equivalent to the constant type
 fn equivalent(format: &CFormatSpec, value: &Expr) -> bool {
-    let format: FormatType = format.format_char.into();
-    let constant: PythonType = value.into();
-    format.is_compatible_with(constant)
+    let format = FormatType::from(format.format_char);
+    match ResolvedPythonType::from(value) {
+        ResolvedPythonType::Atom(atom) => format.is_compatible_with(atom),
+        ResolvedPythonType::Union(atoms) => {
+            atoms.iter().all(|atom| format.is_compatible_with(*atom))
+        }
+        ResolvedPythonType::Unknown => true,
+        ResolvedPythonType::TypeError => true,
+    }
 }
 
-/// Return `true` if the [`Constnat`] aligns with the format type.
+/// Return `true` if the [`Constant`] aligns with the format type.
 fn is_valid_constant(formats: &[CFormatStrOrBytes<String>], value: &Expr) -> bool {
     let formats = collect_specs(formats);
-    // If there is more than one format, this is not valid python and we should
-    // return true so that no error is reported
+    // If there is more than one format, this is not valid Python and we should
+    // return true so that no error is reported.
     let [format] = formats.as_slice() else {
         return true;
     };
@@ -242,8 +249,7 @@ pub(crate) fn bad_string_format_type(checker: &mut Checker, expr: &Expr, right:
             values,
             range: _,
         }) => is_valid_dict(&format_strings, keys, values),
-        Expr::Constant(_) => is_valid_constant(&format_strings, right),
-        _ => true,
+        _ => is_valid_constant(&format_strings, right),
     };
     if !is_valid {
         checker

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

@@ -1,7 +1,7 @@
 use ruff_diagnostics::{Diagnostic, Violation};
 use ruff_macros::{derive_message_formats, violation};
 use ruff_python_ast::{self as ast, Ranged};
-use ruff_python_semantic::analyze::type_inference::PythonType;
+use ruff_python_semantic::analyze::type_inference::{PythonType, ResolvedPythonType};
 
 use crate::checkers::ast::Checker;
 
@@ -46,8 +46,8 @@ pub(crate) fn invalid_envvar_value(checker: &mut Checker, call: &ast::ExprCall)
         };
 
         if matches!(
-            PythonType::from(expr),
-            PythonType::String | PythonType::Unknown
+            ResolvedPythonType::from(expr),
+            ResolvedPythonType::Unknown | ResolvedPythonType::Atom(PythonType::String)
         ) {
             return;
         }

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

@@ -3,7 +3,7 @@ use ruff_python_ast::{Ranged, Stmt};
 use ruff_diagnostics::{Diagnostic, Violation};
 use ruff_macros::{derive_message_formats, violation};
 use ruff_python_ast::{helpers::ReturnStatementVisitor, statement_visitor::StatementVisitor};
-use ruff_python_semantic::analyze::type_inference::PythonType;
+use ruff_python_semantic::analyze::type_inference::{PythonType, ResolvedPythonType};
 
 use crate::checkers::ast::Checker;
 
@@ -42,8 +42,8 @@ pub(crate) fn invalid_str_return(checker: &mut Checker, name: &str, body: &[Stmt
     for stmt in returns {
         if let Some(value) = stmt.value.as_deref() {
             if !matches!(
-                PythonType::from(value),
-                PythonType::String | PythonType::Unknown
+                ResolvedPythonType::from(value),
+                ResolvedPythonType::Unknown | ResolvedPythonType::Atom(PythonType::String)
             ) {
                 checker
                     .diagnostics

crates/ruff/src/rules/pylint/snapshots/ruff__rules__pylint__tests__PLE1307_bad_string_format_type.py.snap

@@ -69,6 +69,16 @@ bad_string_format_type.py:10:1: PLE1307 Format type does not match argument type
 12 | "%d" % ([],)
    |
 
+bad_string_format_type.py:11:1: PLE1307 Format type does not match argument type
+   |
+ 9 | "%x" % 1.1
+10 | "%(key)x" % {"key": 1.1}
+11 | "%d" % []
+   | ^^^^^^^^^ PLE1307
+12 | "%d" % ([],)
+13 | "%(key)d" % {"key": []}
+   |
+
 bad_string_format_type.py:12:1: PLE1307 Format type does not match argument type
    |
 10 | "%(key)x" % {"key": 1.1}
@@ -96,6 +106,7 @@ bad_string_format_type.py:14:7: PLE1307 Format type does not match argument type
 14 | print("%d" % ("%s" % ("nested",),))
    |       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ PLE1307
 15 | "%d" % ((1, 2, 3),)
+16 | "%d" % (1 if x > 0 else [])
    |
 
 bad_string_format_type.py:15:1: PLE1307 Format type does not match argument type
@@ -104,8 +115,17 @@ bad_string_format_type.py:15:1: PLE1307 Format type does not match argument type
 14 | print("%d" % ("%s" % ("nested",),))
 15 | "%d" % ((1, 2, 3),)
    | ^^^^^^^^^^^^^^^^^^^ PLE1307
-16 | 
-17 | # False negatives
+16 | "%d" % (1 if x > 0 else [])
+   |
+
+bad_string_format_type.py:16:1: PLE1307 Format type does not match argument type
+   |
+14 | print("%d" % ("%s" % ("nested",),))
+15 | "%d" % ((1, 2, 3),)
+16 | "%d" % (1 if x > 0 else [])
+   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^ PLE1307
+17 | 
+18 | # False negatives
    |
 
 

crates/ruff/src/rules/pylint/snapshots/ruff__rules__pylint__tests__PLE1507_invalid_envvar_value.py.snap

@@ -31,4 +31,24 @@ invalid_envvar_value.py:8:11: PLE1507 Invalid type for initial `os.getenv` argum
 10 | os.getenv(key=f"foo", default="bar")
    |
 
+invalid_envvar_value.py:12:15: PLE1507 Invalid type for initial `os.getenv` argument; expected `str`
+   |
+10 | os.getenv(key=f"foo", default="bar")
+11 | os.getenv(key="foo" + "bar", default=1)
+12 | os.getenv(key=1 + "bar", default=1)  # [invalid-envvar-value]
+   |               ^^^^^^^^^ PLE1507
+13 | os.getenv("PATH_TEST" if using_clear_path else "PATH_ORIG")
+14 | os.getenv(1 if using_clear_path else "PATH_ORIG")
+   |
+
+invalid_envvar_value.py:14:11: PLE1507 Invalid type for initial `os.getenv` argument; expected `str`
+   |
+12 | os.getenv(key=1 + "bar", default=1)  # [invalid-envvar-value]
+13 | os.getenv("PATH_TEST" if using_clear_path else "PATH_ORIG")
+14 | os.getenv(1 if using_clear_path else "PATH_ORIG")
+   |           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ PLE1507
+15 | 
+16 | AA = "aa"
+   |
+
 

crates/ruff_python_semantic/Cargo.toml

@@ -23,6 +23,7 @@ bitflags = { workspace = true }
 is-macro = { workspace = true }
 num-traits = { workspace = true }
 rustc-hash = { workspace = true }
-
-
 smallvec = { workspace = true }
+
+[dev-dependencies]
+ruff_python_parser = { path = "../ruff_python_parser" }