New codemod: use-set-literal

integration_tests/test_use_set_literal.py

@@ -0,0 +1,29 @@
+from core_codemods.use_set_literal import UseSetLiteral
+from integration_tests.base_test import (
+    BaseIntegrationTest,
+    original_and_expected_from_code_path,
+)
+
+
+class TestUseSetLiteral(BaseIntegrationTest):
+    codemod = UseSetLiteral
+    code_path = "tests/samples/set_literal.py"
+
+    original_code, expected_new_code = original_and_expected_from_code_path(
+        code_path,
+        [(0, "x = {1, 2, 3}\n"), (1, "y = set()\n")],
+    )
+
+    expected_diff = """\
+--- 
++++ 
+@@ -1,2 +1,2 @@
+-x = set([1, 2, 3])
+-y = set([])
++x = {1, 2, 3}
++y = set()
+"""
+
+    expected_line_change = "1"
+    num_changes = 2
+    change_description = UseSetLiteral.CHANGE_DESCRIPTION

src/codemodder/scripts/generate_docs.py

@@ -178,6 +178,10 @@ class DocMetadata:
         importance="Low",
         guidance_explained="Since literals and new objects have their own identities, comparisons against them using `is` operators are most likely a bug and thus we deem the change safe.",
     ),
+    "use-set-literal": DocMetadata(
+        importance="Low",
+        guidance_explained="We believe this change is safe and will not cause any issues.",
+    ),
 }
 
 

src/core_codemods/__init__.py

@@ -27,6 +27,7 @@
 from .url_sandbox import UrlSandbox
 from .use_defused_xml import UseDefusedXml
 from .use_generator import UseGenerator
+from .use_set_literal import UseSetLiteral
 from .use_walrus_if import UseWalrusIf
 from .with_threading_lock import WithThreadingLock
 from .secure_flask_session_config import SecureFlaskSessionConfig
@@ -71,6 +72,7 @@
         UrlSandbox,
         UseDefusedXml,
         UseGenerator,
+        UseSetLiteral,
         UseWalrusIf,
         WithThreadingLock,
         SQLQueryParameterization,

src/core_codemods/docs/pixee_python_use-set-literal.md

@@ -0,0 +1,7 @@
+This codemod converts Python set constructions using literal list arguments into more efficient and readable set literals. It simplifies expressions like `set([1, 2, 3])` to `{1, 2, 3}`, enhancing both performance and code clarity.
+
+Our changes look like this:
+```diff
+-x = set([1, 2, 3])
++x = {1, 2, 3}
+```

src/core_codemods/use_set_literal.py

@@ -0,0 +1,28 @@
+import libcst as cst
+
+from codemodder.codemods.api import BaseCodemod, ReviewGuidance
+from codemodder.codemods.utils_mixin import NameResolutionMixin
+
+
+class UseSetLiteral(BaseCodemod, NameResolutionMixin):
+    NAME = "use-set-literal"
+    SUMMARY = "Use Set Literals Instead of Sets from Lists"
+    REVIEW_GUIDANCE = ReviewGuidance.MERGE_WITHOUT_REVIEW
+    DESCRIPTION = "Replace sets from lists with set literals"
+    REFERENCES: list = []
+
+    def leave_Call(self, original_node: cst.Call, updated_node: cst.Call):
+        match original_node.func:
+            case cst.Name("set"):
+                if self.is_builtin_function(original_node):
+                    match original_node.args:
+                        case [cst.Arg(value=cst.List(elements=elements))]:
+                            # Can't use set literal for empty set
+                            if len(elements) == 0:
+                                self.add_change(original_node, self.CHANGE_DESCRIPTION)
+                                return updated_node.with_changes(args=[])
+
+                            self.add_change(original_node, self.CHANGE_DESCRIPTION)
+                            return cst.Set(elements=elements)
+
+        return updated_node

tests/codemods/test_use_set_literal.py

@@ -0,0 +1,43 @@
+from core_codemods.use_set_literal import UseSetLiteral
+from tests.codemods.base_codemod_test import BaseCodemodTest
+
+
+class TestUseSetLiteral(BaseCodemodTest):
+    codemod = UseSetLiteral
+
+    def test_simple(self, tmpdir):
+        original_code = """
+        x = set([1, 2, 3])
+        """
+        expected_code = """
+        x = {1, 2, 3}
+        """
+        self.run_and_assert(tmpdir, original_code, expected_code)
+
+    def test_empty_list(self, tmpdir):
+        original_code = """
+        x = set([])
+        """
+        expected_code = """
+        x = set()
+        """
+        self.run_and_assert(tmpdir, original_code, expected_code)
+
+    def test_already_empty(self, tmpdir):
+        original_code = """
+        x = set()
+        """
+        self.run_and_assert(tmpdir, original_code, original_code)
+
+    def test_not_builtin(self, tmpdir):
+        original_code = """
+        from whatever import set
+        x = set([1, 2, 3])
+        """
+        self.run_and_assert(tmpdir, original_code, original_code)
+
+    def test_not_list_literal(self, tmpdir):
+        original_code = """
+        x = set(some_previously_defined_list)
+        """
+        self.run_and_assert(tmpdir, original_code, original_code)

tests/samples/set_literal.py

@@ -0,0 +1,2 @@
+x = set([1, 2, 3])
+y = set([])