Add a kwarg to munge to disallow builtins in munged names (#78)

Author: chrisrink10

basilisp/compiler.py

@@ -28,8 +28,8 @@
 import basilisp.reader as reader
 from basilisp.lang.runtime import Var
 from basilisp.lang.typing import LispForm
-from basilisp.lang.util import genname
-from basilisp.util import Maybe, munge
+from basilisp.lang.util import genname, munge
+from basilisp.util import Maybe
 
 _CORE_NS = 'basilisp.core'
 _DEFAULT_FN = '__lisp_expr__'

basilisp/lang/util.py

@@ -1,4 +1,6 @@
+import builtins
 import datetime
+import keyword
 import re
 import uuid
 from typing import Pattern
@@ -30,6 +32,39 @@ def lrepr(f) -> str:
         return repr(f)
 
 
+_MUNGE_REPLACEMENTS = {
+    '+': '__PLUS__',
+    '-': '_',
+    '*': '__STAR__',
+    '/': '__DIV__',
+    '>': '__GT__',
+    '<': '__LT__',
+    '!': '__BANG__',
+    '=': '__EQ__',
+    '?': '__Q__',
+    '\\': '__IDIV__',
+    '&': '__AMP__'
+}
+
+
+def munge(s: str, allow_builtins: bool = True) -> str:
+    """Replace characters which are not valid in Python symbols
+    with valid replacement strings."""
+    new_str = []
+    for c in s:
+        new_str.append(_MUNGE_REPLACEMENTS.get(c, c))
+
+    new_s = ''.join(new_str)
+
+    if keyword.iskeyword(new_s):
+        return f"{new_s}_"
+
+    if not allow_builtins and new_s in builtins.__dict__:
+        return f"{new_s}_"
+
+    return new_s
+
+
 # Use an atomically incremented integer as a suffix for all
 # user-defined function and variable names compiled into Python
 # code so no conflicts occur

basilisp/util.py

@@ -1,6 +1,5 @@
 import functools
 import inspect
-import keyword
 import os.path
 from typing import Optional, Callable, TypeVar, Generic
 
@@ -9,20 +8,6 @@
 
 from basilisp.lang.util import lrepr
 
-_MUNGE_REPLACEMENTS = {
-    '+': '__PLUS__',
-    '-': '_',
-    '*': '__STAR__',
-    '/': '__DIV__',
-    '>': '__GT__',
-    '<': '__LT__',
-    '!': '__BANG__',
-    '=': '__EQ__',
-    '?': '__Q__',
-    '\\': '__IDIV__',
-    '&': '__AMP__'
-}
-
 
 def trace(f):
     @functools.wraps(f)
@@ -102,17 +87,3 @@ def stream(self) -> Sequence:
     @property
     def is_present(self) -> bool:
         return self._inner is not None
-
-
-def munge(s: str) -> str:
-    """Replace characters which are not valid in Python symbols
-    with valid replacement strings."""
-    new_str = []
-    for c in s:
-        new_str.append(_MUNGE_REPLACEMENTS.get(c, c))
-
-    new_s = ''.join(new_str)
-
-    if keyword.iskeyword(new_s):
-        return f"{new_s}_"
-    return new_s

tests/munge_test.py

@@ -1,6 +1,7 @@
+import builtins
 import keyword
 
-from basilisp.util import munge
+from basilisp.lang.util import munge
 
 
 def test_munge_disallows_syms():
@@ -17,6 +18,11 @@ def test_munge_disallows_syms():
     assert "__AMP__form" == munge("&form")
 
 
+def test_munge_disallows_python_builtins():
+    for name in builtins.__dict__.keys():
+        assert f"{name}_" == munge(name, allow_builtins=False)
+
+
 def test_munge_disallows_python_kws():
     for kw in keyword.kwlist:
         assert f"{kw}_" == munge(kw)