refactor: add _as_python_identifier() function

The function takes a PostgreSQL "invalid" name (such as `a-b` or '0' or an
unicode string) and makes a valid Python name.

Currently used to make valid namedtuples from `nametuple_row()` but can
be used to make other valid Python names.

diff --git a/psycopg/psycopg/_encodings.py b/psycopg/psycopg/_encodings.py
index a44f3bda035b70fdde5fc4d804a8a6f0ae93e23b..f672fd521e3dd8214731aa8bb0e0a15b0f65ec01 100644 (file)
--- a/psycopg/psycopg/_encodings.py
+++ b/psycopg/psycopg/_encodings.py
@@ -4,6 +4,8 @@ Mappings between PostgreSQL and Python encodings.
 
 # Copyright (C) 2020 The Psycopg Team
 
+import re
+import string
 import codecs
 from typing import Any, Dict, Optional, TYPE_CHECKING
 
@@ -131,3 +133,23 @@ def pg2pyenc(name: bytes) -> str:
     except KeyError:
         sname = name.decode("utf8", "replace")
         raise NotSupportedError(f"codec not available in Python: {sname!r}")
+
+
+def _as_python_identifier(s: str, prefix: str = "f") -> str:
+    """
+    Reduce a string to a valid Python identifier.
+
+    Replace all non-valid chars with '_' and prefix the value with *prefix* if
+    the first letter is an '_'.
+    """
+    s = _re_clean.sub("_", s)
+    # Python identifier cannot start with numbers, namedtuple fields
+    # cannot start with underscore. So...
+    if s[0] == "_" or "0" <= s[0] <= "9":
+        s = prefix + s
+    return s
+
+
+_re_clean = re.compile(
+    f"[^{string.ascii_lowercase}{string.ascii_uppercase}{string.digits}_]"
+)

diff --git a/psycopg/psycopg/rows.py b/psycopg/psycopg/rows.py
index 99194cd97f117fdc7570b093721f1eb2a15b07b3..2f2d64728ae8bfafde955785beab16d312c73c07 100644 (file)
--- a/psycopg/psycopg/rows.py
+++ b/psycopg/psycopg/rows.py
@@ -4,7 +4,6 @@ psycopg row factories
 
 # Copyright (C) 2021 The Psycopg Team
 
-import re
 import functools
 from typing import Any, Callable, Dict, NamedTuple, NoReturn, Sequence, Tuple
 from typing import TYPE_CHECKING, Type, TypeVar
@@ -12,6 +11,7 @@ from collections import namedtuple
 
 from . import errors as e
 from ._compat import Protocol, TypeAlias
+from ._encodings import _as_python_identifier
 
 if TYPE_CHECKING:
     from .cursor import BaseCursor, Cursor
@@ -138,20 +138,11 @@ def namedtuple_row(
     return nt._make
 
 
-# ascii except alnum and underscore
-_re_clean = re.compile("[" + re.escape(" !\"#$%&'()*+,-./:;<=>?@[\\]^`{|}~") + "]")
-
-
 @functools.lru_cache(512)
 def _make_nt(*key: str) -> Type[NamedTuple]:
     fields = []
     for s in key:
-        s = _re_clean.sub("_", s)
-        # Python identifier cannot start with numbers, namedtuple fields
-        # cannot start with underscore. So...
-        if s[0] == "_" or "0" <= s[0] <= "9":
-            s = "f" + s
-        fields.append(s)
+        fields.append(_as_python_identifier(s))
     return namedtuple("Row", fields)  # type: ignore[return-value]