Refactor canonicalize_version (pypa#793)

jaraco · brettcannon · web-flow · commit 680c31aa25d2 · 2024-09-05T18:55:17.000Z
* In canonicalize_version, re-use Version.__str__.

* Utilize singledispatch to separate concerns in canonicalize_version.

---------

Co-authored-by: Brett Cannon &lt;brett@python.org&gt;
diff --git a/src/packaging/utils.py b/src/packaging/utils.py
@@ -4,11 +4,12 @@
 
 from __future__ import annotations
 
+import functools
 import re
 from typing import NewType, Tuple, Union, cast
 
 from .tags import Tag, parse_tag
-from .version import InvalidVersion, Version
+from .version import InvalidVersion, Version, _TrimmedRelease
 
 BuildTag = Union[Tuple[()], Tuple[int, str]]
 NormalizedName = NewType("NormalizedName", str)
@@ -54,52 +55,40 @@ def is_normalized_name(name: str) -> bool:
     return _normalized_regex.match(name) is not None
 
 
+@functools.singledispatch
 def canonicalize_version(
     version: Version | str, *, strip_trailing_zero: bool = True
 ) -> str:
     """
-    This is very similar to Version.__str__, but has one subtle difference
-    with the way it handles the release segment.
-    """
-    if isinstance(version, str):
-        try:
-            parsed = Version(version)
-        except InvalidVersion:
-            # Legacy versions cannot be normalized
-            return version
-    else:
-        parsed = version
-
-    parts = []
+    Return a canonical form of a version as a string.
 
-    # Epoch
-    if parsed.epoch != 0:
-        parts.append(f"{parsed.epoch}!")
+    >>> canonicalize_version('1.0.1')
+    '1.0.1'
 
-    # Release segment
-    release_segment = ".".join(str(x) for x in parsed.release)
-    if strip_trailing_zero:
-        # NB: This strips trailing '.0's to normalize
-        release_segment = re.sub(r"(\.0)+$", "", release_segment)
-    parts.append(release_segment)
+    Per PEP 625, versions may have multiple canonical forms, differing
+    only by trailing zeros.
 
-    # Pre-release
-    if parsed.pre is not None:
-        parts.append("".join(str(x) for x in parsed.pre))
+    >>> canonicalize_version('1.0.0')
+    '1'
+    >>> canonicalize_version('1.0.0', strip_trailing_zero=False)
+    '1.0.0'
 
-    # Post-release
-    if parsed.post is not None:
-        parts.append(f".post{parsed.post}")
+    Invalid versions are returned unaltered.
 
-    # Development release
-    if parsed.dev is not None:
-        parts.append(f".dev{parsed.dev}")
+    >>> canonicalize_version('foo bar baz')
+    'foo bar baz'
+    """
+    return str(_TrimmedRelease(str(version)) if strip_trailing_zero else version)
 
-    # Local version segment
-    if parsed.local is not None:
-        parts.append(f"+{parsed.local}")
 
-    return "".join(parts)
+@canonicalize_version.register
+def _(version: str, *, strip_trailing_zero: bool = True) -> str:
+    try:
+        parsed = Version(version)
+    except InvalidVersion:
+        # Legacy versions cannot be normalized
+        return version
+    return canonicalize_version(parsed, strip_trailing_zero=strip_trailing_zero)
 
 
 def parse_wheel_filename(
diff --git a/src/packaging/version.py b/src/packaging/version.py
@@ -451,6 +451,23 @@ def micro(self) -> int:
         return self.release[2] if len(self.release) >= 3 else 0
 
 
+class _TrimmedRelease(Version):
+    @property
+    def release(self) -> tuple[int, ...]:
+        """
+        Release segment without any trailing zeros.
+
+        >>> _TrimmedRelease('1.0.0').release
+        (1,)
+        >>> _TrimmedRelease('0.0').release
+        (0,)
+        """
+        rel = super().release
+        nonzeros = (index for index, val in enumerate(rel) if val)
+        last_nonzero = max(nonzeros, default=0)
+        return rel[: last_nonzero + 1]
+
+
 def _parse_letter_version(
     letter: str | None, number: str | bytes | SupportsInt | None
 ) -> tuple[str, int] | None:
