dedupe code for normalization and the actual version parsing

Author: aignas

python/private/version.bzl

@@ -503,40 +503,31 @@ def normalize_pep440(version):
     Returns:
       string containing the normalized version.
     """
-    parser = _new(version.strip())  # PEP 440: Leading and Trailing Whitespace
-    accept(parser, _is("v"), "")  # PEP 440: Preceding v character
-    accept_epoch(parser)
-    accept_release(parser)
-    accept_prerelease(parser)
-    accept_postrelease(parser)
-    accept_devrelease(parser)
-    accept_local(parser)
-    if parser.input[parser.context()["start"]:]:
-        fail(
-            "Failed to parse PEP 440 version identifier '%s'." % parser.input,
-            "Parse error at '%s'" % parser.input[parser.context()["start"]:],
-        )
-    return parser.context()["norm"]
+    return _parse(version, strict = True)["norm"]
 
-def version(version_str, strict = False):
-    """Parse a PEP4408 compliant version
+def _parse(version_str, strict = True):
+    """Escape the version component of a filename.
 
     See https://packaging.python.org/en/latest/specifications/binary-distribution-format/#escaping-and-unicode
     and https://peps.python.org/pep-0440/
 
     Args:
       version_str: version string to be normalized according to PEP 440.
-      strict: fail if the version is invalid.
+      strict: fail if the version is invalid, defaults to True.
 
     Returns:
       string containing the normalized version.
     """
+    if strict:
+        version_str = version_str.strip()  # PEP 440: Leading and Trailing Whitespace
+        is_prefix = False
+    else:
+        version_str = version_str.strip()  # PEP 440: Leading and Trailing Whitespace
+        is_prefix = version_str.endswith(".*")
+        version_str = version_str.strip(" .*")  # PEP 440: Leading and Trailing Whitespace and ".*"
 
-    parser = _new(version_str.strip(" " if strict else " .*"))  # PEP 440: Leading and Trailing Whitespace and .*
-
+    parser = _new(version_str)
     accept(parser, _is("v"), "")  # PEP 440: Preceding v character
-
-    parts = {}
     fns = [
         ("epoch", accept_epoch),
         ("release", accept_release),
@@ -546,15 +537,15 @@ def version(version_str, strict = False):
         ("local", accept_local),
     ]
 
-    for p, fn in fns:
+    parts = {
+        "is_prefix": is_prefix,
+    }
+    for key, fn in fns:
         start = len(parser.context()["norm"])
         fn(parser)
-        parts[p] = parser.context()["norm"][start:]
-
+        parts[key] = parser.context()["norm"][start:]
     parts["norm"] = parser.context()["norm"]
 
-    is_prefix = version_str.endswith(".*")
-    parts["is_prefix"] = is_prefix
     if is_prefix and (parts["local"] or parts["post"] or parts["dev"] or parts["pre"]):
         if strict:
             fail("local version part has been obtained, but only public segments can have prefix matches")
@@ -569,7 +560,26 @@ def version(version_str, strict = False):
                 "Parse error at '%s'" % parser.input[parser.context()["start"]:],
             )
 
-        # If we fail to parse the version return None
+        return None
+
+    return parts
+
+def version(version_str, strict = False):
+    """Parse a PEP4408 compliant version
+
+    See https://packaging.python.org/en/latest/specifications/binary-distribution-format/#escaping-and-unicode
+    and https://peps.python.org/pep-0440/
+
+    Args:
+      version_str: version string to be normalized according to PEP 440.
+      strict: fail if the version is invalid.
+
+    Returns:
+      string containing the normalized version.
+    """
+
+    parts = _parse(version_str, strict = strict)
+    if not parts:
         return None
 
     return _new_version(**parts)
@@ -688,9 +698,9 @@ def _version_ne(left, right):
 def _prefix_err(left, op, right):
     if left.is_prefix or right.is_prefix:
         fail("PEP440: only '==' and '!=' operators can use prefix matching: {} {} {}".format(
-            left,
+            left.str(),
             op,
-            right,
+            right.str(),
         ))
 
 def _version_lt(left, right):