Fix: support Alpine-specific version formats that Gentoo regex rejects (fixes #59)

Root cause: AlpineLinuxVersion.is_valid() delegates directly to gentoo.is_valid(),
which only accepts the Gentoo version grammar. Alpine extends that grammar with
extra patterns: a letter+digit portable-release suffix (e.g. "1.9.5p2"), the
_git/_cvs/_svn snapshot suffixes, dash as a numeric component separator
("1.11-20-r0"), and minor malformations found in real package databases
("0.12.5.-r0", "0.8.21.r2").

Fix: override AlpineLinuxVersion.normalize() with _normalize_alpine_to_gentoo(),
which rewrites these Alpine-only patterns into their Gentoo equivalents before
validation and comparison:
  "1.9.5p2-r0"           -> "1.9.5_p2-r0"
  "5.15.3_git20200401-r0" -> "5.15.3_alpha20200401-r0"
  "1.11-20-r0"           -> "1.11.20-r0"
  "0.12.5.-r0"           -> "0.12.5-r0"
  "0.8.21.r2"            -> "0.8.21-r2"

Author: shivamtiwari3

src/univers/versions.py

@@ -4,6 +4,8 @@
 #
 # Visit https://aboutcode.org and https://github.com/aboutcode-org/univers for support and download.
 
+import re
+
 import attr
 import semantic_version
 from packaging import version as packaging_version
@@ -58,6 +60,64 @@ def is_valid_alpine_version(s):
     return str(i) == left
 
 
+def _normalize_alpine_to_gentoo(string):
+    """
+    Normalize an Alpine Linux version string to a Gentoo-compatible format
+    so that Gentoo's vercmp can compare it correctly.
+
+    Alpine extends Gentoo-style versioning with additional patterns:
+    - A dot immediately before the revision marker ("0.12.5.-r0" -> "0.12.5-r0")
+    - Revision without a leading dash ("0.8.21.r2" -> "0.8.21-r2")
+    - Alpine-only suffix words: git, cvs, svn (mapped to alpha), rev/jdk (mapped to p)
+    - A single letter + digit directly after the dotted version ("1.9.5p2-r0" -> "1.9.5_p2-r0")
+    - A dash as a numeric version separator ("1.11-20-r0" -> "1.11.20-r0")
+
+    For example:
+    >>> _normalize_alpine_to_gentoo("1.9.5p2-r0")
+    '1.9.5_p2-r0'
+    >>> _normalize_alpine_to_gentoo("3.3.3p1-r3")
+    '3.3.3_p1-r3'
+    >>> _normalize_alpine_to_gentoo("5.15.3_git20200401-r0")
+    '5.15.3_alpha20200401-r0'
+    >>> _normalize_alpine_to_gentoo("1.11-20-r0")
+    '1.11.20-r0'
+    >>> _normalize_alpine_to_gentoo("57-1-r2")
+    '57.1-r2'
+    >>> _normalize_alpine_to_gentoo("0.12.5.-r0")
+    '0.12.5-r0'
+    >>> _normalize_alpine_to_gentoo("0.8.21.r2")
+    '0.8.21-r2'
+    >>> _normalize_alpine_to_gentoo("1.2.3-r1")
+    '1.2.3-r1'
+    >>> _normalize_alpine_to_gentoo("1.2.3_alpha1-r1")
+    '1.2.3_alpha1-r1'
+    """
+    # Handle ".rN" (no dash before revision): "0.8.21.r2" -> "0.8.21-r2"
+    string = re.sub(r"\.r(\d+)$", r"-r\1", string)
+
+    # Handle trailing dot before revision: "0.12.5.-r0" -> "0.12.5-r0"
+    string = re.sub(r"\.-r(\d+)$", r"-r\1", string)
+
+    # Map Alpine-only suffix words to Gentoo equivalents.
+    # Must be done before the letter+digit substitution below to avoid mangling them.
+    # _git, _cvs, _svn are snapshot/SCM builds -> treat as pre-release (alpha)
+    string = re.sub(r"_(git|cvs|svn)(\d*)", r"_alpha\2", string)
+    # _rev, _jdk -> treat as patch release (p)
+    string = re.sub(r"_(rev|jdk)(\d*)", r"_p\2", string)
+
+    # Handle single letter+digit suffix: "1.9.5p2-r0" -> "1.9.5_p2-r0"
+    # Matches a letter immediately preceded by a digit, followed by one or more
+    # digits, just before the revision marker "-rN" or end of string.
+    string = re.sub(r"(?<=\d)([a-zA-Z])(\d+)(?=-r\d|$)", r"_\1\2", string)
+
+    # Handle dash as a numeric version-component separator: "1.11-20-r0" -> "1.11.20-r0"
+    # Replaces "-N" only when N is a pure digit run not followed by the revision
+    # marker "rN" (i.e., skip "-r0", "-r1", …).
+    string = re.sub(r"-(?!r\d)(\d+)", r".\1", string)
+
+    return string
+
+
 @attr.s(frozen=True, order=True, eq=True, hash=True)
 class Version:
     """
@@ -431,6 +491,11 @@ def __gt__(self, other):
 
 
 class AlpineLinuxVersion(GentooVersion):
+    @classmethod
+    def normalize(cls, string):
+        string = super().normalize(string)
+        return _normalize_alpine_to_gentoo(string)
+
     @classmethod
     def is_valid(cls, string):
         return is_valid_alpine_version(string) and gentoo.is_valid(string)

tests/test_alpine.py

@@ -39,6 +39,56 @@ def test_alpine_vers_cmp2(test_case):
     avc.assert_result()
 
 
+@pytest.mark.parametrize(
+    ("version", "expected_value"),
+    [
+        # dot immediately before revision marker (issue #59)
+        ("0.12.5.-r0", "0.12.5-r0"),
+        # dot instead of dash before revision number (issue #59)
+        ("0.8.21.r2", "0.8.21-r2"),
+        # dash as numeric version-component separator (issue #59)
+        ("1.11-20-r0", "1.11.20-r0"),
+        ("57-1-r2", "57.1-r2"),
+        # single letter + digit suffix, e.g. OpenSSH portable releases (issue #59)
+        ("1.9.5p2-r0", "1.9.5_p2-r0"),
+        ("3.3.3p1-r3", "3.3.3_p1-r3"),
+        ("6.6.2p1-r0", "6.6.2_p1-r0"),
+        ("6.6.4p1-r1", "6.6.4_p1-r1"),
+        ("6.7.1p1-r1", "6.7.1_p1-r1"),
+        # _git snapshot suffix mapped to _alpha for comparison (issue #59)
+        ("5.15.3_git20200401-r0", "5.15.3_alpha20200401-r0"),
+        ("5.15.3_git20210510-r0", "5.15.3_alpha20210510-r0"),
+    ],
+)
+def test_alpine_extended_version_formats(version, expected_value):
+    """Versions with Alpine-specific patterns must parse and normalise correctly."""
+    v = AlpineLinuxVersion(version)
+    assert v.value == expected_value
+
+
+@pytest.mark.parametrize(
+    ("smaller", "larger"),
+    [
+        # portable-release ordering: p1 < p2
+        ("1.9.5p1-r0", "1.9.5p2-r0"),
+        # git snapshot is a pre-release, comes before the stable release
+        ("5.15.3_git20200401-r0", "5.15.3-r0"),
+        # earlier git snapshot < later git snapshot
+        ("5.15.3_git20200401-r0", "5.15.3_git20210510-r0"),
+        # dash-separated version component ordering
+        ("1.11-20-r0", "1.11-21-r0"),
+        ("57-1-r2", "57-2-r0"),
+        # dot-r vs normal version
+        ("0.8.21.r2", "0.8.22-r0"),
+    ],
+)
+def test_alpine_extended_version_comparison(smaller, larger):
+    """Extended Alpine version formats must compare in the correct order."""
+    v1 = AlpineLinuxVersion(smaller)
+    v2 = AlpineLinuxVersion(larger)
+    assert v1 < v2
+
+
 @pytest.mark.parametrize(
     "test_case",
     [