package_managers: gomod: Allow pre-release IDs & comments in version …

…string

Golang allows arbitrary pre-release version identifiers in version
string, i.e. trying to parse the following in a go.mod file passes:

    # go.mod
    go 1.21.0foo458

    $ go mod edit -json | jq .Go
    1.21.0foo458

Golang also allows go.mod files to include commentaries on the version
string line, i.e. the following passes the built-in parser:

    # go.mod
    go 1.21.0rc1//commentary

    go mod edit -json | jq .Go
    1.21.rc1

This patches enhances our regex to support both.

Signed-off-by: Erik Skultety <[email protected]>

cachi2/core/package_managers/gomod.py

@@ -678,9 +678,15 @@ def _get_gomod_version(go_mod_file: RootedPath) -> Tuple[Optional[str], Optional
     go_version = None
     toolchain_version = None
 
-    version_str_regex = r"(?P<ver>\d+\.\d+(:?\.\d+)?)"
-    go_version_regex = rf"^\s*go\s+{version_str_regex}\s*$"
-    toolchain_version_regex = rf"^\s*toolchain\s+go{version_str_regex}\s*$"
+    # this needs to be able to handle arbitrary pre-release version identifiers and commentaries
+    # as well, since Go itself can parse it
+    # - 'go 1.21.0'
+    # - '   go 1.21.0rc4'
+    # - 'go 1.21beta1//commentary'
+    version_str_regex = r"(?P<ver>\d+\.\d+(:?\.\d+)?(?:[a-zA-Z]+\d+)?)"
+    post_version_chars_regex = r"\s*(?:\/\/.*)?"
+    go_version_regex = rf"^\s*go\s+{version_str_regex}{post_version_chars_regex}$"
+    toolchain_version_regex = rf"^\s*toolchain\s+go{version_str_regex}{post_version_chars_regex}$"
 
     go_pattern = re.compile(go_version_regex)
     toolchain_pattern = re.compile(toolchain_version_regex)

tests/unit/package_managers/test_gomod.py

@@ -1930,6 +1930,11 @@ def test_fetch_tags_fail(repo_remote_with_tag: tuple[RootedPath, RootedPath]) ->
         pytest.param("go 1.21", "1.21", None, id="go_minor"),
         pytest.param("go 1.21.0", "1.21.0", None, id="go_micro"),
         pytest.param("    go    1.21.4    ", "1.21.4", None, id="go_spaces"),
+        pytest.param("go 1.21rc4", "1.21rc4", None, id="go_minor_rc"),
+        pytest.param("go 1.21.0rc4", "1.21.0rc4", None, id="go_micro_rc"),
+        pytest.param("go 1.21.0  // comment", "1.21.0", None, id="go_commentary"),
+        pytest.param("go 1.21.0//commentary", "1.21.0", None, id="go_commentary_no_spaces"),
+        pytest.param("go 1.21.0beta2//comment", "1.21.0beta2", None, id="go_rc_commentary"),
         pytest.param("   toolchain   go1.21.4  ", None, "1.21.4", id="toolchain_spaces"),
         pytest.param("go 1.21\ntoolchain go1.21.6", "1.21", "1.21.6", id="go_and_toolchain"),
     ],
@@ -1949,6 +1954,8 @@ def test_get_gomod_version(
     "go 1.21.0.100",  # non-conforming to the X.Y(.Z)? versioning template
     "1.21",  # missing 'go' at the beginning
     "go 1.21 foo",  # extra characters after version string
+    "go 1.21prerelease",  # pre-release with no number
+    "go 1.21prerelease_4",  # pre-release with non-alphanum character
     "toolchain 1.21",  # missing 'go' prefix for the toolchain spec
 ]