fix: dont strip uppercase prefixes (#589)

Author: a-zw

src/extensions/score_metamodel/checks/check_options.py

@@ -32,25 +32,20 @@ def get_need_type(needs_types: list[ScoreNeedType], directive: str) -> ScoreNeed
     raise ValueError(f"Need type {directive} not found in needs_types")
 
 
-def _get_normalized(need: NeedItem, key: str, remove_prefix: bool = False) -> list[str]:
+def _get_normalized(need: NeedItem, key: str) -> list[str]:
     """Normalize a raw value into a list of strings."""
     raw_value = need.get(key, None)
     if not raw_value:
         return []
     if isinstance(raw_value, str):
-        if remove_prefix:
-            return [_remove_namespace_prefix_(raw_value)]
         return [raw_value]
     if isinstance(raw_value, list):
         # Verify all elements are strings
         raw_list = cast(list[object], raw_value)
         for item in raw_list:
             if not isinstance(item, str):
                 raise ValueError
-        str_list = cast(list[str], raw_value)
-        if remove_prefix:
-            return [_remove_namespace_prefix_(v) for v in str_list]
-        return str_list
+        return cast(list[str], raw_value)
     raise ValueError
 
 
@@ -73,11 +68,6 @@ def _validate_value_pattern(
         ) from e
 
 
-def _remove_namespace_prefix_(word: str) -> str:
-    # If the word starts with uppercase letters followed by an underscore, remove them.
-    return re.sub(r"^[A-Z]+_", "", word)
-
-
 def validate_options(
     log: CheckLogger,
     need_type: ScoreNeedType,
@@ -144,6 +134,22 @@ def _validate(attributes_to_allowed_values: dict[str, str], mandatory: bool):
 #         )
 
 
+def _to_link_pattern(value: "str | ScoreNeedType") -> str:
+    """Convert a link constraint to a regex pattern.
+
+    - A plain type name like ``"stkh_req"`` becomes ``^stkh_req__``
+      (matching IDs that start with the type name followed by ``__``).
+    - A string already starting with ``^`` is treated as an explicit regex
+      and returned unchanged.
+    - A ScoreNeedType dict uses its ``mandatory_options.id`` pattern.
+    """
+    if isinstance(value, str):
+        if value.startswith("^"):
+            return value
+        return f"^{value}__"
+    return value["mandatory_options"]["id"]
+
+
 def validate_links(
     log: CheckLogger,
     need_type: ScoreNeedType,
@@ -159,16 +165,11 @@ def _validate(
         treat_as_info: bool = False,
     ):
         for attribute, allowed_values in attributes_to_allowed_values.items():
-            values = _get_normalized(need, attribute, remove_prefix=True)
+            values = _get_normalized(need, attribute)
             if mandatory and not values:
                 log.warning_for_need(need, f"is missing required link: `{attribute}`.")
 
-            allowed_regex = "|".join(
-                [
-                    v if isinstance(v, str) else v["mandatory_options"]["id"]
-                    for v in allowed_values
-                ]
-            )
+            allowed_regex = "|".join(_to_link_pattern(v) for v in allowed_values)
 
             # regex based validation
             for value in values:

src/extensions/score_metamodel/tests/test_check_options.py

@@ -21,6 +21,7 @@
     check_extra_options,
     check_options,
     parse_milestone,
+    validate_links,
 )
 from score_metamodel.tests import fake_check_logger, need
 from sphinx.application import Sphinx  # type: ignore[import-untyped]
@@ -124,3 +125,125 @@ def test_milestone_parsing():
     assert parse_milestone("v0.5") == (0, 5, 0)
     assert parse_milestone("v1.0") == (1, 0, 0)
     assert parse_milestone("v1.0.1") == (1, 0, 1)
+
+
+class TestValidateLinks:
+    """Tests for validate_links() — specifically the prefix-stripping bug (#588)."""
+
+    NEED_TYPE_WITH_LINK_BY_NAME: ScoreNeedType = {
+        "title": "Test Type",
+        "prefix": "TT",
+        "tags": [],
+        "parts": 1,
+        "directive": "test_type",
+        "mandatory_options": {
+            "id": "^test_type__.*$",
+            "status": "^(draft|valid)$",
+        },
+        "optional_options": {},
+        "mandatory_links": {},
+        "optional_links": {
+            "satisfies": ["product"],
+        },
+    }
+
+    NEED_TYPE_WITH_LINK_BY_REGEX: ScoreNeedType = {
+        "title": "Test Type",
+        "prefix": "TT",
+        "tags": [],
+        "parts": 1,
+        "directive": "test_type",
+        "mandatory_options": {
+            "id": "^test_type__.*$",
+            "status": "^(draft|valid)$",
+        },
+        "optional_options": {},
+        "mandatory_links": {},
+        "optional_links": {
+            "satisfies": ["^product__.*$"],
+        },
+    }
+
+    def test_link_matches_by_type_name_no_prefix(self):
+        """A link value that starts with the type name should pass validation."""
+        need_1 = need(
+            id="test_type__001",
+            type="test_type",
+            satisfies="product__foo",
+        )
+        logger = fake_check_logger()
+        validate_links(
+            cast(CheckLogger, logger),
+            self.NEED_TYPE_WITH_LINK_BY_NAME,
+            need_1,
+        )
+        logger.assert_no_warnings()
+
+    def test_link_with_uppercase_prefix_no_longer_stripped(self):
+        """
+        After the fix for #588, uppercase prefixes are no longer stripped.
+        P_FOO stays as P_FOO and correctly fails to match the 'product' type pattern.
+        """
+        need_1 = need(
+            id="test_type__001",
+            type="test_type",
+            satisfies="P_FOO",
+        )
+        logger = fake_check_logger()
+        validate_links(
+            cast(CheckLogger, logger),
+            self.NEED_TYPE_WITH_LINK_BY_NAME,
+            need_1,
+        )
+        # P_FOO is preserved as-is (no stripping) and doesn't match ^product__
+        logger.assert_warning("references 'P_FOO' as 'satisfies'")
+
+    def test_link_with_uppercase_prefix_not_stripped_for_regex(self):
+        """
+        After the fix for #588, uppercase prefixes are no longer stripped.
+        P_product__foo stays as P_product__foo and does NOT match ^product__.*$
+        because the P_ prefix is preserved.
+        """
+        need_1 = need(
+            id="test_type__001",
+            type="test_type",
+            satisfies="P_product__foo",
+        )
+        logger = fake_check_logger()
+        validate_links(
+            cast(CheckLogger, logger),
+            self.NEED_TYPE_WITH_LINK_BY_REGEX,
+            need_1,
+        )
+        # P_product__foo is preserved as-is and doesn't match ^product__.*$
+        logger.assert_warning("references 'P_product__foo' as 'satisfies'")
+
+    def test_link_without_prefix_matches_by_name(self):
+        """A link value without any prefix that starts with the type name passes."""
+        need_1 = need(
+            id="test_type__001",
+            type="test_type",
+            satisfies="product__bar",
+        )
+        logger = fake_check_logger()
+        validate_links(
+            cast(CheckLogger, logger),
+            self.NEED_TYPE_WITH_LINK_BY_NAME,
+            need_1,
+        )
+        logger.assert_no_warnings()
+
+    def test_link_not_matching_any_type_warns(self):
+        """A link value that doesn't match any allowed type should warn."""
+        need_1 = need(
+            id="test_type__001",
+            type="test_type",
+            satisfies="unknown__thing",
+        )
+        logger = fake_check_logger()
+        validate_links(
+            cast(CheckLogger, logger),
+            self.NEED_TYPE_WITH_LINK_BY_NAME,
+            need_1,
+        )
+        logger.assert_warning("references 'unknown__thing' as 'satisfies'")