feat: classify gear and weapon mods

tools/game-data-capture/src/game_data_capture/domain/models.py

@@ -14,6 +14,39 @@ type MissingReason = typing.Literal[
     "not_applicable",
     "redacted",
 ]
+type ScreenshotFamily = typing.Literal[
+    "weapon",
+    "gear",
+    "skill_chip",
+    "os_protocol",
+    "specialization",
+    "weapon_mod",
+    "gear_mod",
+    "unknown",
+]
+type CanonicalRecordType = typing.Literal[
+    "weapon",
+    "weapon_talent",
+    "weapon_augmentation_profile",
+    "gear",
+    "brand_set",
+    "gear_talent",
+    "gear_augmentation_profile",
+    "skill_chip_set",
+    "skill_chip_set_bonus",
+    "skill_chip_roll_pool",
+    "skill_chip_talent",
+    "os_protocol",
+    "os_protocol_secondary_profile",
+    "os_protocol_talent",
+    "specialization",
+    "specialization_focus",
+    "specialization_ability",
+    "specialization_tree_node",
+    "specialization_tree_rank",
+    "weapon_mod",
+    "gear_mod",
+]
 
 
 class SourceEngine(typing.TypedDict):
@@ -76,14 +109,12 @@ class FieldClaim(typing.TypedDict):
 
 
 class ScreenshotClassification(typing.TypedDict):
-    family: typing.Literal[
-        "weapon", "gear", "skill_chip", "os_protocol", "specialization", "unknown"
-    ]
+    family: ScreenshotFamily
     view: str
     gear_slot: (
         typing.Literal["Mask", "Gloves", "Backpack", "Body Armor", "Holster", "Knee Pads"] | None
     )
-    record_types: list[str]
+    record_types: list[CanonicalRecordType]
     confidence: float
     abstention_reason: str | None
 
@@ -99,7 +130,7 @@ class ObservedItemInstance(typing.TypedDict):
     """Values shown for one configured item; they are not reusable definitions."""
 
     instance_id: str
-    family: str
+    family: ScreenshotFamily
     source_screen_ids: list[str]
     observed_context: ObservedContext
     fields: list[FieldClaim]
@@ -109,7 +140,7 @@ class CanonicalDefinitionCandidate(typing.TypedDict):
     """Evidence-backed reusable claims, kept separate from the observed item."""
 
     candidate_id: str
-    record_type: str
+    record_type: CanonicalRecordType
     source_instance_ids: list[str]
     evidence_refs: list[str]
     fields: list[FieldClaim]

tools/game-data-capture/src/game_data_capture/domain/taxonomy.py

@@ -12,6 +12,8 @@ class Family(StrEnum):
     SKILL_CHIP = "skill_chip"
     OS_PROTOCOL = "os_protocol"
     SPECIALIZATION = "specialization"
+    WEAPON_MOD = "weapon_mod"
+    GEAR_MOD = "gear_mod"
     UNKNOWN = "unknown"
 
 
@@ -57,6 +59,8 @@ class RecordType(StrEnum):
     SPECIALIZATION_ABILITY = "specialization_ability"
     SPECIALIZATION_TREE_NODE = "specialization_tree_node"
     SPECIALIZATION_TREE_RANK = "specialization_tree_rank"
+    WEAPON_MOD = "weapon_mod"
+    GEAR_MOD = "gear_mod"
 
 
 SCREEN_VIEWS: Final[Mapping[Family, tuple[str, ...]]] = MappingProxyType(
@@ -80,6 +84,8 @@ SCREEN_VIEWS: Final[Mapping[Family, tuple[str, ...]]] = MappingProxyType(
             "focus_tree",
             "rank_details",
         ),
+        Family.WEAPON_MOD: ("detail",),
+        Family.GEAR_MOD: ("detail",),
         Family.UNKNOWN: ("unknown",),
     }
 )
@@ -117,6 +123,8 @@ RECORD_TYPES_BY_VIEW: Final[Mapping[tuple[Family, str], tuple[RecordType, ...]]]
             RecordType.SPECIALIZATION_TREE_NODE,
         ),
         (Family.SPECIALIZATION, "rank_details"): (RecordType.SPECIALIZATION_TREE_RANK,),
+        (Family.WEAPON_MOD, "detail"): (RecordType.WEAPON_MOD,),
+        (Family.GEAR_MOD, "detail"): (RecordType.GEAR_MOD,),
         (Family.UNKNOWN, "unknown"): (),
     }
 )

tools/game-data-capture/src/game_data_capture/domain/validation.py

@@ -86,7 +86,13 @@ def validate_classification(classification: Mapping[str, object]) -> None:
         raise ContractValidationError("record_types must contain strings")
     if len(record_types) != len(set(record_types)):
         raise ContractValidationError("record_types must be unique")
-    allowed = {record_type.value for record_type in taxonomy.RECORD_TYPES_BY_VIEW[(family, view)]}
+    if len(record_types) > 8:
+        raise ContractValidationError("record_types cannot contain more than 8 values")
+    try:
+        normalized_record_types = {taxonomy.RecordType(record_type) for record_type in record_types}
+    except ValueError as error:
+        raise ContractValidationError("Unknown canonical record type") from error
+    allowed = set(taxonomy.RECORD_TYPES_BY_VIEW[(family, view)])
     if family is taxonomy.Family.UNKNOWN:
         if record_types:
             raise ContractValidationError("Unknown screens cannot claim canonical record types")
@@ -94,7 +100,7 @@ def validate_classification(classification: Mapping[str, object]) -> None:
         if reason not in taxonomy.ABSTENTION_REASONS:
             raise ContractValidationError("Unknown screens require an abstention reason")
     else:
-        if not record_types or not set(record_types).issubset(allowed):
+        if not record_types or not normalized_record_types.issubset(allowed):
             raise ContractValidationError(
                 f"record_types are incompatible with {family.value}/{view}"
             )