!squash more

src/unihan_etl/validator.py

@@ -18,38 +18,62 @@ class kTGHZ2013Location(pydantic.BaseModel):
     position: int
     # 0 for a main entry and greater than 0 for a parenthesized or bracketed variant #
     # of the main entry
-    entry_type: int
+    entry_type: int = pydantic.Field(
+        description="0 for a main entry and greater than 0 for a parenthesized or bracketed variant of the main entry"
+    )
 
 
-class kTGHZ2013(UCNBaseModel):
+class kTGHZ2013Reading(pydantic.BaseModel):
     """kTGHZ2013 model."""
 
     reading: str
     locations: list[kTGHZ2013Location]
 
-    model_config = pydantic.ConfigDict(validate_assignment=True)
-
-    # @classmethod
-    # def validator(cls, raw: str) -> "kTGHZ2013":
-    #     if len(raw) == 1:
-    #         ucn, field, val = raw[0].split("\t")
-    #         print(ucn, field, val)
-    #         out = expand_kTGHZ2013([val])
-    #
-    #         return cls(
-    #             ucn=ucn, reading=out[0]["reading"], locations=out[0]["locations"]
-    #         )
-    #
-
-    def __init__(self, *args, **kwargs) -> None:
-        if args and len(args) == 1:
-            ucn, field, val = args[0].split("\t")
-            print(ucn, field, val)
-
-            out = expand_kTGHZ2013([val])
-
-            return super().__init__(
-                ucn=ucn, reading=out[0]["reading"], locations=out[0]["locations"]
-            )
 
-        return super().__init__(*args, **kwargs)
+class kTGHZ2013(UCNBaseModel):
+    """kTGHZ2013 model."""
+
+    readings: list[kTGHZ2013Reading]
+
+    model_config = pydantic.ConfigDict(
+        validate_assignment=True,
+        arbitrary_types_allowed=True,
+    )
+
+    @classmethod
+    def from_string(cls, value: str) -> "kTGHZ2013":
+        """Accept csv valdation from UNIHAN."""
+        if isinstance(value, str):
+            ucn, field, val = value.split("\t")
+            outs = expand_kTGHZ2013(val.split(" "))
+
+            return cls(
+                ucn=ucn,
+                readings=[
+                    kTGHZ2013Reading(
+                        reading=out["reading"],
+                        locations=[
+                            kTGHZ2013Location(
+                                page=loc["page"],
+                                position=loc["position"],
+                                entry_type=loc["entry_type"],
+                            )
+                            for loc in out["locations"]
+                        ],
+                    )
+                    for out in outs
+                ],
+            )
+            #     reading=out["reading"],
+            #     locations=[
+            #         kTGHZ2013Location(
+            #             page=location["page"],
+            #             position=location["position"],
+            #             entry_type=location["entry_type"],
+            #         )
+            #         for location in out["locations"]
+            #     ],
+            # )
+        elif isinstance(value, dict):
+            return pydantic.parse_obj_as(cls, value)
+        raise pydantic.ValidationError("Invalid input for kTGHZ2013 model.")  # noqa: TRY003

tests/test_validator.py

@@ -10,5 +10,12 @@
 
 
 def test_kTGHZ2013() -> None:
-    model = validator.kTGHZ2013("U+3447	kTGHZ2013	482.140:zhòu")
+    model = validator.kTGHZ2013.from_string("U+3447	kTGHZ2013	482.140:zhòu")
     assert model.ucn == "U+3447"
+
+    model = validator.kTGHZ2013.from_string(
+        "U+4E07	kTGHZ2013	256.090:mò 379.160:wàn"
+    )
+    assert model.ucn == "U+4E07"
+
+    print(f"\n{model}\n")