Fix Pydantic model parsing (abetlen#1087)

DeNeutoy · web-flow · commit c689ccc7286b · 2024-01-15T10:45:57.000-05:00
diff --git a/llama_cpp/llama_grammar.py b/llama_cpp/llama_grammar.py
@@ -1433,7 +1433,6 @@ def _add_rule(self, name: str, rule: str):
 
     def visit(self, schema: Dict[str, Any], name: str) -> str:
         schema_type: Optional[str] = schema.get("type") # type: ignore
-        assert isinstance(schema_type, str), f"Unrecognized schema: {schema}"
         rule_name = name or "root"
 
         if "$defs" in schema:
diff --git a/tests/test_grammar.py b/tests/test_grammar.py
@@ -1,13 +1,52 @@
 import llama_cpp
+import json
 
 tree = """
 leaf ::= "."
 node ::= leaf | "(" node node ")"
 root ::= node
 """
 
+
 def test_grammar_from_string():
     grammar = llama_cpp.LlamaGrammar.from_string(tree)
     assert grammar._n_rules == 3
     assert grammar._start_rule_index == 2
     assert grammar.grammar is not None
+
+
+def test_composed_pydantic_grammar():
+    """
+    from pydantic import BaseModel
+
+    class A(BaseModel):
+        a: int
+
+    class B(BaseModel):
+        a: A
+        b: int
+    """
+
+    # This schema corresponds to the grammar in the comment above.
+    # We don't use the pydantic models directly to avoid the dependency.
+    schema = {
+        "$defs": {
+            "A": {
+                "properties": {"a": {"title": "A", "type": "integer"}},
+                "required": ["a"],
+                "title": "A",
+                "type": "object",
+            }
+        },
+        "properties": {
+            "a": {"$ref": "#/$defs/A"},
+            "b": {"title": "B", "type": "integer"},
+        },
+        "required": ["a", "b"],
+        "title": "B",
+        "type": "object",
+    }
+
+    grammar = llama_cpp.LlamaGrammar.from_json_schema(json.dumps(schema))
+
+    assert grammar.grammar is not None
