Updated xgrammar backend to not deny supported string formats (#27253)

Signed-off-by: CNE Pierre FICHEPOIL <[email protected]>
Signed-off-by: ExtReMLapin <[email protected]>
Co-authored-by: CNE Pierre FICHEPOIL <[email protected]>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Signed-off-by: sstamenk <[email protected]>

Author: 3 people

tests/v1/structured_output/test_utils.py

@@ -13,7 +13,7 @@
 @pytest.fixture
 def unsupported_string_schemas():
     return [
-        {"type": "string", "format": "email"},
+        {"type": "string", "format": "non_existing_format"},
     ]
 
 
@@ -58,6 +58,7 @@ def supported_schema():
         "properties": {
             "name": {"type": "string"},
             "age": {"type": "integer"},
+            "email": {"type": "string", "format": "email"},
             "status": {"type": "string"},
             "scores": {"type": "array", "items": {"type": "number"}},
             "car_type": {"type": "string", "enum": ["sedan", "suv", "truck"]},

vllm/v1/structured_output/backend_xgrammar.py

@@ -200,6 +200,25 @@ def reset(self):
         self.matcher.reset()
 
 
+# cf https://github.com/mlc-ai/xgrammar/blob/a32ac892676d2eedc0327416105b9b06edfb94b2/cpp/json_schema_converter.cc
+STRING_SUPPORTED_FORMATS = {
+    "email",
+    "date",
+    "time",
+    "date-time",
+    "duration",
+    "ipv4",
+    "ipv6",
+    "hostname",
+    "uuid",
+    "uri",
+    "uri-reference",
+    "uri-template",
+    "json-pointer",
+    "relative-json-pointer",
+}
+
+
 def has_xgrammar_unsupported_json_features(schema: dict[str, Any]) -> bool:
     """Check if JSON schema contains features unsupported by xgrammar."""
 
@@ -219,7 +238,11 @@ def check_object(obj: dict[str, Any]) -> bool:
             return True
 
         # Unsupported keywords for strings
-        if obj.get("type") == "string" and "format" in obj:
+        if (
+            obj.get("type") == "string"
+            and "format" in obj
+            and obj["format"] not in STRING_SUPPORTED_FORMATS
+        ):
             return True
 
         # Unsupported keywords for objects