Fix GLM-4.7 MoE Detector complex JSON Schema type parsing (#15753)
This commit is contained in:
Leoyzen
@@ -12,6 +12,7 @@ from sglang.srt.function_call.core_types import (
|
||||
ToolCallItem,
|
||||
_GetInfoFunc,
|
||||
)
|
||||
from sglang.srt.function_call.utils import infer_type_from_json_schema
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
@@ -31,6 +32,14 @@ def get_argument_type(
|
||||
) -> Optional[str]:
|
||||
"""Get the expected type of a function argument from tool definitions.
|
||||
|
||||
Supports complex JSON Schema definitions including:
|
||||
- Direct type field (including type arrays)
|
||||
- anyOf/oneOf: parameter can be any of multiple types
|
||||
- enum: parameter must be one of enum values
|
||||
- allOf: parameter must satisfy all type definitions
|
||||
- properties: inferred as object type
|
||||
- items: inferred as array type
|
||||
|
||||
Args:
|
||||
func_name: Name of the function/tool
|
||||
arg_key: Name of the argument
|
||||
@@ -58,7 +67,8 @@ def get_argument_type(
|
||||
|
||||
arg_spec = properties.get(arg_key)
|
||||
if isinstance(arg_spec, dict):
|
||||
return arg_spec.get("type")
|
||||
# Use the new type inference function for complex JSON Schema support
|
||||
return infer_type_from_json_schema(arg_spec)
|
||||
|
||||
return None
|
||||
|
||||
@@ -243,25 +253,48 @@ class Glm47MoeDetector(BaseFormatDetector):
|
||||
tools: List of available tools
|
||||
|
||||
Returns:
|
||||
Type string: 'string', 'number', or 'object'
|
||||
Type string: 'string', 'number', 'object', 'array', or 'boolean'
|
||||
"""
|
||||
arg_type = get_argument_type(func_name, key, tools)
|
||||
if arg_type:
|
||||
return arg_type
|
||||
|
||||
# Auto-detect type from value (best effort)
|
||||
first_chars = (
|
||||
self._current_value.strip()[:10]
|
||||
if self._current_value and self._current_value.strip()
|
||||
else ""
|
||||
)
|
||||
if first_chars:
|
||||
first_char = first_chars[0]
|
||||
# Improved auto-detection type from value (best effort)
|
||||
value_content = self._current_value.strip() if self._current_value else ""
|
||||
|
||||
if not value_content:
|
||||
return "string"
|
||||
|
||||
# Try to parse as valid JSON first
|
||||
try:
|
||||
parsed = json.loads(value_content)
|
||||
if isinstance(parsed, dict):
|
||||
return "object"
|
||||
elif isinstance(parsed, list):
|
||||
return "array"
|
||||
elif isinstance(parsed, bool):
|
||||
return "boolean"
|
||||
elif isinstance(parsed, (int, float)):
|
||||
return "number"
|
||||
# For string values, check if they look like numbers
|
||||
elif isinstance(parsed, str):
|
||||
if parsed.isdigit() or (
|
||||
parsed.startswith("-") and parsed[1:].isdigit()
|
||||
):
|
||||
return "number"
|
||||
return "string"
|
||||
except json.JSONDecodeError:
|
||||
# Not valid JSON, try heuristic detection
|
||||
first_char = value_content[0] if value_content else ""
|
||||
|
||||
if first_char.isdigit() or first_char in ["-", "."]:
|
||||
return "number"
|
||||
elif first_char in ["{", "["]:
|
||||
return "object"
|
||||
elif first_char in ['"', "'"]:
|
||||
return "string"
|
||||
|
||||
# Default to string (safest fallback)
|
||||
return "string"
|
||||
|
||||
def _format_value_complete(self, value: str, value_type: str) -> str:
|
||||
|
||||
@@ -12,6 +12,7 @@ from sglang.srt.function_call.core_types import (
|
||||
ToolCallItem,
|
||||
_GetInfoFunc,
|
||||
)
|
||||
from sglang.srt.function_call.utils import infer_type_from_json_schema
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
@@ -31,6 +32,14 @@ def get_argument_type(
|
||||
) -> Optional[str]:
|
||||
"""Get the expected type of a function argument from tool definitions.
|
||||
|
||||
Supports complex JSON Schema definitions including:
|
||||
- Direct type field (including type arrays)
|
||||
- anyOf/oneOf: parameter can be any of multiple types
|
||||
- enum: parameter must be one of enum values
|
||||
- allOf: parameter must satisfy all type definitions
|
||||
- properties: inferred as object type
|
||||
- items: inferred as array type
|
||||
|
||||
Args:
|
||||
func_name: Name of the function/tool
|
||||
arg_key: Name of the argument
|
||||
@@ -48,7 +57,9 @@ def get_argument_type(
|
||||
properties = {}
|
||||
if arg_key not in properties:
|
||||
return None
|
||||
return properties[arg_key].get("type", None)
|
||||
|
||||
# Use new type inference function for complex JSON Schema support
|
||||
return infer_type_from_json_schema(properties[arg_key])
|
||||
|
||||
|
||||
def _convert_to_number(value: str) -> Any:
|
||||
@@ -216,21 +227,48 @@ class Glm4MoeDetector(BaseFormatDetector):
|
||||
tools: List of available tools
|
||||
|
||||
Returns:
|
||||
Type string: 'string', 'number', or 'object'
|
||||
Type string: 'string', 'number', 'object', 'array', or 'boolean'
|
||||
"""
|
||||
arg_type = get_argument_type(func_name, key, tools)
|
||||
if arg_type:
|
||||
return arg_type
|
||||
|
||||
# Auto-detect type from value (best effort)
|
||||
first_chars = self._current_value.strip()[:10] if self._current_value else ""
|
||||
if first_chars:
|
||||
first_char = first_chars[0]
|
||||
# Improved auto-detection type from value (best effort)
|
||||
value_content = self._current_value.strip() if self._current_value else ""
|
||||
|
||||
if not value_content:
|
||||
return "string"
|
||||
|
||||
# Try to parse as valid JSON first
|
||||
try:
|
||||
parsed = json.loads(value_content)
|
||||
if isinstance(parsed, dict):
|
||||
return "object"
|
||||
elif isinstance(parsed, list):
|
||||
return "array"
|
||||
elif isinstance(parsed, bool):
|
||||
return "boolean"
|
||||
elif isinstance(parsed, (int, float)):
|
||||
return "number"
|
||||
# For string values, check if they look like numbers
|
||||
elif isinstance(parsed, str):
|
||||
if parsed.isdigit() or (
|
||||
parsed.startswith("-") and parsed[1:].isdigit()
|
||||
):
|
||||
return "number"
|
||||
return "string"
|
||||
except json.JSONDecodeError:
|
||||
# Not valid JSON, try heuristic detection
|
||||
first_char = value_content[0] if value_content else ""
|
||||
|
||||
if first_char.isdigit() or first_char in ["-", "."]:
|
||||
return "number"
|
||||
elif first_char in ["{", "["]:
|
||||
return "object"
|
||||
elif first_char in ['"', "'"]:
|
||||
return "string"
|
||||
|
||||
# Default to string (safest fallback)
|
||||
return "string"
|
||||
|
||||
def _format_value_complete(self, value: str, value_type: str) -> str:
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
from json import JSONDecodeError, JSONDecoder
|
||||
from json.decoder import WHITESPACE
|
||||
from typing import Any, List, Literal, Optional, Tuple, Union
|
||||
from typing import Any, Dict, List, Literal, Optional, Tuple, Union
|
||||
|
||||
import orjson
|
||||
import partial_json_parser
|
||||
@@ -101,6 +101,109 @@ def _get_tool_schema(tool: Tool) -> dict:
|
||||
}
|
||||
|
||||
|
||||
def infer_type_from_json_schema(schema: Dict[str, Any]) -> Optional[str]:
|
||||
"""
|
||||
Infer the primary type of a parameter from JSON Schema.
|
||||
|
||||
Supports complex JSON Schema structures including:
|
||||
- Direct type field (including type arrays)
|
||||
- anyOf/oneOf: parameter can be any of multiple types
|
||||
- enum: parameter must be one of enum values
|
||||
- allOf: parameter must satisfy all type definitions
|
||||
- properties: inferred as object type
|
||||
- items: inferred as array type
|
||||
|
||||
Args:
|
||||
schema: JSON Schema definition
|
||||
|
||||
Returns:
|
||||
Inferred type ('string', 'number', 'object', 'array', etc.) or None
|
||||
"""
|
||||
if not isinstance(schema, dict):
|
||||
return None
|
||||
|
||||
# Priority 1: Direct type field (including type arrays)
|
||||
if "type" in schema:
|
||||
type_value = schema["type"]
|
||||
if isinstance(type_value, str):
|
||||
return type_value
|
||||
elif isinstance(type_value, list) and type_value:
|
||||
# Handle type arrays: return first non-null type
|
||||
non_null_types = [t for t in type_value if t != "null"]
|
||||
if non_null_types:
|
||||
return non_null_types[0]
|
||||
return "string" # If only null, default to string
|
||||
|
||||
# Priority 2: Handle anyOf/oneOf
|
||||
if "anyOf" in schema or "oneOf" in schema:
|
||||
schemas = schema.get("anyOf") or schema.get("oneOf")
|
||||
types = []
|
||||
|
||||
if isinstance(schemas, list):
|
||||
for sub_schema in schemas:
|
||||
inferred_type = infer_type_from_json_schema(sub_schema)
|
||||
if inferred_type:
|
||||
types.append(inferred_type)
|
||||
|
||||
if types:
|
||||
# If all types are the same, return unified type
|
||||
if len(set(types)) == 1:
|
||||
return types[0]
|
||||
# When types differ, prioritize string (safest)
|
||||
if "string" in types:
|
||||
return "string"
|
||||
# Otherwise return first type
|
||||
return types[0]
|
||||
|
||||
# Priority 3: Handle enum (infer type from enum values)
|
||||
if "enum" in schema and isinstance(schema["enum"], list):
|
||||
if not schema["enum"]:
|
||||
return "string"
|
||||
|
||||
# Infer type from enum values
|
||||
enum_types = set()
|
||||
for value in schema["enum"]:
|
||||
if value is None:
|
||||
enum_types.add("null")
|
||||
elif isinstance(value, bool):
|
||||
enum_types.add("boolean")
|
||||
elif isinstance(value, int):
|
||||
enum_types.add("integer")
|
||||
elif isinstance(value, float):
|
||||
enum_types.add("number")
|
||||
elif isinstance(value, str):
|
||||
enum_types.add("string")
|
||||
elif isinstance(value, list):
|
||||
enum_types.add("array")
|
||||
elif isinstance(value, dict):
|
||||
enum_types.add("object")
|
||||
|
||||
# If type is uniform, return that type
|
||||
if len(enum_types) == 1:
|
||||
return enum_types.pop()
|
||||
# Mixed types, prioritize string
|
||||
return "string"
|
||||
|
||||
# Priority 4: Handle allOf (must satisfy all types)
|
||||
if "allOf" in schema and isinstance(schema["allOf"], list):
|
||||
schemas = schema["allOf"]
|
||||
for sub_schema in schemas:
|
||||
inferred_type = infer_type_from_json_schema(sub_schema)
|
||||
if inferred_type and inferred_type != "string":
|
||||
return inferred_type
|
||||
return "string"
|
||||
|
||||
# Priority 5: Infer object type
|
||||
if "properties" in schema:
|
||||
return "object"
|
||||
|
||||
# Priority 6: Infer array type
|
||||
if "items" in schema:
|
||||
return "array"
|
||||
|
||||
return None
|
||||
|
||||
|
||||
def get_json_schema_constraint(
|
||||
tools: List[Tool], tool_choice: Union[ToolChoice, Literal["required"]]
|
||||
) -> Optional[dict]:
|
||||
|
||||
Reference in New Issue
Block a user