fix(workflow): fix default values in parameter extraction node and incorrect value retrieval in comparison operations

api/app/core/workflow/nodes/operators.py

@@ -260,8 +260,7 @@ class ConditionBase(ABC):
         raise RuntimeError("Unsupported variable type")
 
     def check(self, no_right=False):
-        left = self.pool.get(self.left_selector.variable_selector)
-        if not isinstance(left, self.type_limit):
+        if not isinstance(self.left_value, self.type_limit):
             raise TypeError(f"The variable to be compared on must be of {self.type_limit} type")
         if not no_right:
             right = self.resolve_right_literal_value()

api/app/core/workflow/nodes/parameter_extractor/config.py

@@ -37,7 +37,7 @@ class ParamsConfig(BaseModel):
     )
 
     required: bool = Field(
-        ...,
+        default=False,
         description="Whether the parameter is required"
     )
 
@@ -59,6 +59,6 @@ class ParameterExtractorNodeConfig(BaseNodeConfig):
     )
 
     prompt: str = Field(
-        ...,
+        default="",
         description="User-provided supplemental prompt"
     )

api/app/core/workflow/nodes/parameter_extractor/node.py

@@ -157,9 +157,17 @@ class ParameterExtractorNode(BaseNode):
 
         messages = [
             ("system", system_prompt),
-            ("user", self._render_template(self.typed_config.prompt, state)),
-            ("user", rendered_user_prompt),
+
         ]
+        if self.typed_config.prompt:
+            messages.extend([
+                ("user", self._render_template(self.typed_config.prompt, state)),
+                ("user", rendered_user_prompt),
+            ])
+        else:
+            messages.extend([
+                ("user", rendered_user_prompt),
+            ])
 
         model_resp = await llm.ainvoke(messages)
         result = json_repair.repair_json(model_resp.content, return_objects=True)