fix(llm/nodes.py): Ensure that the output returns without any exceptions

[auxpd]

Summary

Using a workflow for tool encapsulation, when core/workflow/nodes/llm/node.py LLMNode._run is called before the result_text is initialized, it causes the last output dict to reference an uninitialized variable

I directly appended the correct output to the result of the LLM output to ensure that there were no exceptions in the preceding process.

Tip

Close issue syntax: Fixes #<issue number> or Resolves #<issue number>, see documentation for more details.
Fixes #14876

Screenshots

Before	After
...	...

Checklist

Important

Please review the checklist below before submitting your pull request.

• This change requires a documentation update, included: Dify Document
• I understand that this PR may be closed in case there was no previous discussion or issues. (This doesn't apply to typos!)
• I've added a test for each change that was introduced, and I tried as much as possible to make a single atomic change.
• I've updated the documentation accordingly.
• I ran dev/reformat(backend) and cd web && npx lint-staged(frontend) to appease the lint gods

[auxpd]

I have a question, why the source code pulls the outside of the exception catch? I'm not sure if I'm missing any designs

[laipz8200]

Can this change fix #14876? It seems that unless the issue occurs at

outputs = {"text": result_text, "usage": jsonable_encoder(usage), "finish_reason": finish_reason}

there should be no difference between the two.

[auxpd]

This is the previous error stack message:
2025-03-01 18:56:13.046 ERROR [ThreadPoolExecutor-10_5] [graph_engine.py:815] - Node 信息提取 run failed
Traceback (most recent call last):
File "/app/api/core/workflow/graph_engine/graph_engine.py", line 626, in _run_node
for item in generator:
^^^^^^^^^
File "/app/api/core/workflow/nodes/base/node.py", line 84, in run
yield from result
File "/app/api/core/workflow/nodes/llm/node.py", line 212, in _run
outputs = {"text": result_text, "usage": jsonable_encoder(usage), "finish_reason": finish_reason}
^^^^^^^^^^^
UnboundLocalError: cannot access local variable 'result_text' where it is not associated with a value
2025-03-01 18:56:13.047 ERROR [ThreadPoolExecutor-10_5] [graph_engine.py:576] - Unknown Error when generating in parallel
Traceback (most recent call last):
File "/app/api/core/workflow/graph_engine/graph_engine.py", line 553, in _run_parallel_node
for item in generator:
^^^^^^^^^
File "/app/api/core/workflow/graph_engine/graph_engine.py", line 312, in _run
raise e
File "/app/api/core/workflow/graph_engine/graph_engine.py", line 283, in _run
for item in generator:
^^^^^^^^^
File "/app/api/core/workflow/graph_engine/graph_engine.py", line 816, in _run_node
raise e
File "/app/api/core/workflow/graph_engine/graph_engine.py", line 626, in _run_node
for item in generator:
^^^^^^^^^
File "/app/api/core/workflow/nodes/base/node.py", line 84, in run
yield from result
File "/app/api/core/workflow/nodes/llm/node.py", line 212, in _run
outputs = {"text": result_text, "usage": jsonable_encoder(usage), "finish_reason": finish_reason}
^^^^^^^^^^^
UnboundLocalError: cannot access local variable 'result_text' where it is not associated with a value

[auxpd]

If "outputs" dict is expected to be returned with a normal response, would it be more appropriate to put it in the try struct?

[laipz8200]

The root issue is that a return statement is missing in the except block, but your method can also resolve the problem.

[laipz8200]

LGTM