fix(deepseek): deepseek-reasoner model cannot return reasoning_content with tools

[Abhinavexists]

Fixes: #34082

Description

This change restores DeepSeek-specific reasoning that is lost when LangChain routes messages through the OpenAI-style tool-calling pipeline. When tools are bound, the OpenAI tools parser rewrites the model response schema and drops non-OpenAI fields such as reasoning_content and _deepseek_reasoning. This patch re-attaches these fields inside ChatDeepSeek._create_chat_result()

Dependencies

None.

Expected output after fix

(AIMessageChunk(content='', additional_kwargs={}, response_metadata={'finish_reason': 'stop', 'model_name': 'deepseek-chat', 'system_fingerprint': 'fp_ffc7281d48_prod0820_fp8_kvcache', 'model_provider': 'deepseek'}, id='lc_run--1cfb846c-0626-4b96-a078-b1baa8e8b5c9', usage_metadata={'input_tokens': 153, 'output_tokens': 38, 'total_tokens': 191, 'input_token_details': {'cache_read': 128}, 'output_token_details': {}}, chunk_position='last'), {'langgraph_step': 1, 'langgraph_node': 'model', 'langgraph_triggers': ('branch:to:model',), 'langgraph_path': ('__pregel_pull', 'model'), 'langgraph_checkpoint_ns': 'model:2c51adc2-5aba-d990-71f7-325ab7aa624c', 'checkpoint_ns': 'model:2c51adc2-5aba-d990-71f7-325ab7aa624c', 'ls_provider': 'deepseek', 'ls_model_name': 'deepseek-reasoner', 'ls_model_type': 'chat', 'ls_temperature': None})

[codspeed-hq]

CodSpeed Performance Report

Merging #34092 will not alter performance

_{Comparing Abhinavexists:deepseek-reasoning_content-fix (afa4999) with master (880652b)¹}

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

Summary

✅ 13 untouched
⏩ 21 skipped²

Footnotes

1. No successful run was found on master (9c21f83) during the generation of this report, so 880652b was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

2. 21 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩