mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-04-25 00:51:20 +00:00
b61e6c6145
Cycle 2 PR 1 (#14418). Introduces providers/ package with ProviderProfile ABC and auto-discovery registry, then wires ChatCompletionsTransport to delegate to profiles via a clean single-path method. Provider profiles (8 providers): - nvidia: default_max_tokens=16384 - kimi + kimi-cn: OMIT_TEMPERATURE, thinking + top-level reasoning_effort - openrouter: provider_preferences, full reasoning_config passthrough - nous: product tags, reasoning with Nous-specific disabled omission - deepseek: base_url + env_vars - qwen-oauth: vl_high_resolution extra_body, metadata top-level api_kwargs Transport integration: - _build_kwargs_from_profile() replaces the entire legacy flag-based assembly when provider_profile param is passed - Single path: no dual-execution, no overwrites, no legacy fallthrough - build_api_kwargs_extras() returns (extra_body, top_level) tuple to handle Kimi's top-level reasoning_effort vs OpenRouter's extra_body Auth types: api_key | oauth_device_code | oauth_external | copilot | aws (expanded from the lossy 'oauth' to match real Hermes auth modes). 64 new tests: - 30 profile unit tests (registry, all 8 profiles, auth types) - 19 transport parity tests (pin legacy flag-based behavior) - 15 profile wiring tests (verify profile path = legacy path)
70 lines
2.6 KiB
Python
70 lines
2.6 KiB
Python
"""Qwen Portal provider profile."""
|
|
|
|
import copy
|
|
from typing import Any, Dict, List, Tuple
|
|
|
|
from providers.base import ProviderProfile
|
|
from providers import register_provider
|
|
|
|
|
|
class QwenProfile(ProviderProfile):
|
|
"""Qwen Portal — message normalization, vl_high_resolution, metadata top-level."""
|
|
|
|
def prepare_messages(self, messages: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
|
|
"""Normalize content to list-of-dicts format, inject cache_control on system msg.
|
|
|
|
Matches the behavior of run_agent.py:_qwen_prepare_chat_messages().
|
|
"""
|
|
prepared = copy.deepcopy(messages)
|
|
if not prepared:
|
|
return prepared
|
|
|
|
for msg in prepared:
|
|
if not isinstance(msg, dict):
|
|
continue
|
|
content = msg.get("content")
|
|
if isinstance(content, str):
|
|
msg["content"] = [{"type": "text", "text": content}]
|
|
elif isinstance(content, list):
|
|
normalized_parts = []
|
|
for part in content:
|
|
if isinstance(part, str):
|
|
normalized_parts.append({"type": "text", "text": part})
|
|
elif isinstance(part, dict):
|
|
normalized_parts.append(part)
|
|
if normalized_parts:
|
|
msg["content"] = normalized_parts
|
|
|
|
# Inject cache_control on the last part of the system message.
|
|
for msg in prepared:
|
|
if isinstance(msg, dict) and msg.get("role") == "system":
|
|
content = msg.get("content")
|
|
if isinstance(content, list) and content and isinstance(content[-1], dict):
|
|
content[-1]["cache_control"] = {"type": "ephemeral"}
|
|
break
|
|
|
|
return prepared
|
|
|
|
def build_extra_body(self, *, session_id: str = None, **context) -> Dict[str, Any]:
|
|
return {"vl_high_resolution_images": True}
|
|
|
|
def build_api_kwargs_extras(self, *, reasoning_config: dict = None,
|
|
qwen_session_metadata: dict = None,
|
|
**context) -> Tuple[Dict[str, Any], Dict[str, Any]]:
|
|
"""Qwen metadata goes to top-level api_kwargs, not extra_body."""
|
|
top_level = {}
|
|
if qwen_session_metadata:
|
|
top_level["metadata"] = qwen_session_metadata
|
|
return {}, top_level
|
|
|
|
|
|
qwen = QwenProfile(
|
|
name="qwen-oauth",
|
|
aliases=("qwen", "qwen-portal"),
|
|
env_vars=("QWEN_API_KEY",),
|
|
base_url="https://portal.qwen.ai/api/v1",
|
|
auth_type="oauth_external",
|
|
default_max_tokens=65536,
|
|
)
|
|
|
|
register_provider(qwen)
|