# tinyharness.py — 流式 call_model(可运行)
def call_model_streaming(messages, tools):
    with client.messages.stream(
        model="claude-sonnet-5", max_tokens=2048,
        messages=messages, tools=tools,
    ) as stream:
        for text in stream.text_stream:      # 文本增量,实时打印
            print(text, end="", flush=True)
        print()
        return stream.get_final_message()  # SDK 已帮你把碎片拼成完整 message

# tinyharness.py — 重试(可运行)
import time, random
import anthropic

RETRYABLE = (anthropic.RateLimitError, anthropic.APITimeoutError,
             anthropic.InternalServerError)   # 529 也属这类

def call_model(messages, tools, max_retries=5):
    for attempt in range(max_retries + 1):
        try:
            return call_model_streaming(messages, tools)
        except anthropic.BadRequestError as e:
            if "too long" in str(e).lower():
                raise NeedCompaction()      # 交给 CH5 压缩后重发
            raise                          # 其他 400 不可恢复
        except RETRYABLE as e:
            if attempt == max_retries: raise
            delay = 0.5 * (2 ** attempt) + random.random()  # 退避+抖动
            print(f"\n[重试 {attempt+1}] {delay:.1f}s 后…")
            time.sleep(delay)

class NeedCompaction(Exception): pass   # 占位,CH5 实现压缩