使用延伸思考進行構建

weather_tool = {
    "name": "get_weather",
    "description": "Get current weather for a location",
    "input_schema": {
        "type": "object",
        "properties": {
            "location": {"type": "string"}
        },
        "required": ["location"]
    }
}

# 第一個請求 - Claude 以思考和工具請求回應
response = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=16000,
    thinking={
        "type": "enabled",
        "budget_tokens": 10000
    },
    tools=[weather_tool],
    messages=[
        {"role": "user", "content": "What's the weather in Paris?"}
    ]
)

# 提取思考區塊和工具使用區塊
thinking_block = next((block for block in response.content
                      if block.type == 'thinking'), None)
tool_use_block = next((block for block in response.content
                      if block.type == 'tool_use'), None)

# 呼叫您的實際天氣 API，這是您的實際 API 呼叫將進行的地方
# 讓我們假設這是我們得到的回傳
weather_data = {"temperature": 88}

# 第二個請求 - 包括思考區塊和工具結果
# 回應中不會生成新的思考區塊
continuation = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=16000,
    thinking={
        "type": "enabled",
        "budget_tokens": 10000
    },
    tools=[weather_tool],
    messages=[
        {"role": "user", "content": "What's the weather in Paris?"},
        # 注意 thinking_block 與 tool_use_block 一起傳入
        # 如果未傳入，將引發錯誤
        {"role": "assistant", "content": [thinking_block, tool_use_block]},
        {"role": "user", "content": [{
            "type": "tool_result",
            "tool_use_id": tool_use_block.id,
            "content": f"Current temperature: {weather_data['temperature']}°F"
        }]}
    ]
)

import anthropic

client = anthropic.Anthropic()

# 定義工具
calculator_tool = {
    "name": "calculator",
    "description": "Perform mathematical calculations",
    "input_schema": {
        "type": "object",
        "properties": {
            "expression": {
                "type": "string",
                "description": "Mathematical expression to evaluate"
            }
        },
        "required": ["expression"]
    }
}

database_tool = {
    "name": "database_query",
    "description": "Query product database",
    "input_schema": {
        "type": "object",
        "properties": {
            "query": {
                "type": "string",
                "description": "SQL query to execute"
            }
        },
        "required": ["query"]
    }
}

# 第一個請求 - Claude 在所有工具呼叫之前思考一次
response = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=16000,
    thinking={
        "type": "enabled",
        "budget_tokens": 10000
    },
    tools=[calculator_tool, database_tool],
    messages=[{
        "role": "user",
        "content": "What's the total revenue if we sold 150 units of product A at $50 each, and how does this compare to our average monthly revenue from the database?"
    }]
)

# 回應包括思考，然後是工具使用
# 注意：Claude 在開始時思考一次，然後做出所有工具決定
print("First response:")
for block in response.content:
    if block.type == "thinking":
        print(f"Thinking (summarized): {block.thinking}")
    elif block.type == "tool_use":
        print(f"Tool use: {block.name} with input {block.input}")
    elif block.type == "text":
        print(f"Text: {block.text}")

# 您將執行工具並返回結果...
# 在獲得兩個工具結果後，Claude 直接回應而不進行額外思考

import anthropic

client = anthropic.Anthropic()

# 與之前相同的工具定義
calculator_tool = {
    "name": "calculator",
    "description": "Perform mathematical calculations",
    "input_schema": {
        "type": "object",
        "properties": {
            "expression": {
                "type": "string",
                "description": "Mathematical expression to evaluate"
            }
        },
        "required": ["expression"]
    }
}

database_tool = {
    "name": "database_query",
    "description": "Query product database",
    "input_schema": {
        "type": "object",
        "properties": {
            "query": {
                "type": "string",
                "description": "SQL query to execute"
            }
        },
        "required": ["query"]
    }
}

# 第一個請求，啟用交錯思考
response = client.beta.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=16000,
    thinking={
        "type": "enabled",
        "budget_tokens": 10000
    },
    tools=[calculator_tool, database_tool],
    betas=["interleaved-thinking-2025-05-14"],
    messages=[{
        "role": "user",
        "content": "What's the total revenue if we sold 150 units of product A at $50 each, and how does this compare to our average monthly revenue from the database?"
    }]
)

print("Initial response:")
thinking_blocks = []
tool_use_blocks = []

for block in response.content:
    if block.type == "thinking":
        thinking_blocks.append(block)
        print(f"Thinking: {block.thinking}")
    elif block.type == "tool_use":
        tool_use_blocks.append(block)
        print(f"Tool use: {block.name} with input {block.input}")
    elif block.type == "text":
        print(f"Text: {block.text}")

# 第一個工具結果（計算機）
calculator_result = "7500"  # 150 * 50

# 使用第一個工具結果繼續
response2 = client.beta.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=16000,
    thinking={
        "type": "enabled",
        "budget_tokens": 10000
    },
    tools=[calculator_tool, database_tool],
    betas=["interleaved-thinking-2025-05-14"],
    messages=[
        {
            "role": "user",
            "content": "What's the total revenue if we sold 150 units of product A at $50 each, and how does this compare to our average monthly revenue from the database?"
        },
        {
            "role": "assistant",
            "content": [thinking_blocks[0], tool_use_blocks[0]]
        },
        {
            "role": "user",
            "content": [{
                "type": "tool_result",
                "tool_use_id": tool_use_blocks[0].id,
                "content": calculator_result
            }]
        }
    ]
)

print("\nAfter calculator result:")
# 使用交錯思考，Claude 可以思考計算機結果
# 在決定查詢資料庫之前
for block in response2.content:
    if block.type == "thinking":
        thinking_blocks.append(block)
        print(f"Interleaved thinking: {block.thinking}")
    elif block.type == "tool_use":
        tool_use_blocks.append(block)
        print(f"Tool use: {block.name} with input {block.input}")

# 第二個工具結果（資料庫）
database_result = "5200"  # 範例平均月收入

# 使用第二個工具結果繼續
response3 = client.beta.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=16000,
    thinking={
        "type": "enabled",
        "budget_tokens": 10000
    },
    tools=[calculator_tool, database_tool],
    betas=["interleaved-thinking-2025-05-14"],
    messages=[
        {
            "role": "user",
            "content": "What's the total revenue if we sold 150 units of product A at $50 each, and how does this compare to our average monthly revenue from the database?"
        },
        {
            "role": "assistant",
            "content": [thinking_blocks[0], tool_use_blocks[0]]
        },
        {
            "role": "user",
            "content": [{
                "type": "tool_result",
                "tool_use_id": tool_use_blocks[0].id,
                "content": calculator_result
            }]
        },
        {
            "role": "assistant",
            "content": thinking_blocks[1:] + tool_use_blocks[1:]
        },
        {
            "role": "user",
            "content": [{
                "type": "tool_result",
                "tool_use_id": tool_use_blocks[1].id,
                "content": database_result
            }]
        }
    ]
)

print("\nAfter database result:")
# 使用交錯思考，Claude 可以思考兩個結果
# 在制定最終回應之前
for block in response3.content:
    if block.type == "thinking":
        print(f"Final thinking: {block.thinking}")
    elif block.type == "text":
        print(f"Final response: {block.text}")

from anthropic import Anthropic
import requests
from bs4 import BeautifulSoup

client = Anthropic()

def fetch_article_content(url):
    response = requests.get(url)
    soup = BeautifulSoup(response.content, 'html.parser')

    # 移除指令碼和樣式元素
    for script in soup(["script", "style"]):
        script.decompose()

    # 取得文字
    text = soup.get_text()

    # 分解成行並移除每行上的前導和尾隨空格
    lines = (line.strip() for line in text.splitlines())
    # 將多個標題分解成一行
    chunks = (phrase.strip() for line in lines for phrase in line.split("  "))
    # 刪除空行
    text = '\n'.join(chunk for chunk in chunks if chunk)

    return text

# 取得文章內容
book_url = "https://www.gutenberg.org/cache/epub/1342/pg1342.txt"
book_content = fetch_article_content(book_url)
# 使用足夠的文字進行快取（前幾章）
LARGE_TEXT = book_content[:5000]

SYSTEM_PROMPT=[
    {
        "type": "text",
        "text": "You are an AI assistant that is tasked with literary analysis. Analyze the following text carefully.",
    },
    {
        "type": "text",
        "text": LARGE_TEXT,
        "cache_control": {"type": "ephemeral"}
    }
]

MESSAGES = [
    {
        "role": "user",
        "content": "Analyze the tone of this passage."
    }
]

# 第一個請求 - 建立快取
print("First request - establishing cache")
response1 = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=20000,
    thinking={
        "type": "enabled",
        "budget_tokens": 4000
    },
    system=SYSTEM_PROMPT,
    messages=MESSAGES
)

print(f"First response usage: {response1.usage}")

MESSAGES.append({
    "role": "assistant",
    "content": response1.content
})
MESSAGES.append({
    "role": "user",
    "content": "Analyze the characters in this passage."
})
# 第二個請求 - 相同的思考參數（預期快取命中）
print("\nSecond request - same thinking parameters (cache hit expected)")
response2 = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=20000,
    thinking={
        "type": "enabled",
        "budget_tokens": 4000
    },
    system=SYSTEM_PROMPT,
    messages=MESSAGES
)

print(f"Second response usage: {response2.usage}")

# 第三個請求 - 不同的思考參數（訊息快取未命中）
print("\nThird request - different thinking parameters (cache miss for messages)")
response3 = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=20000,
    thinking={
        "type": "enabled",
        "budget_tokens": 8000  # 已變更思考預算
    },
    system=SYSTEM_PROMPT,  # 系統提示保持快取
    messages=MESSAGES  # 訊息快取已失效
)

print(f"Third response usage: {response3.usage}")

from anthropic import Anthropic
import requests
from bs4 import BeautifulSoup

client = Anthropic()

def fetch_article_content(url):
    response = requests.get(url)
    soup = BeautifulSoup(response.content, 'html.parser')

    # 移除指令碼和樣式元素
    for script in soup(["script", "style"]):
        script.decompose()

    # 取得文字
    text = soup.get_text()

    # 分解成行並移除每行上的前導和尾隨空格
    lines = (line.strip() for line in text.splitlines())
    # 將多個標題分解成一行
    chunks = (phrase.strip() for line in lines for phrase in line.split("  "))
    # 刪除空行
    text = '\n'.join(chunk for chunk in chunks if chunk)

    return text

# 取得文章內容
book_url = "https://www.gutenberg.org/cache/epub/1342/pg1342.txt"
book_content = fetch_article_content(book_url)
# 使用足夠的文字進行快取（前幾章）
LARGE_TEXT = book_content[:5000]

# 無系統提示 - 改為在訊息中快取
MESSAGES = [
    {
        "role": "user",
        "content": [
            {
                "type": "text",
                "text": LARGE_TEXT,
                "cache_control": {"type": "ephemeral"},
            },
            {
                "type": "text",
                "text": "Analyze the tone of this passage."
            }
        ]
    }
]

# 第一個請求 - 建立快取
print("First request - establishing cache")
response1 = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=20000,
    thinking={
        "type": "enabled",
        "budget_tokens": 4000
    },
    messages=MESSAGES
)

print(f"First response usage: {response1.usage}")

MESSAGES.append({
    "role": "assistant",
    "content": response1.content
})
MESSAGES.append({
    "role": "user",
    "content": "Analyze the characters in this passage."
})
# 第二個請求 - 相同的思考參數（預期快取命中）
print("\nSecond request - same thinking parameters (cache hit expected)")
response2 = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=20000,
    thinking={
        "type": "enabled",
        "budget_tokens": 4000  # 相同的思考預算
    },
    messages=MESSAGES
)

print(f"Second response usage: {response2.usage}")

MESSAGES.append({
    "role": "assistant",
    "content": response2.content
})
MESSAGES.append({
    "role": "user",
    "content": "Analyze the setting in this passage."
})

# 第三個請求 - 不同的思考預算（預期快取未命中）
print("\nThird request - different thinking budget (cache miss expected)")
response3 = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=20000,
    thinking={
        "type": "enabled",
        "budget_tokens": 8000  # 不同的思考預算破壞快取
    },
    messages=MESSAGES
)

print(f"Third response usage: {response3.usage}")

import anthropic

client = anthropic.Anthropic()

# 使用觸發 redacted thinking 的特殊提示（僅用於演示目的）
response = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=16000,
    thinking={
        "type": "enabled",
        "budget_tokens": 10000
    },
    messages=[{
        "role": "user",
        "content": "ANTHROPIC_MAGIC_STRING_TRIGGER_REDACTED_THINKING_46C9A13E193C177646C7398A98432ECCCE4C1253D5E2D82641AC0E52CC2876CB"
    }]
)

# 識別 redacted thinking 區塊
has_redacted_thinking = any(
    block.type == "redacted_thinking" for block in response.content
)

if has_redacted_thinking:
    print("Response contains redacted thinking blocks")
    # 這些區塊在後續請求中仍然可用

    # 提取所有區塊（redacted 和非 redacted）
    all_thinking_blocks = [
        block for block in response.content
        if block.type in ["thinking", "redacted_thinking"]
    ]

    # 在傳遞到後續請求時，包括所有區塊而不進行修改
    # 這保留了 Claude 推理的完整性

    print(f"Found {len(all_thinking_blocks)} thinking blocks total")
    print(f"These blocks are still billable as output tokens")