从零搭建RAG应用：掌握文本分块、向量检索、指代消解等核心技术实现，大模型入门到精通，收藏这篇就足够了！

平常用ChatGPT只能基于训练数据回答问题，但RAG可以让它查阅你的专有文档——不管是内部报告、技术文档还是业务资料，都能成为AI的参考资源。

Python老猿

464人浏览 · 2025-09-24 11:38:47

Python老猿 · 2025-09-24 11:38:47 发布

RAG（检索增强生成）本质上就是给AI模型外挂一个知识库。平常用ChatGPT只能基于训练数据回答问题，但RAG可以让它查阅你的专有文档——不管是内部报告、技术文档还是业务资料，都能成为AI的参考资源。

很多人第一反应是用LangChain或LlamaIndex这些现成框架，确实能快速搭起来。但自己实现的核心价值在于：你能清楚知道文档是怎么被切分的、向量是怎么生成的、检索逻辑具体怎么跑的。

当系统出现检索不准确、回答质量差、成本过高这些问题时，你能精确定位到是哪个环节的问题。比如是分块策略不合适，还是embedding模型选择有问题，或者是检索参数需要调整。用框架的话，很多时候只能盲目调参数，治标不治本。

另外业务场景往往有特殊需求：PDF表格要特殊处理、某些文档类型需要提取特定元数据、检索结果要按业务规则重排序等等。自己实现就能在任何环节做针对性优化，而不是被框架的设计限制住。

下面我们开始一步一步的进行：

文档解析：让机器能读懂你的文件

做RAG第一步是把各种格式的文档统一处理成纯文本。PDF、Word、txt这些常见格式各有各的解析方式。

import os
import PyPDF2
import docx

defload_plain_text(file_path: str) -> str:
"""Load and return the full contents of a .txt file."""
with open(file_path, 'r', encoding='utf-8') as fp:
return fp.read()

defextract_text_from_pdf(file_path: str) -> str:
"""Read every page of a PDF and stitch the text together."""
    texts = []
with open(file_path, 'rb') as fp:
        reader = PyPDF2.PdfReader(fp)
for pg in reader.pages:
# ensure we don't end up with None
            page_txt = pg.extract_text() or""
            texts.append(page_txt)
# separate pages with a newline
return"\n".join(texts)

defextract_text_from_docx(file_path: str) -> str:
"""Grab all paragraphs from a .docx document."""
    doc = docx.Document(file_path)
    paras = [p.text for p in doc.paragraphs]
return"\n".join(paras)

然后写个统一的路由器，根据文件后缀调用对应的解析函数：

import os

defload_document(file_path: str):
"""Load a document's text based on its file extension."""
    _, extension = os.path.splitext(file_path)
    extension = extension.lower()

if extension == '.txt':
return read_text_file(file_path)
elif extension == '.pdf':
return read_pdf_file(file_path)
elif extension == '.docx':
return read_docx_file(file_path)
else:
raise ValueError(f"Unsupported file type: {extension}")

这样设计的好处是：文档结构保持完整（用换行符分隔段落），支持各种大小写的文件扩展名，不支持的格式会给出明确提示。

文本分块：把长文档切成合适的片段

因为有上下文的的限制，所以直接把整篇文档丢给LLM就像让人一口吃下整个pizza——不现实。所以需要把文档切成适当大小的块。

defchunk_sentences(text: str, max_length: int = 500) -> list[str]:
"""Split text into size-limited chunks, breaking only at sentence boundaries."""
# Normalize whitespace and split on basic sentence delimiter
    segments = text.replace('\n', ' ').split('. ')
    blocks = []
    buffer = []
    buffer_len = 0

for segment in segments:
        seg = segment.strip()
ifnot seg:
continue# skip empty strings

# Make sure each segment ends with a period
ifnot seg.endswith('.'):
            seg += '.'

        seg_len = len(seg)

# If adding this segment would exceed max_length, flush the buffer first
if buffer and buffer_len + seg_len > max_length:
            blocks.append(' '.join(buffer))
            buffer = [seg]
            buffer_len = seg_len
else:
            buffer.append(seg)
            buffer_len += seg_len

# Append any remaining sentences
if buffer:
        blocks.append(' '.join(buffer))

return blocks

块大小选择是个权衡问题。小块（200-500字符）适合精确匹配，像索引卡一样；中等块（500-1000字符）能保留更多上下文；大块（1000+字符）上下文丰富但可能模糊焦点。技术文档通常用小块效果更好，叙述性内容可以用大一些的块，并且分块还可以有很多策略可选，在以前的文章中都有总结。

搭建向量数据库：用ChromaDB存储语义信息

文档切块后需要存到向量数据库里，这样才能做语义搜索。ChromaDB是个不错的选择，轻量但功能够用。

import chromadb
from chromadb.utils import embedding_functions

# Persistent storage - saves data between sessions
client = chromadb.PersistentClient(path="chroma_db")

# Our "brain" for understanding text meaning
sentence_transformer_ef = embedding_functions.SentenceTransformerEmbeddingFunction(
    model_name="all-MiniLM-L6-v2"# Compact but powerful
)

# Create our knowledge repository
collection = client.get_or_create_collection(
    name="documents_collection",
    embedding_function=sentence_transformer_ef
 )

这里用了几个核心组件：PersistentClient确保数据持久化存储，重启程序后数据还在；SentenceTransformerEmbeddingFunction把文本转成向量，让机器理解语义；all-MiniLM-L6-v2是个轻量级但效果不错的embedding模型。

文档索引：批量处理和元数据管理

接下来要把文档批量处理成可检索的格式。每个文本块都需要唯一ID和元数据，方便后续溯源。

defbuild_knowledge_units(path: str):
"""Ingest a file, break it into chunks, and tag each piece with metadata."""
try:
# Pull in the raw text
        raw = load_document(path)
# Break the text into bite-sized segments
        segments = partition_text(raw)
# Grab just the filename for provenance
        name = os.path.basename(path)

# Assemble metadata dicts for each segment
        metadata_records = [
            {"source_file": name, "segment_index": idx}
for idx in range(len(segments))
        ]

# Generate a stable ID for each piece
        unique_keys = [
f"{name}_seg_{idx}"
for idx in range(len(segments))
        ]

return unique_keys, segments, metadata_records

except Exception as err:
        print(f"Failed to process '{path}': {err}")
# Return empty lists so downstream code can continue safely
return [], [], []

为了提高效率，批量插入比单条插入快很多：

defbatch_insert_into_store(store, record_ids, contents, metadata_list):
"""Insert items into the vector store in optimized batches."""
    batch_size = 100# tuned for ChromaDB throughput
for start_idx in range(0, len(contents), batch_size):
        stop_idx = min(start_idx + batch_size, len(contents))
        store.add(
            documents=contents[start_idx:stop_idx],     # text chunks
            metadatas=metadata_list[start_idx:stop_idx],# chunk metadata
            ids=record_ids[start_idx:stop_idx]          # unique chunk IDs
        )

defingest_folder(store, directory: str):
"""Walk a directory, process each file, and load into the store."""
# gather only regular files
    entries = [
        os.path.join(directory, name)
for name in os.listdir(directory)
if os.path.isfile(os.path.join(directory, name))
    ]

for path in entries:
        filename = os.path.basename(path)
        print(f"► Processing {filename} …")
        ids, contents, metadata_list = process_document(path)
if contents:
            batch_insert_into_store(store, ids, contents, metadata_list)
             print(f"✔ Loaded {len(contents)} chunks from {filename}")

实际运行时你会看到这样的输出：

 ► Processing customer_faqs.pdf …
 ✔ Loaded 51 chunks from customer_faqs.pdf

 ► Processing onboarding_guide.docx …
 ✔ Loaded 20 chunks from onboarding_guide.docx

语义检索：找到最相关的内容块

有了向量数据库，就能进行语义搜索了。不是简单的关键词匹配，而是理解查询的语义含义。

defrun_semantic_query(collection, query: str, top_k: int = 2):
"""Run a semantic search to find the most relevant chunks."""
return collection.query(
        query_texts=[query],      # The actual search query
        n_results=top_k           # Number of matches to return
    )

defbuild_context_and_citations(results):
"""Combine matched chunks and reference their original sources."""
    combined_text = "\n\n".join(results['documents'][0])
    references = [
f"{meta['source']} (chunk {meta['chunk']})"
for meta in results['metadatas'][0]
    ]
return combined_text, references

想了解底层发生了什么，可以看看搜索结果的详细信息：

defdisplay_search_hits(results):
"""Nicely formatted display of search output for readability."""
    print("\nTop Matches\n" + "=" * 50)

    hits = results['documents'][0]
    metadata = results['metadatas'][0]
    scores = results['distances'][0]

for idx in range(len(hits)):
        snippet = hits[idx]
        info = metadata[idx]
        score = scores[idx]

        print(f"\nMatch #{idx + 1}")
        print(f"From: {info['source']} — Chunk {info['chunk']}")
        print(f"Similarity Score: {1 - score:.2f} / 1.00")
         print(f"Excerpt: {snippet[:150]}...\n")

搜索结果会显示相似度分数和来源文档，帮你判断检索质量。

接入LLM：让模型基于检索内容回答

检索到相关内容后，就要让LLM基于这些内容生成回答。关键是构造好的prompt，确保模型只基于提供的上下文回答。

import os
from openai import OpenAI

# Initialize the OpenAI client
client = OpenAI()

# Set your OpenAI API key
os.environ["OPENAI_API_KEY"] = "your_api_key"# Replace this with your actual key

defbuild_prompt(context: str, question: str) -> str:
"""Construct a focused prompt using context and a user question."""
returnf"""You are a helpful assistant. Use only the context provided below to answer.
If the answer cannot be found in the context, reply with "I don't have that information."

Context:
{context}

Question: {question}

Answer:"""

defask_openai(question: str, context: str) -> str:
"""Send the prompt to OpenAI and return the generated response."""
    prompt = build_prompt(context, question)

try:
        reply = client.chat.completions.create(
            model="gpt-4-turbo",
            messages=[
                {"role": "system", "content": "You answer based strictly on the context provided."},
                {"role": "user", "content": prompt}
            ],
            temperature=0.3,
            max_tokens=300
        )
return reply.choices[0].message.content
except Exception as err:
returnf"Error: {str(err)}"

temperature参数控制生成的随机性：0.0最保守只输出事实，0.5平衡事实和表达，1.0最有创造性。对RAG来说，0.0到0.3比较合适，保证回答基于文档内容。

对话记忆：让AI记住聊天历史

ChatGPT能记住对话上下文，我们的RAG系统也需要这个能力。实现起来其实不复杂，就是维护一个会话状态。

import uuid
from datetime import datetime

# In-memory chat log (swap with persistent storage in production)
chat_sessions = {}

defstart_new_session() -> str:
"""Initialize a fresh conversation session with a unique ID."""
    session_id = str(uuid.uuid4())
    chat_sessions[session_id] = []
return session_id

deflog_message(session_id: str, sender: str, message: str):
"""Add a message to the session history."""
if session_id notin chat_sessions:
        chat_sessions[session_id] = []

    chat_sessions[session_id].append({
"role": sender,
"content": message,
"timestamp": datetime.now().isoformat()
    })

deffetch_recent_messages(session_id: str, limit: int = 5):
"""Return the last few messages from a session."""
    msgs = chat_sessions.get(session_id, [])
return msgs[-limit:]

defprepare_history_for_model(messages: list) -> str:
"""Convert messages into a single formatted string."""
return"\n".join(
f"{msg['role'].capitalize()}: {msg['content']}"
for msg in messages
     )

这样设计后，每个用户的对话都有独立的session_id，每条消息都会记录到历史中，生成回答时可以参考最近的几条消息作为上下文。

解决指代消解：理解"它"、"那个"指的是什么

用户经常会问一些不完整的问题，比如先问"LaunchPad项目是什么"，接着问"它什么时候开始的"。这里的"它"显然指LaunchPad，但AI不知道。需要把后续问题重写成独立完整的问题。

defrewrite_query_with_context(query: str, chat_log: str, client: OpenAI) -> str:
"""Rewrites a follow-up query as a full standalone question using prior conversation."""

  prompt = f"""Rephrase follow-up questions to be fully self-contained.
Refer to the chat history as needed. Return only the rewritten question.

Chat History:
{chat_log}

Follow-up: {query}
Standalone Question:"""

try:
      response = client.chat.completions.create(
          model="gpt-3.5-turbo",  # Fast, cheap, reliable
          messages=[{"role": "user", "content": prompt}],
          temperature=0# Keep output consistent
      )
return response.choices[0].message.content
except Exception as err:
      print(f"Failed to contextualize query: {err}")
return query  # Return original if there's an error

这样"它什么时候开始的"就能自动变成"LaunchPad项目什么时候开始的"，确保搜索能找到正确的内容。

完整的对话流程：把所有组件串起来

最后把所有功能整合成一个完整的对话式RAG系统：

defhandle_conversational_query(
    collection,
    query: str,
    session_id: str,
    n_chunks: int = 3
):
"""Orchestrates the full RAG-based QA flow in a chat session."""

# Step 1: Pull session history and prep it for context injection
    chat_log = get_conversation_history(session_id)
    prior_messages = format_history(chat_log)

# Step 2: Resolve pronouns or unclear references in the query
    refined_query = contextualize_query(query, prior_messages, client)
    print(f"[Refined Query] {refined_query}")

# Step 3: Retrieve relevant knowledge from the vector DB
    search_results = run_semantic_query(collection, refined_query, n_chunks)
    retrieved_text, citations = build_context_and_citations(search_results)

# Step 4: Generate an answer grounded in retrieved content
    answer = generate_response(refined_query, retrieved_text)

# Step 5: Save both user input and AI reply into memory
    add_message(session_id, "user", query)
    add_message(session_id, "assistant", answer)

return answer, citations

实际使用时的对话流程：

session = start_conversation()

# 初始查询
q1 = "LaunchPad是做什么的？"
reply, refs = smart_retrieval(collection, q1, session)
print(f"Answer: {reply}\nSources: {refs}")

# 后续查询
q2 = "它什么时候开始的？"
reply, refs = smart_retrieval(collection, q2, session)
 print(f"Answer: {reply}\nSources: {refs}")

系统会自动处理指代消解，输出类似这样：

 Contextualized：LaunchPad是做什么的？
 Answer：LaunchPad帮助初创公司快速原型设计和验证产品想法。
 Sources：['program_overview.pdf (chunk 1)']

 Contextualized：LaunchPad项目什么时候开始的？
 Answer：LaunchPad项目始于2018年。
 Sources：['timeline_notes.txt (chunk 3)']

几个实用的优化技巧

做完基础功能后，还有一些进阶优化可以考虑。

1、混合搜索能结合语义搜索和元数据过滤：

# 在语义搜索中添加元数据过滤
 collection.query(
     query_texts=[query],
     n_results=5,
     where={"department": "HR"}  # 只搜索HR部门的文档
 )

2、自动添加来源引用让用户知道答案的出处：

# 在回答后自动添加来源链接
defenhance_response(response, sources):
returnf"{response}\n\n来源:\n" + "\n".join(
f"- {source}"for source in sources
     )

3、根据文档类型调整分块策略：

# 根据文档类型动态调整块大小
if"financial"in file_name:
     chunks = split_text(content, chunk_size=300)  # 财务文档用小块
else:
     chunks = split_text(content, chunk_size=600)  # 其他文档用大块

4、长对话需要压缩历史：

# 当对话历史太长时自动摘要
defsummarize_history(history):
    prompt = f"总结以下对话的关键信息:\n{history}"
return client.chat.completions.create(/*...*/).choices[0].message.content