使用KNN及tensorflow进行中文pdf关键词搜索，类似AutoGPT或ChatPDF实现原理

接上回，如何使用AI模型（如GPT、LLaMA），训练某一考试的教材、历年试题？

直接上代码，结合该功能与GPT进行搞基：

PDF文本搜索

import os
import re
import shutil
import urllib.request
from pathlib import Path
from tempfile import NamedTemporaryFile

import fitz
import numpy as np
import openai
import tensorflow_hub as hub
from sklearn.neighbors import NearestNeighbors

# 对每页PDF进行预处理，生成一个text_list
def preprocess(text):
    text = text.replace('\n', ' ')
    text = re.sub('\s+', ' ', text)
    return text


def pdf_to_text(path, start_page=1, end_page=None):
    doc = fitz.open(path)
    total_pages = doc.page_count

    if end_page is None:
        end_page = total_pages

    text_list = []

    for i in range(start_page - 1, end_page):
        text = doc.load_page(i).get_text("text")
        text = preprocess(text)
        text_list.append(text)

    doc.close()
    return text_list

def text_to_chunks(texts, word_length=150, start_page=1):
    text_toks = [t.split(' ') for t in texts]
    page_nums = []
    chunks = []

    for idx, words in enumerate(text_toks):
        for i in range(0, len(words), word_length):
            chunk = words[i : i + word_length]
            if (
                (i + word_length) > len(words)
                and (len(chunk) < word_length)
                and (len(text_toks) != (idx + 1))
            ):
                text_toks[idx + 1] = chunk + text_toks[idx + 1]
                continue
            chunk = ' '.join(chunk).strip()
            chunk = f'[Page no. {idx+start_page}]' + ' ' + '"' + chunk + '"'
#             print({idx+start_page})
            chunks.append(chunk)
    return chunks

class SemanticSearch:
    def __init__(self):
        self.use = hub.load("F:/*******") # 中文 https://www.intumu.com/article/203
        
        self.fitted = False

    def fit(self, data, batch=100, n_neighbors=3): # batch=1000, n_neighbors=5
        self.data = data
        self.embeddings = self.get_text_embedding(data, batch=batch)
        n_neighbors = min(n_neighbors, len(self.embeddings))
        self.nn = NearestNeighbors(n_neighbors=n_neighbors)
        self.nn.fit(self.embeddings)
        self.fitted = True

    def __call__(self, text, return_data=True):
        inp_emb = self.use([text])
        neighbors = self.nn.kneighbors(inp_emb, return_distance=False)[0]

        if return_data:
            return [self.data[i] for i in neighbors]
        else:
            return neighbors

    def get_text_embedding(self, texts, batch=1000):
        embeddings = []
        for i in range(0, len(texts), batch):
            text_batch = texts[i : (i + batch)]
            emb_batch = self.use(text_batch)
            embeddings.append(emb_batch)
        embeddings = np.vstack(embeddings)
        return embeddings


def load_recommender(path, start_page=1):
    global recommender
    texts = pdf_to_text(path, start_page=start_page)
    chunks = text_to_chunks(texts, start_page=start_page)
    recommender.fit(chunks) 
    return 'Corpus Loaded.'

# 开始训练语料库
pdf_path='第3章  岩土工程勘察.pdf'
recommender = SemanticSearch()
load_recommender(pdf_path) # 使用fit生成语料库 https://www.intumu.com/article/203
question='钻孔深度相关规定？'
topn_chunks = recommender(question)
print(topn_chunks)

GPT查询代码

def generate_answer(question, openAI_key):
    topn_chunks = recommender(question)  
    prompt = ""
    prompt += 'search results:\n\n'
    for c in topn_chunks:
        prompt += c + '\n\n'

    prompt += (
        "Instructions: Compose a comprehensive reply to the query using the search results given. "
        "Cite each reference using [ Page Number] notation (every result has this number at the beginning). "
        "Citation should be done at the end of each sentence. If the search results mention multiple subjects "
        "with the same name, create separate answers for each. Only include information found in the results and "
        "don't add any additional information. Make sure the answer is correct and don't output false content. "
        "If the text does not relate to the query, simply state 'Text Not Found in PDF'. Ignore outlier "
        "search results which has nothing to do with the question. Only answer what is asked. The "
        "answer should be short and concise. Answer step-by-step. \n\nQuery: {question}\nAnswer: "
    )

    prompt += f"Query: {question}\nAnswer:"
    answer = generate_text(openAI_key, prompt, "text-davinci-003")
#     answer = handle_message(prompt)
    return answer

def generate_text(openAI_key, prompt, engine="text-davinci-003"):
    openai.api_key = openAI_key
    completions = openai.Completion.create(
        engine=engine,
        prompt=prompt,
        max_tokens=512,
        n=1,
        stop=None,
        temperature=0.7,
    )
    message = completions.choices[0].text
    return message

openAI_key = 'sk-zo59kJ9gV7yx8xgsn8jrT3BlbkFJT******' #https://www.intumu.com/article/203
generate_answer(question, openAI_key)

结语

以上类似于AutoGPT或chatPDF的实现原理，感兴趣的读者可以试试。