Directly upload the code (supports all APIs of the openai package. It is recommended to use deepseek, which is cheap and easy to use)

Analyze multiple pdf files

I use it to analyze blackhat

import os
import PyPDF2
import openai
import time
from tqdm import tqdm
import argparse

初始系统消息

system_message = """
你是一名信息安全专家 帮我分析我提供的ppt中的内容并分别生成标题和 摘要
"""

标记字符串

MARK_STRING = "auto_aied"

读取PDF文件并提取文本内容

def extract_text_from_pdf(pdf_path):
with open(pdf_path, 'rb') as file:
reader = PyPDF2.PdfReader(file)
pages_text = []

    for page_num in tqdm(range(len(reader.pages)), desc="Extracting text from PDF"):
        page = reader.pages[page_num]
        page_text = page.extract_text().strip()
        pages_text.append(page_text)

return pages_text

使用OpenAI API生成摘要

def generate_summary_with_openai(text):
messages = [
{"role": "system", "content": system_message},
{"role": "user", "content": text}
]

start_time = time.time()
response = openai.ChatCompletion.create(
    model="gpt-4.0-o",
    messages=messages
)
end_time = time.time()
elapsed_time = end_time - start_time

result = response.choices[0].message.content.strip()
print(f"Generated summary in {elapsed_time:.2f} seconds")
return result

生成Markdown内容

def generate_markdown(pages_text):
markdown_content = ""

# 将多个页面的内容合并成一个请求
batch_size = 15  # 每15页合并成一个请求
summaries = []

for i in tqdm(range(0, len(pages_text), batch_size), desc="Processing pages"):
    batch_text = "\n".join(pages_text[i:i + batch_size])
    summary = generate_summary_with_openai(batch_text)
    summaries.append(summary)

# 合并所有摘要
combined_summary = "\n".join(summaries)
final_summary = generate_summary_with_openai(combined_summary)

# 生成Markdown内容
markdown_content += f"{final_summary}\n\n"

return markdown_content

保存Markdown文件

def save_markdown_file(markdown_content, output_path):
with open(output_path, "a", encoding="utf-8") as md_file:
md_file.write(markdown_content)

重命名文件

def rename_file(pdf_path, mark_string):
base_name, ext = os.path.splitext(pdf_path)
new_name = f"{base_name}_{mark_string}{ext}"
os.rename(pdf_path, new_name)
return new_name

主函数

def main(pdf_folder_path, output_path, api_key):
# 设置OpenAI API密钥
openai.api_key = api_key

# 清空输出文件
with open(output_path, "w", encoding="utf-8") as md_file:
    md_file.write("")

# 遍历文件夹中的所有PDF文件
for pdf_file in tqdm(os.listdir(pdf_folder_path), desc="Processing PDFs"):
    if pdf_file.endswith(".pdf"):
        pdf_path = os.path.normpath(os.path.join(pdf_folder_path, pdf_file))
        if MARK_STRING in pdf_file:
            print(f"Skipping already processed file: {pdf_file}")
            continue

        print(f"Starting analysis for {pdf_file}...")
        pages_text = extract_text_from_pdf(pdf_path)
        print(f"PDF extraction completed for {pdf_file}.")

        print(f"Starting Markdown generation for {pdf_file}...")
        markdown_content = generate_markdown(pages_text)
        print(f"Markdown generation completed for {pdf_file}.")

        # 添加PDF文件名作为大标题
        markdown_content = f"# {pdf_file}\n\n{markdown_content}"

        print(f"Saving Markdown content for {pdf_file}...")
        save_markdown_file(markdown_content, output_path)
        print(f"Markdown content saved for {pdf_file}.")

        # 重命名文件
        new_name = rename_file(pdf_path, MARK_STRING)
        print(f"Renamed file to: {new_name}")

if name == "main":
parser = argparse.ArgumentParser(description="Analyze PDFs and generate summaries.")
parser.add_argument("pdf_folder_path", type=str, help="Path to the folder containing PDFs")
parser.add_argument("output_path", type=str, help="Path to the output Markdown file")
parser.add_argument("api_key", type=str, help="OpenAI API key")

args = parser.parse_args()

main(args.pdf_folder_path, args.output_path, args.api_key)</code></pre><p>How to use</p><p>python pdf_to_markdown.py "C:\Users\Administrator\Desktop\blackhat-conferences-2024\BlackHat ASIA 2024-Slides" <a href="http://output.md/">output.md</a> "your_openai_api_key"</p><p><br></p><h2>Analyze local images</h2><pre><code class="language-JavaScript">import pytesseract

from PIL import Image
import openai
import argparse

设置Tesseract可执行文件路径（如果需要）

pytesseract.pytesseract.tesseract_cmd = r'C:\Program Files (x86)\Tesseract-OCR\tesseract.exe'

初始系统消息

system_message = """
你现在是一网络安全专家。

要求: 用中文回复
"""

使用OCR解析图片中的文本

def extract_text_from_image(image_path):
image = Image.open(image_path)
text = pytesseract.image_to_string(image)
return text

使用OpenAI API生成回复

def generate_response_with_openai(text, api_key):
openai.api_key = api_key
messages = [
{"role": "system", "content": system_message},
{"role": "user", "content": text}
]

response = openai.ChatCompletion.create(
    model="gpt-4.0-o",
    messages=messages
)
return response.choices[0].message.content.strip()

主函数

def main(image_path, api_key):
# 提取图片中的文本
text = extract_text_from_image(image_path)
print("OCR解析的文本：")
print(text)

# 使用OpenAI API生成回复
response = generate_response_with_openai(text, api_key)
print("OpenAI生成的回复：")
print(response)

if name == "main":
parser = argparse.ArgumentParser(description="Extract text from image using OCR and generate response using OpenAI API.")
parser.add_argument("image_path", type=str, help="Path to the image file")
parser.add_argument("api_key", type=str, help="OpenAI API key")

args = parser.parse_args()

main(args.image_path, args.api_key)</code></pre><p>python image_to_text.py "C:\Users\Administrator\Desktop\1.jpg" "your_openai_api_key”</p><p><br></p><h2>Analyze local ppt</h2><pre><code class="language-JavaScript">from pptx import Presentation

import openai
import argparse

初始系统消息

system_message = """
你是一名信息安全专家 帮我分析我提供的ppt中的内容并分别生成标题和 摘要
"""

读取PPT文件并提取文本内容

def extract_text_from_ppt(ppt_path):
prs = Presentation(ppt_path)
slides_text = []

for slide in prs.slides:
    slide_text = ""
    for shape in slide.shapes:
        if hasattr(shape, "text"):
            slide_text += shape.text + " "
    slides_text.append(slide_text.strip())

return slides_text

使用OpenAI API生成标题和摘要

def generate_title_and_summary_with_openai(text, api_key):
openai.api_key = api_key
messages = [
{"role": "system", "content": system_message},
{"role": "user", "content": text}
]

response = openai.ChatCompletion.create(
    model="gpt-4-o",
    messages=messages
)
result = response.choices[0].message.content.strip()
return result

生成Markdown内容

def generate_markdown(slides_text, api_key):
markdown_content = ""

for i, text in enumerate(slides_text):
    # 生成标题和摘要
    result = generate_title_and_summary_with_openai(text, api_key)
    lines = result.split("\n")
    if lines:
        title = lines[0].strip()
        summary = "\n".join(lines[1:]).strip()

        # 生成Markdown内容
        markdown_content += f"## {title}\n\n"
        markdown_content += f"{summary}\n\n"

return markdown_content

保存Markdown文件

def save_markdown_file(markdown_content, output_path):
with open(output_path, "w", encoding="utf-8") as md_file:
md_file.write(markdown_content)

主函数

def main(ppt_path, output_path, api_key):
slides_text = extract_text_from_ppt(ppt_path)
markdown_content = generate_markdown(slides_text, api_key)
save_markdown_file(markdown_content, output_path)

if name == "main":
parser = argparse.ArgumentParser(description="Analyze PPT content and generate summaries using OpenAI API.")
parser.add_argument("ppt_path", type=str, help="Path to the PPT file")
parser.add_argument("output_path", type=str, help="Path to the output Markdown file")
parser.add_argument("api_key", type=str, help="OpenAI API key")

args = parser.parse_args()

main(args.ppt_path, args.output_path, args.api_key)</code></pre><p><br></p>