langchain-chat-with-milvus/import_from_wordpress.py
2023-11-21 23:20:50 +08:00

45 lines
850 B
Python

import os
import html2text
import requests
wordpress_url = "https://ivampiresp.com"
api_url = wordpress_url + "/wp-json/wp/v2/posts"
leaf_api_url = "http://localhost:8080/api/documents"
libraryId = 1
jwt = os.getenv("JWT")
# 获取全部文章
res = requests.get(api_url)
res_json = res.json()
for i in range(len(res_json)):
title = res_json[i]["title"]["rendered"]
post_id = res_json[i]["id"]
url = res_json[i]["link"]
text = html2text.HTML2Text().handle(res_json[i]["content"]["rendered"])
content = f"""
文章ID: {post_id}
链接: {url}
---
{text}
"""
result = requests.post(leaf_api_url, json={
"Title": title,
"Content": content,
"LibraryId": libraryId
}, headers={
# "Authorization": f"Bearer {jwt}"
"X-Jwt-Payload": f"{jwt}"
})
print(result.json())