43 lines
1.4 KiB
Python
43 lines
1.4 KiB
Python
"""抓取 kline8.com 缠论《教你炒股票》系列的配置常量。"""
|
||
|
||
from pathlib import Path
|
||
|
||
# —— 源站点 ——
|
||
BASE_URL = "https://www.kline8.com"
|
||
INDEX_URL = f"{BASE_URL}/chanlun?p=0" # p=0 是目录页
|
||
|
||
|
||
def article_url(p: int) -> str:
|
||
"""第 p 篇文章的 URL（p=1 即第一篇）。"""
|
||
return f"{BASE_URL}/chanlun?p={p}"
|
||
|
||
|
||
# —— 请求参数 ——
|
||
USER_AGENT = (
|
||
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
|
||
"AppleWebKit/537.36 (KHTML, like Gecko) "
|
||
"Chrome/124.0 Safari/537.36"
|
||
)
|
||
REQUEST_TIMEOUT = 20 # 秒
|
||
REQUEST_DELAY = 0.35 # 每次请求间隔（秒），礼貌抓取
|
||
MAX_RETRIES = 3
|
||
|
||
# 源站个别篇目 <h5> 日期残缺（仅写"2006"），用已知准确时间纠偏（精确到分钟）
|
||
DATE_OVERRIDES = {
|
||
91: "2007-12-17 21:40",
|
||
107: "2008-08-19 16:10",
|
||
}
|
||
|
||
# —— 本项目产物（归档） ——
|
||
PROJECT_ROOT = Path(__file__).resolve().parent
|
||
DATA_DIR = PROJECT_ROOT / "data"
|
||
RAW_DIR = DATA_DIR / "raw" # 原始 HTML 归档
|
||
MARKDOWN_DIR = DATA_DIR / "markdown" # 每篇 markdown 归档
|
||
JSON_PATH = DATA_DIR / "articles.json" # 全量结构化数据
|
||
|
||
# —— 目标 Vue 项目（写入数据与图片） ——
|
||
CHAN_PROJECT = Path(r"D:\code\chan")
|
||
CHAN_DATA_FILE = CHAN_PROJECT / "src" / "data" / "chanlun.ts"
|
||
CHAN_IMG_DIR = CHAN_PROJECT / "public" / "chanlun-img"
|
||
CHAN_IMG_URL_PREFIX = "/chanlun-img"
|