From bd483ff77c714f6047231e951513bacaf79335d6 Mon Sep 17 00:00:00 2001
From: Dnomd343 <i@343.re>
Date: Sun, 16 Oct 2022 00:48:46 +0800
Subject: [PATCH] feat: crawler script for `108shu.com`

---
 src/108shu.com/crawler.sh | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
 create mode 100755 src/108shu.com/crawler.sh

diff --git a/src/108shu.com/crawler.sh b/src/108shu.com/crawler.sh
new file mode 100755
index 0000000..4f53aca
--- /dev/null
+++ b/src/108shu.com/crawler.sh
@@ -0,0 +1,17 @@
+#!/usr/bin/env bash
+
+cd `dirname $0`
+mkdir -p ./data/html/
+
+python3 catalog.py > ./data/catalog.json
+python3 fetch.py ./data/catalog.json ./data/html/
+python3 extract.py ./data/catalog.json ./data/html/ > ./data/xxrs.json
+
+cd ./data/
+xz -k9 catalog.json
+tar cJf html.tar.xz html/
+xz -k9 xxrs.json
+
+mkdir -p ../archive/
+mv *.xz ../archive/
+cd ../