From 63dfca00f965c41f269f64100e6a802181292e1c Mon Sep 17 00:00:00 2001
From: Dnomd343 <i@343.re>
Date: Sat, 15 Oct 2022 16:49:05 +0800
Subject: [PATCH] feat: crawler script for `wxsy.net`

---
 src/wxsy.net/crawler.sh | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
 create mode 100644 src/wxsy.net/crawler.sh

diff --git a/src/wxsy.net/crawler.sh b/src/wxsy.net/crawler.sh
new file mode 100644
index 0000000..299b207
--- /dev/null
+++ b/src/wxsy.net/crawler.sh
@@ -0,0 +1,20 @@
+#!/usr/bin/env bash
+
+cd `dirname $0`
+mkdir -p ./data/html/
+mkdir -p ./data/json/
+
+python3 catalog.py > ./data/catalog.json
+python3 fetch.py ./data/catalog.json ./data/html/
+python3 extract.py ./data/catalog.json ./data/html/ ./data/json
+python3 release.py ./data/catalog.json ./data/json/ > ./data/xxrs.json
+
+cd ./data/
+xz -k9 catalog.json
+tar cJf html.tar.xz html/
+tar cJf json.tar.xz json/
+xz -k9 xxrs.json
+
+mkdir -p ../archive/
+mv *.xz ../archive/
+cd ../