|
|
#!/usr/bin/env bash
|
|
|
# fetch.sh — WebFetch 的本地替代。DeepSeek 后端下 WebFetch 域名安全校验失败，
|
|
|
# 且 hooks 无法拦截（校验发生在 PreToolUse 之前），故创建此脚本作为完整替代。
|
|
|
#
|
|
|
# 用法：bash scripts/fetch.sh <url> [--raw]
|
|
|
# --raw 输出原始 HTML/响应，不处理
|
|
|
#
|
|
|
# 输出：
|
|
|
# - JSON → pretty-print（python3 -m json.tool）
|
|
|
# - HTML → 基本标签剥离，输出纯文本
|
|
|
# - 其他/text → 原样输出
|
|
|
# - 非文本 → 打印 Content-Type + 大小，不输出二进制
|
|
|
|
|
|
set -euo pipefail
|
|
|
|
|
|
usage() {
|
|
|
echo "用法: bash scripts/fetch.sh <url> [--raw]"
|
|
|
echo " --raw 输出原始响应，不做格式处理"
|
|
|
exit 1
|
|
|
}
|
|
|
|
|
|
[ $# -ge 1 ] || usage
|
|
|
|
|
|
URL="$1"
|
|
|
RAW=false
|
|
|
[ "${2:-}" = "--raw" ] && RAW=true
|
|
|
|
|
|
TMPFILE=$(mktemp)
|
|
|
trap 'rm -f "$TMPFILE"' EXIT
|
|
|
|
|
|
# 下载；同时获取 HTTP 状态码和 Content-Type
|
|
|
HTTP_INFO=$(curl -sSL -w '\n%{http_code}\n%{content_type}' -o "$TMPFILE" \
|
|
|
-H 'User-Agent: Mozilla/5.0 (compatible; fetch.sh/1.0)' \
|
|
|
--max-time 30 \
|
|
|
"$URL" 2>/dev/null || echo -e "\n000\n")
|
|
|
|
|
|
HTTP_CODE=$(echo "$HTTP_INFO" | tail -2 | head -1)
|
|
|
SERVER_CT=$(echo "$HTTP_INFO" | tail -1 | sed 's/;.*//') # 去掉 charset
|
|
|
|
|
|
if [ "$RAW" = true ]; then
|
|
|
cat "$TMPFILE"
|
|
|
exit 0
|
|
|
fi
|
|
|
|
|
|
# 确定内容类型：优先服务器 Content-Type，其次 file 命令
|
|
|
detect_content_type() {
|
|
|
# 服务器声明的类型
|
|
|
case "$SERVER_CT" in
|
|
|
application/json|text/json|application/*+json)
|
|
|
echo "json"; return ;;
|
|
|
esac
|
|
|
|
|
|
# 用 file 命令检测
|
|
|
local ft=$(file -b --mime-type "$TMPFILE" 2>/dev/null)
|
|
|
case "$ft" in
|
|
|
application/json|text/json|application/*+json)
|
|
|
echo "json"; return ;;
|
|
|
esac
|
|
|
|
|
|
# 尝试 JSON 解析（检测 API 返回的 text/plain JSON）
|
|
|
if python3 -c "import json; json.load(open('$TMPFILE'))" 2>/dev/null; then
|
|
|
echo "json"; return
|
|
|
fi
|
|
|
|
|
|
# 服务器 HTML
|
|
|
case "$SERVER_CT" in
|
|
|
text/html|application/xhtml*)
|
|
|
echo "html"; return ;;
|
|
|
esac
|
|
|
|
|
|
case "$ft" in
|
|
|
text/html|application/xhtml*)
|
|
|
echo "html"; return ;;
|
|
|
text/*)
|
|
|
echo "text"; return ;;
|
|
|
esac
|
|
|
|
|
|
echo "binary"
|
|
|
}
|
|
|
|
|
|
CT=$(detect_content_type)
|
|
|
|
|
|
case "$CT" in
|
|
|
json)
|
|
|
python3 -m json.tool "$TMPFILE" 2>/dev/null || cat "$TMPFILE"
|
|
|
;;
|
|
|
html)
|
|
|
# 基本 HTML→文本：删除 script/style，删除标签，压缩空白
|
|
|
sed -E 's/<script[^>]*>[^<]*<\/script>//gi
|
|
|
s/<style[^>]*>[^<]*<\/style>//gi' "$TMPFILE" \
|
|
|
| sed -E 's/<[^>]+>//g' \
|
|
|
| sed -E '/^[[:space:]]*$/d' \
|
|
|
| sed -E 's/&/\&/g; s/</</g; s/>/>/g; s/"/"/g; s/'/'"'"'/g; s/ / /g' \
|
|
|
| head -500
|
|
|
;;
|
|
|
text)
|
|
|
cat "$TMPFILE"
|
|
|
;;
|
|
|
*)
|
|
|
FILE_SIZE=$(wc -c < "$TMPFILE")
|
|
|
echo "[fetch.sh] 非文本响应 (server:${SERVER_CT:-none}, ${FILE_SIZE} bytes)"
|
|
|
echo "[fetch.sh] 使用 'bash scripts/fetch.sh \"$URL\" --raw' 查看原始内容"
|
|
|
echo "[fetch.sh] 或使用 Chrome browser 工具 (navigate) 渲染页面"
|
|
|
;;
|
|
|
esac
|
