seo-analyzer.sh

#!/bin/bash
# SEO 分析器
# 分析网页 SEO 优化情况

set -e

# 颜色定义
RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
BLUE='\033[0;34m'
NC='\033[0m'

URL="$1"

if [ -z "$URL" ]; then
    echo -e "${RED}用法: $0 <URL>${NC}"
    exit 1
fi

echo -e "${BLUE}🦞 SEO 分析器${NC}"
echo -e "${YELLOW}分析目标: $URL${NC}\n"

# 检查 HTTP 状态
check_status() {
    echo -e "${GREEN}=== HTTP 状态 ===${NC}"
    STATUS=$(curl -s -o /dev/null -w "%{http_code}" "$URL")
    if [ "$STATUS" = "200" ]; then
        echo -e "状态码: ${GREEN}$STATUS${NC} ✅"
    else
        echo -e "状态码: ${RED}$STATUS${NC} ❌"
    fi
    
    # 响应时间
    TIME=$(curl -s -o /dev/null -w "%{time_total}" "$URL")
    echo -e "响应时间: ${TIME}s"
}

# 检查 Meta 标签
check_meta() {
    echo -e "\n${GREEN}=== Meta 标签 ===${NC}"
    
    # 获取页面内容
    CONTENT=$(curl -s "$URL")
    
    # Title
    TITLE=$(echo "$CONTENT" | grep -oP '(?<=<title>).*?(?=</title>)' | head -1)
    if [ -n "$TITLE" ]; then
        LEN=${#TITLE}
        echo -e "标题: ${GREEN}$TITLE${NC} ($LEN 字符)"
        if [ $LEN -lt 30 ] || [ $LEN -gt 60 ]; then
            echo -e "  ${YELLOW}⚠ 建议标题长度 30-60 字符${NC}"
        fi
    else
        echo -e "标题: ${RED}未找到${NC} ❌"
    fi
    
    # Description
    DESC=$(echo "$CONTENT" | grep -oP '(?<=<meta name="description" content=").*?(?=")' | head -1)
    if [ -n "$DESC" ]; then
        LEN=${#DESC}
        echo -e "描述: ${GREEN}$DESC${NC} ($LEN 字符)"
        if [ $LEN -lt 120 ] || [ $LEN -gt 160 ]; then
            echo -e "  ${YELLOW}⚠ 建议描述长度 120-160 字符${NC}"
        fi
    else
        echo -e "描述: ${RED}未找到${NC} ❌"
    fi
    
    # Canonical
    CANONICAL=$(echo "$CONTENT" | grep -oP '(?<=<link rel="canonical" href=").*?(?=")' | head -1)
    if [ -n "$CANONICAL" ]; then
        echo -e "Canonical: ${GREEN}$CANONICAL${NC} ✅"
    else
        echo -e "Canonical: ${YELLOW}未设置${NC} ⚠"
    fi
}

# 检查标题结构
check_headings() {
    echo -e "\n${GREEN}=== 标题结构 ===${NC}"
    
    CONTENT=$(curl -s "$URL")
    
    H1_COUNT=$(echo "$CONTENT" | grep -c '<h1' || true)
    if [ "$H1_COUNT" = "1" ]; then
        echo -e "H1 标签: ${GREEN}1 个${NC} ✅"
    else
        echo -e "H1 标签: ${YELLOW}$H1_COUNT 个${NC} ⚠ 建议每页只有一个 H1"
    fi
    
    H2_COUNT=$(echo "$CONTENT" | grep -c '<h2' || true)
    echo -e "H2 标签: $H2_COUNT 个"
    
    H3_COUNT=$(echo "$CONTENT" | grep -c '<h3' || true)
    echo -e "H3 标签: $H3_COUNT 个"
}

# 检查图片
check_images() {
    echo -e "\n${GREEN}=== 图片检查 ===${NC}"
    
    CONTENT=$(curl -s "$URL")
    
    IMG_TOTAL=$(echo "$CONTENT" | grep -c '<img' || true)
    IMG_ALT=$(echo "$CONTENT" | grep '<img' | grep -c 'alt=' || true)
    IMG_NO_ALT=$((IMG_TOTAL - IMG_ALT))
    
    echo -e "总图片数: $IMG_TOTAL"
    echo -e "有 alt 属性: ${GREEN}$IMG_ALT${NC}"
    
    if [ $IMG_NO_ALT -gt 0 ]; then
        echo -e "缺少 alt: ${YELLOW}$IMG_NO_ALT${NC} ⚠"
    fi
}

# 检查链接
check_links() {
    echo -e "\n${GREEN}=== 链接检查 ===${NC}"
    
    CONTENT=$(curl -s "$URL")
    
    INTERNAL=$(echo "$CONTENT" | grep -oP '(?<=href=")[^"]*' | grep -c 'miaoquai.com' || true)
    EXTERNAL=$(echo "$CONTENT" | grep -oP '(?<=href=")[^"]*' | grep -cE '^https?://' | grep -v 'miaoquai.com' || true)
    
    echo -e "内链数: $INTERNAL"
    echo -e "外链数: $EXTERNAL"
}

# 生成报告
generate_report() {
    REPORT_FILE="/tmp/seo-report-$(date +%Y%m%d%H%M%S).md"
    
    cat > "$REPORT_FILE" << EOF
# SEO 分析报告

**URL**: $URL  
**分析时间**: $(date '+%Y-%m-%d %H:%M:%S')

## 检查项目

- [x] HTTP 状态
- [x] Meta 标签
- [x] 标题结构
- [x] 图片优化
- [x] 链接检查

---
🦞 Generated by Miaoquai OpenClaw Tools
EOF
    
    echo -e "\n${GREEN}✅ 报告已生成: $REPORT_FILE${NC}"
}

# 运行所有检查
check_status
check_meta
check_headings
check_images
check_links
generate_report

echo -e "\n${GREEN}✅ SEO 分析完成${NC}"