[3] 利用搜索引擎高级语法获取特征网站的目标搜索url

获取特定搜索目标特征url

利用百度、谷歌、bing搜索引擎高级语法生成 指定特征结果的搜索url.

import urllib.parse
import sys

"""
example:
    query = "nginx reverse proxy"
    accurate_query = "accurate_query"
    site = "nginx.org"
    filetype = "pdf"
    inurl = "config"
    intitle = "setup"
    after = "2022-01-01"
    before = "2023-01-01"
    ip = "8.8.8.8",
    page_num = 1
"""


def generate_bing_search_url(query=None, accurate_query=None, site=None, filetype=None, inurl=None, intitle=None,
                             ip=None,
                             after=None, before=None, page_num = 1):
    # 基本查询词
    base_url = "https://www.bing.com/search"

    search_query = ""

    # 构建查询字符串,
    if query:
        search_query = query

    # 处理精确搜索
    elif accurate_query:
        search_query = f'"{query}"'

        # 处理 site 限定
    if site:
        search_query += f" site:{site}"
    # 处理文件类型 filetype
    if filetype:
        search_query += f" filetype:{filetype}"
    # 处理 URL 包含特定字符串
    if inurl:
        search_query += f" inurl:{inurl}"

    # 处理标题包含特定关键词
    if intitle:
        search_query += f" intitle:{intitle}"

    # 处理指定ip托管网站
    if ip:
        search_query += f" ip:{ip}"

    # 处理时间范围
    if after:
        search_query += f" after:{after}"
    if before:
        search_query += f" before:{before}"

    if not search_query:
        return ""

    # URL 编码
    params = {'q': search_query}

    # 分页
    params["first"] = (int(page_num) - 1) * 10

    url = f"{base_url}?{urllib.parse.urlencode(params)}"
    return url


def generate_baidu_search_url(query=None, site=None, filetype=None, inurl=None, intitle=None, accurate_query=None,
                              ip=None, after=None, before=None, page_num = 1):
    # 基本查询词
    base_url = "https://www.baidu.com/s"
    if query:
        search_query = query
    # 如果需要精确匹配，将查询词放在双引号中
    elif accurate_query:
        search_query = f'"{query}"'

    # 处理 site 限定
    if site:
        search_query += f" site:{site}"

    # 处理文件类型 filetype
    if filetype:
        search_query += f" filetype:{filetype}"

    # 处理 URL 包含特定字符串
    if inurl:
        search_query += f" inurl:{inurl}"

    # 处理标题包含特定关键词
    if intitle:
        search_query += f" intitle:{intitle}"

    # 处理 IP 地址
    if ip:
        search_query += f" {ip}"

    # 处理时间范围
    if after:
        search_query += f" after:{after}"
    if before:
        search_query += f" before:{before}"

    if not search_query:
        return ""

    # URL 编码
    params = {'wd': search_query}

    # 分页
    params["pn"] = (int(page_num) - 1) * 10
    url = f"{base_url}?{urllib.parse.urlencode(params)}"
    return url


def generate_google_search_url(query=None, site=None, filetype=None, inurl=None, intitle=None, after=None, before=None, ip = None,
                               accurate_query=None, page_num = 1):
    # 基本查询词
    base_url = "https://www.google.com/search"

    if query:
        search_query = query
    # 如果需要精确匹配，将查询词放在双引号中
    elif accurate_query:
        search_query = f'"{query}"'

    # 处理 site 限定
    if site:
        search_query += f" site:{site}"

    # 处理文件类型 filetype
    if filetype:
        search_query += f" filetype:{filetype}"

    # 处理 URL 包含特定字符串
    if inurl:
        search_query += f" inurl:{inurl}"

    # 处理标题包含特定关键词
    if intitle:
        search_query += f" intitle:{intitle}"

    # 处理时间范围
    if after:
        search_query += f" after:{after}"
    if before:
        search_query += f" before:{before}"

    # 处理 IP 地址
    if ip:
        search_query += f" {ip}"

    # 处理时间范围
    if after:
        search_query += f" after:{after}"
    if before:
        search_query += f" before:{before}"

    if not search_query:
        return ""

    # URL 编码
    params = {'q': search_query}
    # 分页
    params["start"] = (int(page_num) - 1) * 10

    url = f"{base_url}?{urllib.parse.urlencode(params)}"
    return url


if __name__ == "__main__":
    type = input("选择搜索引擎 0 bing 1 baidu 2 google")

    engine_type = None

    if type == "0":
        engine_type = generate_bing_search_url
    elif type == "1":
        engine_type = generate_baidu_search_url
    elif type == "2":
        engine_type = generate_google_search_url
    else:
        print("未指定搜索引擎")
        sys.exit(0)

    # 示例使用
    query = input("指定搜索字符串")
    accurate_query = input("指定精确搜索字符串")
    page_num = input("指定获取页码")
    site = input("指定 目标搜索限制域名")
    filetype = input("指定 目标页面的文件格式")
    inurl = input("指定搜索咪表URL中包含的关键词")
    intitle = input("指定目标网站标题包含指定关键词")
    after = input("指定 目标网站发布的起始时间")
    before = input("指定网页发布的终止时间")
    ip = input("指定目标网站托管的IP")

    page_num = page_num if page_num else 1

    # 生成搜索 URL
    search_url = engine_type(query, site=site, filetype=filetype, inurl=inurl, intitle=intitle, ip=ip,
                             after=after,
                             before=before, page_num= page_num)
    print(search_url)