browserSpider/program/spider/scripts/inc/util.lua

--公共文件
--获取table长度
function table_length(t)
	if type(t)~="table" then
		return 0
	end
	local len=0
	for _,_ in pairs(t) do
		len=len+1
	end
	return len
end

--保存网页为pdf文件
function save_page_to_pdf(tab_title,tab_href,dir,title)
	local filename = ""
	if string.find(title,"2024年")~=nil then
		filename = "2024_"..title.."__浏览器生成.pdf"
	elseif string.find(title,"2023年")~=nil then
		filename = "2023_"..title.."__浏览器生成.pdf"
	else
		filename = "2022_"..title.."__浏览器生成.pdf"
	end
	print("准备生成网页pdf",filename)
	local ok2 = browser_print2pdf(tab_href,1000*60*20,dir..filename)
	print("生成pdf",ok2)
end

--页面清理元素，只保留中心元素
local remote_body_el_without_js=[[
	var hold_el = document.querySelector("%s");
	function removeAllChildren(parent) {
	  var children = parent.childNodes;	
	  for (var i = children.length - 1; i >= 0; i--) {
		parent.removeChild(children[i]);
	  }
	}
	removeAllChildren(document.body);
	document.body.appendChild(hold_el);
]]

--页面清理元素，只保留中心元素
function clear_page_el(tab_title,tab_href,timeout,css_selector)
	local js = string.format(remote_body_el_without_js, css_selector)
	local ok, ret = browser_executejs(tab_title,tab_href,timeout,0, js)
	return ok,ret
end