import std/httpclient import std/htmlparser import std/xmltree import std/strtabs import std/os import scrap import std/json import std/strutils var client = newHttpClient() var html: string #var node: XmlNode proc do_another_loop( node : JsonNode, level : int, searchString : string = "",path :string = "H") = if node.len() > 0: for a in node.keys: if node.kind == JObject: #for i in 0 .. level: if searchString == "" or a.contains(searchString): #for x in 0..level: # stdout.write " " stdout.write "|", path & "-" & a if node[a].kind == JObject: stdout.write("\n") do_another_loop(node[a],level+1,searchString,path & "-" & a) else: if searchString == "" or a.contains(searchString): stdout.write " [", node[a].kind, "]" stdout.write("\n") if node[a].kind == JArray: if searchString == "" or a.contains(searchString): echo "Array Elements: ", node[a].getElems() echo "Array Length", node[a].len() echo "WWWWWWWWWWWWWWW" for b in node[a]: echo b do_another_loop(b,level,searchString,path & "-" & a) #stdout.write "\n" else: echo "\n--------" echo "URL:" var url: string = "https://www.comparis.ch/immobilien/marktplatz/lenzburg/mieten?page=2"#readLine(stdin) echo "given url is: ",url try: html = client.getContent(url) let node = parseHtml(html) var htmlnode: XmlNode var entry : Entry add(entry.desc,Descriptor(name : "test", html_context_tag : "div",html_context_attrs : "class",html_context_key : "css-19re50j", html_tag : "p",aattrs : "class",attrs_key : "css-svet6u")) #comparis stores part of the data in a script tag as a json-string, so we get that part: add(entry.desc,Descriptor(name : "script-json", html_context_tag : "script",html_context_attrs : "type",html_context_key : r"application/json", html_tag : "",aattrs : "",attrs_key : "")) echo("###########") discard entry.getEntryFromHtml(node) #echo entry.desc[1].content[0] let jsonString = replaceWord(entry.desc[1].content[0],sub = r"\"",by = r""") let jsonNode = parseJson(jsonString) #echo "jsonString: ",jsonString #echo "JSON Node:", jsonNode #echo "node fields: ",jsonNode["props"]["pageProps"]["pageTitle"].getStr() let field = jsonNode.getFields() do_another_loop(jsonNode,0,searchString = "resultItems") #echo "resultitem", jsonNode["resultItems"] # for a in jsonNode.keys: # echo a # if jsonNode[a].len() > 0: # for b in jsonNode[a].keys: # echo ">>>" # do_another_loop(jsonNode[a][b]) # else: # echo "---------------" finally: client.close()