public void TestPaging2()
{
var crawler = new RuiJiCrawler();
var request = new Request("https://3w.huanqiu.com/a/4e2d56fd7f51/7DHitRASkPC?p=1&agt=8");
var response = crawler.Request(request);
var content = response.Data.ToString();
var exp = @"
[meta]
#title
css h1.a-title
#date_dt
css .time:text
#content
css .a-con:ohtml
[paging]
css .a-page
css a[href]";
var block = RuiJiBlockParser.ParserBlock(exp);
var result = RuiJiExtractor.Extract(content, block);
if (result.Paging != null && result.Paging.Count > 0 && result.Metas != null && result.Metas.ContainsKey("content"))
{
result = PagingExtractor.MergeContent(request.Uri, result, block);
}
Assert.True(true);
}
public void TestExtractMeta()
{
var crawler = new RuiJiCrawler();
var request = new Request("https://my.oschina.net/zhupingqi/blog/1826317");
var response = crawler.Request(request);
var content = response.Data.ToString();
var parser = new RuiJiParser();
var eb = parser.ParseExtract(@"
[meta]
#title
css h1.header:text
#author
css div.blog-meta .avatar + span:text
#date
css div.blog-meta > div.item:first:text
regS /发布于/ 1
#words_i
css div.blog-meta > div.item:eq(1):text
regS / / 1
#content
css #articleContent:html" );
var result = RuiJiExtractor.Extract(content, eb.Result);
Assert.True(true);
}
public void TestPaging()
{
var crawler = new RuiJiCrawler();
var request = new Request("https://www.kuaidaili.com/free/inha/1/");
var response = crawler.Request(request);
var content = response.Data.ToString();
var block = new ExtractBlock();
var s = RuiJiBlockParser.ParserBlock(@"
[tile]
css table.table-bordered tr:gt(0):ohtml
[meta]
#ip
css td[data-title='IP']:text
# port
css td[data-title='PORT']:text
[paging]
css #listnav a:[href]
");
var result = RuiJiExtractor.Extract(content, s);
Assert.True(true);
}
public static List <ExtractResult> Extract(ExtractRequest request)
{
if (NodeConfigurationSection.Standalone)
{
var result = RuiJiExtractor.Extract(request);
return(result);
}
else
{
var proxyUrl = ProxyManager.Instance.Elect(NodeProxyTypeEnum.FEEDPROXY);
if (string.IsNullOrEmpty(proxyUrl))
{
throw new Exception("no available Extractor proxy servers");
}
proxyUrl = IPHelper.FixLocalUrl(proxyUrl);
var client = new RestClient("http://" + proxyUrl);
var restRequest = new RestRequest("api/ep/extract");
restRequest.Method = Method.POST;
restRequest.JsonSerializer = new NewtonJsonSerializer();
var json = JsonConvert.SerializeObject(request);
restRequest.AddJsonBody(json);
restRequest.Timeout = 15000;
var restResponse = client.Execute(restRequest);
var response = JsonConvert.DeserializeObject <List <ExtractResult> >(restResponse.Content);
return(response);
}
}
public object TestFeed(FeedModel feed, [FromUri] bool down, [FromUri] bool debug = false)
{
try
{
var compile = new UrlCompile();
var addrs = compile.GetResult(feed.Address);
var results = new List <ExtractResult>();
foreach (var addr in addrs)
{
feed.Address = addr.ToString();
var job = new FeedJob();
var snap = job.DoTask(feed, false);
if (string.IsNullOrEmpty(feed.RuiJiExpression))
{
results.Add(new ExtractResult());
continue;
}
var block = RuiJiBlockParser.ParserBlock(feed.RuiJiExpression);
var result = RuiJiExtractor.Extract(snap.Content, block);
if (!debug)
{
CrawlTaskFunc.ClearContent(result);
}
if (down)
{
var s = new FileStorage(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "www", "download"));
var files = result.Content.ToString().Replace("\r\n", "\n").Split('\n');
foreach (var file in files)
{
if (!string.IsNullOrEmpty(file) && Uri.IsWellFormedUriString(file, UriKind.Absolute))
{
var res = Crawler.Request(file);
var c = new DownloadContentModel();
c.Url = file.Trim();
c.IsRaw = res.IsRaw;
c.Data = res.Data;
s.Insert(c);
}
}
}
results.Add(result);
}
return(results);
}
catch (Exception ex)
{
return(ex);
}
}
public void TestLocalExtract()
{
var crawler = new RuiJiCrawler();
var request = new Request("http://www.ruijihg.com/%e5%bc%80%e5%8f%91/");
var response = crawler.Request(request);
var content = response.Data.ToString();
var block = new ExtractBlock();
block.Selectors = new List <ISelector>
{
new CssSelector(".entry-content", CssTypeEnum.INNERHTML)
};
block.TileSelector = new ExtractTile
{
Selectors = new List <ISelector>
{
new CssSelector(".pt-cv-content-item", CssTypeEnum.INNERHTML)
}
};
//block.TileSelector.Metas.AddMeta(new ExtractBase {
// Name = "title",
// Selectors = new List<ISelector> {
// new CssSelector(".pt-cv-title")
// }
//});
//block.TileSelector.Metas.AddMeta(new ExtractBase
//{
// Name = "url",
// Selectors = new List<ISelector> {
// new CssSelector(".pt-cv-readmore","href")
// }
//});
block.TileSelector.Metas.AddMeta("title", new List <ISelector> {
new CssSelector(".pt-cv-title")
});
block.TileSelector.Metas.AddMeta("url", new List <ISelector> {
new CssSelector(".pt-cv-readmore", "href")
});
var r = RuiJiExtractor.Extract(content, block);
Assert.IsTrue(r.Content.ToString().Length > 0);
Assert.IsTrue(r.Tiles.Count > 0);
}
public void TestExtract2()
{
var crawler = new RuiJiCrawler();
var request = new Request("https://www.oschina.net/blog");
var response = crawler.Request(request);
var content = response.Data.ToString();
var parser = new RuiJiParser();
var eb = parser.ParseExtract("css a.blog-title-link:[href]\nexp https://my.oschina.net/*/blog/*");
var result = RuiJiExtractor.Extract(content, eb.Result);
Assert.True(true);
}
public List <ExtractResult> Extract([FromBody] string json)
{
var node = ServerManager.Get(Request.RequestUri.Authority);
var request = JsonConvert.DeserializeObject <ExtractRequest>(json);
if (node.NodeType == Node.NodeTypeEnum.Extractor)
{
var result = RuiJiExtractor.Extract(request);
return(result);
}
else
{
return(Extractor.Extract(request));
}
}
public List <string> ExtractAddress(FeedSnapshot feed)
{
var block = new ExtractBlock();
block.TileSelector.Selectors.Add(new CssSelector("a", "href"));
if (feed.UseBlock)
{
if (!string.IsNullOrEmpty(feed.BlockExpression))
{
block = JsonConvert.DeserializeObject <ExtractBlock>(feed.BlockExpression);
}
}
else
{
if (!string.IsNullOrEmpty(feed.RuiJiExpression))
{
block.TileSelector.Selectors.Clear();
var parser = new RuiJiParser();
var s = RuiJiBlockParser.ParserBase(feed.RuiJiExpression).Selectors;
block.TileSelector.Selectors.AddRange(s);
}
}
var result = RuiJiExtractor.Extract(feed.Content, block);
var results = new List <string>();
if (result.Tiles != null)
{
foreach (var item in result.Tiles)
{
var href = item.Content.ToString();
if (href.Contains("#"))
{
href = href.Substring(0, href.IndexOf('#'));
}
if (Uri.IsWellFormedUriString(href, UriKind.Relative))
{
href = new Uri(new Uri(feed.Url), href).AbsoluteUri.ToString();
}
results.Add(href);
}
}
return(results.Distinct().ToList());
}
public void TestExtract()
{
var crawler = new RuiJiCrawler();
var request = new Request("http://www.ruijihg.com/%e5%bc%80%e5%8f%91/");
var response = crawler.Request(request);
var content = response.Data.ToString();
var block = new ExtractBlock();
var s = RuiJiBlockParser.ParserBase("css a:[href]").Selectors;
block.TileSelector.Selectors.AddRange(s);
var result = RuiJiExtractor.Extract(content, block);
Assert.True(true);
}
public void TestJsonPExtract()
{
var url = "http://app.cannews.com.cn/roll.php?do=query&callback=jsonp1475197217819&_={# ticks() #}&date={# now(\"yyyy-MM-dd\") #}&size=20&page=1";
var f = new UrlCompile();
//url = f.Compile(url);
var c = new RuiJiCrawler();
var response = c.Request(new Request(url));
var expression = @"
reg /jsonp[\d]+?\((.*)\)/ 1
jpath $..url
";
var b = RuiJiBlockParser.ParserBlock(expression);
var result = RuiJiExtractor.Extract(response.Data.ToString(), b);
Assert.IsTrue(result.Content.ToString().Length > 0);
}
public void TestPaging()
{
var crawler = new RuiJiCrawler();
var request = new Request("https://www.kuaidaili.com/free/inha/10");
var response = crawler.Request(request);
var content = response.Data.ToString();
var exp = @"
[tile]
css table.table-bordered tr:gt(0):ohtml
[meta]
#ip
css td[data-title='IP']:text
#port
css td[data-title='PORT']:text
[paging]
css #listnav a[href]";
var block = RuiJiBlockParser.ParserBlock(exp);
var result = RuiJiExtractor.Extract(content, block);
if (result.Paging != null && result.Paging.Count > 0 && result.Tiles != null)
{
var storage = new FileStorage(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "www", "download"));
PagingExtractor.CrawlPage(request.Uri, result, block, (u, res) => {
var c = new DownloadContentModel();
c.Url = u.AbsolutePath.Trim();
c.IsRaw = false;
c.Data = JsonConvert.SerializeObject(res.Tiles);
storage.Insert(c);
}, int.MaxValue);
}
Assert.True(true);
}
protected List <string> ExtractFeedAddress(Snapshot snapshot)
{
var block = new ExtractBlock();
block.TileSelector.Selectors.Add(new CssSelector("a", "href"));
if (!string.IsNullOrEmpty(snapshot.Expression))
{
block.TileSelector.Selectors.Clear();
var parser = new RuiJiParser();
var s = RuiJiBlockParser.ParserBase(snapshot.Expression).Selectors;
block.TileSelector.Selectors.AddRange(s);
}
var result = RuiJiExtractor.Extract(snapshot.Content, block);
var results = new List <string>();
if (result.Tiles != null)
{
foreach (var item in result.Tiles)
{
var href = item.Content.ToString();
if (href.Contains("#"))
{
href = href.Substring(0, href.IndexOf('#'));
}
if (Uri.IsWellFormedUriString(href, UriKind.Relative))
{
href = new Uri(new Uri(snapshot.RequestUrl), href).AbsoluteUri.ToString();
}
results.Add(href);
}
}
return(results.Distinct().ToList());
}
public void TestExtractTile()
{
var crawler = new RuiJiCrawler();
var request = new Request("http://www.ruijihg.com/archives/category/tech/bigdata");
var response = crawler.Request(request);
var content = response.Data.ToString();
var parser = new RuiJiParser();
var eb = parser.ParseExtract(@"[tile]
css article:html
[meta]
#title
css .entry-header:text
#summary
css .entry-header + p:text
ex /Read more »/ -e" );
var result = RuiJiExtractor.Extract(content, eb.Result);
Assert.True(true);
}
public object Run(object t, ParallelTask task)
{
var model = t as CrawlTaskModel;
var results = new List <object>();
var reporter = task.Progress as IProgress <string>;
reporter.Report("正在读取Feed记录");
var feed = FeedLiteDb.GetFeed(model.FeedId);
reporter.Report("正在下载 Feed");
var compile = new UrlCompile();
var addrs = compile.GetResult(feed.Address);
foreach (var addr in addrs)
{
feed.Address = addr.ToString();
var job = new FeedJob();
var snap = job.DoTask(feed, false);
reporter.Report("Feed 下载完成");
var block = RuiJiBlockParser.ParserBlock(feed.RuiJiExpression);
var feedResult = RuiJiExtractor.Extract(snap.Content, block);
results.Add(feedResult);
reporter.Report("正在提取Feed地址");
var j = new FeedExtractJob();
var urls = j.ExtractAddress(snap);
reporter.Report("Feed地址提取完成");
if (!string.IsNullOrEmpty(snap.RuiJiExpression))
{
foreach (var url in urls)
{
reporter.Report("正在提取地址 " + url);
var result = Cooperater.GetResult(url);
if (result != null)
{
var cm = new ContentModel();
cm.Id = model.FeedId;
cm.Url = url;
cm.Metas = result.Metas;
cm.CDate = DateTime.Now;
results.Add(cm);
}
}
}
reporter.Report("计算完成");
if (!model.IncludeContent)
{
results.ForEach((m) =>
{
ClearContent(m);
});
}
}
return(results);
}
