From 47a03607836f0806dbff6c541614e6de72a695dd Mon Sep 17 00:00:00 2001 From: "yihua.huang" Date: Tue, 12 Nov 2013 11:51:34 +0800 Subject: [PATCH] #35 add status code to page --- .../src/main/java/us/codecraft/webmagic/Page.java | 10 ++++++++++ .../webmagic/downloader/HttpClientDownloader.java | 1 + 2 files changed, 11 insertions(+) diff --git a/webmagic-core/src/main/java/us/codecraft/webmagic/Page.java b/webmagic-core/src/main/java/us/codecraft/webmagic/Page.java index 0821e6d..a623518 100644 --- a/webmagic-core/src/main/java/us/codecraft/webmagic/Page.java +++ b/webmagic-core/src/main/java/us/codecraft/webmagic/Page.java @@ -33,6 +33,8 @@ public class Page { private Selectable url; + private int statusCode; + private List targetRequests = new ArrayList(); public Page() { @@ -162,6 +164,14 @@ public class Page { return resultItems; } + public int getStatusCode() { + return statusCode; + } + + public void setStatusCode(int statusCode) { + this.statusCode = statusCode; + } + @Override public String toString() { return "Page{" + diff --git a/webmagic-core/src/main/java/us/codecraft/webmagic/downloader/HttpClientDownloader.java b/webmagic-core/src/main/java/us/codecraft/webmagic/downloader/HttpClientDownloader.java index 4286054..389633f 100644 --- a/webmagic-core/src/main/java/us/codecraft/webmagic/downloader/HttpClientDownloader.java +++ b/webmagic-core/src/main/java/us/codecraft/webmagic/downloader/HttpClientDownloader.java @@ -165,6 +165,7 @@ public class HttpClientDownloader implements Downloader { page.setHtml(new Html(UrlUtils.fixAllRelativeHrefs(content, request.getUrl()))); page.setUrl(new PlainText(request.getUrl())); page.setRequest(request); + page.setStatusCode(httpResponse.getStatusLine().getStatusCode()); return page; }