Mobai
/
magic-Dependency
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

enable blank space in quotes in UrlUtils.fixAllRelativeHrefs #80

master
yihua.huang 2014-04-01 20:35:11 +08:00
parent 97b6f46280
commit 2780423e60
2 changed files with 54 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
webmagic-core/src/main/java/us/codecraft/webmagic/utils/UrlUtils.java
View File

@ -73,18 +73,37 @@ public class UrlUtils {
return domain; return domain;
} }
private static Pattern patternForHref = Pattern.compile("(<a[^<>]*href=)[\"']{0,1}([^\"'<>\\s]*)[\"']{0,1}", Pattern.CASE_INSENSITIVE); /**
* allow blank space in quote
*/
private static Pattern patternForHrefWithQuote = Pattern.compile("(<a[^<>]*href=)[\"']([^\"'<>]*)[\"']", Pattern.CASE_INSENSITIVE);
/**
* disallow blank space without quote
*/
private static Pattern patternForHrefWithoutQuote = Pattern.compile("(<a[^<>]*href=)([^\"'<>\\s]+)", Pattern.CASE_INSENSITIVE);
public static String fixAllRelativeHrefs(String html, String url) { public static String fixAllRelativeHrefs(String html, String url) {
html = replaceByPattern(html, url, patternForHrefWithQuote);
html = replaceByPattern(html, url, patternForHrefWithoutQuote);
return html;
}
public static String replaceByPattern(String html, String url, Pattern pattern) {
StringBuilder stringBuilder = new StringBuilder(); StringBuilder stringBuilder = new StringBuilder();
Matcher matcher = patternForHref.matcher(html); Matcher matcher = pattern.matcher(html);
int lastEnd = 0; int lastEnd = 0;
boolean modified = false;
while (matcher.find()) { while (matcher.find()) {
modified = true;
stringBuilder.append(StringUtils.substring(html, lastEnd, matcher.start())); stringBuilder.append(StringUtils.substring(html, lastEnd, matcher.start()));
stringBuilder.append(matcher.group(1)); stringBuilder.append(matcher.group(1));
stringBuilder.append("\"").append(canonicalizeUrl(matcher.group(2), url)).append("\""); stringBuilder.append("\"").append(canonicalizeUrl(matcher.group(2), url)).append("\"");
lastEnd = matcher.end(); lastEnd = matcher.end();
} }
if (!modified) {
return html;
}
stringBuilder.append(StringUtils.substring(html, lastEnd)); stringBuilder.append(StringUtils.substring(html, lastEnd));
return stringBuilder.toString(); return stringBuilder.toString();
} }

44
webmagic-core/src/test/java/us/codecraft/webmagic/utils/UrlUtilsTest.java
View File

@ -3,6 +3,8 @@ package us.codecraft.webmagic.utils;
import org.junit.Assert; import org.junit.Assert;
import org.junit.Test; import org.junit.Test;
import static org.assertj.core.api.Assertions.assertThat;
/** /**
* @author code4crafter@gmail.com <br> * @author code4crafter@gmail.com <br>
* Date: 13-4-21 * Date: 13-4-21
@ -12,19 +14,39 @@ public class UrlUtilsTest {
@Test @Test
public void testFixRelativeUrl() { public void testFixRelativeUrl() {
String fixrelativeurl = UrlUtils.canonicalizeUrl("aa", "http://www.dianping.com/sh/ss/com"); String absoluteUrl = UrlUtils.canonicalizeUrl("aa", "http://www.dianping.com/sh/ss/com");
System.out.println("fix: " + fixrelativeurl); assertThat(absoluteUrl).isEqualTo("http://www.dianping.com/sh/ss/aa");
Assert.assertEquals("http://www.dianping.com/sh/ss/aa", fixrelativeurl);
fixrelativeurl = UrlUtils.canonicalizeUrl("../aa", "http://www.dianping.com/sh/ss/com"); absoluteUrl = UrlUtils.canonicalizeUrl("../aa", "http://www.dianping.com/sh/ss/com");
Assert.assertEquals("http://www.dianping.com/sh/aa", fixrelativeurl); assertThat(absoluteUrl).isEqualTo("http://www.dianping.com/sh/aa");
fixrelativeurl = UrlUtils.canonicalizeUrl("..aa", "http://www.dianping.com/sh/ss/com"); absoluteUrl = UrlUtils.canonicalizeUrl("..aa", "http://www.dianping.com/sh/ss/com");
Assert.assertEquals("http://www.dianping.com/sh/ss/..aa", fixrelativeurl); assertThat(absoluteUrl).isEqualTo("http://www.dianping.com/sh/ss/..aa");
fixrelativeurl = UrlUtils.canonicalizeUrl("../../aa", "http://www.dianping.com/sh/ss/com/");
Assert.assertEquals("http://www.dianping.com/sh/aa", fixrelativeurl); absoluteUrl = UrlUtils.canonicalizeUrl("../../aa", "http://www.dianping.com/sh/ss/com/");
fixrelativeurl = UrlUtils.canonicalizeUrl("../../aa", "http://www.dianping.com/sh/ss/com"); assertThat(absoluteUrl).isEqualTo("http://www.dianping.com/sh/aa");
Assert.assertEquals("http://www.dianping.com/aa", fixrelativeurl);
absoluteUrl = UrlUtils.canonicalizeUrl("../../aa", "http://www.dianping.com/sh/ss/com");
assertThat(absoluteUrl).isEqualTo("http://www.dianping.com/aa");
}
@Test
public void testFixAllRelativeHrefs() {
String originHtml = "<a href=\"/start\">";
String replacedHtml = UrlUtils.fixAllRelativeHrefs(originHtml, "http://www.dianping.com/");
assertThat(replacedHtml).isEqualTo("<a href=\"http://www.dianping.com/start\">");
originHtml = "<a href=\"/start a\">";
replacedHtml = UrlUtils.fixAllRelativeHrefs(originHtml, "http://www.dianping.com/");
assertThat(replacedHtml).isEqualTo("<a href=\"http://www.dianping.com/start a\">");
originHtml = "<a href='/start a'>";
replacedHtml = UrlUtils.fixAllRelativeHrefs(originHtml, "http://www.dianping.com/");
assertThat(replacedHtml).isEqualTo("<a href=\"http://www.dianping.com/start a\">");
originHtml = "<a href=/start tag>";
replacedHtml = UrlUtils.fixAllRelativeHrefs(originHtml, "http://www.dianping.com/");
assertThat(replacedHtml).isEqualTo("<a href=\"http://www.dianping.com/start\" tag>");
} }
@Test @Test