0ct0pu5/fess
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix #1959 add isPruned option

Browse source
This commit is contained in:
Shinsuke Sugaya 2019-01-10 12:35:10 +09:00
parent df18a72a4b
commit 8c3c76c22a
2 changed files with 14 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

13
src/main/java/org/codelibs/fess/crawler/transformer/FessXpathTransformer.java
View file

@ -106,6 +106,8 @@ public class FessXpathTransformer extends XpathTransformer implements FessTransf
protected boolean useGoogleOffOn = true;
protected Map<String, Boolean> fieldPrunedRuleMap = new HashMap<>();
@PostConstruct
public void init() {
fessConfig = ComponentUtil.getFessConfig();
@ -170,7 +172,11 @@ public class FessXpathTransformer extends XpathTransformer implements FessTransf
case XObject.CLASS_RTREEFRAG:
case XObject.CLASS_UNRESOLVEDVARIABLE:
default:
final Node value = getXPathAPI().selectSingleNode(document, entry.getValue());
final Boolean isPruned = fieldPrunedRuleMap.get(entry.getKey());
Node value = getXPathAPI().selectSingleNode(document, entry.getValue());
if (isPruned != null && isPruned.booleanValue()) {
value = pruneNode(value);
}
putResultDataBody(dataMap, entry.getKey(), value != null ? value.getTextContent() : null);
break;
}
@ -913,4 +919,9 @@ public class FessXpathTransformer extends XpathTransformer implements FessTransf
}
return null;
}
public void addFieldRule(final String name, final String xpath, final boolean isPruned) {
addFieldRule(name, xpath);
fieldPrunedRuleMap.put(name, isPruned);
}
}

2
src/main/resources/crawler/transformer.xml
View file

@ -19,10 +19,12 @@
<postConstruct name="addFieldRule">
<arg>"title"</arg>
<arg>"//TITLE"</arg>
<arg>true</arg>
</postConstruct>
<postConstruct name="addFieldRule">
<arg>"important_content"</arg>
<arg>"//*[self::H1 or self::H2 or self::H3]"</arg>
<arg>true</arg>
</postConstruct>
</component>