lee/heritrix3
1
0
Fork 0
mirror of https://github.com/internetarchive/heritrix3.git synced 2026-09-23 22:26:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

* ExtractorMultipleRegex.java

Browse Source 
remove the "fooIndex" thing from available bindings, since it's kinda hacky and turned out not to be needed for our use case
This commit is contained in:
Noah Levitt
2012-11-15 18:56:54 -08:00
parent 4af58f38e5
commit 22ea489c57
1 changed files with 4 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/src/main/java/org/archive/modules/extractor/ExtractorMultipleRegex.java
+4 -10
View File
@@ -90,7 +90,7 @@ public class ExtractorMultipleRegex extends Extractor {
public MatchList(String regex, CharSequence cs) {
Matcher matcher = TextUtils.getMatcher(regex, cs);
while (matcher.find()) {
add(new GroupList(matcher.toMatchResult()));
add(new GroupList(matcher));
}
}
public MatchList(GroupList... groupList) {
@@ -133,13 +133,13 @@ public class ExtractorMultipleRegex extends Extractor {
}
// run all the regexes on the content and cache results
for (String patternName: getContentRegexes().keySet()) {
String regex = getContentRegexes().get(patternName);
for (String regexName: getContentRegexes().keySet()) {
String regex = getContentRegexes().get(regexName);
MatchList matchList = new MatchList(regex, cs);
if (matchList.isEmpty()) {
return; // no match found for regex, so we can stop now
}
matchLists.put(patternName, matchList);
matchLists.put(regexName, matchList);
}
Template groovyTemplate;
@@ -176,14 +176,8 @@ public class ExtractorMultipleRegex extends Extractor {
int tmp = outlinkIndex;
for (int regexIndex = 0; regexIndex < regexNames.length; regexIndex++) {
MatchList matchList = matchLists.get(regexNames[regexIndex]);
int matchIndex = tmp % matchList.size();
bindings.put(regexNames[regexIndex], matchList.get(matchIndex));
// make the index of this match available to the template as well
bindings.put(regexNames[regexIndex] + "Index", matchIndex);
tmp = tmp / matchList.size();
}