[
https://issues.apache.org/jira/browse/LUCENE-5356?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13838022#comment-13838022
]
Dawid Weiss commented on LUCENE-5356:
-------------------------------------
A quick look at the patch:
{code}
/** Schema attribute. */
- @Deprecated
public static final String DICTIONARY_SCHEMA_ATTRIBUTE = "dictionary";
{code}
We should not un-deprecate this property, especially that its new meaning is
different to what it was before. The custom dictionary should be a separate
property, with a new semantics.
All the logic in MorfologikLemmatizer seems awkward to me:
{code}
+ @Override
+ public Iterator<WordData> iterator() {
+ if (delegate.size() == 1) {
+ return delegate.get(0).iterator();
+ } else {
+ throw new RuntimeException("No iteration over compound stemmer
forms: "
+ + Arrays.toString(delegate.toArray()));
+ }
+ }
{code}
How can this ever be != 1 if the only place you add a delegate in is in the
constructor?
> more generic lucene-morfologik integration
> ------------------------------------------
>
> Key: LUCENE-5356
> URL: https://issues.apache.org/jira/browse/LUCENE-5356
> Project: Lucene - Core
> Issue Type: Improvement
> Components: modules/analysis
> Affects Versions: 4.6
> Reporter: Michal Hlavac
> Assignee: Dawid Weiss
> Priority: Minor
> Labels: newbie, patch
> Fix For: 5.0, 4.7
>
> Attachments: LUCENE-5356.patch
>
>
> I have little proposal for morfologik lucene module. Current module is
> tightly coupled with polish DICTIONARY enumeration.
> But other people (like me) can build own dictionaries to FSA and use it with
> lucene.
> You can find proposal in attachment and also example usage in analyzer
> (SlovakLemmaAnalyzer).
> It uses dictionary property as String resource from classpath, not
> enumeration.
> One change is, that dictionary variable must be set in MofologikFilterFactory
> (no default value).
--
This message was sent by Atlassian JIRA
(v6.1#6144)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
