Improve custom analyzer config

[adamretter]

Counterpart to eXist-db/exist#4082.

[duncdrum]

this needs some explication. the difference between String and String[ ] is far from self-evident.

the prose mentions javadocs but lacks a link, its also ambiguous if refers to the javadoc of Lucene 4 which might not be easy to find anymore, or exist-db 5.4.0 which similarly are hard to find in a publicly available location.

also please resolve the conflict, needs a rebase

[line-o]

I would like to see an example for at least one of the new types added to the documentation.
As you just added this you will likely have one at hand.

[adamretter]

@line-o Here you go:

<analyzer class="com.evolvedbinary.lucene.analyzer.PLMultiAnalyzer">
  <param name="minimumTermLength" type="int" value="2" />
  <param name="punctuationDictionary" type="java.lang.char[]">
    <value>'</value>
    <value>-</value>
    <value>’</value>
  </param>
<analyzer>

[line-o]

@adamretter org.apache.lucene.analysis.util.CharArraySet is already supported.

[line-o]

@adamretter My original request was also to add the example as a listing to the documentation for others to be able to see how to use that feature.

[adamretter]

Sorry that wasn't clear at all @line-o . Also that analyser above is private so I don't think it makes sense as an example in the docs. We don't have a public analyser that takes such arguments.

[line-o]

To describe an interface you do not need to publish the implementation. Additional argument types are useless until it is documented how to declare them in the index configuration.

I see you edited your example to now pass in java.lang.char[] (edit: char is literal type)
. So that would be really helpful to have in there like a so:

<analyzer class="my.PunctuationAnalyzer">
  <param name="stringArray" type="char[]">
    <value>'</value>
    <value>-</value>
    <value>’</value>
  </param>
<analyzer>

[line-o]

Please correct the types in the keyword list.

[joewiz]

@duncdrum Are there still outstanding changes you're requesting? If not, could you mark this as approved?

[dizzzz]

@duncdrum ?

[adamretter]

@joewiz @dizzzz Do you think we can get this merged now?

[joewiz]

According to the commit messages @adamretter addressed @duncdrum's concerns, and @duncdrum has had over a month to raise objections, so I will override the block and merge.