Hi,
Sebi wrote:
Does anyone read this message? Can anyone give an answer?
Sorry for the delay, I had a flu and didn't have a chance to answer you.
I have some new questions:
1. I have problems when I search with phrase queries using query parser or direct
binding to the $index->find function. But If I construct a phrase query with
Zend_search_lucene API, $index->find function will return a result.
The following will return no results:
$result = $index->find ('title:"kirchner kurt"') (this is identical with
the following query)
or
$query = Zend_Search_Lucene_Search_QueryParser::parse('title:"kirchner
kurt"');
$result = $index->find ($query);
print $query->__toString(); //outputs (title:"kirchner kurt")
But with API construnction the search works fine:
$query = new Zend_Search_Lucene_Search_Query_Phrase(array('kirchner',
'kurt'), null, 'title');
$result = $index->find($query);
print $query->__toString(); //outputs title:"kirchner kurt"
Please Alexander, can you check if it is bug or something? Thank you.
I just checked it and found a bug in phrase query construction.
'true' parameter was used instead of optional integer "term position". I
just fixed this and committed into SVN.
2. What is the difference between a Binary field and an UnIndexed Field?
> Does UnIndexed type change the stored text? I ask this because
> I saved a text with some latin non-ascii characters as UnIndexed
field and
> this text was changed. Maybe iconv() function was applied on
> stored text? What is the difference after all?
Yes. Only Binary fields are stored "as is".
UnIndexed non-binary field may be transformed by iconv().
Text is transformed to "ASCII//TRANSLIT" now and will be transformed to
"UTF-8" in future.
3. In the documentation is mentioned that, by default, the search will look through all fileds.
> What types must have this fields? Keyword? Text? UnStored?
Search is performed through all indexed fields (Keyword, Text, UnStored).
4. I understand how the automatic optimize process merge segments into new larger ones. But I'm curious about the call of optimize function. I read that it does not functiuon on the same algorithm like auto-optimize. It just merge all segments, no metter of their size or their group into new one. Is it true? In this case the first danger is to reach a maximum size for the segment file.
optimize() call merges all segments into new one.
Yes, if it exceed 2GB, then index will be damaged. I reopened "index
size" issue some time ago
(http://framework.zend.com/issues/browse/ZF-527), so it's in a queue.
5. I met some classes which I don't understand because of the lack of
explications. I need a very short explication for each:
- Zend_Search_Lucene_Search_QueryEntry
Query entry :)
It may be phrase, term or subquery.
Actually it's a boolean query elements.
Each boolean query/subquery has its own context.
- Zend_Search_Lucene_Search_Lexer
It splits query into lexemes (query syntax lexemes, words and phrases).
6. Does Zend_Search_Lucene support steming like Java Lucene? Or maybe this
options will be added in the future?
Yes. Filters are intended for this (take a look on
Zend_Search_Lucene_Analysis_TokenFilter class).
With best regards,
Alexander Veremyev.
Thank you for all your answers. All my respect,
Sebi.
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
