lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ilya Zavorin <izavo...@caci.com>
Subject RE: need to find locations of query hits in doc: works fine for regular text but not for phone numbers
Date Thu, 14 Jun 2012 16:49:23 GMT
OK, so I figured out what the problem was. It wasn't with the digits but rather with the various
delimiters like "(" and "-" that I use.

Essentially, the statement 

	String[] subTerms = qstr.split("\\s+");

Does not split a query the same way as the query parser would do it. And thanks, query.toString(),
helped me see that.

My question now is this: is there a way of easily extracting a sequence of substrings from
query to use in place of the subTerms array I get from split?

I see that sometimes query.toString() returns things like 

	"contents:800 contents:555 contents:1212" 

but other times it's somehting like

	"contents:800 (contents:555 contents:1212)"

So instead of trying to guess what other formats query.toString can produce and trying to
parse those, can I somehow extract the substrings of the query reliably?

Thanks!


-----Original Message-----
From: Jack Krupansky [mailto:jack@basetechnology.com] 
Sent: Wednesday, June 13, 2012 11:42 PM
To: java-user@lucene.apache.org
Subject: Re: need to find locations of query hits in doc: works fine for regular text but
not for phone numbers

Try putting the phone number in quotes in the query:

String qstr = "\"800-555-1212\"";

And check query.toString to see how the query parser analyzed the term, bot with and without
quotes.

And make sure you initialized the query parser with "contents" as the default field.

-- Jack Krupansky

-----Original Message-----
From: Ilya Zavorin
Sent: Wednesday, June 13, 2012 10:52 PM
To: java-user@lucene.apache.org
Subject: need to find locations of query hits in doc: works fine for regular text but not
for phone numbers

Hello All,

I am using 3.4. I need to find locations of query hits in a document. What I've implemented
works fine for textual queries but does not work for phone numbers.

Here's how I index my docs:

String oc = "Joe dialed 800-555-1212 but got a busy signal"; doc.add(new Field("contents",
oc, Field.Store.NO, Field.Index.ANALYZED, Field.TermVector.WITH_POSITIONS_OFFSETS));


Now, here how I find locations. I search for a query. If I get a hit, I split my query (in
case it's multi-word) into words and search for each of them using TermFreqVector like this:


//String qstr = "my multiword query"; // for queries like this it works fine...
String qstr = "800-555-1212"; // ...but not for ones like this Query query = parser.parse(qstr);
TopDocs results = searcher.search(query, Integer.MAX_VALUE); ScoreDoc[] hits = results.scoreDocs;

String[] subTerms = qstr.split("\\s+"); // phone string stays intact here

for (int i = 0; i < hits.length; i++) {
int docId = hits[i].doc;
Document doc = searcher.doc(docId);

TermFreqVector tfvector = reader.getTermFreqVector(docId, "contents"); TermPositionVector
tpvector = (TermPositionVector)tfvector;

for (String subTerm : subTerms)
{
String subq = subTerm.toLowerCase();
int termidx = tfvector.indexOf(subq);  // get termidx = -1 here

TermVectorOffsetInfo[] tvoffsetinfo = tpvector.getOffsets(termidx);
            for (int j=0;j<tvoffsetinfo.length;j++) {
            int offsetStart = tvoffsetinfo[j].getStartOffset();
            int offsetEnd = tvoffsetinfo[j].getEndOffset(); // ...

For a query like "800-555-1212", tfvector.indexOf returns -1. What am I doing wrong?

Thanks,

Ilya Zavorin


---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org 


---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org


Mime
View raw message