Groonga 3.0.4 has been released

BloGroonga

2013-05-29

Groonga 3.0.4 has been released

Groonga 3.0.4 has been released!

How to install: Install

There are three topics for this release.

Supported multiple index section in match_columns
Supported tokenize command experimentally
Supported Debian 8.0 (jessie)

Supported multiple index section in match_columns

In this release, we began to support multiple index section in match_columns.

Here is the sample schema:

table_create Memos TABLE_NO_KEY
column_create Memos title COLUMN_SCALAR ShortText
column_create Memos content COLUMN_SCALAR ShortText

table_create Lexicon TABLE_PAT_KEY ShortText --default_tokenizer TokenBigram
column_create Lexicon title_index COLUMN_INDEX|WITH_POSITION|WITH_SECTION Memos title
column_create Lexicon content_index COLUMN_INDEX|WITH_POSITION|WITH_SECTION Memos content

In the previous release, you need to execute following query if you want to use column index of title with weight.

select Memos 
  --match_columns "Lexicon.title_index * 10 || Lexicon.content_index" 
  --query rroonga --output_columns "title, content, _score"

In this release, you can use almost same thing for multiple column index. To use this feature, you need change schema a bit.

table_create Lexicon TABLE_PAT_KEY ShortText --default_tokenizer TokenBigram
column_create Lexicon memo_index COLUMN_INDEX|WITH_POSITION|WITH_SECTION Memos title,content

Then specify memo_index.(COLUMN_NAME) in the query. Here is the revised query using multiple column index:

select Memos 
  --match_columns "Lexicon.memo_index.title * 10 || Lexicon.memo_index.content" 
  --query rroonga 
  --output_columns "title, content, _score"

There are many case which you want to apply multiple tokenizer or normalizer for each column.

If you specify just column name, there is a case that multiple index is used and unwanted search results returns.

In this release, you can specify specific column which is a member of multiple column index.

Supported tokenize command experimentally

In this release, we began to support tokenize command experimentally.

There is a normalize command to confirm behaviour of normalizer, on the other hand, there is no way to confirm behaviour of tokenizer.

In this release, you can confirm both of normalizer and tokenizer behaviour now.

Here is the syntax of tokenize command:

tokenize tokenizer string [normalizer] [flag]

Here is a sample of tokenize command:

> tokenize TokenBigram "groonga3.0.4"
[
  [0,1369290009.77065,0.000670194625854492],
  [
    {"value":"gr","position":0},
    {"value":"ro","position":1},
    {"value":"oo","position":2},
    {"value":"on","position":3},
    {"value":"ng","position":4},
    {"value":"ga","position":5},
    {"value":"a3","position":6},
    {"value":"3.","position":7},
    {"value":".0","position":8},
    {"value":"0.","position":9},
    {"value":".4","position":10},
    {"value":"4","position":11}
  ]
]

See the following documentation about details.

Supported Debian 8.0 (jessie)

In this release, we began to support Debian 8.0 (jessie).

See install documentation about jessie

Note that squeeze support will be discontinued for a while.

Conclusion

See Release 3.0.4 2013/05/29 about detailed changes since 3.0.3.

Let's search by groonga!

BloGroonga