Difference between revisions of "Simple autotagging"
From OpenKM Documentation
| Line 1: | Line 1: | ||
The script assign keywords to documents name based in database metadata values present into document content. | The script assign keywords to documents name based in database metadata values present into document content. | ||
| − | ''' | + | |
| − | Database metadata:''' | + | '''Description:''' |
| + | * There's a database metadata table '''doc_type''' with document type values. | ||
| + | * There's a '''script''' which looking for key values into documents. | ||
| + | * There's an '''automation task''' - based in scripting - executed after uploaded document which tags documents. | ||
| + | |||
| + | '''Database metadata:''' | ||
<source lang="sql"> | <source lang="sql"> | ||
-- DOCS TYPE | -- DOCS TYPE | ||
Revision as of 13:11, 24 May 2013
The script assign keywords to documents name based in database metadata values present into document content.
Description:
- There's a database metadata table doc_type with document type values.
- There's a script which looking for key values into documents.
- There's an automation task - based in scripting - executed after uploaded document which tags documents.
Database metadata:
-- DOCS TYPE
DELETE FROM OKM_DB_METADATA_TYPE WHERE DMT_TABLE='doc_type';
INSERT INTO OKM_DB_METADATA_TYPE (DMT_TABLE, DMT_REAL_CoLUMN, DMT_TYPE, DMT_VIRTUAL_CoLUMN) VALUES ('doc_type', 'col00', 'text', 'dt_id');
INSERT INTO OKM_DB_METADATA_TYPE (DMT_TABLE, DMT_REAL_CoLUMN, DMT_TYPE, DMT_VIRTUAL_CoLUMN) VALUES ('doc_type', 'col01', 'text', 'dt_description');
-- VALUES
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','1','Article');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','25','Audio');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','6','Broker Note');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','8','Case Study');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','5','Company Information');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','4','Conference Report');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','3','Course Material');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','7','Dissertation');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','12','Form');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','11','Image');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','26','Infographics');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','16','Interview');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','17','Presentation');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','19','Report');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','14','Video');
INSERT INTO OKM_DB_METADATA_VALUE (DMV_TABLE, DMV_COL00, DMV_COL01) VALUES ('doc_type','0','_Other');
Code:
import com.openkm.dao.bean.NodeDocumentVersion;
import com.openkm.dao.bean.NodeDocument;
import com.openkm.extractor.TextExtractorWork;
import com.openkm.api.OKMRepository;
import com.openkm.dao.NodeDocumentVersionDAO;
import com.openkm.dao.NodeDocumentDAO;
import org.hibernate.Session;
import com.openkm.dao.HibernateUtil;
import org.hibernate.Query;
import com.openkm.dao.bean.DatabaseMetadataValue;
import com.openkm.module.db.stuff.DbSessionManager;
import com.openkm.api.OKMProperty;
import com.openkm.dao.HibernateUtil;
import com.openkm.core.DatabaseException;
import org.hibernate.HibernateException;
String systemToken = DbSessionManager.getInstance().getSystemToken();
//String uuid = "7f8b48ee-5efd-48dc-82ec-70053f8ab709";
// Getting path
String docPath = OKMRepository.getInstance().getNodePath(null, uuid);
// Getting doc version uuid
NodeDocumentVersion currentVersion = NodeDocumentVersionDAO.getInstance().findCurrentVersion(uuid);
String docVerUuuid = currentVersion.getUuid();
// Document extractor
TextExtractorWork tew = new TextExtractorWork();
tew.setDocUuid(uuid);
tew.setDocPath(docPath);
tew.setDocVerUuid(docVerUuuid);
//tew.setTenant(1);
// Execute extractor
NodeDocumentDAO.getInstance().textExtractorHelper(tew);
// Getting extracted text
NodeDocument docNode = NodeDocumentDAO.getInstance().findByPk(uuid);
// Text to lowercase
String text = docNode.getText().toLowerCase();
// Looking for metadata description values
String qs = "from DatabaseMetadataValue";
Session session = HibernateUtil.getSessionFactory().openSession();
try {
Query q = session.createQuery(qs);
List ret = q.list();
for (DatabaseMetadataValue dmv : ret ) {
if (text.contains(dmv.getCol01().toLowerCase())) {
OKMProperty.getInstance().addKeyword(systemToken, docPath, dmv.getCol01().toLowerCase());
}
}
} catch (HibernateException e) {
throw new DatabaseException(e.getMessage(), e);
} finally {
HibernateUtil.close(session);
}
