es-rebuild

es-rebuild

#!/usr/bin/env python

import sys, MySQLdb, cPickle, base64
sys.path.append("/usr/lib/archivematica/archivematicaCommon")
import databaseInterface
sys.path.append("/usr/lib/archivematica/archivematicaCommon/externals")
import pyes

def index_documents_from_database(index, type):
    indexed = 0

    sql = "SELECT docId, data FROM ElasticsearchIndexBackup WHERE indexName='%s' AND typeName='%s'" % (index, type)
    c, sqlLock = databaseInterface.querySQL(sql)
    row = c.fetchone()
    while row != None:
        sys.stdout.write('.')
        docId = row[0].__str__()
        data  = cPickle.loads(base64.decodestring(str(row[1])))
        result = conn.index(data, index, type, docId)
        indexed = indexed + 1
        row = c.fetchone()
    sqlLock.release()

    return indexed

# establish allows index names and document types
indexTypes = {
    'transfers': 'transfer',
    'aips': 'aip'
}

# make sure command-line usage and specified index name are valid
if len(sys.argv) != 2:
    print 'Usage: %s <aips|transfers>' % (sys.argv[0])
    sys.exit(1)
else:
    index = sys.argv[1]
    if index not in indexTypes:
        print 'Index name must be "aips" or "transfers".'
        sys.exit(1)
    else:
        type = indexTypes[index]

conn = pyes.ES('127.0.0.1:9200')

# delete index if it exists
try:
    conn.delete_index(index)
except:
    print "Error deleting %s index or index already deleted." % (index)

# create new index
conn.create_index(index)

print 'Indexing data.'

indexed = index_documents_from_database(index, type)
print
print '%d documents indexed.' % (indexed)