| 1 | #!/usr/bin/env python
|
|---|
| 2 |
|
|---|
| 3 | import xapian
|
|---|
| 4 | import os
|
|---|
| 5 | import random
|
|---|
| 6 |
|
|---|
| 7 | os.environ['XAPIAN_FLUSH_THRESHOLD'] = "100000000"
|
|---|
| 8 |
|
|---|
| 9 | db = xapian.WritableDatabase('/tmp/test_index', xapian.DB_CREATE_OR_OPEN)
|
|---|
| 10 |
|
|---|
| 11 | some_data = 'one two three four five six seven eight nine ten'.split()
|
|---|
| 12 |
|
|---|
| 13 | for id in xrange(500000):
|
|---|
| 14 | print id
|
|---|
| 15 |
|
|---|
| 16 | if id % 50000 == 0:
|
|---|
| 17 | print "flushing index"
|
|---|
| 18 | db.flush()
|
|---|
| 19 |
|
|---|
| 20 | value = ''
|
|---|
| 21 | for num in xrange(10):
|
|---|
| 22 | value += chr(65 + random.randint(0, 25))
|
|---|
| 23 |
|
|---|
| 24 | doc = xapian.Document()
|
|---|
| 25 | doc.add_term("Q%d" % id)
|
|---|
| 26 | doc.add_term("XFOO" + some_data[random.randint(0, 9)])
|
|---|
| 27 | doc.add_term("XBAR" + some_data[random.randint(0, 1)])
|
|---|
| 28 |
|
|---|
| 29 | if db.term_exists("Q%d" % id):
|
|---|
| 30 | db.replace_document("Q%d" % id, doc)
|
|---|
| 31 | else:
|
|---|
| 32 | db.add_document(doc)
|
|---|
