Searching after update

Question

If I create a CONTEXT index on a column and then update the value in the column for a single row, that row is no longer found when I do a CONTAINS() search, even if the new and old value both match the CONTAIN() condition. Is there any way to include updated rows in the search results before a sync_index is performed?

Answer 1

No, as far as I know, there is no way to include updated rows in a contains query until after synchronization or rebuild or optimization or drop and recreate or some such thing. When the row is updated, it is removed from the ...$k table and marked as unused in the ...$n table. The demonstration below shows what happens in these tables after update, after synchronization and after dropping and recreating the index.
SCOTT@10gXE> CREATE TABLE test_tab (test_col VARCHAR2 (30))
  2  /
Table created.
SCOTT@10gXE> INSERT INTO test_tab VALUES ('word1 word2')
  2  /
1 row created.
SCOTT@10gXE> CREATE INDEX test_idx ON test_tab (test_col) INDEXTYPE IS CTXSYS.CONTEXT
  2  /
Index created.
SCOTT@10gXE> SELECT ROWID, t.* FROM test_tab t
  2  /
ROWID              TEST_COL
AAAH6CAABAAAKwiAAA word1 word2
SCOTT@10gXE> SELECT ROWID, t.* FROM test_tab t WHERE CONTAINS (test_col, 'word1') > 0
  2  /
ROWID              TEST_COL
AAAH6CAABAAAKwiAAA word1 word2
SCOTT@10gXE> SELECT token_text, token_first, token_last FROM dr$test_idx$i
  2  /
TOKEN_TEXT                                                       TOKEN_FIRST TOKEN_LAST
WORD1                                                                      1          1
WORD2                                                                      1          1
SCOTT@10gXE> SELECT * FROM dr$test_idx$k
  2  /
     DOCID TEXTKEY
         1 AAAH6CAABAAAKwiAAA
SCOTT@10gXE> SELECT * FROM dr$test_idx$n
  2  /
no rows selected
SCOTT@10gXE> UPDATE test_tab
  2  SET    test_col = 'word1 word3'
  3  WHERE  test_col = 'word1 word2'
  4  /
1 row updated.
SCOTT@10gXE> COMMIT
  2  /
Commit complete.
SCOTT@10gXE> SELECT ROWID, t.* FROM test_tab t
  2  /
ROWID              TEST_COL
AAAH6CAABAAAKwiAAA word1 word3
SCOTT@10gXE> SELECT ROWID, t.* FROM test_tab t WHERE CONTAINS (test_col, 'word1') > 0
  2  /
no rows selected
SCOTT@10gXE> SELECT token_text, token_first, token_last FROM dr$test_idx$i
  2  /
TOKEN_TEXT                                                       TOKEN_FIRST TOKEN_LAST
WORD1                                                                      1          1
WORD2                                                                      1          1
SCOTT@10gXE> SELECT * FROM dr$test_idx$k
  2  /
no rows selected
SCOTT@10gXE> SELECT * FROM dr$test_idx$n
  2  /
NLT_DOCID N
         1 U
SCOTT@10gXE> EXEC CTX_DDL.SYNC_INDEX ('TEST_IDX')
PL/SQL procedure successfully completed.
SCOTT@10gXE> SELECT ROWID, t.* FROM test_tab t
  2  /
ROWID              TEST_COL
AAAH6CAABAAAKwiAAA word1 word3
SCOTT@10gXE> SELECT ROWID, t.* FROM test_tab t WHERE CONTAINS (test_col, 'word1') > 0
  2  /
ROWID              TEST_COL
AAAH6CAABAAAKwiAAA word1 word3
SCOTT@10gXE> SELECT token_text, token_first, token_last FROM dr$test_idx$i
  2  /
TOKEN_TEXT                                                       TOKEN_FIRST TOKEN_LAST
WORD1                                                                      1          1
WORD1                                                                      2          2
WORD2                                                                      1          1
WORD3                                                                      2          2
SCOTT@10gXE> SELECT * FROM dr$test_idx$k
  2  /
     DOCID TEXTKEY
         2 AAAH6CAABAAAKwiAAA
SCOTT@10gXE> SELECT * FROM dr$test_idx$n
  2  /
NLT_DOCID N
         1 U
SCOTT@10gXE> DROP INDEX test_idx
  2  /
Index dropped.
SCOTT@10gXE> CREATE INDEX test_idx ON test_tab (test_col) INDEXTYPE IS CTXSYS.CONTEXT
  2  /
Index created.
SCOTT@10gXE> SELECT ROWID, t.* FROM test_tab t
  2  /
ROWID              TEST_COL
AAAH6CAABAAAKwiAAA word1 word3
SCOTT@10gXE> SELECT ROWID, t.* FROM test_tab t WHERE CONTAINS (test_col, 'word1') > 0
  2  /
ROWID              TEST_COL
AAAH6CAABAAAKwiAAA word1 word3
SCOTT@10gXE> SELECT token_text, token_first, token_last FROM dr$test_idx$i
  2  /
TOKEN_TEXT                                                       TOKEN_FIRST TOKEN_LAST
WORD1                                                                      1          1
WORD3                                                                      1          1
SCOTT@10gXE> SELECT * FROM dr$test_idx$k
  2  /
     DOCID TEXTKEY
         1 AAAH6CAABAAAKwiAAA
SCOTT@10gXE> SELECT * FROM dr$test_idx$n
  2  /
no rows selected
SCOTT@10gXE>