get the table names from file using UNIX Script

I am having a sample file as given below. This is an SQL Loader control file:

LOAD DATA
APPEND
INTO TABLE XXWIN_TMP_LOADER_TAB
(  seq         POSITION(1:10)  INTEGER EXTERNAL
  ,h_record    POSITION(11:20) CHAR
  ,h_file_name POSITION(21:55) CHAR
)
APPEND
INTO TABLE XXWIN_SQL_LOADER_TAB
(  seq         POSITION(1:10)  INTEGER EXTERNAL
  ,h_record    POSITION(11:20) CHAR
  ,h_file_name POSITION(21:55) CHAR
)
APPEND
INTO TABLE XXWIN_SQL_LOADER_TAB
(  seq         POSITION(1:10)  INTEGER EXTERNAL
  ,h_record    POSITION(11:20) CHAR
  ,h_file_name POSITION(21:55) CHAR
)

I would like to select 开发者_运维百科any number of table names occurring in the file which are starting with 'XX_' and ending with '_TAB' and store it into an array using an UNIX script.

Please advice.

Thanks, Arun

If the file syntax is not changing (the table names start with XX, not XX_):

tnames=`grep -o "TABLE XX[^ ].*_TAB" <file_name> | sed 's/TABLE //g'`
for tn in $tnames; do echo $tn; done

Change the <file_name> to the name of the file.

You don't say which shell, but since sh doesn't support arrays I'm assuming Bash.

tables=($(sed -n '/TABLE /s/TABLE \(XX[^ ]*TAB\) *$/\1/p' inputfile))
for table in ${tables[@]}
do
    echo "$table"
done