Is there a better implementation to extract a multiline text structure from one file based on the index from the other file? - perl

Question

I have two files. One is a file of table creation strings dumped from database, the other is the name of tables, with a "prompt" as a prefix and a "..." as suffix. Just as below:
file A (index):
prompt branch...
prompt branch_param...
prompt branch_pre_param...
prompt business...
prompt business_map...
prompt business_type...
file B (dump):
CREATE TABLE "KS"."BRANCH"
("BRANCH_CODE" CHARACTER(3) NOT NULL DEFAULT '',
"BRANCH_NAME" CHARACTER(40) NOT NULL DEFAULT '',
"PARAM_LEVEL" INTEGER NOT NULL DEFAULT 0
)
DATA CAPTURE NONE
IN "LONG_DATA_TBS";
CREATE TABLE "KS"."BRANCH2BANK"
("BRANCH_CODE" CHARACTER(3) NOT NULL DEFAULT '',
"BANK_CODE" CHARACTER(6) NOT NULL DEFAULT '',
"ACC_COMP_RESULT" CHARACTER(1) NOT NULL DEFAULT ''
)
DATA CAPTURE NONE
IN "SMALL_TBS";
CREATE TABLE "KS"."BRANCH2BOND"
("BRANCH_CODE" CHARACTER(3) NOT NULL DEFAULT '',
"BOND_CODE" CHARACTER(8) NOT NULL DEFAULT '',
"BOND_NAME" CHARACTER(20) NOT NULL DEFAULT '',
"TOTAL_AMT" DECIMAL(19, 4) NOT NULL DEFAULT 0,
"FINANCING_CUST_NO" CHARACTER(10) NOT NULL DEFAULT '',
"SET_DATE" CHARACTER(8) NOT NULL DEFAULT '',
"SET_TIME" CHARACTER(8) NOT NULL DEFAULT '',
"SET_EMP" CHARACTER(6) NOT NULL DEFAULT '',
"SPARE1" CHARACTER(20) NOT NULL DEFAULT '',
"SPARE2" CHARACTER(20) NOT NULL DEFAULT ''
)
DATA CAPTURE NONE
IN "SMALL_TBS";
CREATE TABLE "KS"."BRANCH_PARAM"
("BRANCH_CODE" CHARACTER(3) NOT NULL DEFAULT '',
"PARAM_CODE" CHARACTER(4) NOT NULL DEFAULT '',
"SET_DATE" CHARACTER(8) NOT NULL DEFAULT '',
"SET_TIME" CHARACTER(8) NOT NULL DEFAULT ''
)
DATA CAPTURE NONE
IN "SMALL_TBS";
CREATE TABLE "KS"."BRANCH_RESERVE_CREDIT_STOCK"
("BRANCH_CODE" CHARACTER(3) NOT NULL DEFAULT '',
"SET_TIME" CHARACTER(8) NOT NULL DEFAULT ''
)
DATA CAPTURE NONE
IN "TX_DATA_TBS"
INDEX IN "TX_INDEX_TBS";
I have written a perl implementation, but I think it is much too ugly and inefficient. Is there a better way to improve this?
my code: (rewritten with Richard and lilydjwg's advice) (last version)
#!/usr/bin/perl
use 5.016;
my (%hash,$cont);
open IN,'<',shift;
while(<IN>){
chomp;
$hash{$1}=1 if /prompt (\w+)\.\.\./;
}
close IN;
open IN,'<',shift;
while(<IN>){
chomp;
$cont = (defined $hash{lc $1}?say "prompt $1..." : 0) if /CREATE TABLE "KS"\."(\w+)"/;
say if $cont == 1;
}
close IN;

Answer 1

Presumably it's the repeated reads you don't like.
So - read the CREATE TABLE file once, checking for:
CREATE TABLE "KS"."(\w+)"
Then you can build up the table definition until the next CREATE TABLE at which point you put the table-definition into a hash keyed by the table-name.
Then, read your prompts and grab the definitions one by one from the hash printing them out.
Alternatively, you could just read the CREATE TABLE file into a single string and search+replace the table-name part since that's all you seem to be changing at the moment. The first approach is more flexible though.
Edit:
You could make the defined bit a little clearer perhaps with:
while ($line=<IN>) {
chomp($line);
if (/CREATE TABLE "KS"\."(\w+)"/ && $hash{lc $1}) {
$line = ...
}
say $line;
}
I like to use an explicit variable in my while-loops once I get beyond a couple of lines too.

Answer 2

It seems that file A is relatively small. You can read and build a set (or the like) from it which contains all the table names. Then read and identify on the SQL dump file, for each table creation statement, check if that table name is in your set.
I don't quite know Perl, but this Python code seems what you want:
import sys
tableNames = {x[7:-3] for l in open(sys.argv[1]) if x.startswith('prompt ')}
for l in open(sys.argv[2]):
if l.startswith('CREATE TABLE "KS"."'):
name = l.split('"')[4].lower()
if name in tableNames:
print("prompt {0}...\nCreate table{0}(".format(name))
print(l, end='')