3

This is my first post, and I have no idea how I managed anything before StackExchage, Google, Wiki, GNU, Internet, the list goes on :)

I am trying to find a way to construct a SQL database generator bash script, which currently looks like this...

renice -n 19 $$;

idx=32768;
dbt='Radix_en';

cat Domains_en.txt;
cat Tables_en.txt;

while read; 
do
checks="$(echo -n $REPLY | md5sum)";
checks=${checks%"  -"};

echo "insert into $dbt values ($idx,'$(uuidgen)','${checks}',$REPLY);";
idx=$((idx+1));

done < Data.txt;
echo "commit;";

The data comes from Data.txt, currently in the form of:

'NUMBER','US_EN','LATIN','GREEK','GERMAN'
0,'zero','nulla','μηδέν','Null'
1,'one','Unum','ένα','ein'

The output is valid SQL (for Firebird):

create domain ...;
create domain ...;
commit;

create table ( ... );
create table ( ... );
commit;

insert into Radix_en values (32768,'dff0207a-591f-4435-9f8b-7b9b3e6ba2c1','d1f77359b3f7236806489ba3108c771f','NUMBER','US_EN','LATIN','GREEK','GERMAN');
insert into Radix_en values (32769,'5ef0e634-5c96-4ae4-92a8-0d68c02ffeb6','4e3f710600230cf0520bf32269511062',0,'zero','nulla','μηδέν','Null');
insert into Radix_en values (32770,'eae9cacc-3ee3-4471-afad-e5af201da435','9ab2f782988416431238ec63277b11df',1,'one','Unum','ένα','ein');
commit;

I would like to find a way to generate the MD5 checksum for every field, instead of the entire line of text including delimiters.

Data.txt format is not yet finalized, and I may change its format to anything which would makes this possible or easier.

Also, if there has to be several separate steps - fine, since the entire process should be scripted and automated. I was considering processing Data.txt first, then run it through this script, but I would like to simplify the process as much as possible. The number of different Data.txt files could be large, and I still have numerous other processors to include.

As the matter of fact I am also trying to learn more about bash scripting, and I am rather hooping to find expert approach and advise to this problem more than a specific solution.

I am not even sure if my post title is the solution I need, and thus it is related to my question. I was not sure if I should post this on Superuser where I usually visit or here. So I post here first, and sorry if I am off a bit.

Thanks!

Sandor

... Edited to add more on 08/23/2014 3:00 AM

Thanks to mikeserv using IFS is working, so my scripts now looks like this:

renice -n 19 $$ > /dev/null; #for now

idx=32768;
dbt='Radix_en';

cat Domains_en.txt;
cat Tables_en.txt;

while read; 
do

gid="$idx,'$(uuidgen)'";

IFS=,; set -f # no gobbling!
echo "insert into $dbt values ($gid";
    for field in $REPLY
    do  
    printf '%s' ",$field,'";
    printf '%s' "$field" | md5sum;
    done | cut -d\  -f1;


echo "$var);";

idx=$((idx+1));

done < Data.txt;

The output is great, the line breaks are making text edit/search much easier while Firebird is still happy, except one thing..

Here is the output:

create domain ...;
create domain ...;
commit;

create table ( ... );
create table ( ... );
commit;

insert into Radix_en values (32768,'303f8957-57cf-4485-ace4-d21c7cf144e6'
,'NUMBER','722d79c16b51fe86610972b8d950038c
,'US_EN','b63fb39e32b062c76694bec58c4f8c67
,'LATIN','fd6f27a3c59111fc2a0b5e452595ef3d
,'GREEK','c081310697bb6b7d7bed5034824e2530
,'GERMAN','15db1d0e1b0861d8ac1f391db801493a
);
insert into Radix_en values (32769,'e7fdf095-d31c-4c59-a23b-7ea67db7aefb'
,0,'cfcd208495d565ef66e7dff9f98764da
,'zero','01b40535afbfd9611e910f58f4ab5146
,'nulla','584edd0b6638798dee53e2c23e84e2d1
,'μηδέν','cd3ed2f1039ed8668b4d48e742bd2e5b
,'Null','e0a93a9e6b0eb1688837d8bab9b4badb
);
insert into Radix_en values (32770,'a21916b5-2a05-4656-ad4e-c8cfee1abfcc'
,1,'c4ca4238a0b923820dcc509a6f75849b
,'one','7e31533231a12e4a560a18ac8cd02112
,'Unum','05d92bcbffbf59b375f25945e9af2dd0
,'ένα','826f5e2d5ba7ace48f4d6fe3c5e2925f
,'ein','dcc09a2cb665ca332d1689cb11aff592
);
commit;

The md5 hash is missing a delimiter at the end, and I have no idea how to negotiate the output with the pipes. What is it I am not understanding here?

Since in this particular case the data fields are going to hold code for programmable ICs no extra characters are going to be acceptable in the checksum between the delimiters, and so far it looks like so. Again, the code is in ASCII and my delimiter is going be something that is not part of ASCII so it is safe. Firebird also going to reject anything not ASCII.

If you would be so kind to point me to how to finish this script, as I am already banging my head against some new issues IFS just showed me (yes, file paths on Windows). I'll try and see how this script is going to work with pure ASCII, then I would like to move on and make another post about some more issues.

Thanks again for your help!

Sandor

... Edited to final on 08/30/2014 7:00 PM

Replacing cut with sed seems to work. Firebird field input still needs to be escaped for semi-colons (') with another of the same added, and the current comma IFS delimiter in data files still has to be replaced with non-ASCII. Instead of recursive file lists this script is still a single-file input. echo should probably be replaced by printf, and a whole lot of more...

Here is the final script excluding the shebang:

renice -n 19 $$ >> Radix_en_log.txt;

idx=32768; dbt='Radix_en';
cat Domains_en.txt; cat Tables_en.txt;

while read;     do
gid="$idx,'$(uuidgen)'";

IFS=,; set -f;
echo "insert into $dbt values ($gid";
    for field in $REPLY
    do  
        printf '%s' ",$field,'"; printf '%s' "$field" | md5sum;
    done | sed "s/[ ][ ][-]/\'/g"; printf '%s\n' ");";
    idx=$((idx+1));
        done < Data.txt;
echo "commit;";

Here is the output:

create domain ...;
create domain ...;
commit;

create table ( ... );
create table ( ... );
commit;

insert into Radix_en values (32768,'2f675b86-b2b4-4e52-b000-e6a8cf0f3dca'
,'NUMBER','722d79c16b51fe86610972b8d950038c'
,'US_EN','b63fb39e32b062c76694bec58c4f8c67'
,'LATIN','fd6f27a3c59111fc2a0b5e452595ef3d'
,'GREEK','c081310697bb6b7d7bed5034824e2530'
,'GERMAN','15db1d0e1b0861d8ac1f391db801493a'
);
insert into Radix_en values (32769,'e2afcd65-9a1b-49e3-baf1-74b0619a4776'
,0,'cfcd208495d565ef66e7dff9f98764da'
,'zero','01b40535afbfd9611e910f58f4ab5146'
,'nulla','584edd0b6638798dee53e2c23e84e2d1'
,'μηδέν','cd3ed2f1039ed8668b4d48e742bd2e5b'
,'Null','e0a93a9e6b0eb1688837d8bab9b4badb'
);
insert into Radix_en values (32770,'f51b72eb-d64f-4e9e-ab49-8954df9505cd'
,1,'c4ca4238a0b923820dcc509a6f75849b'
,'one','7e31533231a12e4a560a18ac8cd02112'
,'Unum','05d92bcbffbf59b375f25945e9af2dd0'
,'ένα','826f5e2d5ba7ace48f4d6fe3c5e2925f'
,'ein','dcc09a2cb665ca332d1689cb11aff592'
);
commit;

Thanks!

Sandor

5
  • 1
    are the comma delimiters safe? like - is there a chance of them occurring within the quoted strings?
    – mikeserv
    Commented Aug 22, 2014 at 18:06
  • @mikeserv - The majority of the data does not yet exists, so I can use anything. Even when exporting from another DBS the structure too, may be changed. edit I guess what I mean is the delimiters do not yet exists, so there is no chance of not being safe.
    – arch-abit
    Commented Aug 22, 2014 at 18:27
  • I realize this is more of a design issue for now, but I add: as far as delimiters go the data being created is probably going to be audited for the selected delimiter of the chosen character set, and it is going to have a representable substitute when it is used inside the data.
    – arch-abit
    Commented Aug 22, 2014 at 18:48
  • 1
    Oh cool. Then it's easy. Yeah - you can just IFS. It can be switched around or whatever - I'm doing an answer if that's ok.
    – mikeserv
    Commented Aug 22, 2014 at 18:49
  • @mikeserv - Yes, thanks, any suggestions, the more ways the better :) I still have more to add to this project, so flexibility is good.
    – arch-abit
    Commented Aug 22, 2014 at 18:51

1 Answer 1

5

The shell has a built-in variable expansion field separator. So if you have a string and your delimiter is solid you can do:

var=32768,'dff0207a-591f-4435-9f8b-7b9b3e6ba2c1','d1f77359b3f7236806489ba3108c771f','NUMBER','US_EN','LATIN','GREEK','GERMAN'
(   IFS=,; set -f
    for field in $var
    do  printf '\n%s\n\t' "$field - md5:" >&2
        printf %s "$field" |
        md5sum
    done |
    cut -d\  -f1
)

32768 - md5:
        f43764367fa4b73ba947fae71b0223a4

dff0207a-591f-4435-9f8b-7b9b3e6ba2c1 - md5:
        0983e6c45209f390461c1b1df9320674

d1f77359b3f7236806489ba3108c771f - md5:
        07d82ab57ba81f991ab996bd7c5a0441

NUMBER - md5:
        34f55eca38e0605a84f169ff61a2a396

US_EN - md5:
        c9d3e580b7b102e864d9aea8703486ab

LATIN - md5:
        0e869135050d24ea6e7a30fc6edbac6c

GREEK - md5:
        d4cacc28e56302bcec9d7af4bba8c9a7

GERMAN - md5:
        ed73cca110623766d7a2457331a4f373

That should give you a newline separated list of md5s - as it did me.

IFS=, is used to specify that when any variable type shell expansion is performed the shell should split it out on the , character rather than <space><newline><tab> - which is the default. set -f is used to specify that if the shell should encounter any file globs within an unquoted expansion it should not expand them - so echo * would print only * regardless of the contents of the current directory.

For every comma separated field in $var the shell does printf "$field" | md5sum - so once per field without separator strings as I take the question to mean. And last cut trims the few spaces and the - at the end of each output line as it receives them. Most of the output is actually to stderr - cut only ever sees the md5sums.

10
  • 1
    @arch-abit : am I understanding this correctly? Am I splitting the right stuff?
    – mikeserv
    Commented Aug 22, 2014 at 19:00
  • 1
    @arch-abit : Ok. maybe I explained it better now?
    – mikeserv
    Commented Aug 22, 2014 at 19:22
  • 1
    I had to go to work, but I did some reading and the script works pretty good.. I updated my post. Thanks for your help!
    – arch-abit
    Commented Aug 23, 2014 at 8:32
  • 1
    @arch-abit as youve posted them your scripts will break on the shell globs. you need set -f to ensure you dont expand filenames into your unquoted expansions.
    – mikeserv
    Commented Aug 23, 2014 at 9:05
  • 1
    it's a simple thing. its just in case one of those vars contains the string * which the shell will then fill with all of the files in the current directory. you turnit back on with set +f.
    – mikeserv
    Commented Aug 23, 2014 at 9:23

This site is temporarily in read-only mode and not accepting new answers.

Not the answer you're looking for? Browse other questions tagged .