Parsing an Oracle connect string

[fbicknel]

Fun stuff.

Say you have an oracle connect string that looks like:

['] [userid] [/ [password]] [@net_service_name] [']

Now say you want to parse that. I came up with this regex that seems to do it, at least for some of the simple tests below.

I humbly ask your opinions:

my ($userid, $password, $netServiceName) = /'?([^@\/']+)?(\/[^@\/']*)?
+(@[^@\/']+)?'?/;

Samples.

>>string<< is parsed as <userid> </password> <@net_service_name>

Note, btw, the special case of '/' which needs to parse as simply '/'.

>>SYS/change_on_install@prod1<<
<SYS> </change_on_install> <@prod1>
>>rman/rman@rcat<<
<rman> </rman> <@rcat>
>>SYS<<
<SYS> <> <>
>>/<<
<> </> <>
>>/fooey<<
<> </fooey> <>
>>/fooey@foobar<<
<> </fooey> <@foobar>
>>SYS@foobar<<
<SYS> <> <@foobar>
>>'simple/fooey@foobar'<<
<simple> </fooey> <@foobar>

[a]

Hmm, not really clear on what you've got here but:

while (<DATA>) {
    chomp;
    my ($userid, $password, $net_service_name) 
             = m#'?([^@/']+)?(/[^@/']+)?(@[^@/']+)?'?#;
    print "]]$_\[\[ 
  \tis parsed as \[$userid] \[$password] [$net_service_name]\n"
}   # while 

__DATA__
>>SYS/change_on_install@prod1<< <SYS> 
</change_on_install> <@prod1> 
>>rman/rman@rcat<< 
<rman> </rman> <@rcat> 
>>SYS<< <SYS> <> <> >>/<< <> </> <> 
>>/fooey<< <> </fooey> <> 
>>/fooey@foobar<< 
<> </fooey> <@foobar> 
>>SYS@foobar<< 
<SYS> <> <@foobar> 
>>'simple/fooey@foobar'<< 
<simple> </fooey> <@foobar>

I get a different result, seems you might want to consider white space, for one thing. That's not allowed, right? Are the pointy brackets legit Oracle chars?

[fbicknel]

You're right: whitespace wasn't considered, I'll check to see if it's allowed. For now it's not. (edit: I checked.... it's allowed. I need to modify my r.e. to allow them then. Ick: quotes.) As to the pointy stuff... it's just a way to show each of the fields one by one. Sorry for the confusion.

Ok, so I diddled a bit with this and came up with the following. It allows spaces in passwords (that's allowed, but not the other two fields.) It also allows spaces within the string, so long as they are next the delimiters, '/' and '@'. First, here's the code fragment:

 10     $string =~ s/ *([@\/]) */$1/g;
 11     my ($userid, $diag, $password, $netServiceName) = $string =~ /
+'?([^@\/']+)?(?:(\/)'?([^@\/']*)'?)?(@[^@\/']+)?'?/;

Since I added the ability to surround the password in 'quotes', it unfortunately removed that bit from the '/' discovery. So that's now captured in $diag.

Also (line 10) removes any optional spaces near the delimiters,

@\/

. This seems to parse ok in my test program:

SYS / foo_pw @ stuff
SYS / ' foo_pw1' @ stuff
SYS /' foo_pw2' @ stuff
SYS / foo sp ace @ stuff
SYS/change_on_install@prod1
LAST
rman/rman@rcat
SYS
/
/fooey
/fooey@foobar
SYS@foobar
'simple/fooey@foobar'

Parses as:

<     SYS> <     /foo_pw> [  @stuff]
<     SYS> <   / foo_pw1> [  @stuff]
<     SYS> <   / foo_pw2> [  @stuff]
<     SYS> < /foo sp ace> [  @stuff]
<     SYS> </change_on_i> [  @prod1]
<    LAST> <            > [        ]
<    rman> <       /rman> [   @rcat]
<     SYS> <            > [        ]
<        > <           /> [        ]
<        > <      /fooey> [        ]
<        > <      /fooey> [ @foobar]
<     SYS> <            > [ @foobar]
<  simple> <      /fooey> [ @foobar]

(Note that my test program truncates some longer passwords in the printout.)