Zeev,
It is not a hasard that I provided the XML stuff. Searching in XML using
XPath and nagivating in an XML is far easier that with Perl's blessed
structures.
- printout the ast using
MarpaX::Languages::SQL2003::AST->new()->asXML($input)->toString(1) instead
of trying to digest the grammar
- you will see the hierarchy of what you are looking for
- use XPath
Example in attachement on how to find the table names.
If you insist on the blessed mode, perl people would suggest modules like
Data::Find or Data::Walk, I'd personnaly revisit
MarpaX::Languages::C::AST::Util::Data::Find
<https://github.com/jddurand/MarpaX-Languages-C-AST/blob/master/lib/MarpaX/Languages/C/AST/Util/Data/Find.pm>,
taking into account that with MarpaX::Languages::SQL2003::AST, a lexeme is
a blessed hash that is guaranteed to have keys "start", "length", "text"
and "value", when an lhs is blessed array.
Jean-Damien.
Le jeudi 18 décembre 2014 16:11:37 UTC+1, Zeev Atlas a écrit :
>
> Thank you again for advice and support
> Is there any 'accessor' or a template for such an accessor for the blessed
> structure that is available anywhere in the Marpa world? (before I code
> such an accessor myself and risk being incompatible with next version.)
> ZA
>
> On Tuesday, December 16, 2014 12:39:00 PM UTC-5, Durand Jean-Damien wrote:
>
>> Ok, so I do nothing / JD.
>>
>> Le mardi 16 décembre 2014 07:16:38 UTC+1, [email protected] a écrit :
>>>
>>> Thank you... regardless, I tried your suggestion and will move to
>>> strawberry anyway, since I can install XML::LibXML and I guess anything else
>>> In any case, for my purpose, the AST is all what I need. I intend to
>>> write an application that analyze SQL statements and deduce source tables
>>> and where/join-on clauses, basically find the flow of data and put it in
>>> csv/excel format.
>>> ZA
>>>
>>> On Monday, December 15, 2014 11:30:00 PM UTC-5, Durand Jean-Damien wrote:
>>>>
>>>> Yes, this is easy... will keep you informed
>>>>
>>>> Le mardi 16 décembre 2014 04:46:43 UTC+1, Zeev Atlas a écrit :
>>>>>
>>>>> XML::LibXML is not available (yet) for ActiveState Perl 5.16 and above
>>>>> and I use 5.18. Is it possible to create a package that produces AST
>>>>> version only and does not have 'use XML::LibXML;' at all. I do not have
>>>>> too much experience with Linux (barely have one machine with Ubuntu). I
>>>>> will go for strawberry only if I have no other choice (because it
>>>>> resembles
>>>>> the Linux environment)
>>>>> Thanks
>>>>> ZA
>>>>>
>>>>> On Sunday, December 14, 2014 10:48:22 AM UTC-5, Durand Jean-Damien
>>>>> wrote:
>>>>>>
>>>>>> ActiveState will certainly work on that, I bet users of this
>>>>>> dsitribution can't stand without XML::LibXML for a while, usual bribes
>>>>>> or
>>>>>> whatever PPDs can perhaps help, although I guess you (perhaps) already
>>>>>> tried.
>>>>>> I personnaly stopped using this distribution in favour of strawberry,
>>>>>> that has both the advantage and drawback that it is not using nmake/cl.
>>>>>>
>>>>>> http://code.activestate.com/ppm/XML-LibXML/
>>>>>>
>>>>>> JD.
>>>>>>
>>>>>> Le dimanche 14 décembre 2014 02:00:20 UTC+1, Zeev Atlas a écrit :
>>>>>>>
>>>>>>> Thanks again... I work on a Windows platform (active state Perl) and
>>>>>>> there is no XML::LibXML. I'll have to make some adjustments in the
>>>>>>> next
>>>>>>> few days or just look at the module and do my own.
>>>>>>>
>>>>>>> On Saturday, December 13, 2014 9:40:54 AM UTC-5, Durand Jean-Damien
>>>>>>> wrote:
>>>>>>>>
>>>>>>>> Zeev,
>>>>>>>>
>>>>>>>> Instead I invite you to try the sql2003toxml binary distributed
>>>>>>>> with MarpaX::Languages::SQL2003::AST, c.f. my g+ post
>>>>>>>> <https://plus.google.com/105243759955561553019/posts/Mn7DQSJuKc8>.
>>>>>>>>
>>>>>>>> Jean-Damien.
>>>>>>>>
>>>>>>>> Le vendredi 12 décembre 2014 14:10:07 UTC+1, Zeev Atlas a écrit :
>>>>>>>>>
>>>>>>>>> COuld you please migrate the POC as well
>>>>>>>>> ZA
>>>>>>>>>
>>>>>>>>>
--
You received this message because you are subscribed to the Google Groups
"marpa parser" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
#!env perl
use strict;
use diagnostics;
use MarpaX::Languages::SQL2003::AST;
use Data::Dumper;
my $input = do {local $/; <DATA>};
my %tables = ();
my $ast = MarpaX::Languages::SQL2003::AST->new();
my $xml = eval {$ast->asXML($input)};
print STDERR $xml->toString(1) . "\n";
if ($@) {
print STDERR "Failed to parse $input: $@";
} else {
#
# Get table.column information
#
our $xpath = XML::LibXML::XPathExpression->new("//Table_Primary");
#
# Find the table names and their eventual aliases
#
foreach ($xml->findnodes($xpath)) {
my $name = $_
->firstChild() # Table_Or_Query_Name
->firstChild() # Table_Name
-> firstChild() # Local_Or_Schema_Qualified_Name
->getAttribute('value');
my $alias = $_
->firstChild() # Table_Or_Query_Name
->nextSibling() # Gen1649_Maybe
->firstChild();
if (defined($alias)) {
$alias = $alias->firstChild() # As_Maybe
->nextSibling() # Correlation_Name
->firstChild() # Identifier
->getAttribute('value');
}
$alias ||= $name;
print STDOUT "Table alias $alias <=> name $name\n";
$tables{$alias} = $name;
}
}
print Dumper(\%tables);
__DATA__
select t.i, t.j, u.k, v.m, v.n, w.o from myTableT as t, myTableU as u, myTableV as v, w
