Added options for DBI parser.
[dbsrgits/SQL-Translator.git] / lib / SQL / Translator / XMI / Parser.pm
CommitLineData
f42065cb 1package SQL::Translator::XMI::Parser;
2
93f4a354 3# -------------------------------------------------------------------
42b5b9b6 4# $Id: Parser.pm,v 1.6 2003-10-01 17:45:47 grommit Exp $
93f4a354 5# -------------------------------------------------------------------
6# Copyright (C) 2003 Mark Addison <mark.addison@itn.co.uk>,
7#
8# This program is free software; you can redistribute it and/or
9# modify it under the terms of the GNU General Public License as
10# published by the Free Software Foundation; version 2.
11#
12# This program is distributed in the hope that it will be useful, but
13# WITHOUT ANY WARRANTY; without even the implied warranty of
14# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15# General Public License for more details.
16#
17# You should have received a copy of the GNU General Public License
18# along with this program; if not, write to the Free Software
19# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
20# 02111-1307 USA
21# -------------------------------------------------------------------
22
f42065cb 23=pod
24
25=head1 NAME
26
93f4a354 27SQL::Translator::XMI::Parser - XMI Parser class for use in SQL Fairy's XMI
28parser.
f42065cb 29
30=cut
31
32use strict;
33use 5.006_001;
93f4a354 34use vars qw/$VERSION/;
42b5b9b6 35$VERSION = sprintf "%d.%02d", q$Revision: 1.6 $ =~ /(\d+)\.(\d+)/;
f42065cb 36
93f4a354 37use Data::Dumper;
f42065cb 38use XML::XPath;
39use XML::XPath::XMLParser;
40use Storable qw/dclone/;
41
42# Spec
93f4a354 43#------
44# See SQL::Translator::XMI::Parser::V12 and SQL::Translator::XMI::Parser:V10
45# for examples.
f42065cb 46#
93f4a354 47# Hash ref used to describe the 2 xmi formats 1.2 and 1.0. Neither is complete!
f42065cb 48#
49# NB The names of the data keys MUST be the same for both specs so the
50# data structures returned are the same.
51#
93f4a354 52# TODO
53#
54# * There is currently no way to set the data key name for attrib_data, it just
f42065cb 55# uses the attribute name from the XMI. This isn't a problem at the moment as
56# xmi1.0 names all these things with tags so we don't need the attrib data!
57# Also use of names seems to be consistant between the versions.
58#
f42065cb 59#
93f4a354 60# XmiSpec( $spec )
f42065cb 61#
93f4a354 62# Call as class method to set up the parser from a spec (see above). This
63# generates the get_ methods for the version of XMI the spec is for. Called by
64# the sub-classes (e.g. V12 and V10) to create parsers for each version.
f42065cb 65#
93f4a354 66sub XmiSpec {
67 my ($me,$spec) = @_;
42b5b9b6 68 _init_specs($spec);
69 $me->_mk_gets($spec);
93f4a354 70}
f42065cb 71
72# Build lookups etc. Its important that each spec item becomes self contained
b4b9f867 73# so we can build good closures, therefore we do all the lookups 1st.
42b5b9b6 74sub _init_specs {
f42065cb 75 my $specs = shift;
76
77 foreach my $spec ( values %$specs ) {
b4b9f867 78 # Look up for kids get method
79 foreach ( @{$spec->{kids}} ) {
f42065cb 80 $_->{get_method} = "get_".$specs->{$_->{class}}{plural};
81 }
0b3f94e0 82
b4b9f867 83 # Add xmi.id ti all specs. Everything we want at the moment (in both
84 # versions) has an id. The tags that don't seem to be used for
85 # structure.
86 my $attrib_data = $spec->{attrib_data} ||= [];
87 push @$attrib_data, "xmi.id";
f42065cb 88 }
89
90}
91
93f4a354 92# Create get methods from spec
93#
42b5b9b6 94sub _mk_gets {
93f4a354 95 my ($proto,$specs) = @_;
96 my $class = ref($proto) || $proto;
97 foreach ( values %$specs ) {
98 # Clone from specs and sort out the lookups into it so we get a
99 # self contained spec to use as a proper closure.
100 my $spec = dclone($_);
101
102 # Create _get_* method with get_* as an alias unless the user has
103 # defined it. Allows for override. Note the alias is in this package
104 # so we can add overrides to both specs.
105 no strict "refs";
106 my $meth = "_get_$spec->{plural}";
42b5b9b6 107 *{$meth} = _mk_get($spec);
93f4a354 108 *{__PACKAGE__."::get_$spec->{plural}"} = sub {shift->$meth(@_);}
109 unless $class->can("get_$spec->{plural}");
f42065cb 110 }
111}
112
93f4a354 113#
114# Sets up the XML::XPath object and then checks the version of the XMI file and
115# blesses its self into either the V10 or V12 class.
116#
f42065cb 117sub new {
118 my $proto = shift;
119 my $class = ref($proto) || $proto;
120 my %args = @_;
121 my $me = {};
93f4a354 122
f42065cb 123 # Create the XML::XPath object
124 # TODO Docs recommend we only use 1 XPath object per application
125 my $xp;
126 foreach (qw/filename xml ioref/) {
127 if ($args{$_}) {
128 $xp = XML::XPath->new( $_ => $args{$_});
129 $xp->set_namespace("UML", "org.omg.xmi.namespace.UML");
130 last;
131 }
132 }
133 $me = { xml_xpath => $xp };
93f4a354 134
135 # Work out the version of XMI we have and return as that sub class
136 my $xmiv = $args{xmi_version}
b4b9f867 137 || "".$xp->findvalue('/XMI/@xmi.version')
138 || die "Can't find XMI version";
93f4a354 139 $xmiv =~ s/[.]//g;
140 $class = __PACKAGE__."::V$xmiv";
141 eval "use $class;";
142 die "Failed to load version sub class $class : $@" if $@;
f42065cb 143
93f4a354 144 return bless $me, $class;
f42065cb 145}
146
93f4a354 147#
42b5b9b6 148# _mk_get
f42065cb 149#
93f4a354 150# Generates and returns a get_ sub for the spec given.
151# So, if you want to change how the get methods (e.g. get_classes) work do it
152# here!
f42065cb 153#
154# The get methods made have the args described in the docs and 2 private args
155# used internally, to call other get methods from paths in the spec.
f42065cb 156# NB: DO NOT use publicly as you will break the version independance. e.g. When
157# using _xpath you need to know which version of XMI to use. This is handled by
158# the use of different paths in the specs.
b2a00f50 159#
f42065cb 160# _context => The context node to use, if not given starts from root.
b2a00f50 161#
f42065cb 162# _xpath => The xpath to use for finding stuff.
b2a00f50 163#
42b5b9b6 164sub _mk_get {
f42065cb 165 my $spec = shift;
b2a00f50 166
f42065cb 167 # get_* closure using $spec
168 return sub {
169 my ($me, %args) = @_;
170 my $xp = delete $args{_context} || $me->{xml_xpath};
171 my $things;
172
173 my $xpath = $args{_xpath} ||= $spec->{default_path};
174#warn "Searching for $spec->{plural} using:$xpath\n";
175
176 my @nodes = $xp->findnodes($xpath);
0b3f94e0 177#warn "None.\n" unless @nodes;
f42065cb 178 return unless @nodes;
179
180 for my $node (@nodes) {
0b3f94e0 181#warn " Found $spec->{name} xmi.id=".$node->getAttribute("xmi.id")." name=".$node->getAttribute("name")."\n";
f42065cb 182 my $thing = {};
183 # my $thing = { xpNode => $node };
b2a00f50 184
0b3f94e0 185 # Have we seen this before? If so just use the ref we have.
186 if ( my $id = $node->getAttribute("xmi.id") ) {
187 if ( my $foo = $me->{model}{things}{$id} ) {
188#warn " Reffing from model **********************\n";
189 push @$things, $foo;
190 next;
191 }
192 }
193
f42065cb 194 # Get the Tag attributes
195 foreach ( @{$spec->{attrib_data}} ) {
196 $thing->{$_} = $node->getAttribute($_);
197 }
b2a00f50 198
f42065cb 199 # Add the path data
200 foreach ( @{$spec->{path_data}} ) {
0b3f94e0 201#warn " $spec->{name} - $_->{name} using:$_->{path}\n";
f42065cb 202 my @nodes = $node->findnodes($_->{path});
203 $thing->{$_->{name}} = @nodes ? $nodes[0]->getData
204 : (exists $_->{default} ? $_->{default} : undef);
205 }
b2a00f50 206
207 # Run any filters set
208 #
f42065cb 209 # Should we do this after the kids as we may want to test them?
210 # e.g. test for number of attribs
211 if ( my $filter = $args{filter} ) {
212 local $_ = $thing;
213 next unless $filter->($thing);
214 }
b2a00f50 215
0b3f94e0 216 # Add anything with an id to the things lookup
217 push @$things, $thing;
218 if ( exists $thing->{"xmi.id"} and defined $thing->{"xmi.id"}
219 and my $id = $thing->{"xmi.id"}
220 ) {
221 $me->{model}{things}{$id} = $thing; }
222
f42065cb 223 # Kids
224 #
225 foreach ( @{$spec->{kids}} ) {
b2a00f50 226 my $data;
f42065cb 227 my $meth = $_->{get_method};
b2a00f50 228 my $path = $_->{path};
0b3f94e0 229
230 # Variable subs on the path from thing
231 $path =~ s/\$\{(.*?)\}/$thing->{$1}/g;
b2a00f50 232 $data = $me->$meth( _context => $node, _xpath => $path,
f42065cb 233 filter => $args{"filter_$_->{name}"} );
b2a00f50 234
f42065cb 235 if ( $_->{multiplicity} eq "1" ) {
236 $thing->{$_->{name}} = shift @$data;
237 }
238 else {
b2a00f50 239 my $kids = $thing->{$_->{name}} = $data || [];
240 if ( my $key = $_->{"map"} ) {
241 $thing->{"_map_$_->{name}"} = _mk_map($kids,$key);
242 }
f42065cb 243 }
244 }
0b3f94e0 245 }
f42065cb 246
0b3f94e0 247 if ( $spec->{isRoot} ) {
248 push(@{$me->{model}{$spec->{plural}}}, $_) foreach @$things;
f42065cb 249 }
0b3f94e0 250 return $things;
f42065cb 251} # /closure sub
252
42b5b9b6 253} # /_mk_get
f42065cb 254
b2a00f50 255sub _mk_map {
256 my ($kids,$key) = @_;
257 my $map = {};
258 foreach (@$kids) {
259 $map->{$_->{$key}} = $_ if exists $_->{$key};
260 }
261 return $map;
262}
263
42b5b9b6 264sub get_associations {
265 my $assoc = shift->_get_associations(@_);
266 foreach (@$assoc) {
267 next unless defined $_->{ends}; # Wait until we get all of an association
268 my @ends = @{$_->{ends}};
269 if (@ends != 2) {
270 warn "Sorry can't handle otherEnd associations with more than 2 ends";
271 return $assoc;
272 }
273 $ends[0]{otherEnd} = $ends[1];
274 $ends[1]{otherEnd} = $ends[0];
275 }
276 return $assoc;
277}
278
f42065cb 2791; #===========================================================================
280
281
282package XML::XPath::Function;
283
284#
285# May need to look at doing deref on all paths just to be on the safe side!
286#
287# Will also want some caching as these calls are expensive as the whole doc
288# is used but the same ref will likley be requested lots of times.
289#
290sub xmiDeref {
291 my $self = shift;
292 my ($node, @params) = @_;
293 if (@params > 1) {
294 die "xmiDeref() function takes one or no parameters\n";
295 }
296 elsif (@params) {
297 my $nodeset = shift(@params);
298 return $nodeset unless $nodeset->size;
299 $node = $nodeset->get_node(1);
300 }
301 die "xmiDeref() needs an Element node."
302 unless $node->isa("XML::XPath::Node::Element");
303
304 my $id = $node->getAttribute("xmi.idref") or return $node;
305 return $node->getRootNode->find('//*[@xmi.id="'.$id.'"]');
306}
307
308
309# compile please
3101;
311
312__END__
313
314=head1 SYNOPSIS
315
316 use SQL::Translator::XMI::Parser;
317 my $xmip = SQL::Translator::XMI::Parser->new( xml => $xml );
318 my $classes = $xmip->get_classes();
319
320=head1 DESCRIPTION
321
322Parses XMI files (XML version of UML diagrams) to perl data structures and
323provides hooks to filter the data down to what you want.
324
325=head2 new
326
b4b9f867 327Pass in name/value arg of either C<filename>, C<xml> or C<ioref> for the XMI
328data you want to parse.
329
330The version of XMI to use either 1.0 or 1.2 is worked out from the file. You
331can also use a C<xmi_version> arg to set it explicitley.
f42065cb 332
333=head2 get_* methods
334
335Doc below is for classes method, all the other calls follow this form.
336
337=head2 get_classes( ARGS )
338
339 ARGS - Name/Value list of args.
340
341 filter => A sub to filter the node to see if we want it. Has the nodes data,
342 before kids are added, referenced to $_. Should return true if you
343 want it, false otherwise.
344
345 e.g. To find only classes with a "Foo" stereotype.
346
347 filter => sub { return $_->{stereotype} eq "Foo"; }
348
349 filter_attributes => A filter sub to pass onto get_attributes.
350
351 filter_operations => A filter sub to pass onto get_operations.
352
353Returns a perl data structure including all the kids. e.g.
354
355 {
356 'name' => 'Foo',
357 'visibility' => 'public',
358 'isActive' => 'false',
359 'isAbstract' => 'false',
360 'isSpecification' => 'false',
361 'stereotype' => 'Table',
362 'isRoot' => 'false',
363 'isLeaf' => 'false',
364 'attributes' => [
365 {
366 'name' => 'fooid',
367 'stereotype' => 'PK',
368 'datatype' => 'int'
369 'ownerScope' => 'instance',
370 'visibility' => 'public',
371 'initialValue' => undef,
372 'isSpecification' => 'false',
373 },
374 {
375 'name' => 'name',
376 'stereotype' => '',
377 'datatype' => 'varchar'
378 'ownerScope' => 'instance',
379 'visibility' => 'public',
380 'initialValue' => '',
381 'isSpecification' => 'false',
382 },
383 ]
384 'operations' => [
385 {
386 'name' => 'magic',
387 'isQuery' => 'false',
388 'ownerScope' => 'instance',
389 'visibility' => 'public',
390 'isSpecification' => 'false',
391 'stereotype' => '',
392 'isAbstract' => 'false',
393 'isLeaf' => 'false',
394 'isRoot' => 'false',
395 'concurrency' => 'sequential'
396 'parameters' => [
397 {
398 'kind' => 'inout',
399 'isSpecification' => 'false',
400 'stereotype' => '',
401 'name' => 'arg1',
402 'datatype' => undef
403 },
404 {
405 'kind' => 'inout',
406 'isSpecification' => 'false',
407 'stereotype' => '',
408 'name' => 'arg2',
409 'datatype' => undef
410 },
411 {
412 'kind' => 'return',
413 'isSpecification' => 'false',
414 'stereotype' => '',
415 'name' => 'return',
416 'datatype' => undef
417 }
418 ],
419 }
420 ],
421 }
422
423=head1 XMI XPath Functions
424
93f4a354 425The Parser adds the following extra XPath functions for use in the Specs.
f42065cb 426
427=head2 xmiDeref
428
429Deals with xmi.id/xmi.idref pairs of attributes. You give it an
430xPath e.g 'UML:ModelElement.stereotype/UML:stereotype' if the the
431tag it points at has an xmi.idref it looks up the tag with that
432xmi.id and returns it.
433
434If it doesn't have an xmi.id, the path is returned as normal.
435
436e.g. given
437
438 <UML:ModelElement.stereotype>
439 <UML:Stereotype xmi.idref = 'stTable'/>
440 </UML:ModelElement.stereotype>
441 ...
442 <UML:Stereotype xmi.id='stTable' name='Table' visibility='public'
443 isAbstract='false' isSpecification='false' isRoot='false' isLeaf='false'>
444 <UML:Stereotype.baseClass>Class</UML:Stereotype.baseClass>
445 </UML:Stereotype>
446
447Using xmideref(//UML:ModelElement.stereotype/UML:stereotype) would return the
448<UML:Stereotype xmi.id = '3b4b1e:f762a35f6b:-7fb6' ...> tag.
449
450Using xmideref(//UML:ModelElement.stereotype/UML:stereotype)/@name would give
451"Table".
452
453=head1 SEE ALSO
454
455perl(1).
456
457=head1 TODO
458
459=head1 BUGS
460
461=head1 VERSION HISTORY
462
463=head1 AUTHOR
464
465grommit <mark.addison@itn.co.uk>
466
f42065cb 467=cut