1 | ###########################################################################
|
---|
2 | #
|
---|
3 | # AZCompactList.pm --
|
---|
4 | # A component of the Greenstone digital library software
|
---|
5 | # from the New Zealand Digital Library Project at the
|
---|
6 | # University of Waikato, New Zealand.
|
---|
7 | #
|
---|
8 | # Copyright (C) 1999 New Zealand Digital Library Project
|
---|
9 | #
|
---|
10 | # This program is free software; you can redistribute it and/or modify
|
---|
11 | # it under the terms of the GNU General Public License as published by
|
---|
12 | # the Free Software Foundation; either version 2 of the License, or
|
---|
13 | # (at your option) any later version.
|
---|
14 | #
|
---|
15 | # This program is distributed in the hope that it will be useful,
|
---|
16 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
17 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
18 | # GNU General Public License for more details.
|
---|
19 | #
|
---|
20 | # You should have received a copy of the GNU General Public License
|
---|
21 | # along with this program; if not, write to the Free Software
|
---|
22 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
23 | #
|
---|
24 | ###########################################################################
|
---|
25 | # classifier plugin for sorting alphabetically
|
---|
26 | # options are:
|
---|
27 |
|
---|
28 | package AZCompactList;
|
---|
29 |
|
---|
30 | use BasClas;
|
---|
31 | use sorttools;
|
---|
32 |
|
---|
33 | use strict;
|
---|
34 | no strict 'refs'; # allow filehandles to be variables and viceversa
|
---|
35 |
|
---|
36 | sub BEGIN {
|
---|
37 | @AZCompactList::ISA = ('BasClas');
|
---|
38 | }
|
---|
39 |
|
---|
40 | my $doclevel_list =
|
---|
41 | [ { 'name' => "top",
|
---|
42 | 'desc' => "{AZCompactList.doclevel.top}" },
|
---|
43 | { 'name' => "section",
|
---|
44 | 'desc' => "{AZCompactList.doclevel.section}" } ];
|
---|
45 |
|
---|
46 | my $arguments =
|
---|
47 | [ { 'name' => "metadata",
|
---|
48 | 'desc' => "{AZCompactList.metadata}",
|
---|
49 | 'type' => "metadata",
|
---|
50 | 'reqd' => "yes" },
|
---|
51 | { 'name' => "firstvalueonly",
|
---|
52 | 'desc' => "{AZCompactList.firstvalueonly}",
|
---|
53 | 'type' => "flag",
|
---|
54 | 'reqd' => "no" },
|
---|
55 | { 'name' => "allvalues",
|
---|
56 | 'desc' => "{AZCompactList.allvalues}",
|
---|
57 | 'type' => "flag",
|
---|
58 | 'reqd' => "no" },
|
---|
59 | { 'name' => "buttonname",
|
---|
60 | 'desc' => "{BasClas.buttonname}",
|
---|
61 | 'type' => "string",
|
---|
62 | 'deft' => "",
|
---|
63 | 'reqd' => "no" },
|
---|
64 | { 'name' => "sort",
|
---|
65 | 'desc' => "{AZCompactList.sort}",
|
---|
66 | 'type' => "string",
|
---|
67 | # 'deft' => "Title",
|
---|
68 | 'reqd' => "no" },
|
---|
69 | { 'name' => "removeprefix",
|
---|
70 | 'desc' => "{BasClas.removeprefix}",
|
---|
71 | 'type' => "regexp",
|
---|
72 | 'deft' => "",
|
---|
73 | 'reqd' => "no" },
|
---|
74 | { 'name' => "removesuffix",
|
---|
75 | 'desc' => "{BasClas.removesuffix}",
|
---|
76 | 'type' => "regexp",
|
---|
77 | 'deft' => "",
|
---|
78 | 'reqd' => "no" },
|
---|
79 | { 'name' => "mingroup",
|
---|
80 | 'desc' => "{AZCompactList.mingroup}",
|
---|
81 | 'type' => "int",
|
---|
82 | 'deft' => "2",
|
---|
83 | 'range' => "1,",
|
---|
84 | 'reqd' => "no" },
|
---|
85 | { 'name' => "minnesting",
|
---|
86 | 'desc' => "{AZCompactList.minnesting}",
|
---|
87 | 'type' => "int",
|
---|
88 | 'deft' => "20",
|
---|
89 | 'range' => "2,",
|
---|
90 | 'reqd' => "no" },
|
---|
91 | { 'name' => "mincompact",
|
---|
92 | 'desc' => "{AZCompactList.mincompact}",
|
---|
93 | 'type' => "int",
|
---|
94 | 'deft' => "10",
|
---|
95 | 'range' => "1,",
|
---|
96 | 'reqd' => "no" },
|
---|
97 | { 'name' => "maxcompact",
|
---|
98 | 'desc' => "{AZCompactList.maxcompact}",
|
---|
99 | 'type' => "int",
|
---|
100 | 'deft' => "30",
|
---|
101 | 'range' => "1,",
|
---|
102 | 'reqd' => "no" },
|
---|
103 | { 'name' => "doclevel",
|
---|
104 | 'desc' => "{AZCompactList.doclevel}",
|
---|
105 | 'type' => "enum",
|
---|
106 | 'list' => $doclevel_list,
|
---|
107 | 'deft' => "top",
|
---|
108 | 'reqd' => "no" },
|
---|
109 | { 'name' => "freqsort",
|
---|
110 | 'desc' => "{AZCompactList.freqsort}",
|
---|
111 | 'type' => "flag"},
|
---|
112 | { 'name' => "recopt",
|
---|
113 | 'desc' => "{AZCompactList.recopt}",
|
---|
114 | 'type' => "flag",
|
---|
115 | 'reqd' => "no" } ];
|
---|
116 |
|
---|
117 | my $options =
|
---|
118 | { 'name' => "AZCompactList",
|
---|
119 | 'desc' => "{AZCompactList.desc}",
|
---|
120 | 'abstract' => "no",
|
---|
121 | 'inherits' => "yes",
|
---|
122 | 'args' => $arguments };
|
---|
123 |
|
---|
124 |
|
---|
125 | sub new {
|
---|
126 | my ($class) = shift (@_);
|
---|
127 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
---|
128 | push(@$pluginlist, $class);
|
---|
129 |
|
---|
130 | if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
|
---|
131 | if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
|
---|
132 |
|
---|
133 | my $self = (defined $hashArgOptLists)? new BasClas($pluginlist,$inputargs,$hashArgOptLists): new BasClas($pluginlist,$inputargs);
|
---|
134 |
|
---|
135 | if ($self->{'info_only'}) {
|
---|
136 | # don't worry about any options etc
|
---|
137 | return bless $self, $class;
|
---|
138 | }
|
---|
139 |
|
---|
140 | if (!$self->{"metadata"}) {
|
---|
141 | my $outhandle = $self->{'outhandle'};
|
---|
142 | print $outhandle "AZCompactList Error: required option -metadata not supplied\n";
|
---|
143 | $self->print_txt_usage("");
|
---|
144 | die "AZCompactList Error: required option -metadata not supplied\n";
|
---|
145 | }
|
---|
146 |
|
---|
147 | # Manually set $self parameters.
|
---|
148 | $self->{'list'} = {};
|
---|
149 | $self->{'listmetavalue'} = {};
|
---|
150 | $self->{'list_mvpair'} = {};
|
---|
151 | $self->{'reclassify'} = {};
|
---|
152 | $self->{'reclassifylist'} = {};
|
---|
153 |
|
---|
154 | $self->{'buttonname'} = $self->generate_title_from_metadata($self->{'metadata'}) unless ($self->{'buttonname'});
|
---|
155 |
|
---|
156 | if (defined($self->{"removeprefix"}) && $self->{"removeprefix"}) {
|
---|
157 | $self->{"removeprefix"} =~ s/^\^//; # don't need a leading ^
|
---|
158 | }
|
---|
159 | if (defined($self->{"removesuffix"}) && $self->{"removesuffix"}) {
|
---|
160 | $self->{"removesuffix"} =~ s/\$$//; # don't need a trailing $
|
---|
161 | }
|
---|
162 |
|
---|
163 | $self->{'recopt'} = ($self->{'recopt'} == 0) ? undef : "on";
|
---|
164 |
|
---|
165 | # Clean out the unused keys
|
---|
166 | if($self->{"removeprefix"} eq "") {delete $self->{"removeprefix"};}
|
---|
167 | if($self->{"removesuffix"} eq "") {delete $self->{"removesuffix"};}
|
---|
168 |
|
---|
169 | return bless $self, $class;
|
---|
170 | }
|
---|
171 |
|
---|
172 | sub init
|
---|
173 | {
|
---|
174 | my $self = shift (@_);
|
---|
175 |
|
---|
176 | $self->{'list'} = {};
|
---|
177 | $self->{'listmetavalue'} = {};
|
---|
178 | $self->{'list_mvpair'} = {};
|
---|
179 | $self->{'reclassify'} = {};
|
---|
180 | $self->{'reclassifylist'} = {};
|
---|
181 | }
|
---|
182 |
|
---|
183 | my $tmp = 0;
|
---|
184 |
|
---|
185 | sub classify
|
---|
186 | {
|
---|
187 | my $self = shift (@_);
|
---|
188 | my ($doc_obj) = @_;
|
---|
189 |
|
---|
190 | my $doc_OID = $doc_obj->get_OID();
|
---|
191 |
|
---|
192 | my @sectionlist = ();
|
---|
193 | my $topsection = $doc_obj->get_top_section();
|
---|
194 | my $metaname = $self->{'metadata'};
|
---|
195 | my $outhandle = $self->{'outhandle'};
|
---|
196 |
|
---|
197 | $metaname =~ s/(\/|\|).*//; # grab first name in n1/n2/n3 or n1|n2|n3 list
|
---|
198 | my @commameta_list = split(/,/, $metaname);
|
---|
199 |
|
---|
200 | if ($self->{'doclevel'} =~ /^top(level)?/i)
|
---|
201 | {
|
---|
202 | push(@sectionlist,$topsection);
|
---|
203 | }
|
---|
204 | else
|
---|
205 | {
|
---|
206 | my $thissection = $doc_obj->get_next_section($topsection);
|
---|
207 | while (defined $thissection)
|
---|
208 | {
|
---|
209 | push(@sectionlist,$thissection);
|
---|
210 | $thissection = $doc_obj->get_next_section ($thissection);
|
---|
211 | }
|
---|
212 | }
|
---|
213 |
|
---|
214 | my $thissection;
|
---|
215 | foreach $thissection (@sectionlist)
|
---|
216 | {
|
---|
217 | my $full_doc_OID
|
---|
218 | = ($thissection ne "") ? "$doc_OID.$thissection" : $doc_OID;
|
---|
219 | if (defined $self->{'list_mvpair'}->{$full_doc_OID})
|
---|
220 | {
|
---|
221 | print $outhandle "WARNING: AZCompactList::classify called multiple times for $full_doc_OID\n";
|
---|
222 | }
|
---|
223 | $self->{'list'}->{$full_doc_OID} = [];
|
---|
224 | $self->{'listmetavalue'}->{$full_doc_OID} = [];
|
---|
225 | $self->{'list_mvpair'}->{$full_doc_OID} = [];
|
---|
226 |
|
---|
227 | my $metavalues = [];
|
---|
228 | foreach my $cmn (@commameta_list) {
|
---|
229 | my $cmvalues = $doc_obj->get_metadata($thissection,$cmn);
|
---|
230 | push(@$metavalues,@$cmvalues) if (@{$cmvalues});
|
---|
231 | last if (@{$cmvalues} && !$self->{'allvalues'});
|
---|
232 | }
|
---|
233 |
|
---|
234 | my $metavalue;
|
---|
235 | foreach $metavalue (@$metavalues)
|
---|
236 | {
|
---|
237 | # Tidy up use of white space in metavalue for classifying
|
---|
238 | $metavalue =~ s/^\s*//s;
|
---|
239 | $metavalue =~ s/\s*$//s;
|
---|
240 | $metavalue =~ s/\n/ /s;
|
---|
241 | $metavalue =~ s/\s{2,}/ /s;
|
---|
242 |
|
---|
243 | # if this document doesn't contain the metadata element we're
|
---|
244 | # sorting by we won't include it in this classification
|
---|
245 | if (defined $metavalue && $metavalue =~ /\w/)
|
---|
246 | {
|
---|
247 | if (defined($self->{'removeprefix'}) &&
|
---|
248 | length($self->{'removeprefix'})) {
|
---|
249 | $metavalue =~ s/^$self->{'removeprefix'}//;
|
---|
250 |
|
---|
251 | # check that it's not now empty
|
---|
252 | if (!$metavalue) {next;}
|
---|
253 | }
|
---|
254 |
|
---|
255 | if (defined($self->{'removesuffix'}) &&
|
---|
256 | length($self->{'removesuffix'})) {
|
---|
257 | $metavalue =~ s/$self->{'removesuffix'}$//;
|
---|
258 |
|
---|
259 | # check that it's not now empty
|
---|
260 | if (!$metavalue) {next;}
|
---|
261 | }
|
---|
262 |
|
---|
263 | my $formatted_metavalue;
|
---|
264 | if ($self->{'no_metadata_formatting'}) {
|
---|
265 | $formatted_metavalue = $metavalue;
|
---|
266 | } else {
|
---|
267 | $formatted_metavalue = &sorttools::format_metadata_for_sorting($self->{'metadata'}, $metavalue, $doc_obj);
|
---|
268 | }
|
---|
269 |
|
---|
270 | #### prefix-str
|
---|
271 | if (! defined($formatted_metavalue)) {
|
---|
272 | print $outhandle "Warning: AZCompactList: metavalue is ";
|
---|
273 | print $outhandle "empty\n";
|
---|
274 | $formatted_metavalue="";
|
---|
275 | }
|
---|
276 |
|
---|
277 | my $mv_pair = { 'mv' => $metavalue, 'fmv' => $formatted_metavalue };
|
---|
278 | push(@{$self->{'list'}->{$full_doc_OID}},$formatted_metavalue);
|
---|
279 | push(@{$self->{'listmetavalue'}->{$full_doc_OID}} ,$metavalue);
|
---|
280 | push(@{$self->{'list_mvpair'}->{$full_doc_OID}},$mv_pair);
|
---|
281 |
|
---|
282 |
|
---|
283 | last if ($self->{'firstvalueonly'});
|
---|
284 | }
|
---|
285 | }
|
---|
286 |
|
---|
287 | # This is used in reclassify below for AZCompactSectionList
|
---|
288 | my $sortmeta = $doc_obj->get_metadata_element($thissection, $self->{'sort'});
|
---|
289 | $self->{'reclassify'}->{$full_doc_OID} = [$doc_obj,$sortmeta];
|
---|
290 | }
|
---|
291 | }
|
---|
292 |
|
---|
293 | sub reinit
|
---|
294 | {
|
---|
295 | my ($self,$classlist_ref) = @_;
|
---|
296 | my $outhandle = $self->{'outhandle'};
|
---|
297 |
|
---|
298 | my %mtfreq = ();
|
---|
299 | my @single_classlist = ();
|
---|
300 | my @multiple_classlist = ();
|
---|
301 |
|
---|
302 | # find out how often each metavalue occurs
|
---|
303 | map
|
---|
304 | {
|
---|
305 | foreach my $mvp (@{$self->{'list_mvpair'}->{$_}} )
|
---|
306 | {
|
---|
307 | ### print STDERR "*** plain mv = $mvp->{'mv'}\n";
|
---|
308 | ### print STDERR "*** format mv = $mvp->{'fmv'}\n";
|
---|
309 |
|
---|
310 | my $metavalue = $mvp->{'mv'};
|
---|
311 | $metavalue =~ s!^-!\\-!; # in case it starts with "-"
|
---|
312 | $mtfreq{$metavalue}++;
|
---|
313 | }
|
---|
314 | } @$classlist_ref;
|
---|
315 |
|
---|
316 | # use this information to split the list: single metavalue/repeated value
|
---|
317 | map
|
---|
318 | {
|
---|
319 | my $i = 1;
|
---|
320 | my $metavalue;
|
---|
321 | foreach my $mvp (@{$self->{'list_mvpair'}->{$_}})
|
---|
322 | {
|
---|
323 | my $metavalue = $mvp->{'mv'};
|
---|
324 | $metavalue =~ s!^-!\\-!; # in case it starts with "-"
|
---|
325 | my $cs_metavalue = $mvp->{'mv'}; # case sensative
|
---|
326 | if ($mtfreq{$metavalue} >= $self->{'mingroup'})
|
---|
327 | {
|
---|
328 | ### print STDERR "*** pushing on $cs_metavalue\n";
|
---|
329 | push(@multiple_classlist,[$_,$i,$metavalue]);
|
---|
330 | }
|
---|
331 | else
|
---|
332 | {
|
---|
333 | push(@single_classlist,[$_,$cs_metavalue]);
|
---|
334 | $metavalue =~ tr/[A-Z]/[a-z]/;
|
---|
335 | $self->{'reclassifylist'}->{"Metavalue_$i.$_"} = $metavalue;
|
---|
336 | }
|
---|
337 | $i++;
|
---|
338 | }
|
---|
339 | } @$classlist_ref;
|
---|
340 |
|
---|
341 |
|
---|
342 | # Setup sub-classifiers for multiple list
|
---|
343 |
|
---|
344 | $self->{'classifiers'} = {};
|
---|
345 |
|
---|
346 | my $pm;
|
---|
347 | foreach $pm ("List", "SectionList")
|
---|
348 | {
|
---|
349 | my $listname
|
---|
350 | = &util::filename_cat($ENV{'GSDLHOME'},"perllib/classify/$pm.pm");
|
---|
351 | if (-e $listname) { require $listname; }
|
---|
352 | else
|
---|
353 | {
|
---|
354 | print $outhandle "AZCompactList ERROR - couldn't find classifier \"$listname\"\n";
|
---|
355 | die "\n";
|
---|
356 | }
|
---|
357 | }
|
---|
358 |
|
---|
359 | # Create classifiers objects for each entry >= mingroup
|
---|
360 | my $metavalue;
|
---|
361 | my $doclevel = $self->{'doclevel'};
|
---|
362 | my $mingroup = $self->{'mingroup'};
|
---|
363 | my @metaname_list = split(/\/|\|/,$self->{'metadata'});
|
---|
364 | my $metaname = shift(@metaname_list);
|
---|
365 | my $hierarchical = 0;
|
---|
366 | if (scalar(@metaname_list) > 1) {
|
---|
367 | $hierarchical = 1;
|
---|
368 | $metaname = join('/',@metaname_list);
|
---|
369 | }
|
---|
370 | foreach $metavalue (keys %mtfreq)
|
---|
371 | {
|
---|
372 | if ($mtfreq{$metavalue} >= $mingroup)
|
---|
373 | {
|
---|
374 | # occurs more often than minimum required to compact into a group
|
---|
375 | my $listclassobj;
|
---|
376 |
|
---|
377 | if (!$hierarchical)
|
---|
378 | {
|
---|
379 | my @args;
|
---|
380 | push @args, ("-metadata", "$metaname");
|
---|
381 | # buttonname is also used for the node's title
|
---|
382 | push @args, ("-buttonname", "$metavalue");
|
---|
383 | push @args, ("-sort", $self->{'sort'});
|
---|
384 |
|
---|
385 | my $ptArgs = \@args;
|
---|
386 | if ($doclevel =~ m/^top(level)?/i)
|
---|
387 | {
|
---|
388 | eval ("\$listclassobj = new List([],\$ptArgs)");
|
---|
389 | }
|
---|
390 | else
|
---|
391 | {
|
---|
392 | eval ("\$listclassobj = new SectionList([],\$ptArgs)");
|
---|
393 | }
|
---|
394 | }
|
---|
395 | else
|
---|
396 | {
|
---|
397 |
|
---|
398 | my @args;
|
---|
399 | push @args, ("-metadata", "$metaname");
|
---|
400 | # buttonname is also used for the node's title
|
---|
401 | push @args, ("-buttonname", "$metavalue");
|
---|
402 | push @args, ("-sort", $self->{'sort'});
|
---|
403 |
|
---|
404 | if (defined $self->{'removeprefix'}) {
|
---|
405 | push @args, ("-removeprefix", $self->{'removeprefix'});
|
---|
406 | }
|
---|
407 | if (defined $self->{'removesuffix'}) {
|
---|
408 | push @args, ("-removesuffix", $self->{'removesuffix'});
|
---|
409 | }
|
---|
410 |
|
---|
411 | push @args, ("-doclevel", "$doclevel");
|
---|
412 | push @args, ("-mingroup", $mingroup);
|
---|
413 | push @args, "-recopt ";
|
---|
414 |
|
---|
415 | my $ptArgs = \@args;
|
---|
416 | eval ("\$listclassobj = new AZCompactList([],\$ptArgs)");
|
---|
417 | }
|
---|
418 |
|
---|
419 | if ($@) {
|
---|
420 | print $outhandle "$@";
|
---|
421 | die "\n";
|
---|
422 | }
|
---|
423 |
|
---|
424 | $listclassobj->init();
|
---|
425 |
|
---|
426 | if (defined $metavalue && $metavalue =~ /\w/)
|
---|
427 | {
|
---|
428 | my $formatted_node = $metavalue;
|
---|
429 |
|
---|
430 | if (defined($self->{'removeprefix'}) &&
|
---|
431 | length($self->{'removeprefix'})) {
|
---|
432 | $formatted_node =~ s/^$self->{'removeprefix'}//;
|
---|
433 | # check that it's not now empty
|
---|
434 | if (!$formatted_node) {next;}
|
---|
435 | }
|
---|
436 | if (defined($self->{'removesuffix'}) &&
|
---|
437 | length($self->{'removesuffix'})) {
|
---|
438 | $formatted_node =~ s/$self->{'removesuffix'}$//;
|
---|
439 | # check that it's not now empty
|
---|
440 | if (!$formatted_node) {next;}
|
---|
441 | }
|
---|
442 |
|
---|
443 | $formatted_node = &sorttools::format_metadata_for_sorting($self->{'metadata'}, $formatted_node) unless $self->{'no_metadata_formatting'};
|
---|
444 |
|
---|
445 | # In case our formatted string is empty...
|
---|
446 | if (! defined($formatted_node)) {
|
---|
447 | print $outhandle "Warning: AZCompactList: metavalue is ";
|
---|
448 | print $outhandle "empty\n";
|
---|
449 | $formatted_node="";
|
---|
450 | }
|
---|
451 |
|
---|
452 | # use the lower case, for speed of lookup.
|
---|
453 | my $meta_lc=lc($metavalue);
|
---|
454 | $self->{'classifiers'}->{$meta_lc}
|
---|
455 | = { 'classifyobj' => $listclassobj,
|
---|
456 | 'formattednode' => $formatted_node };
|
---|
457 | }
|
---|
458 | }
|
---|
459 | }
|
---|
460 |
|
---|
461 |
|
---|
462 | return (\@single_classlist,\@multiple_classlist);
|
---|
463 | }
|
---|
464 |
|
---|
465 |
|
---|
466 | sub reclassify
|
---|
467 | {
|
---|
468 | my ($self,$multiple_cl_ref) = @_;
|
---|
469 |
|
---|
470 | # Entries in the current classify list that are "book nodes"
|
---|
471 | # should be recursively classified.
|
---|
472 | #--
|
---|
473 | foreach my $dm_pair (@$multiple_cl_ref)
|
---|
474 | {
|
---|
475 | my ($doc_OID,$mdoffset,$metavalue,$cs_metavalue) = @$dm_pair;
|
---|
476 | my $listclassobj;
|
---|
477 |
|
---|
478 | # find metavalue in list of sub-classifiers
|
---|
479 | # check if we have a key (lower case) for this metadata value
|
---|
480 | my $node_name=lc($metavalue);
|
---|
481 | if (exists $self->{'classifiers'}->{$node_name})
|
---|
482 | {
|
---|
483 | my ($doc_obj, $sortmeta) = @{$self->{'reclassify'}->{$doc_OID}};
|
---|
484 |
|
---|
485 | # record the metadata value offset temporarily, so eg AZList can
|
---|
486 | # get the correct metadata value (for multi-valued metadata fields)
|
---|
487 | $doc_obj->{'mdoffset'}=$mdoffset;
|
---|
488 |
|
---|
489 | if ($doc_OID =~ m/^[^\.]*\.([\d\.]+)$/)
|
---|
490 | {
|
---|
491 | my $section=$1;
|
---|
492 | if ($self->{'doclevel'} =~ m/^top/i) { # toplevel
|
---|
493 | $self->{'classifiers'}->{$node_name}->{'classifyobj'}
|
---|
494 | ->classify($doc_obj, "Section=$section");
|
---|
495 | } else { # section level
|
---|
496 | # Thanks to Don Gourley for this...
|
---|
497 | # classify can't handle multi-level section
|
---|
498 | $self->{'classifiers'}->{$node_name}->{'classifyobj'}
|
---|
499 | ->classify_section($section, $doc_obj, $sortmeta);
|
---|
500 | }
|
---|
501 | }
|
---|
502 | else
|
---|
503 | {
|
---|
504 | $self->{'classifiers'}->{$node_name}->{'classifyobj'}
|
---|
505 | ->classify($doc_obj);
|
---|
506 | }
|
---|
507 | } else { # this key is not in the hash
|
---|
508 | my $outhandle=$self->{outhandle};
|
---|
509 | print $outhandle "Warning: AZCompactList::reclassify ";
|
---|
510 | print $outhandle "could not find sub-node for metadata=`$metavalue' with doc_OID $doc_OID\n";
|
---|
511 | }
|
---|
512 | }
|
---|
513 | }
|
---|
514 |
|
---|
515 |
|
---|
516 |
|
---|
517 | sub get_reclassify_info
|
---|
518 | {
|
---|
519 | my $self = shift (@_);
|
---|
520 |
|
---|
521 | my $node_name;
|
---|
522 | foreach $node_name (keys %{$self->{'classifiers'}})
|
---|
523 | {
|
---|
524 | my $classifyinfo
|
---|
525 | = $self->{'classifiers'}->{$node_name}->{'classifyobj'}
|
---|
526 | ->get_classify_info();
|
---|
527 | $self->{'classifiers'}->{$node_name}->{'classifyinfo'}
|
---|
528 | = $classifyinfo;
|
---|
529 | $self->{'reclassifylist'}->{"CLASSIFY.$node_name"}
|
---|
530 | = $self->{'classifiers'}->{$node_name}->{'formattednode'};
|
---|
531 | }
|
---|
532 | }
|
---|
533 |
|
---|
534 |
|
---|
535 | sub alpha_numeric_cmp
|
---|
536 | {
|
---|
537 | my ($self,$a,$b) = @_;
|
---|
538 |
|
---|
539 | my $title_a = $self->{'reclassifylist'}->{$a};
|
---|
540 | my $title_b = $self->{'reclassifylist'}->{$b};
|
---|
541 |
|
---|
542 | if ($title_a =~ m/^(\d+(\.\d+)?)/)
|
---|
543 | {
|
---|
544 | my $val_a = $1;
|
---|
545 | if ($title_b =~ m/^(\d+(\.\d+)?)/)
|
---|
546 | {
|
---|
547 | my $val_b = $1;
|
---|
548 | if ($val_a != $val_b)
|
---|
549 | {
|
---|
550 | return ($val_a <=> $val_b);
|
---|
551 | }
|
---|
552 | }
|
---|
553 | }
|
---|
554 |
|
---|
555 | return ($title_a cmp $title_b);
|
---|
556 | }
|
---|
557 |
|
---|
558 | sub frequency_cmp
|
---|
559 | {
|
---|
560 | my ($self,$a,$b) = @_;
|
---|
561 |
|
---|
562 |
|
---|
563 | my $title_a = $self->{'reclassifylist'}->{$a};
|
---|
564 | my $title_b = $self->{'reclassifylist'}->{$b};
|
---|
565 |
|
---|
566 | my $a_freq = 1;
|
---|
567 | my $b_freq = 1;
|
---|
568 |
|
---|
569 | if ($a =~ m/^CLASSIFY\.(.*)$/)
|
---|
570 | {
|
---|
571 | my $a_node = $1;
|
---|
572 | my $a_nodeinfo = $self->{'classifiers'}->{$a_node}->{'classifyinfo'};
|
---|
573 | $a_freq = scalar(@{$a_nodeinfo->{'contains'}});
|
---|
574 | }
|
---|
575 |
|
---|
576 | if ($b =~ m/^CLASSIFY\.(.*)$/)
|
---|
577 | {
|
---|
578 | my $b_node = $1;
|
---|
579 | my $b_nodeinfo = $self->{'classifiers'}->{$b_node}->{'classifyinfo'};
|
---|
580 | $b_freq = scalar(@{$b_nodeinfo->{'contains'}});
|
---|
581 | }
|
---|
582 |
|
---|
583 | return $b_freq <=> $a_freq;
|
---|
584 | }
|
---|
585 |
|
---|
586 | sub get_classify_info {
|
---|
587 | my $self = shift (@_);
|
---|
588 |
|
---|
589 | my @classlist =keys %{$self->{'list_mvpair'}}; # list all doc oids
|
---|
590 |
|
---|
591 | my ($single_cl_ref,$multiple_cl_ref) = $self->reinit(\@classlist);
|
---|
592 | $self->reclassify($multiple_cl_ref);
|
---|
593 | $self->get_reclassify_info();
|
---|
594 |
|
---|
595 | my @reclassified_classlist;
|
---|
596 | if ($self->{'freqsort'})
|
---|
597 | {
|
---|
598 | @reclassified_classlist
|
---|
599 | = sort { $self->frequency_cmp($a,$b) } keys %{$self->{'reclassifylist'}};
|
---|
600 | # supress sub-grouping by alphabet
|
---|
601 | map { $self->{'reclassifylist'}->{$_} = "A".$self->{'reclassifylist'}; } keys %{$self->{'reclassifylist'}};
|
---|
602 | }
|
---|
603 | else
|
---|
604 | {
|
---|
605 | # @reclassified_classlist
|
---|
606 | # = sort {$self->{'reclassifylist'}->{$a} cmp $self->{'reclassifylist'}->{$b};} keys %{$self->{'reclassifylist'}};
|
---|
607 |
|
---|
608 | # alpha_numeric_cmp is slower than "cmp" but handles numbers better ...
|
---|
609 |
|
---|
610 | @reclassified_classlist
|
---|
611 | = sort { $self->alpha_numeric_cmp($a,$b) } keys %{$self->{'reclassifylist'}};
|
---|
612 |
|
---|
613 | }
|
---|
614 |
|
---|
615 | return $self->splitlist (\@reclassified_classlist);
|
---|
616 | }
|
---|
617 |
|
---|
618 | sub get_entry {
|
---|
619 | my $self = shift (@_);
|
---|
620 | my ($title, $childtype, $metaname, $thistype) = @_;
|
---|
621 | # organise into classification structure
|
---|
622 | my %classifyinfo = ('childtype'=>$childtype,
|
---|
623 | 'Title'=>$title,
|
---|
624 | 'contains'=>[],
|
---|
625 | 'mdtype'=>$metaname);
|
---|
626 |
|
---|
627 | $classifyinfo{'thistype'} = $thistype
|
---|
628 | if defined $thistype && $thistype =~ /\w/;
|
---|
629 |
|
---|
630 | return \%classifyinfo;
|
---|
631 | }
|
---|
632 |
|
---|
633 |
|
---|
634 |
|
---|
635 | # splitlist takes an ordered list of classifications (@$classlistref) and
|
---|
636 | # splits it up into alphabetical sub-sections.
|
---|
637 | sub splitlist {
|
---|
638 | my $self = shift (@_);
|
---|
639 | my ($classlistref) = @_;
|
---|
640 | my $classhash = {};
|
---|
641 |
|
---|
642 | # top level
|
---|
643 | my @metanames = split(/\/|\|/,$self->{'metadata'});
|
---|
644 | my $metaname = shift(@metanames);
|
---|
645 |
|
---|
646 | my $childtype = "HList";
|
---|
647 | $childtype = "VList" if (scalar (@$classlistref) <= $self->{'minnesting'});
|
---|
648 |
|
---|
649 | my $title = $self->{'buttonname'}; # should always be defined by now.
|
---|
650 | my $classifyinfo;
|
---|
651 | if (!defined($self->{'recopt'}))
|
---|
652 | {
|
---|
653 | $classifyinfo
|
---|
654 | = $self->get_entry ($title, $childtype, $metaname, "Invisible");
|
---|
655 | }
|
---|
656 | else
|
---|
657 | {
|
---|
658 | $classifyinfo
|
---|
659 | = $self->get_entry ($title, $childtype, $metaname, "VList");
|
---|
660 | }
|
---|
661 |
|
---|
662 | # don't need to do any splitting if there are less than 'minnesting' classifications
|
---|
663 | if ((scalar @$classlistref) <= $self->{'minnesting'}) {
|
---|
664 | foreach my $subOID (@$classlistref) {
|
---|
665 | if ($subOID =~ /^CLASSIFY\.(.*)$/
|
---|
666 | && defined $self->{'classifiers'}->{$1})
|
---|
667 | {
|
---|
668 | push (@{$classifyinfo->{'contains'}},
|
---|
669 | $self->{'classifiers'}->{$1}->{'classifyinfo'});
|
---|
670 | }
|
---|
671 | else
|
---|
672 | {
|
---|
673 | $subOID =~ s/^Metavalue_(\d+)\.//;
|
---|
674 | my $metaname_offset = $1 -1;
|
---|
675 | my $oid_rec = {'OID'=>$subOID, 'offset'=>$metaname_offset};
|
---|
676 | push (@{$classifyinfo->{'contains'}}, $oid_rec);
|
---|
677 | }
|
---|
678 | }
|
---|
679 | return $classifyinfo;
|
---|
680 | }
|
---|
681 |
|
---|
682 | # first split up the list into separate A-Z and 0-9 classifications
|
---|
683 | foreach my $classification (@$classlistref) {
|
---|
684 | my $title = $self->{'reclassifylist'}->{$classification};
|
---|
685 | $title =~ s/&(.){2,4};//g; # remove any HTML special chars
|
---|
686 | $title =~ s/^\W+//g; # remove leading non-word chars
|
---|
687 |
|
---|
688 | # only want first character for classification
|
---|
689 | $title =~ m/^(.)/; # always a char, or first byte of wide char?
|
---|
690 | if (defined($1) && $1 ne "") {
|
---|
691 | $title=$1;
|
---|
692 | } else {
|
---|
693 | print STDERR "no first character found for \"$title\" - \"" .
|
---|
694 | $self->{'reclassifylist'}->{$classification} . "\"\n";
|
---|
695 | }
|
---|
696 | $title =~ tr/[a-z]/[A-Z]/;
|
---|
697 |
|
---|
698 | if ($title =~ /^[0-9]$/) {$title = '0-9';}
|
---|
699 | elsif ($title !~ /^[A-Z]$/) {
|
---|
700 | my $outhandle = $self->{'outhandle'};
|
---|
701 | print $outhandle "AZCompactList: WARNING $classification has badly formatted title ($title)\n";
|
---|
702 | }
|
---|
703 | $classhash->{$title} = [] unless defined $classhash->{$title};
|
---|
704 | push (@{$classhash->{$title}}, $classification);
|
---|
705 | }
|
---|
706 | $classhash = $self->compactlist ($classhash);
|
---|
707 |
|
---|
708 | my @tmparr = ();
|
---|
709 | foreach my $subsection (sort keys (%$classhash)) {
|
---|
710 | push (@tmparr, $subsection);
|
---|
711 | }
|
---|
712 |
|
---|
713 | # if there's a 0-9 section it will have been sorted to the beginning
|
---|
714 | # but we want it at the end
|
---|
715 | if ($tmparr[0] eq '0-9') {
|
---|
716 | shift @tmparr;
|
---|
717 | push (@tmparr, '0-9');
|
---|
718 | }
|
---|
719 | foreach my $subclass (@tmparr)
|
---|
720 | {
|
---|
721 | my $tempclassify
|
---|
722 | = (scalar(@tmparr)==1)
|
---|
723 | ? ($self->get_entry(" ", "VList", $metaname))
|
---|
724 | : ($self->get_entry($subclass, "VList", $metaname));
|
---|
725 |
|
---|
726 |
|
---|
727 | foreach my $subsubOID (@{$classhash->{$subclass}})
|
---|
728 | {
|
---|
729 | if ($subsubOID =~ /^CLASSIFY\.(.*)$/
|
---|
730 | && defined $self->{'classifiers'}->{$1})
|
---|
731 | {
|
---|
732 | # this is a "bookshelf" node... >1 entry compacted
|
---|
733 | push (@{$tempclassify->{'contains'}},
|
---|
734 | $self->{'classifiers'}->{$1}->{'classifyinfo'});
|
---|
735 | # set the metadata field name, for mdoffset to work
|
---|
736 | $self->{'classifiers'}->{$1}->{'classifyinfo'}->{'mdtype'}=
|
---|
737 | $metaname;
|
---|
738 | }
|
---|
739 | else
|
---|
740 | {
|
---|
741 | $subsubOID =~ s/^Metavalue_(\d+)\.//;
|
---|
742 | # record the offset if this metadata type has multiple values
|
---|
743 | my $metaname_offset = $1 -1;
|
---|
744 | my $oid_rec = {'OID'=>$subsubOID, 'offset'=>$metaname_offset};
|
---|
745 | push (@{$tempclassify->{'contains'}}, $oid_rec);
|
---|
746 | }
|
---|
747 | }
|
---|
748 | push (@{$classifyinfo->{'contains'}}, $tempclassify);
|
---|
749 | }
|
---|
750 |
|
---|
751 | return $classifyinfo;
|
---|
752 | }
|
---|
753 |
|
---|
754 | sub compactlist {
|
---|
755 | my $self = shift (@_);
|
---|
756 | my ($classhashref) = @_;
|
---|
757 | my $compactedhash = {};
|
---|
758 | my @currentOIDs = ();
|
---|
759 | my $currentfirstletter = "";
|
---|
760 | my $currentlastletter = "";
|
---|
761 | my $lastkey = "";
|
---|
762 |
|
---|
763 | # minimum and maximum documents to be displayed per page.
|
---|
764 | # the actual maximum will be max + (min-1).
|
---|
765 | # the smallest sub-section is a single letter at present
|
---|
766 | # so in this case there may be many times max documents
|
---|
767 | # displayed on a page.
|
---|
768 | my $min = $self->{'mincompact'};
|
---|
769 | my $max = $self->{'maxcompact'};
|
---|
770 |
|
---|
771 | foreach my $subsection (sort keys %$classhashref) {
|
---|
772 | if ($subsection eq '0-9') {
|
---|
773 | @{$compactedhash->{$subsection}} = @{$classhashref->{$subsection}};
|
---|
774 | next;
|
---|
775 | }
|
---|
776 | $currentfirstletter = $subsection if $currentfirstletter eq "";
|
---|
777 | if ((scalar (@currentOIDs) < $min) ||
|
---|
778 | ((scalar (@currentOIDs) + scalar (@{$classhashref->{$subsection}})) <= $max)) {
|
---|
779 | push (@currentOIDs, @{$classhashref->{$subsection}});
|
---|
780 | $currentlastletter = $subsection;
|
---|
781 | } else {
|
---|
782 |
|
---|
783 | if ($currentfirstletter eq $currentlastletter) {
|
---|
784 | @{$compactedhash->{$currentfirstletter}} = @currentOIDs;
|
---|
785 | $lastkey = $currentfirstletter;
|
---|
786 | } else {
|
---|
787 | @{$compactedhash->{"$currentfirstletter-$currentlastletter"}} = @currentOIDs;
|
---|
788 | $lastkey = "$currentfirstletter-$currentlastletter";
|
---|
789 | }
|
---|
790 | if (scalar (@{$classhashref->{$subsection}}) >= $max) {
|
---|
791 | $compactedhash->{$subsection} = $classhashref->{$subsection};
|
---|
792 | @currentOIDs = ();
|
---|
793 | $currentfirstletter = "";
|
---|
794 | $lastkey=$subsection;
|
---|
795 | } else {
|
---|
796 | @currentOIDs = @{$classhashref->{$subsection}};
|
---|
797 | $currentfirstletter = $subsection;
|
---|
798 | $currentlastletter = $subsection;
|
---|
799 | }
|
---|
800 | }
|
---|
801 | }
|
---|
802 |
|
---|
803 | # add final OIDs to last sub-classification if there aren't many otherwise
|
---|
804 | # add final sub-classification
|
---|
805 |
|
---|
806 | # don't add if there aren't any oids
|
---|
807 | if (! scalar (@currentOIDs)) {return $compactedhash;}
|
---|
808 |
|
---|
809 | if (scalar (@currentOIDs) < $min) {
|
---|
810 | my ($newkey) = $lastkey =~ /^(.)/;
|
---|
811 | @currentOIDs = (@{$compactedhash->{$lastkey}}, @currentOIDs);
|
---|
812 | delete $compactedhash->{$lastkey};
|
---|
813 | @{$compactedhash->{"$newkey-$currentlastletter"}} = @currentOIDs;
|
---|
814 | } else {
|
---|
815 | if ($currentfirstletter eq $currentlastletter) {
|
---|
816 | @{$compactedhash->{$currentfirstletter}} = @currentOIDs;
|
---|
817 | } else {
|
---|
818 | @{$compactedhash->{"$currentfirstletter-$currentlastletter"}} = @currentOIDs;
|
---|
819 | }
|
---|
820 | }
|
---|
821 |
|
---|
822 | return $compactedhash;
|
---|
823 | }
|
---|
824 |
|
---|
825 | 1;
|
---|
826 |
|
---|
827 |
|
---|