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