/[webpac]/branches/ffzg/all2xml.pl
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /branches/ffzg/all2xml.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 164 - (hide annotations)
Sat Nov 22 22:04:05 2003 UTC (20 years, 4 months ago) by dpavlin
Original Path: trunk/all2xml.pl
File MIME type: text/plain
File size: 16888 byte(s)
implemented filter which can replace (or be used together with) unac_string
from Text::Unaccent

1 dpavlin 1 #!/usr/bin/perl -w
2    
3     use strict;
4     use OpenIsis;
5     use Getopt::Std;
6     use Data::Dumper;
7     use XML::Simple;
8 dpavlin 5 use Text::Unaccent 1.02; # 1.01 won't compile on my platform,
9 dpavlin 10 use Text::Iconv;
10 dpavlin 13 use Config::IniFiles;
11 dpavlin 40 use Encode;
12 dpavlin 1
13 dpavlin 10 $|=1;
14 dpavlin 9
15 dpavlin 13 my $config_file = $0;
16     $config_file =~ s/\.pl$/.conf/;
17     die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file);
18    
19 dpavlin 10 my $config;
20    
21 dpavlin 58 #use index_DBI; # default DBI module for index
22     use index_DBI_cache; # faster DBI module using memory cache
23 dpavlin 50 my $index;
24 dpavlin 10
25 dpavlin 1 my %opts;
26    
27 dpavlin 7 # usage:
28     # -d directory name
29     # -m multiple directories
30     # -q quiet
31     # -s run swish
32 dpavlin 1
33 dpavlin 7 getopts('d:m:qs', \%opts);
34    
35 dpavlin 43 my $path; # this is name of database
36 dpavlin 1
37 dpavlin 57 Text::Iconv->raise_error(0); # Conversion errors don't raise exceptions
38 dpavlin 1
39 dpavlin 40 # this is encoding of all files on disk, including import_xml/*.xml file and
40     # filter/*.pm files! It will be used to store strings in perl internally!
41     my $codepage = 'ISO-8859-2';
42 dpavlin 1
43 dpavlin 40 my $utf2cp = Text::Iconv->new('UTF-8',$codepage);
44     # this function will convert data from XML files to local encoding
45     sub x {
46     return $utf2cp->convert($_[0]);
47     }
48 dpavlin 3
49 dpavlin 54 # decode isis/excel or other import codepage
50     my $import2cp;
51 dpavlin 10
52 dpavlin 40 # outgoing xml must be in UTF-8
53     my $cp2utf = Text::Iconv->new($codepage,'UTF-8');
54 dpavlin 29
55 dpavlin 54 # mapping between data type and tag which specify
56     # format in XML file
57     my %type2tag = (
58     'isis' => 'isis',
59 dpavlin 62 'excel' => 'column',
60     'marc' => 'marc',
61 dpavlin 67 'feed' => 'feed'
62 dpavlin 54 );
63 dpavlin 3
64 dpavlin 54 sub data2xml {
65    
66 dpavlin 10 use xmlify;
67    
68 dpavlin 54 my $type = shift @_;
69 dpavlin 3 my $row = shift @_;
70 dpavlin 13 my $add_xml = shift @_;
71 dpavlin 59 # needed to read values from configuration file
72     my $cfg = shift @_;
73     my $database = shift @_;
74 dpavlin 3
75     my $xml;
76    
77 dpavlin 10 use parse_format;
78 dpavlin 3
79 dpavlin 13 my $html = ""; # html formatted display output
80 dpavlin 10
81 dpavlin 13 my %field_usage; # counter for usage of each field
82    
83 dpavlin 32 # sort subrouting using order="" attribute
84     sub by_order {
85 dpavlin 98 my $va = $config->{indexer}->{$a}->{order} ||
86     $config->{indexer}->{$a};
87     my $vb = $config->{indexer}->{$b}->{order} ||
88     $config->{indexer}->{$b};
89 dpavlin 29
90 dpavlin 98 return $va <=> $vb;
91 dpavlin 32 }
92 dpavlin 3
93 dpavlin 32 foreach my $field (sort by_order keys %{$config->{indexer}}) {
94    
95 dpavlin 40 $field=x($field);
96 dpavlin 13 $field_usage{$field}++;
97    
98 dpavlin 10 my $swish_data = "";
99 dpavlin 163 my $swish_exact_data = "";
100 dpavlin 3 my $display_data = "";
101 dpavlin 35 my $line_delimiter;
102 dpavlin 3
103 dpavlin 34 my ($swish,$display);
104    
105 dpavlin 54 my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
106     foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
107 dpavlin 3
108 dpavlin 40 my $format = x($x->{content});
109     my $delimiter = x($x->{delimiter}) || ' ';
110 dpavlin 3
111 dpavlin 54 my $repeat_off = 0; # repeatable offset
112 dpavlin 29
113 dpavlin 163 my ($s,$se,$d,$i) = (1,0,1,0); # swish, display default
114 dpavlin 29 $s = 0 if (lc($x->{type}) eq "display");
115     $d = 0 if (lc($x->{type}) eq "swish");
116 dpavlin 163 $se = 1 if (lc($x->{type}) eq "swish_exact");
117 dpavlin 29 ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
118    
119 dpavlin 35 # what will separate last line from this one?
120     if ($display_data && $x->{append} && $x->{append} eq "1") {
121 dpavlin 34 $line_delimiter = ' ';
122 dpavlin 35 } elsif ($display_data) {
123 dpavlin 34 $line_delimiter = '<br/>';
124     }
125 dpavlin 29
126 dpavlin 34 # init vars so that we go into while...
127     ($swish,$display) = (1,1);
128 dpavlin 29
129 dpavlin 97 # placeholder for all repeatable entries for index
130     my @index_data;
131     my $index_filter;
132    
133 dpavlin 153 sub mkformat {
134     my $x = shift || die "mkformat needs tag reference";
135     my $data = shift || return;
136     my $format_name = x($x->{format_name}) || return $data;
137     my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!";
138     my $format_delimiter = x($x->{format_delimiter});
139     my @data;
140     if ($format_delimiter) {
141     @data = split(/$format_delimiter/,$data);
142     } else {
143     push @data,$data;
144     }
145    
146     if ($fmt) {
147     my $nr = scalar $fmt =~ s/%s/%s/g;
148     if (($#data+1) == $nr) {
149     return sprintf($fmt,@data);
150     } else {
151     print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
152     return $data;
153     }
154     } else {
155     print STDERR "usage of link '$format_name' without defined format (<link> tag)\n";
156     }
157     }
158    
159 dpavlin 90 # while because of repeatable fields
160     while ($swish || $display) {
161 dpavlin 54 ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
162 dpavlin 90 if ($repeat_off > 1000) {
163     print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n";
164     last;
165     }
166    
167 dpavlin 29 # filter="name" ; filter this field through
168     # filter/[name].pm
169     my $filter = $x->{filter};
170 dpavlin 20 if ($filter) {
171 dpavlin 29 require "filter/".$filter.".pm";
172 dpavlin 20 }
173 dpavlin 29 # type="swish" ; field for swish
174 dpavlin 163 if ($swish) {
175     if ($filter && ($s || $se)) {
176 dpavlin 29 no strict 'refs';
177 dpavlin 163 my $tmp = join(" ",&$filter($swish)) if ($s || $se);
178     $swish_data .= $tmp if ($s);
179     $swish_exact_data .= $tmp if ($se);
180    
181 dpavlin 29 } else {
182 dpavlin 163 $swish_data .= $swish if ($s);
183     $swish_exact_data .= $swish if ($se);
184 dpavlin 29 }
185     }
186 dpavlin 17
187 dpavlin 29 # type="display" ; field for display
188     if ($d && $display) {
189 dpavlin 35 if ($line_delimiter && $display_data) {
190     $display_data .= $line_delimiter;
191     undef $line_delimiter;
192     }
193 dpavlin 29 if ($filter) {
194     no strict 'refs';
195 dpavlin 138 if ($display_data) {
196 dpavlin 153 $display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display)));
197 dpavlin 138 } else {
198 dpavlin 153 $display_data = join($delimiter,mkformat($x,&$filter($display)));
199 dpavlin 138 }
200 dpavlin 29 } else {
201 dpavlin 138 if ($display_data) {
202 dpavlin 153 $display_data .= $delimiter.mkformat($x,$display);
203 dpavlin 138 } else {
204 dpavlin 153 $display_data = mkformat($x,$display);
205 dpavlin 138 }
206 dpavlin 29 }
207 dpavlin 20 }
208 dpavlin 29
209     # type="index" ; insert into index
210     if ($i && $display) {
211 dpavlin 97 push @index_data, $display;
212     $index_filter = $filter if ($filter);
213     }
214     }
215    
216     # fill data in index
217     if (@index_data) {
218     if ($index_filter) {
219     no strict 'refs';
220 dpavlin 135 foreach my $d (@index_data) {
221     $index->insert($field, &$index_filter($d), $path);
222 dpavlin 20 }
223 dpavlin 97 } else {
224     foreach my $d (@index_data) {
225     $index->insert($field, $d, $path);
226     }
227 dpavlin 20 }
228 dpavlin 17 }
229 dpavlin 3 }
230 dpavlin 9
231 dpavlin 59 # now try to parse variables from configuration file
232     foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) {
233 dpavlin 13
234 dpavlin 62 my $delimiter = x($x->{delimiter}) || ' ';
235 dpavlin 59 my $val = $cfg->val($database, x($x->{content}));
236    
237     my ($s,$d,$i) = (1,1,0); # swish, display default
238     $s = 0 if (lc($x->{type}) eq "display");
239     $d = 0 if (lc($x->{type}) eq "swish");
240 dpavlin 163 # no support for swish exact in config.
241     # IMHO, it's useless
242 dpavlin 59 ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
243    
244     if ($val) {
245 dpavlin 62 $display_data .= $delimiter.$val if ($d);
246 dpavlin 59 $swish_data .= $val if ($s);
247     $index->insert($field, $val, $path) if ($i);
248     }
249    
250     }
251    
252    
253 dpavlin 10 if ($display_data) {
254 dpavlin 29
255 dpavlin 13 if ($field eq "headline") {
256     $xml .= xmlify("headline", $display_data);
257     } else {
258    
259     # find field name (signular, plural)
260     my $field_name = "";
261     if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) {
262     $field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
263     } elsif ($config->{indexer}->{$field}->{name_plural}) {
264     $field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
265 dpavlin 20 } elsif ($config->{indexer}->{$field}->{name}) {
266     $field_name = $config->{indexer}->{$field}->{name}."#-#";
267 dpavlin 13 } else {
268 dpavlin 20 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
269 dpavlin 13 }
270     if ($field_name) {
271 dpavlin 40 $html .= x($field_name);
272 dpavlin 13 }
273     $html .= $display_data."###\n";
274     }
275 dpavlin 10 }
276     if ($swish_data) {
277 dpavlin 20 # remove extra spaces
278     $swish_data =~ s/ +/ /g;
279     $swish_data =~ s/ +$//g;
280    
281 dpavlin 40 $xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
282 dpavlin 10 }
283    
284 dpavlin 163 if ($swish_exact_data) {
285     $swish_exact_data =~ s/ +/ /g;
286     $swish_exact_data =~ s/ +$//g;
287 dpavlin 9
288 dpavlin 163 # add delimiters before and after word.
289     # That is required to produce exact match
290     $xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx'));
291     }
292    
293    
294 dpavlin 3 }
295 dpavlin 13
296     # dump formatted output in <html>
297     if ($html) {
298 dpavlin 81 #$xml .= xmlify("html",$html);
299     $xml .= "<html><![CDATA[ $html ]]></html>";
300 dpavlin 13 }
301    
302 dpavlin 3 if ($xml) {
303 dpavlin 13 $xml .= $add_xml if ($add_xml);
304 dpavlin 10 return "<xml>\n$xml</xml>\n";
305 dpavlin 3 } else {
306     return;
307     }
308     }
309    
310     ##########################################################################
311    
312 dpavlin 54 # read configuration for this script
313 dpavlin 13 my $cfg = new Config::IniFiles( -file => $config_file );
314 dpavlin 1
315 dpavlin 54 # read global.conf configuration
316     my $cfg_global = new Config::IniFiles( -file => 'global.conf' );
317    
318 dpavlin 50 # open index
319     $index = new index_DBI(
320 dpavlin 54 $cfg_global->val('global', 'dbi_dbd'),
321     $cfg_global->val('global', 'dbi_dsn'),
322     $cfg_global->val('global', 'dbi_user'),
323     $cfg_global->val('global', 'dbi_passwd') || '',
324 dpavlin 50 );
325    
326 dpavlin 97 my $show_progress = $cfg_global->val('global', 'show_progress');
327    
328 dpavlin 164 my $unac_filter = $cfg_global->val('global', 'unac_filter');
329     if ($unac_filter) {
330     require $unac_filter;
331     }
332    
333 dpavlin 13 foreach my $database ($cfg->Sections) {
334 dpavlin 1
335 dpavlin 54 my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
336 dpavlin 40 my $add_xml = $cfg -> val($database, 'xml'); # optional
337 dpavlin 1
338 dpavlin 54 print STDERR "reading ./import_xml/$type.xml\n";
339 dpavlin 1
340 dpavlin 58 # extract just type basic
341     my $type_base = $type;
342     $type_base =~ s/_.+$//g;
343 dpavlin 40
344 dpavlin 153 $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config', 'format' ], forcecontent => 1);
345 dpavlin 58
346 dpavlin 54 # output current progress indicator
347     my $last_p = 0;
348     sub progress {
349 dpavlin 101 return if (! $show_progress);
350 dpavlin 54 my $current = shift;
351 dpavlin 57 my $total = shift || 1;
352 dpavlin 54 my $p = int($current * 100 / $total);
353     if ($p != $last_p) {
354     printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p );
355     $last_p = $p;
356     }
357 dpavlin 3 }
358    
359 dpavlin 67 my $fake_dir = 1;
360     sub fakeprogress {
361 dpavlin 104 return if (! $show_progress);
362 dpavlin 67 my $current = shift @_;
363    
364     my @ind = ('-','\\','|','/','-','\\','|','/', '-');
365    
366     $last_p += $fake_dir;
367     $fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0);
368     if ($last_p % 10 == 0) {
369     printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]);
370     }
371     }
372    
373 dpavlin 54 # now read database
374     print STDERR "using: $type...\n";
375 dpavlin 1
376 dpavlin 58 if ($type_base eq "isis") {
377    
378 dpavlin 54 my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!";
379 dpavlin 1
380 dpavlin 54 $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
381     my $db = OpenIsis::open( $isis_db );
382 dpavlin 10
383 dpavlin 106 # check if .txt database for OpenIsis is zero length,
384     # if so, erase it and re-open database
385     sub check_txt_db {
386     my $isis_db = shift || die "need isis database name";
387 dpavlin 109 my $reopen = 0;
388    
389 dpavlin 106 if (-e $isis_db.".TXT") {
390 dpavlin 109 print STDERR "WARNING: removing $isis_db.TXT OpenIsis database...\n";
391 dpavlin 108 unlink $isis_db.".TXT" || warn "FATAL: unlink error on '$isis_db.TXT': $!";
392 dpavlin 109 $reopen++;
393 dpavlin 106 }
394 dpavlin 109 if (-e $isis_db.".PTR") {
395     print STDERR "WARNING: removing $isis_db.PTR OpenIsis database...\n";
396     unlink $isis_db.".PTR" || warn "FATAL: unlink error on '$isis_db.PTR': $!";
397     $reopen++;
398     }
399     return OpenIsis::open( $isis_db ) if ($reopen);
400 dpavlin 106 }
401    
402     # EOF error
403     if ($db == -1) {
404     $db = check_txt_db($isis_db);
405 dpavlin 108 if ($db == -1) {
406 dpavlin 106 print STDERR "FATAL: OpenIsis can't open zero size file $isis_db\n";
407     next;
408     }
409     }
410    
411 dpavlin 101 # OpenIsis::ERR_BADF
412     if ($db == -4) {
413     print STDERR "FATAL: OpenIsis can't find file $isis_db\n";
414     next;
415     # OpenIsis::ERR_IO
416     } elsif ($db == -5) {
417     print STDERR "FATAL: OpenIsis can't access file $isis_db\n";
418     next;
419     } elsif ($db < 0) {
420     print STDERR "FATAL: OpenIsis unknown error $db with file $isis_db\n";
421     next;
422     }
423    
424 dpavlin 54 my $max_rowid = OpenIsis::maxRowid( $db );
425 dpavlin 3
426 dpavlin 106 # if 0 records, try to rease isis .txt database
427     if ($max_rowid == 0) {
428     # force removal of database
429     $db = check_txt_db($isis_db);
430     $max_rowid = OpenIsis::maxRowid( $db );
431     }
432    
433 dpavlin 54 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
434    
435     my $path = $database;
436    
437     for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
438     my $row = OpenIsis::read( $db, $row_id );
439     if ($row && $row->{mfn}) {
440    
441     progress($row->{mfn}, $max_rowid);
442    
443     my $swishpath = $path."#".int($row->{mfn});
444    
445 dpavlin 59 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
446 dpavlin 54 $xml = $cp2utf->convert($xml);
447     use bytes; # as opposed to chars
448     print "Path-Name: $swishpath\n";
449     print "Content-Length: ".(length($xml)+1)."\n";
450     print "Document-Type: XML\n\n$xml\n";
451     }
452 dpavlin 3 }
453 dpavlin 54 }
454 dpavlin 108 # for this to work with current version of OpenIsis (0.9.0)
455     # you might need my patch from
456     # http://www.rot13.org/~dpavlin/projects/openisis-0.9.0-perl_close.diff
457     OpenIsis::close($db);
458 dpavlin 54 print STDERR "\n";
459 dpavlin 3
460 dpavlin 58 } elsif ($type_base eq "excel") {
461 dpavlin 54 use Spreadsheet::ParseExcel;
462     use Spreadsheet::ParseExcel::Utility qw(int2col);
463    
464     $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
465     my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
466 dpavlin 43
467 dpavlin 54 my $sheet = x($config->{sheet}) || die "no sheet in $type.xml";
468 dpavlin 56 my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml";
469 dpavlin 54
470     my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'";
471    
472     my $sheet_nr = 0;
473     foreach my $oWks (@{$oBook->{Worksheet}}) {
474     #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n";
475     last if ($oWks->{Name} eq $sheet);
476     $sheet_nr++;
477     }
478    
479     my $oWorksheet = $oBook->{Worksheet}[$sheet_nr];
480    
481     print STDERR "using sheet: ",$oWorksheet->{Name},"\n";
482     defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file";
483     my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow};
484    
485 dpavlin 56 for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) {
486 dpavlin 54 my $row;
487     for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
488     my $cell = $oWorksheet->{Cells}[$iR][$iC];
489     if ($cell) {
490     $row->{int2col($iC)} = $cell->Value;
491     }
492     }
493    
494     progress($iR, $end_row);
495    
496     # print "row[$iR/$end_row] ";
497     # foreach (keys %{$row}) {
498     # print "$_: ",$row->{$_},"\t";
499     # }
500     # print "\n";
501    
502     my $swishpath = $database."#".$iR;
503    
504     next if (! $row);
505    
506 dpavlin 59 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
507 dpavlin 44 $xml = $cp2utf->convert($xml);
508 dpavlin 35 use bytes; # as opposed to chars
509 dpavlin 43 print "Path-Name: $swishpath\n";
510 dpavlin 10 print "Content-Length: ".(length($xml)+1)."\n";
511     print "Document-Type: XML\n\n$xml\n";
512 dpavlin 3 }
513 dpavlin 1 }
514 dpavlin 62 } elsif ($type_base eq "marc") {
515 dpavlin 67
516 dpavlin 62 use MARC;
517    
518     $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
519     my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
520    
521     # optional argument is format
522     my $format = x($config->{format}) || 'usmarc';
523    
524     print STDERR "Reading MARC file '$marc_file'\n";
525    
526     my $marc = new MARC;
527     my $nr = $marc->openmarc({
528     file=>$marc_file, format=>$format
529     }) || die "Can't open MARC file '$marc_file'";
530    
531     my $i=0; # record nr.
532    
533     my $rec;
534    
535     while ($marc->nextmarc(1)) {
536    
537     # XXX
538 dpavlin 67 fakeprogress($i++);
539 dpavlin 62
540     my $swishpath = $database."#".$i;
541    
542     if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) {
543     $xml = $cp2utf->convert($xml);
544     use bytes; # as opposed to chars
545     print "Path-Name: $swishpath\n";
546     print "Content-Length: ".(length($xml)+1)."\n";
547     print "Document-Type: XML\n\n$xml\n";
548     }
549     }
550 dpavlin 67 } elsif ($type_base eq "feed") {
551    
552     $import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);
553     my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!";
554    
555     print STDERR "Reading feed from program '$prog'\n";
556    
557     open(FEED,"feeds/$prog |") || die "can't start $prog: $!";
558    
559 dpavlin 74 my $i=1; # record nr.
560 dpavlin 67
561     my $data;
562     my $line=1;
563    
564     while (<FEED>) {
565     chomp;
566    
567     if (/^$/) {
568 dpavlin 74 my $swishpath = $database."#".$i++;
569 dpavlin 67
570     if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) {
571     $xml = $cp2utf->convert($xml);
572     use bytes; # as opposed to chars
573     print "Path-Name: $swishpath\n";
574     print "Content-Length: ".(length($xml)+1)."\n";
575     print "Document-Type: XML\n\n$xml\n";
576     }
577     $line = 1;
578     $data = {};
579     next;
580     }
581    
582 dpavlin 74 $line = $1 if (s/^(\d+):\s*//);
583 dpavlin 67 $data->{$line++} = $_;
584    
585 dpavlin 74 fakeprogress($i);
586 dpavlin 67
587     }
588 dpavlin 1 }
589     }
590 dpavlin 3
591 dpavlin 10 # call this to commit index
592     $index->close;
593 dpavlin 3
594     1;
595     __END__
596     ##########################################################################
597    
598     =head1 NAME
599    
600 dpavlin 81 all2xml.pl - read various file formats and dump XML for SWISH-E
601 dpavlin 3
602     =head1 DESCRIPTION
603    
604 dpavlin 81 This command will read ISIS data file using OpenIsis perl module, MARC
605     records using MARC module and optionally Micro$oft Excel files to
606     create one XML file for usage with I<SWISH-E> indexer. Dispite it's name,
607     this script B<isn't general xml generator> from isis files (isis allready
608     has something like that). Output of this script is tailor-made for SWISH-E.
609 dpavlin 3
610 dpavlin 81 =head1 BUGS
611    
612     Documentation is really lacking. However, in true Open Source spirit, source
613     is best documentation. I even made considerable effort to comment parts
614     which are not intuitively clear, so...
615    
616 dpavlin 3 =head1 AUTHOR
617    
618     Dobrica Pavlinusic <dpavlin@rot13.org>
619    
620     =head1 COPYRIGHT
621    
622     GNU Public License (GPL) v2 or later
623    
624     =head1 SEE ALSO
625    
626     SWISH-E web site at http://www.swish-e.org
627    
628     =cut

Properties

Name Value
cvs2svn:cvs-rev 1.46
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26