/[webpac]/trunk/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 /trunk/all2xml.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 62 - (hide annotations)
Fri Jul 4 20:11:48 2003 UTC (20 years, 8 months ago) by dpavlin
File MIME type: text/plain
File size: 11691 byte(s)
added MARC file import

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 54 );
62 dpavlin 3
63 dpavlin 54 sub data2xml {
64    
65 dpavlin 10 use xmlify;
66    
67 dpavlin 54 my $type = shift @_;
68 dpavlin 3 my $row = shift @_;
69 dpavlin 13 my $add_xml = shift @_;
70 dpavlin 59 # needed to read values from configuration file
71     my $cfg = shift @_;
72     my $database = shift @_;
73 dpavlin 3
74     my $xml;
75    
76 dpavlin 10 use parse_format;
77 dpavlin 3
78 dpavlin 13 my $html = ""; # html formatted display output
79 dpavlin 10
80 dpavlin 13 my %field_usage; # counter for usage of each field
81    
82 dpavlin 32 # sort subrouting using order="" attribute
83     sub by_order {
84     return 0 if (! $config->{indexer}->{$a}->{order});
85     return 0 if (! $config->{indexer}->{$b}->{order});
86 dpavlin 29
87 dpavlin 32 return $config->{indexer}->{$a}->{order} <=>
88     $config->{indexer}->{$b}->{order} ;
89     }
90 dpavlin 3
91 dpavlin 32 foreach my $field (sort by_order keys %{$config->{indexer}}) {
92    
93 dpavlin 40 $field=x($field);
94    
95 dpavlin 13 $field_usage{$field}++;
96    
97 dpavlin 10 my $swish_data = "";
98 dpavlin 3 my $display_data = "";
99 dpavlin 35 my $line_delimiter;
100 dpavlin 3
101 dpavlin 34 my ($swish,$display);
102    
103 dpavlin 54 my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
104     foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
105 dpavlin 3
106 dpavlin 40 my $format = x($x->{content});
107     my $delimiter = x($x->{delimiter}) || ' ';
108 dpavlin 3
109 dpavlin 54 my $repeat_off = 0; # repeatable offset
110 dpavlin 29
111     my ($s,$d,$i) = (1,1,0); # swish, display default
112     $s = 0 if (lc($x->{type}) eq "display");
113     $d = 0 if (lc($x->{type}) eq "swish");
114     ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
115    
116 dpavlin 35 # what will separate last line from this one?
117     if ($display_data && $x->{append} && $x->{append} eq "1") {
118 dpavlin 34 $line_delimiter = ' ';
119 dpavlin 35 } elsif ($display_data) {
120 dpavlin 34 $line_delimiter = '<br/>';
121     }
122 dpavlin 29
123 dpavlin 34 # init vars so that we go into while...
124     ($swish,$display) = (1,1);
125 dpavlin 29
126 dpavlin 54 if ($swish || $display) {
127     ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
128 dpavlin 29 # filter="name" ; filter this field through
129     # filter/[name].pm
130     my $filter = $x->{filter};
131 dpavlin 20 if ($filter) {
132 dpavlin 29 require "filter/".$filter.".pm";
133 dpavlin 20 }
134 dpavlin 29 # type="swish" ; field for swish
135     if ($s && $swish) {
136     if ($filter) {
137     no strict 'refs';
138     $swish_data .= join(" ",&$filter($swish));
139     } else {
140     $swish_data .= $swish;
141     }
142     }
143 dpavlin 17
144 dpavlin 29 # type="display" ; field for display
145     if ($d && $display) {
146 dpavlin 35 if ($line_delimiter && $display_data) {
147     $display_data .= $line_delimiter;
148     undef $line_delimiter;
149     }
150 dpavlin 29 if ($filter) {
151     no strict 'refs';
152     $display_data .= join($delimiter,&$filter($display));
153     } else {
154     if ($display_data) {
155     $display_data .= $delimiter.$display;
156     } else {
157     $display_data .= $display;
158     }
159     }
160 dpavlin 20 }
161 dpavlin 29
162     # type="index" ; insert into index
163     if ($i && $display) {
164 dpavlin 40 my $index_data = $display;
165 dpavlin 29 if ($filter) {
166     no strict 'refs';
167     foreach my $d (&$filter($index_data)) {
168 dpavlin 43 $index->insert($field, $d, $path);
169 dpavlin 29 }
170     } else {
171 dpavlin 43 $index->insert($field, $index_data, $path);
172 dpavlin 20 }
173     }
174 dpavlin 17 }
175 dpavlin 3 }
176 dpavlin 9
177 dpavlin 59 # now try to parse variables from configuration file
178     foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) {
179 dpavlin 13
180 dpavlin 62 my $delimiter = x($x->{delimiter}) || ' ';
181 dpavlin 59 my $val = $cfg->val($database, x($x->{content}));
182    
183     my ($s,$d,$i) = (1,1,0); # swish, display default
184     $s = 0 if (lc($x->{type}) eq "display");
185     $d = 0 if (lc($x->{type}) eq "swish");
186     ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
187    
188     if ($val) {
189 dpavlin 62 $display_data .= $delimiter.$val if ($d);
190 dpavlin 59 $swish_data .= $val if ($s);
191     $index->insert($field, $val, $path) if ($i);
192     }
193    
194     }
195    
196    
197 dpavlin 10 if ($display_data) {
198 dpavlin 29
199 dpavlin 13 if ($field eq "headline") {
200     $xml .= xmlify("headline", $display_data);
201     } else {
202    
203     # find field name (signular, plural)
204     my $field_name = "";
205     if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) {
206     $field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
207     } elsif ($config->{indexer}->{$field}->{name_plural}) {
208     $field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
209 dpavlin 20 } elsif ($config->{indexer}->{$field}->{name}) {
210     $field_name = $config->{indexer}->{$field}->{name}."#-#";
211 dpavlin 13 } else {
212 dpavlin 20 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
213 dpavlin 13 }
214     if ($field_name) {
215 dpavlin 40 $html .= x($field_name);
216 dpavlin 13 }
217     $html .= $display_data."###\n";
218     }
219 dpavlin 10 }
220     if ($swish_data) {
221 dpavlin 20 # remove extra spaces
222     $swish_data =~ s/ +/ /g;
223     $swish_data =~ s/ +$//g;
224    
225 dpavlin 40 $xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
226 dpavlin 10 }
227    
228 dpavlin 9
229 dpavlin 3 }
230 dpavlin 13
231     # dump formatted output in <html>
232     if ($html) {
233     $xml .= xmlify("html",$html);
234     }
235    
236 dpavlin 3 if ($xml) {
237 dpavlin 13 $xml .= $add_xml if ($add_xml);
238 dpavlin 10 return "<xml>\n$xml</xml>\n";
239 dpavlin 3 } else {
240     return;
241     }
242     }
243    
244     ##########################################################################
245    
246 dpavlin 54 # read configuration for this script
247 dpavlin 13 my $cfg = new Config::IniFiles( -file => $config_file );
248 dpavlin 1
249 dpavlin 54 # read global.conf configuration
250     my $cfg_global = new Config::IniFiles( -file => 'global.conf' );
251    
252 dpavlin 50 # open index
253     $index = new index_DBI(
254 dpavlin 54 $cfg_global->val('global', 'dbi_dbd'),
255     $cfg_global->val('global', 'dbi_dsn'),
256     $cfg_global->val('global', 'dbi_user'),
257     $cfg_global->val('global', 'dbi_passwd') || '',
258 dpavlin 50 );
259    
260 dpavlin 13 foreach my $database ($cfg->Sections) {
261 dpavlin 1
262 dpavlin 54 my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
263 dpavlin 40 my $add_xml = $cfg -> val($database, 'xml'); # optional
264 dpavlin 1
265 dpavlin 54 print STDERR "reading ./import_xml/$type.xml\n";
266 dpavlin 1
267 dpavlin 58 # extract just type basic
268     my $type_base = $type;
269     $type_base =~ s/_.+$//g;
270 dpavlin 40
271 dpavlin 59 $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1);
272 dpavlin 58
273 dpavlin 54 # output current progress indicator
274     my $last_p = 0;
275     sub progress {
276     #return if (! $opts{q}); # FIXME
277     my $current = shift;
278 dpavlin 57 my $total = shift || 1;
279 dpavlin 54 my $p = int($current * 100 / $total);
280     if ($p != $last_p) {
281     printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p );
282     $last_p = $p;
283     }
284 dpavlin 3 }
285    
286 dpavlin 54 # now read database
287     print STDERR "using: $type...\n";
288 dpavlin 1
289 dpavlin 58 if ($type_base eq "isis") {
290    
291 dpavlin 54 my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!";
292 dpavlin 1
293 dpavlin 54 $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
294     my $db = OpenIsis::open( $isis_db );
295 dpavlin 10
296 dpavlin 54 my $max_rowid = OpenIsis::maxRowid( $db );
297 dpavlin 3
298 dpavlin 54 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
299    
300     my $path = $database;
301    
302     for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
303     my $row = OpenIsis::read( $db, $row_id );
304     if ($row && $row->{mfn}) {
305    
306     progress($row->{mfn}, $max_rowid);
307    
308     my $swishpath = $path."#".int($row->{mfn});
309    
310 dpavlin 59 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
311 dpavlin 54 $xml = $cp2utf->convert($xml);
312     use bytes; # as opposed to chars
313     print "Path-Name: $swishpath\n";
314     print "Content-Length: ".(length($xml)+1)."\n";
315     print "Document-Type: XML\n\n$xml\n";
316     }
317 dpavlin 3 }
318 dpavlin 54 }
319     print STDERR "\n";
320 dpavlin 3
321 dpavlin 58 } elsif ($type_base eq "excel") {
322 dpavlin 54 use Spreadsheet::ParseExcel;
323     use Spreadsheet::ParseExcel::Utility qw(int2col);
324    
325     $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
326     my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
327 dpavlin 43
328 dpavlin 54 my $sheet = x($config->{sheet}) || die "no sheet in $type.xml";
329 dpavlin 56 my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml";
330 dpavlin 54
331     my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'";
332    
333     my $sheet_nr = 0;
334     foreach my $oWks (@{$oBook->{Worksheet}}) {
335     #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n";
336     last if ($oWks->{Name} eq $sheet);
337     $sheet_nr++;
338     }
339    
340     my $oWorksheet = $oBook->{Worksheet}[$sheet_nr];
341    
342     print STDERR "using sheet: ",$oWorksheet->{Name},"\n";
343     defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file";
344     my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow};
345    
346 dpavlin 56 for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) {
347 dpavlin 54 my $row;
348     for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
349     my $cell = $oWorksheet->{Cells}[$iR][$iC];
350     if ($cell) {
351     $row->{int2col($iC)} = $cell->Value;
352     }
353     }
354    
355     progress($iR, $end_row);
356    
357     # print "row[$iR/$end_row] ";
358     # foreach (keys %{$row}) {
359     # print "$_: ",$row->{$_},"\t";
360     # }
361     # print "\n";
362    
363     my $swishpath = $database."#".$iR;
364    
365     next if (! $row);
366    
367 dpavlin 59 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
368 dpavlin 44 $xml = $cp2utf->convert($xml);
369 dpavlin 35 use bytes; # as opposed to chars
370 dpavlin 43 print "Path-Name: $swishpath\n";
371 dpavlin 10 print "Content-Length: ".(length($xml)+1)."\n";
372     print "Document-Type: XML\n\n$xml\n";
373 dpavlin 3 }
374 dpavlin 1 }
375 dpavlin 62 } elsif ($type_base eq "marc") {
376     ## XXX
377     use MARC;
378    
379     $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
380     my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
381    
382     # optional argument is format
383     my $format = x($config->{format}) || 'usmarc';
384    
385     my %id_stored; # to aviod duplicates
386    
387     print STDERR "Reading MARC file '$marc_file'\n";
388    
389     my $marc = new MARC;
390     my $nr = $marc->openmarc({
391     file=>$marc_file, format=>$format
392     }) || die "Can't open MARC file '$marc_file'";
393    
394     my $i=0; # record nr.
395     my $inc=1;
396     my $max_i=1000;
397    
398     my $rec;
399    
400     while ($marc->nextmarc(1)) {
401    
402     # XXX
403     progress($i, $max_i);
404     $i += $inc;
405     $inc = -$inc if ($i > $max_i || $i < 0);
406    
407     my $swishpath = $database."#".$i;
408    
409     if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) {
410     $xml = $cp2utf->convert($xml);
411     use bytes; # as opposed to chars
412     print "Path-Name: $swishpath\n";
413     print "Content-Length: ".(length($xml)+1)."\n";
414     print "Document-Type: XML\n\n$xml\n";
415     }
416     }
417 dpavlin 1 }
418     }
419 dpavlin 3
420 dpavlin 10 # call this to commit index
421     $index->close;
422 dpavlin 3
423     1;
424     __END__
425     ##########################################################################
426    
427     =head1 NAME
428    
429     isis2xml.pl - read isis file and dump XML
430    
431     =head1 DESCRIPTION
432    
433     This command will read ISIS data file using OpenIsis perl module and
434     create XML file for usage with I<SWISH-E>
435     indexer. Dispite it's name, this script B<isn't general xml generator>
436     from isis files (isis allready has something like that). Output of this
437     script is tailor-made for SWISH-E.
438    
439     =head1 AUTHOR
440    
441     Dobrica Pavlinusic <dpavlin@rot13.org>
442    
443     =head1 COPYRIGHT
444    
445     GNU Public License (GPL) v2 or later
446    
447     =head1 SEE ALSO
448    
449     SWISH-E web site at http://www.swish-e.org
450    
451     =cut

Properties

Name Value
cvs2svn:cvs-rev 1.26
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26