/[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 58 - (hide annotations)
Fri Jul 4 16:56:40 2003 UTC (16 years, 3 months ago) by dpavlin
File MIME type: text/plain
File size: 9895 byte(s)
support type and sub-types (in form type_subtype)

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     'excel' => 'column'
60     );
61 dpavlin 3
62 dpavlin 54 sub data2xml {
63    
64 dpavlin 10 use xmlify;
65    
66 dpavlin 54 my $type = shift @_;
67 dpavlin 3 my $row = shift @_;
68 dpavlin 13 my $add_xml = shift @_;
69 dpavlin 3
70     my $xml;
71    
72 dpavlin 10 use parse_format;
73 dpavlin 3
74 dpavlin 13 my $html = ""; # html formatted display output
75 dpavlin 10
76 dpavlin 13 my %field_usage; # counter for usage of each field
77    
78 dpavlin 32 # sort subrouting using order="" attribute
79     sub by_order {
80     return 0 if (! $config->{indexer}->{$a}->{order});
81     return 0 if (! $config->{indexer}->{$b}->{order});
82 dpavlin 29
83 dpavlin 32 return $config->{indexer}->{$a}->{order} <=>
84     $config->{indexer}->{$b}->{order} ;
85     }
86 dpavlin 3
87 dpavlin 32 foreach my $field (sort by_order keys %{$config->{indexer}}) {
88    
89 dpavlin 40 $field=x($field);
90    
91 dpavlin 13 $field_usage{$field}++;
92    
93 dpavlin 10 my $swish_data = "";
94 dpavlin 3 my $display_data = "";
95 dpavlin 35 my $line_delimiter;
96 dpavlin 3
97 dpavlin 34 my ($swish,$display);
98    
99 dpavlin 54 my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
100     foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
101 dpavlin 3
102 dpavlin 40 my $format = x($x->{content});
103     my $delimiter = x($x->{delimiter}) || ' ';
104 dpavlin 3
105 dpavlin 54 my $repeat_off = 0; # repeatable offset
106 dpavlin 29
107     my ($s,$d,$i) = (1,1,0); # swish, display default
108     $s = 0 if (lc($x->{type}) eq "display");
109     $d = 0 if (lc($x->{type}) eq "swish");
110     ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
111    
112 dpavlin 35 # what will separate last line from this one?
113     if ($display_data && $x->{append} && $x->{append} eq "1") {
114 dpavlin 34 $line_delimiter = ' ';
115 dpavlin 35 } elsif ($display_data) {
116 dpavlin 34 $line_delimiter = '<br/>';
117     }
118 dpavlin 29
119 dpavlin 34 # init vars so that we go into while...
120     ($swish,$display) = (1,1);
121 dpavlin 29
122 dpavlin 54 if ($swish || $display) {
123     ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
124 dpavlin 29 # filter="name" ; filter this field through
125     # filter/[name].pm
126     my $filter = $x->{filter};
127 dpavlin 20 if ($filter) {
128 dpavlin 29 require "filter/".$filter.".pm";
129 dpavlin 20 }
130 dpavlin 29 # type="swish" ; field for swish
131     if ($s && $swish) {
132     if ($filter) {
133     no strict 'refs';
134     $swish_data .= join(" ",&$filter($swish));
135     } else {
136     $swish_data .= $swish;
137     }
138     }
139 dpavlin 17
140 dpavlin 29 # type="display" ; field for display
141     if ($d && $display) {
142 dpavlin 35 if ($line_delimiter && $display_data) {
143     $display_data .= $line_delimiter;
144     undef $line_delimiter;
145     }
146 dpavlin 29 if ($filter) {
147     no strict 'refs';
148     $display_data .= join($delimiter,&$filter($display));
149     } else {
150     if ($display_data) {
151     $display_data .= $delimiter.$display;
152     } else {
153     $display_data .= $display;
154     }
155     }
156 dpavlin 20 }
157 dpavlin 29
158     # type="index" ; insert into index
159     if ($i && $display) {
160 dpavlin 40 my $index_data = $display;
161 dpavlin 29 if ($filter) {
162     no strict 'refs';
163     foreach my $d (&$filter($index_data)) {
164 dpavlin 43 $index->insert($field, $d, $path);
165 dpavlin 29 }
166     } else {
167 dpavlin 43 $index->insert($field, $index_data, $path);
168 dpavlin 20 }
169     }
170 dpavlin 17 }
171 dpavlin 3 }
172 dpavlin 9
173 dpavlin 13
174 dpavlin 10 if ($display_data) {
175 dpavlin 29
176 dpavlin 13 if ($field eq "headline") {
177     $xml .= xmlify("headline", $display_data);
178     } else {
179    
180     # find field name (signular, plural)
181     my $field_name = "";
182     if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) {
183     $field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
184     } elsif ($config->{indexer}->{$field}->{name_plural}) {
185     $field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
186 dpavlin 20 } elsif ($config->{indexer}->{$field}->{name}) {
187     $field_name = $config->{indexer}->{$field}->{name}."#-#";
188 dpavlin 13 } else {
189 dpavlin 20 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
190 dpavlin 13 }
191     if ($field_name) {
192 dpavlin 40 $html .= x($field_name);
193 dpavlin 13 }
194     $html .= $display_data."###\n";
195     }
196 dpavlin 10 }
197     if ($swish_data) {
198 dpavlin 20 # remove extra spaces
199     $swish_data =~ s/ +/ /g;
200     $swish_data =~ s/ +$//g;
201    
202 dpavlin 40 $xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
203 dpavlin 10 }
204    
205 dpavlin 9
206 dpavlin 3 }
207 dpavlin 13
208     # dump formatted output in <html>
209     if ($html) {
210     $xml .= xmlify("html",$html);
211     }
212    
213 dpavlin 3 if ($xml) {
214 dpavlin 13 $xml .= $add_xml if ($add_xml);
215 dpavlin 10 return "<xml>\n$xml</xml>\n";
216 dpavlin 3 } else {
217     return;
218     }
219     }
220    
221     ##########################################################################
222    
223 dpavlin 54 # read configuration for this script
224 dpavlin 13 my $cfg = new Config::IniFiles( -file => $config_file );
225 dpavlin 1
226 dpavlin 54 # read global.conf configuration
227     my $cfg_global = new Config::IniFiles( -file => 'global.conf' );
228    
229 dpavlin 50 # open index
230     $index = new index_DBI(
231 dpavlin 54 $cfg_global->val('global', 'dbi_dbd'),
232     $cfg_global->val('global', 'dbi_dsn'),
233     $cfg_global->val('global', 'dbi_user'),
234     $cfg_global->val('global', 'dbi_passwd') || '',
235 dpavlin 50 );
236    
237 dpavlin 13 foreach my $database ($cfg->Sections) {
238 dpavlin 1
239 dpavlin 54 my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
240 dpavlin 40 my $add_xml = $cfg -> val($database, 'xml'); # optional
241 dpavlin 1
242 dpavlin 54 print STDERR "reading ./import_xml/$type.xml\n";
243 dpavlin 1
244 dpavlin 58 # extract just type basic
245     my $type_base = $type;
246     $type_base =~ s/_.+$//g;
247 dpavlin 40
248 dpavlin 58 $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base} ], forcecontent => 1);
249    
250 dpavlin 54 # output current progress indicator
251     my $last_p = 0;
252     sub progress {
253     #return if (! $opts{q}); # FIXME
254     my $current = shift;
255 dpavlin 57 my $total = shift || 1;
256 dpavlin 54 my $p = int($current * 100 / $total);
257     if ($p != $last_p) {
258     printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p );
259     $last_p = $p;
260     }
261 dpavlin 3 }
262    
263 dpavlin 54 # now read database
264     print STDERR "using: $type...\n";
265 dpavlin 1
266 dpavlin 58 if ($type_base eq "isis") {
267    
268 dpavlin 54 my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!";
269 dpavlin 1
270 dpavlin 54 $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
271     my $db = OpenIsis::open( $isis_db );
272 dpavlin 10
273 dpavlin 54 my $max_rowid = OpenIsis::maxRowid( $db );
274 dpavlin 3
275 dpavlin 54 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
276    
277     my $path = $database;
278    
279     for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
280     my $row = OpenIsis::read( $db, $row_id );
281     if ($row && $row->{mfn}) {
282    
283     progress($row->{mfn}, $max_rowid);
284    
285     my $swishpath = $path."#".int($row->{mfn});
286    
287 dpavlin 58 if (my $xml = data2xml($type_base,$row,$add_xml)) {
288 dpavlin 54 $xml = $cp2utf->convert($xml);
289     use bytes; # as opposed to chars
290     print "Path-Name: $swishpath\n";
291     print "Content-Length: ".(length($xml)+1)."\n";
292     print "Document-Type: XML\n\n$xml\n";
293     }
294 dpavlin 3 }
295 dpavlin 54 }
296     print STDERR "\n";
297 dpavlin 3
298 dpavlin 58 } elsif ($type_base eq "excel") {
299 dpavlin 54 use Spreadsheet::ParseExcel;
300     use Spreadsheet::ParseExcel::Utility qw(int2col);
301    
302     $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
303     my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
304 dpavlin 43
305 dpavlin 54 my $sheet = x($config->{sheet}) || die "no sheet in $type.xml";
306 dpavlin 56 my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml";
307 dpavlin 54
308     my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'";
309    
310     my $sheet_nr = 0;
311     foreach my $oWks (@{$oBook->{Worksheet}}) {
312     #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n";
313     last if ($oWks->{Name} eq $sheet);
314     $sheet_nr++;
315     }
316    
317     my $oWorksheet = $oBook->{Worksheet}[$sheet_nr];
318    
319     print STDERR "using sheet: ",$oWorksheet->{Name},"\n";
320     defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file";
321     my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow};
322    
323 dpavlin 56 for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) {
324 dpavlin 54 my $row;
325     for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
326     my $cell = $oWorksheet->{Cells}[$iR][$iC];
327     if ($cell) {
328     $row->{int2col($iC)} = $cell->Value;
329     }
330     }
331    
332     progress($iR, $end_row);
333    
334     # print "row[$iR/$end_row] ";
335     # foreach (keys %{$row}) {
336     # print "$_: ",$row->{$_},"\t";
337     # }
338     # print "\n";
339    
340     my $swishpath = $database."#".$iR;
341    
342     next if (! $row);
343    
344     if (my $xml = data2xml($type,$row,$add_xml)) {
345 dpavlin 44 $xml = $cp2utf->convert($xml);
346 dpavlin 35 use bytes; # as opposed to chars
347 dpavlin 43 print "Path-Name: $swishpath\n";
348 dpavlin 10 print "Content-Length: ".(length($xml)+1)."\n";
349     print "Document-Type: XML\n\n$xml\n";
350 dpavlin 3 }
351 dpavlin 1 }
352     }
353     }
354 dpavlin 3
355 dpavlin 10 # call this to commit index
356     $index->close;
357 dpavlin 3
358     1;
359     __END__
360     ##########################################################################
361    
362     =head1 NAME
363    
364     isis2xml.pl - read isis file and dump XML
365    
366     =head1 DESCRIPTION
367    
368     This command will read ISIS data file using OpenIsis perl module and
369     create XML file for usage with I<SWISH-E>
370     indexer. Dispite it's name, this script B<isn't general xml generator>
371     from isis files (isis allready has something like that). Output of this
372     script is tailor-made for SWISH-E.
373    
374     =head1 AUTHOR
375    
376     Dobrica Pavlinusic <dpavlin@rot13.org>
377    
378     =head1 COPYRIGHT
379    
380     GNU Public License (GPL) v2 or later
381    
382     =head1 SEE ALSO
383    
384     SWISH-E web site at http://www.swish-e.org
385    
386     =cut

Properties

Name Value
cvs2svn:cvs-rev 1.24
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26