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