18 |
|
|
19 |
my $config; |
my $config; |
20 |
|
|
21 |
use index_DBI; # there is no other, right now ;-) |
#use index_DBI; # default DBI module for index |
22 |
|
use index_DBI_cache; # faster DBI module using memory cache |
23 |
my $index; |
my $index; |
24 |
|
|
25 |
my %opts; |
my %opts; |
34 |
|
|
35 |
my $path; # this is name of database |
my $path; # this is name of database |
36 |
|
|
37 |
Text::Iconv->raise_error(1); # Conversion errors raise exceptions |
Text::Iconv->raise_error(0); # Conversion errors don't raise exceptions |
38 |
|
|
39 |
# this is encoding of all files on disk, including import_xml/*.xml file and |
# 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! |
# filter/*.pm files! It will be used to store strings in perl internally! |
66 |
my $type = shift @_; |
my $type = shift @_; |
67 |
my $row = shift @_; |
my $row = shift @_; |
68 |
my $add_xml = shift @_; |
my $add_xml = shift @_; |
69 |
|
# needed to read values from configuration file |
70 |
|
my $cfg = shift @_; |
71 |
|
my $database = shift @_; |
72 |
|
|
73 |
my $xml; |
my $xml; |
74 |
|
|
173 |
} |
} |
174 |
} |
} |
175 |
|
|
176 |
|
# now try to parse variables from configuration file |
177 |
|
foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) { |
178 |
|
|
179 |
|
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 |
if ($display_data) { |
if ($display_data) { |
196 |
|
|
262 |
|
|
263 |
print STDERR "reading ./import_xml/$type.xml\n"; |
print STDERR "reading ./import_xml/$type.xml\n"; |
264 |
|
|
265 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type} ], forcecontent => 1); |
# extract just type basic |
266 |
|
my $type_base = $type; |
267 |
|
$type_base =~ s/_.+$//g; |
268 |
|
|
269 |
|
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1); |
270 |
|
|
271 |
# output current progress indicator |
# output current progress indicator |
272 |
my $last_p = 0; |
my $last_p = 0; |
273 |
sub progress { |
sub progress { |
274 |
#return if (! $opts{q}); # FIXME |
#return if (! $opts{q}); # FIXME |
275 |
my $current = shift; |
my $current = shift; |
276 |
my $total = shift; |
my $total = shift || 1; |
277 |
my $p = int($current * 100 / $total); |
my $p = int($current * 100 / $total); |
278 |
if ($p != $last_p) { |
if ($p != $last_p) { |
279 |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); |
284 |
# now read database |
# now read database |
285 |
print STDERR "using: $type...\n"; |
print STDERR "using: $type...\n"; |
286 |
|
|
287 |
if ($type eq "isis") { |
if ($type_base eq "isis") { |
288 |
|
|
289 |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |
290 |
|
|
291 |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
305 |
|
|
306 |
my $swishpath = $path."#".int($row->{mfn}); |
my $swishpath = $path."#".int($row->{mfn}); |
307 |
|
|
308 |
if (my $xml = data2xml($type,$row,$add_xml)) { |
if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) { |
309 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
310 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
311 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |
316 |
} |
} |
317 |
print STDERR "\n"; |
print STDERR "\n"; |
318 |
|
|
319 |
} elsif ($type eq "excel") { |
} elsif ($type_base eq "excel") { |
320 |
use Spreadsheet::ParseExcel; |
use Spreadsheet::ParseExcel; |
321 |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
322 |
|
|
324 |
my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!"; |
my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!"; |
325 |
|
|
326 |
my $sheet = x($config->{sheet}) || die "no sheet in $type.xml"; |
my $sheet = x($config->{sheet}) || die "no sheet in $type.xml"; |
327 |
my $start_row = x($config->{start_row}) || die "no start_row in $type.xml"; |
my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml"; |
328 |
|
|
329 |
my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'"; |
my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'"; |
330 |
|
|
341 |
defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file"; |
defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file"; |
342 |
my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow}; |
my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow}; |
343 |
|
|
344 |
for(my $iR = $oWorksheet->{MinRow} ; defined $end_row && $iR <= $end_row ; $iR++) { |
for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) { |
345 |
my $row; |
my $row; |
346 |
for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { |
for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { |
347 |
my $cell = $oWorksheet->{Cells}[$iR][$iC]; |
my $cell = $oWorksheet->{Cells}[$iR][$iC]; |
362 |
|
|
363 |
next if (! $row); |
next if (! $row); |
364 |
|
|
365 |
if (my $xml = data2xml($type,$row,$add_xml)) { |
if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) { |
366 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
367 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
368 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |