40 |
|
|
41 |
Text::Iconv->raise_error(1); # Conversion errors raise exceptions |
Text::Iconv->raise_error(1); # Conversion errors raise exceptions |
42 |
|
|
|
#my $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8'); |
|
|
#my $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
|
43 |
my $isis_codepage; |
my $isis_codepage; |
44 |
my $index_codepage; |
my $index_codepage; |
45 |
my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1'); |
my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1'); |
46 |
my $xml_codepage; |
my $xml_codepage; |
47 |
|
|
48 |
|
my $XML_CHARSET = 'UTF8'; |
49 |
|
|
50 |
|
|
51 |
sub isis2xml { |
sub isis2xml { |
52 |
|
|
53 |
use xmlify; |
use xmlify; |
63 |
|
|
64 |
my %field_usage; # counter for usage of each field |
my %field_usage; # counter for usage of each field |
65 |
|
|
66 |
foreach my $field (keys %{$config->{indexer}}) { |
# sort subrouting using order="" attribute |
67 |
|
sub by_order { |
68 |
|
return 0 if (! $config->{indexer}->{$a}->{order}); |
69 |
|
return 0 if (! $config->{indexer}->{$b}->{order}); |
70 |
|
|
71 |
|
return $config->{indexer}->{$a}->{order} <=> |
72 |
|
$config->{indexer}->{$b}->{order} ; |
73 |
|
} |
74 |
|
|
75 |
|
foreach my $field (sort by_order keys %{$config->{indexer}}) { |
76 |
|
|
77 |
$field_usage{$field}++; |
$field_usage{$field}++; |
78 |
|
|
79 |
my $swish_data = ""; |
my $swish_data = ""; |
80 |
my $display_data = ""; |
my $display_data = ""; |
81 |
|
my $line_delimiter = ""; |
82 |
|
|
83 |
|
my ($swish,$display); |
84 |
|
|
85 |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
86 |
|
|
87 |
my $format = $x->{content}; |
my $format = $x->{content}; |
88 |
|
my $delimiter = $x->{delimiter} || ' '; |
89 |
|
|
90 |
|
# FIX: this is ugly, UGLY, cludge string is returned |
91 |
|
# in UTF8 encoding , but as if source charset |
92 |
|
# is ISO8859-1 and not some other. This break other |
93 |
|
# character encodings, so we convert it first |
94 |
|
# back to ISO8859-1 |
95 |
|
$format = $xml_codepage->convert($format); |
96 |
|
$delimiter = $xml_codepage->convert($delimiter) if ($delimiter); |
97 |
|
|
98 |
|
my $isis_i = 0; # isis repeatable offset |
99 |
|
|
100 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
my ($s,$d,$i) = (1,1,0); # swish, display default |
101 |
$s = 0 if (lc($x->{type}) eq "display"); |
$s = 0 if (lc($x->{type}) eq "display"); |
102 |
$d = 0 if (lc($x->{type}) eq "swish"); |
$d = 0 if (lc($x->{type}) eq "swish"); |
103 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
104 |
#print STDERR "## s: $s d: $d i: $i ## $format ##\n"; |
#print STDERR "## s: $s d: $d i: $i ## $format ##\n"; |
105 |
|
|
106 |
$format = $cludge_codepage->convert($format); |
if ($x->{append}) { |
107 |
my ($swish,$display) = parse_format($format,$row); |
$line_delimiter = ' '; |
108 |
|
} else { |
109 |
|
$line_delimiter = '<br/>'; |
110 |
|
} |
111 |
|
|
112 |
|
$display_data .= $line_delimiter if ($display_data && $display); |
113 |
|
# init vars so that we go into while... |
114 |
|
($swish,$display) = (1,1); |
115 |
|
|
116 |
|
while ($swish || $display) { |
117 |
|
($swish,$display) = parse_format($format,$row,$isis_i++); |
118 |
#print STDERR "s: $swish\nd: $display\n" if ($swish); |
#print STDERR "s: $swish\nd: $display\n" if ($swish); |
119 |
|
|
120 |
#print STDERR "swish: $swish<-- display: $display<--\n"; |
#print STDERR "swish: $swish<-- display: $display<--\n"; |
121 |
# FIX: this is ugly, UGLY, cludge: OpenIsis return |
|
122 |
# UTF8 encoding of strings, but as if source charset |
# filter="name" ; filter this field through |
123 |
# is ISO8859-1 and not some other. This breaks our |
# filter/[name].pm |
124 |
# isis character encoding, so we convert it first |
my $filter = $x->{filter}; |
125 |
# back to ISO8859-1 (which can actually be different |
if ($filter) { |
126 |
# encoding in isis) |
require "filter/".$filter.".pm"; |
127 |
|
} |
128 |
$swish_data .= $swish if ($s && $swish); |
# type="swish" ; field for swish |
129 |
$display_data .= $display if ($d && $display); |
if ($s && $swish) { |
130 |
|
if ($filter) { |
131 |
# insert into index |
#print STDERR "using filter '$filter'\n"; |
132 |
if ($i && $display) { |
no strict 'refs'; |
133 |
my $index_data = $index_codepage->convert($display) || $display; |
$swish_data .= join(" ",&$filter($swish)); |
134 |
$index->insert($field, $index_data, $db_dir); |
} else { |
135 |
|
$swish_data .= $swish; |
136 |
|
} |
137 |
|
} |
138 |
|
|
139 |
|
# type="display" ; field for display |
140 |
|
if ($d && $display) { |
141 |
|
if ($filter) { |
142 |
|
no strict 'refs'; |
143 |
|
$display_data .= join($delimiter,&$filter($display)); |
144 |
|
} else { |
145 |
|
if ($display_data) { |
146 |
|
$display_data .= $delimiter.$display; |
147 |
|
} else { |
148 |
|
$display_data .= $display; |
149 |
|
} |
150 |
|
} |
151 |
|
} |
152 |
|
|
153 |
|
# type="index" ; insert into index |
154 |
|
if ($i && $display) { |
155 |
|
my $index_data = $index_codepage->convert($display) || $display; |
156 |
|
if ($filter) { |
157 |
|
no strict 'refs'; |
158 |
|
foreach my $d (&$filter($index_data)) { |
159 |
|
$index->insert($field, $d, $db_dir); |
160 |
|
} |
161 |
|
} else { |
162 |
|
$index->insert($field, $index_data, $db_dir); |
163 |
|
} |
164 |
|
} |
165 |
} |
} |
166 |
} |
} |
167 |
|
|
168 |
|
|
169 |
#print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data); |
#print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data); |
170 |
if ($display_data) { |
if ($display_data) { |
171 |
|
|
172 |
|
# remove last <br> |
173 |
|
$display_data =~ s/$line_delimiter$//; |
174 |
|
|
175 |
$display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !"; |
$display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !"; |
176 |
# FIX: this is removed and replaced by html tag. |
# FIX: this is removed and replaced by html tag. |
177 |
#$xml .= xmlify($field."_display", $display_data); |
#$xml .= xmlify($field."_display", $display_data); |
186 |
$field_name = $config->{indexer}->{$field}->{name_singular}."#-#"; |
$field_name = $config->{indexer}->{$field}->{name_singular}."#-#"; |
187 |
} elsif ($config->{indexer}->{$field}->{name_plural}) { |
} elsif ($config->{indexer}->{$field}->{name_plural}) { |
188 |
$field_name = $config->{indexer}->{$field}->{name_plural}."#-#"; |
$field_name = $config->{indexer}->{$field}->{name_plural}."#-#"; |
189 |
} else { |
} elsif ($config->{indexer}->{$field}->{name}) { |
190 |
$field_name = $config->{indexer}->{$field}->{name}."#-#"; |
$field_name = $config->{indexer}->{$field}->{name}."#-#"; |
191 |
|
} else { |
192 |
|
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
193 |
} |
} |
194 |
if ($field_name) { |
if ($field_name) { |
195 |
$html .= $xml_codepage->convert($field_name); |
$html .= $xml_codepage->convert($field_name); |
196 |
|
# $html .= "-->".$field_name."<--"; |
197 |
} |
} |
198 |
$html .= $display_data."###\n"; |
$html .= $display_data."###\n"; |
199 |
} |
} |
200 |
} |
} |
201 |
if ($swish_data) { |
if ($swish_data) { |
202 |
my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2'); |
my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2'); |
203 |
|
# remove extra spaces |
204 |
|
$swish_data =~ s/ +/ /g; |
205 |
|
$swish_data =~ s/ +$//g; |
206 |
|
|
207 |
$swish_data = $i->convert($swish_data); |
$swish_data = $i->convert($swish_data); |
208 |
$xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data)); |
$xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data)); |
209 |
#$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data; |
#$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data; |
239 |
|
|
240 |
# read configuration for this type |
# read configuration for this type |
241 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1); |
$config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1); |
242 |
$isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8'); |
$isis_codepage = Text::Iconv->new($config->{isis_codepage},$XML_CHARSET); |
243 |
$index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
$index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
244 |
$xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),'UTF8'); |
$xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),$XML_CHARSET); |
245 |
|
|
246 |
my $db = OpenIsis::open( $isis_db ); |
my $db = OpenIsis::open( $isis_db ); |
247 |
if (0) { |
if (0) { |