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 |
|
|
67 |
foreach my $field (keys %{$config->{indexer}}) { |
foreach my $field (keys %{$config->{indexer}}) { |
68 |
|
|
69 |
$field_usage{$field}++; |
$field_usage{$field}++; |
70 |
|
|
71 |
my $swish_data = ""; |
my $swish_data = ""; |
72 |
my $display_data = ""; |
my $display_data = ""; |
73 |
|
my $line_delimiter = ""; |
74 |
|
|
75 |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
76 |
|
|
77 |
my $format = $x->{content}; |
my $format = $x->{content}; |
78 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
my $delimiter = $x->{delimiter} || ' '; |
|
$s = 0 if (lc($x->{type}) eq "display"); |
|
|
$d = 0 if (lc($x->{type}) eq "swish"); |
|
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
|
|
#print STDERR "## s: $s d: $d i: $i ## $format ##\n"; |
|
79 |
|
|
80 |
# FIX: this is ugly, UGLY, cludge string is returned |
# FIX: this is ugly, UGLY, cludge string is returned |
81 |
# in UTF8 encoding , but as if source charset |
# in UTF8 encoding , but as if source charset |
82 |
# is ISO8859-1 and not some other. This break other |
# is ISO8859-1 and not some other. This break other |
83 |
# character encodings, so we convert it first |
# character encodings, so we convert it first |
84 |
# back to ISO8859-1 |
# back to ISO8859-1 |
85 |
$format = $cludge_codepage->convert($format); |
$format = $xml_codepage->convert($format); |
86 |
|
$delimiter = $xml_codepage->convert($delimiter) if ($delimiter); |
87 |
|
|
88 |
|
my $isis_i = 0; # isis repeatable offset |
89 |
|
|
90 |
|
my ($s,$d,$i) = (1,1,0); # swish, display default |
91 |
|
$s = 0 if (lc($x->{type}) eq "display"); |
92 |
|
$d = 0 if (lc($x->{type}) eq "swish"); |
93 |
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
94 |
|
#print STDERR "## s: $s d: $d i: $i ## $format ##\n"; |
95 |
|
|
96 |
|
$display_data .= $line_delimiter if ($display_data && $display_data !~ /$line_delimiter$/); |
97 |
|
|
98 |
my ($swish,$display) = parse_format($format,$row); |
my ($swish,$display) = (1,1); |
99 |
|
|
100 |
|
while ($swish || $display) { |
101 |
|
($swish,$display) = parse_format($format,$row,$isis_i++); |
102 |
#print STDERR "s: $swish\nd: $display\n" if ($swish); |
#print STDERR "s: $swish\nd: $display\n" if ($swish); |
103 |
|
|
104 |
#print STDERR "swish: $swish<-- display: $display<--\n"; |
#print STDERR "swish: $swish<-- display: $display<--\n"; |
105 |
|
|
106 |
# filter="name" ; filter this field through |
# filter="name" ; filter this field through |
107 |
# filter/[name].pm |
# filter/[name].pm |
108 |
my $filter; |
my $filter = $x->{filter}; |
|
if ($x->{filter}) { |
|
|
$filter = "filter/".$x->{filter}.".pm"; |
|
|
require $filter; |
|
|
} |
|
|
# type="swish" ; field for swish |
|
|
if ($s && $swish) { |
|
109 |
if ($filter) { |
if ($filter) { |
110 |
$swish_data .= join(" ",&filter($swish)); |
require "filter/".$filter.".pm"; |
111 |
} else { |
} |
112 |
$swish_data .= $swish if ($s && $swish); |
# type="swish" ; field for swish |
113 |
|
if ($s && $swish) { |
114 |
|
if ($filter) { |
115 |
|
print STDERR "using filter '$filter'\n"; |
116 |
|
no strict 'refs'; |
117 |
|
$swish_data .= join(" ",&$filter($swish)); |
118 |
|
} else { |
119 |
|
$swish_data .= $swish; |
120 |
|
} |
121 |
} |
} |
|
} |
|
122 |
|
|
123 |
# type="display" ; field for display |
# type="display" ; field for display |
124 |
if ($d && $display) { |
if ($d && $display) { |
125 |
if ($filter) { |
if ($filter) { |
126 |
$display_data .= join(" ",&filter($display)); |
no strict 'refs'; |
127 |
} else { |
$display_data .= join($delimiter,&$filter($display)); |
128 |
$display_data .= $display if ($s && $display); |
} else { |
129 |
|
if ($display_data) { |
130 |
|
$display_data .= $delimiter.$display; |
131 |
|
} else { |
132 |
|
$display_data .= $display; |
133 |
|
} |
134 |
|
} |
135 |
} |
} |
136 |
} |
|
137 |
|
# type="index" ; insert into index |
138 |
# type="index" ; insert into index |
if ($i && $display) { |
139 |
if ($i && $display) { |
my $index_data = $index_codepage->convert($display) || $display; |
140 |
my $index_data = $index_codepage->convert($display) || $display; |
if ($filter) { |
141 |
if ($filter) { |
no strict 'refs'; |
142 |
foreach my $d (&filter($index_data)) { |
foreach my $d (&$filter($index_data)) { |
143 |
$index->insert($field, $d, $db_dir); |
$index->insert($field, $d, $db_dir); |
144 |
|
} |
145 |
|
} else { |
146 |
|
$index->insert($field, $index_data, $db_dir); |
147 |
} |
} |
|
} else { |
|
|
$index->insert($field, $index_data, $db_dir); |
|
148 |
} |
} |
149 |
} |
} |
150 |
|
if ($x->{append}) { |
151 |
|
$line_delimiter = ' '; |
152 |
|
} else { |
153 |
|
$line_delimiter = '<br/>'; |
154 |
|
} |
155 |
} |
} |
156 |
|
|
157 |
|
|
158 |
#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); |
159 |
if ($display_data) { |
if ($display_data) { |
160 |
|
|
161 |
|
# remove last <br> |
162 |
|
$display_data =~ s/$line_delimiter$//; |
163 |
|
|
164 |
$display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !"; |
$display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !"; |
165 |
# FIX: this is removed and replaced by html tag. |
# FIX: this is removed and replaced by html tag. |
166 |
#$xml .= xmlify($field."_display", $display_data); |
#$xml .= xmlify($field."_display", $display_data); |
182 |
} |
} |
183 |
if ($field_name) { |
if ($field_name) { |
184 |
$html .= $xml_codepage->convert($field_name); |
$html .= $xml_codepage->convert($field_name); |
185 |
|
# $html .= "-->".$field_name."<--"; |
186 |
} |
} |
187 |
$html .= $display_data."###\n"; |
$html .= $display_data."###\n"; |
188 |
} |
} |
228 |
|
|
229 |
# read configuration for this type |
# read configuration for this type |
230 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1); |
$config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1); |
231 |
$isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8'); |
$isis_codepage = Text::Iconv->new($config->{isis_codepage},$XML_CHARSET); |
232 |
$index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
$index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
233 |
$xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),'UTF8'); |
$xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),$XML_CHARSET); |
234 |
|
|
235 |
my $db = OpenIsis::open( $isis_db ); |
my $db = OpenIsis::open( $isis_db ); |
236 |
if (0) { |
if (0) { |