1 |
#------------------------------------------------------------- |
2 |
# |
3 |
# parse_format(...) |
4 |
# |
5 |
|
6 |
|
7 |
sub parse_format { |
8 |
my $type = shift || die "parset_format must be called with type!"; |
9 |
my $format = shift || die "parse_format must be called with format!"; |
10 |
my $row = shift || die "parse_format must be called with row!"; |
11 |
my $i = shift || 0; # isis repeatable number |
12 |
my $codepage = shift || die "parse_format must be called with codepage!"; |
13 |
if ($type eq "isis") { |
14 |
return parse_iso_format($format,$row,$i,$codepage,'isis_sf'); |
15 |
} elsif ($type eq "excel") { |
16 |
return parse_excel_format($format,$row,$i,$codepage); |
17 |
} elsif ($type eq "marc") { |
18 |
return parse_iso_format($format,$row,$i,$codepage,'marc_sf'); |
19 |
} elsif ($type eq "feed") { |
20 |
return parse_feed_format($format,$row,$i,$codepage); |
21 |
} |
22 |
} |
23 |
|
24 |
#------------------------------------------------------------- |
25 |
|
26 |
sub parse_iso_format { |
27 |
|
28 |
my $format = shift; |
29 |
my $row = shift; |
30 |
my $i = shift; |
31 |
my $codepage = shift; |
32 |
|
33 |
my $func = shift || die "need to know which sub-field function to use"; |
34 |
|
35 |
require $func.".pm"; |
36 |
|
37 |
my $out; |
38 |
my $out_swish; |
39 |
|
40 |
my $prefix = ""; |
41 |
if ($format =~ s/^([^\d]+)//) { |
42 |
$prefix = $1; |
43 |
} |
44 |
|
45 |
my $display; |
46 |
my $swish; |
47 |
|
48 |
sub cnv_cp { |
49 |
my $codepage = shift; |
50 |
my $tmp = shift || return; |
51 |
if ($codepage) { |
52 |
$tmp = $codepage->convert($tmp) || print STDERR "iso: '$tmp' can't convert\n"; |
53 |
} |
54 |
return $tmp; |
55 |
} |
56 |
|
57 |
while ($format) { |
58 |
#print STDERR "\n#### $format"; |
59 |
# this is EBSCO special to support numeric subfield in |
60 |
# form of 856#3 |
61 |
if ($format =~ s/^(\d\d\d)#*(\w?)//) { |
62 |
my $tmp = cnv_cp($codepage,&$func($row,$1,$2,$i)); |
63 |
if ($tmp) { |
64 |
$display .= $prefix.$tmp; |
65 |
$swish .= $tmp." "; |
66 |
#print STDERR " == $tmp"; |
67 |
} |
68 |
$prefix = ""; |
69 |
# this might be our local scpeciality -- fields 10 and 11 |
70 |
# (as opposed to 010 and 011) so they are strictly listed |
71 |
# here |
72 |
} elsif ($format =~ s/^(1[01])//) { |
73 |
my $tmp = cnv_cp($codepage,&$func($row,$1,undef,$i)); |
74 |
if ($tmp) { |
75 |
$display .= $prefix.$tmp; |
76 |
$swish .= $tmp." "; |
77 |
} |
78 |
$prefix = ""; |
79 |
} elsif ($format =~ s/^mfn//i) { |
80 |
$display .= $prefix . $row->{mfn}; |
81 |
$prefix = ""; |
82 |
} elsif ($format =~ s/^([^\d]+)(\d{0,3})/$2/) { |
83 |
$prefix .= $1 if ($display); |
84 |
} elsif ($format =~ s/^([^\d]+\d{0,2})//) { |
85 |
$prefix .= $1 if ($display); |
86 |
} elsif ($format =~ s/^(\d{1,2})//) { |
87 |
$prefix .= $1 if ($display); |
88 |
} else { |
89 |
print STDERR "unparsed format: $format\n"; |
90 |
$prefix .= $format; |
91 |
$format = ""; |
92 |
} |
93 |
} |
94 |
# add suffix |
95 |
$display .= $prefix if ($display); |
96 |
|
97 |
return ($swish,$display); |
98 |
} |
99 |
|
100 |
#------------------------------------------------------------- |
101 |
|
102 |
sub parse_excel_format { |
103 |
my $format = shift; |
104 |
my $row = shift; |
105 |
my $i = shift; |
106 |
my $codepage = shift; |
107 |
|
108 |
return if ($i > 0); # Excel doesn't support repeatable fields |
109 |
|
110 |
my $out; |
111 |
my $out_swish; |
112 |
|
113 |
my $prefix = ""; |
114 |
if ($format =~ s/^([^A-Z\|]{1,3})//) { |
115 |
$prefix = $1; |
116 |
} |
117 |
|
118 |
my $display; |
119 |
my $swish; |
120 |
|
121 |
while ($format && length($format) > 0) { |
122 |
#print STDERR "\n#### $format #"; |
123 |
if ($format =~ s/^\|([A-Z]{1,2})\|//) { |
124 |
#print STDERR "--$1-> $format -[",length($format),"] "; |
125 |
if ($row->{$1}) { |
126 |
my $tmp = $row->{$1}; |
127 |
if ($codepage) { |
128 |
$tmp = $codepage->convert($tmp) || warn "excel: $1 '$tmp' can't convert"; |
129 |
} |
130 |
$display .= $prefix . $tmp; |
131 |
$swish .= $tmp." "; |
132 |
#print STDERR " == $tmp"; |
133 |
} |
134 |
$prefix = ""; |
135 |
} elsif ($format =~ s/^([^A-Z\|]+)(\|[A-Z]{1,2}\|)/$2/) { |
136 |
$prefix .= $1 if ($display); |
137 |
} else { |
138 |
print STDERR "unparsed format: $format\n"; |
139 |
$prefix .= $format; |
140 |
$format = ""; |
141 |
} |
142 |
#print STDERR " display: $display swish: $swish [format: $format]"; |
143 |
} |
144 |
# add suffix |
145 |
$display .= $prefix if ($display); |
146 |
|
147 |
return ($swish,$display); |
148 |
} |
149 |
|
150 |
#------------------------------------------------------------- |
151 |
|
152 |
sub parse_feed_format { |
153 |
my $format = shift; |
154 |
my $data = shift; |
155 |
my $i = shift; |
156 |
my $codepage = shift; |
157 |
|
158 |
# XXX feed doesn't support repeatable fields, but they really |
159 |
# should, This is a bug. It should be fixed! |
160 |
return if ($i > 0); |
161 |
|
162 |
my $out; |
163 |
my $out_swish; |
164 |
|
165 |
my $prefix = ""; |
166 |
if ($format =~ s/^([^\d\|]{1,3})//) { |
167 |
$prefix = $1; |
168 |
} |
169 |
|
170 |
my $display; |
171 |
my $swish; |
172 |
|
173 |
while ($format && length($format) > 0) { |
174 |
#print STDERR "\n#### $format #"; |
175 |
if ($format =~ s/^\|(\d+)\|//) { |
176 |
#print STDERR "--$1-> $format -[",length($format),"] "; |
177 |
if ($data->{$1}) { |
178 |
my $tmp = $data->{$1}; |
179 |
if ($codepage) { |
180 |
$tmp = $codepage->convert($tmp) || warn "feed: $1 '$tmp' can't convert\n"; |
181 |
} |
182 |
$display .= $prefix . $tmp; |
183 |
$swish .= $tmp." "; |
184 |
#print STDERR " == $tmp"; |
185 |
} |
186 |
$prefix = ""; |
187 |
} elsif ($format =~ s/^([^\d\|]+)(\|\d+\|)/$2/) { |
188 |
$prefix .= $1 if ($display); |
189 |
} else { |
190 |
print STDERR "unparsed format: $format\n"; |
191 |
$prefix .= $format; |
192 |
$format = ""; |
193 |
} |
194 |
#print STDERR " display: $display swish: $swish [format: $format]"; |
195 |
} |
196 |
# add suffix |
197 |
$display .= $prefix if ($display); |
198 |
|
199 |
return ($swish,$display); |
200 |
} |
201 |
|
202 |
#------------------------------------------------------------- |
203 |
|
204 |
1; |