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