1 |
#!/usr/bin/perl |
2 |
|
3 |
use warnings; |
4 |
use strict; |
5 |
|
6 |
use Text::CSV_XS; |
7 |
use Text::CSV::Separator qw(get_separator); |
8 |
use Carp qw/confess/; |
9 |
use LWP::Simple; |
10 |
use Number::Bytes::Human qw/format_bytes/; |
11 |
|
12 |
use JSON::Syck; |
13 |
use Data::Dump qw/dump/; |
14 |
|
15 |
$|++; |
16 |
|
17 |
my $csv_path = 'links.csv'; |
18 |
my $img_path = 'img'; |
19 |
my $first_line_labels = 1; |
20 |
my $split_fields = { |
21 |
label => sub { return split(/,\s*/,$_[0]) }, |
22 |
}; |
23 |
|
24 |
my @char_list = get_separator( path => $csv_path ); |
25 |
|
26 |
my $separator; |
27 |
if (@char_list) { |
28 |
if (@char_list == 1) { |
29 |
$separator = $char_list[0]; |
30 |
} else { |
31 |
$separator = $char_list[0]; |
32 |
} |
33 |
} else { |
34 |
die "Couldn't detect the field separator.\n"; |
35 |
} |
36 |
|
37 |
warn "Separator: $separator\n"; |
38 |
|
39 |
my $csv_parser = Text::CSV_XS->new({ |
40 |
sep_char => $separator, |
41 |
# binary => '1', |
42 |
# always_quote => '1' |
43 |
}); |
44 |
|
45 |
open my $csv_fh, '<', $csv_path; |
46 |
|
47 |
my @dump; |
48 |
|
49 |
my @labels; |
50 |
|
51 |
my $split_stats; |
52 |
|
53 |
my $multiplier = { |
54 |
kb => 1024, |
55 |
mb => 1024 * 1024, |
56 |
gb => 1024 * 1024 * 1024, |
57 |
}; |
58 |
|
59 |
my $multiplier_regex = join('|',keys %$multiplier); |
60 |
|
61 |
sub clean { |
62 |
my @out; |
63 |
foreach my $l ( @_ ) { |
64 |
my $o = $l; |
65 |
$l =~ s/^(['"])(.*)\1/$2/; |
66 |
$l =~ s/^\s+//s; |
67 |
$l =~ s/\s+$//s; |
68 |
push @out, $l; |
69 |
warn "clean '$o' -> '$l'\n" if ( $o ne $l ); |
70 |
} |
71 |
return @out if wantarray; |
72 |
return shift @out; |
73 |
} |
74 |
|
75 |
sub human { |
76 |
my $s = shift; |
77 |
|
78 |
if ( $s =~ m/(\d+)\s*($multiplier_regex)/i) { |
79 |
my ( $v, $m ) = ( $1, lc($2) ); |
80 |
my $factor = $multiplier->{$m}; |
81 |
confess "can't find multiplier $m" unless defined $factor; |
82 |
my $new = format_bytes( $v * $factor, bs => 1024 ); |
83 |
warn "## [$s] $v * $factor ($m) -> $new\n"; |
84 |
return $new; |
85 |
} |
86 |
return; |
87 |
} |
88 |
|
89 |
while (<$csv_fh>) { |
90 |
$csv_parser->parse($_); |
91 |
my @fields = $csv_parser->fields; |
92 |
|
93 |
if ( $first_line_labels && $. == 1 ) { |
94 |
@labels = @fields; |
95 |
next; |
96 |
} |
97 |
|
98 |
my $h; |
99 |
foreach my $i ( 0 .. $#fields ) { |
100 |
my $l = $labels[$i]; |
101 |
die "no label for field $i '$fields[$i]'" unless $l; |
102 |
|
103 |
my $v = clean( $fields[$i] ); |
104 |
# FIXME reject some values? |
105 |
|
106 |
$h->{ $l } = $v; |
107 |
|
108 |
if ( my $split = $split_fields->{$l} ) { |
109 |
confess "expected CODE for \$split_files->{$l}" unless ref($split) eq 'CODE'; |
110 |
|
111 |
my @sv = $split->( $v ); |
112 |
|
113 |
# warn "sv = ",dump( @sv ); |
114 |
|
115 |
foreach my $j ( 0 .. $#sv ) { |
116 |
|
117 |
my $v = clean( $sv[$j] ); |
118 |
|
119 |
if ( $j == 0 ) { |
120 |
$h->{ $l . '_short' } = $v; |
121 |
} |
122 |
|
123 |
if ( my $human = human( $v ) ) { |
124 |
$h->{ $l . '_' . $j . '_human' } = $human; |
125 |
} else { |
126 |
$h->{ $l . '_' . $j } = $v; |
127 |
} |
128 |
|
129 |
$split_stats->{$v}->{$j}++; |
130 |
$split_stats->{$v}->{sum}++; |
131 |
push @{ $split_stats->{$v}->{rec}->{$#dump + 1} }, $j; |
132 |
} |
133 |
} |
134 |
} |
135 |
warn "\nRecord #$. ",dump($h),"\n"; |
136 |
|
137 |
my $id = $h->{id}; |
138 |
|
139 |
if ( ! defined($id) || $id eq '' ) { |
140 |
warn "## skipped: $_"; |
141 |
next; |
142 |
} |
143 |
|
144 |
my $url = "http://www.links.hr/photo/big/$id.jpg"; |
145 |
my $img_thumb_path = "$img_path/t/$id.jpg"; |
146 |
my $img_orig_path = "$img_path/$id.jpg"; |
147 |
|
148 |
if ( mirror( $url, $img_orig_path ) != RC_NOT_MODIFIED ) { |
149 |
warn "$url -> $img_orig_path\n"; |
150 |
} |
151 |
system('convert', '-geometry', '320x200', $img_orig_path, $img_thumb_path ) if ! -e $img_thumb_path; |
152 |
|
153 |
$h->{'image-url'} = $img_orig_path; |
154 |
$h->{'image-thumb-url'} = $img_thumb_path; |
155 |
|
156 |
push @dump, $h; |
157 |
} |
158 |
|
159 |
close $csv_fh; |
160 |
|
161 |
foreach my $v ( keys %$split_stats ) { |
162 |
|
163 |
if ( $split_stats->{$v}->{sum} == 1 ) { |
164 |
delete( $split_stats->{$v} ); |
165 |
next; |
166 |
} |
167 |
|
168 |
foreach my $i ( keys %{ $split_stats->{$v}->{rec} } ) { |
169 |
push @{ $dump[ $i ]->{feature} }, $v; |
170 |
} |
171 |
} |
172 |
|
173 |
#warn "split_stats = ", dump( $split_stats ), "\n"; |
174 |
|
175 |
warn "dump = ", dump( @dump ), "\n"; |
176 |
|
177 |
print "features: .", join(', .', keys %$split_stats), "\n"; |
178 |
|
179 |
my $js_path = $csv_path; |
180 |
$js_path =~ s/\.csv/.js/gi; |
181 |
|
182 |
open my $fh, '>', $js_path || die "can't open $js_path: $!"; |
183 |
print $fh JSON::Syck::Dump( { items => \@dump } ); |
184 |
close $fh; |
185 |
|