| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package WWW::YahooJapan::Baseball::Parser; |
|
2
|
|
|
|
|
|
|
|
|
3
|
2
|
|
|
2
|
|
43279
|
use Web::Scraper; |
|
|
2
|
|
|
|
|
170946
|
|
|
|
2
|
|
|
|
|
16
|
|
|
4
|
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
sub parse_games_page { |
|
6
|
0
|
|
|
0
|
0
|
0
|
my $date = shift; |
|
7
|
0
|
|
|
|
|
0
|
my $league = shift; |
|
8
|
0
|
|
|
|
|
0
|
my %params = @_; |
|
9
|
|
|
|
|
|
|
my $day_scraper = scraper { |
|
10
|
0
|
|
|
0
|
|
0
|
process '//*[@id="gm_sch"]/div[contains(@class, "' . $league . '")]/following-sibling::div[position() <= 2 and contains(@class, "NpbScoreBg")]//a[starts-with(@href, "/npb/game/' . $date . '") and not(contains(@href, "/top"))]', 'uris[]' => '@href'; |
|
11
|
0
|
|
|
|
|
0
|
}; |
|
12
|
0
|
0
|
|
|
|
0
|
my $res = $day_scraper->scrape(defined $params{html} ? ($params{html}, $params{uri}) : $params{uri}); |
|
13
|
0
|
|
|
|
|
0
|
@{$res->{uris}}; |
|
|
0
|
|
|
|
|
0
|
|
|
14
|
|
|
|
|
|
|
} |
|
15
|
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
sub parse_game_player_row { |
|
17
|
31
|
|
|
31
|
0
|
42
|
my $cells = shift; |
|
18
|
31
|
|
|
|
|
38
|
my $stats = {}; |
|
19
|
|
|
|
|
|
|
|
|
20
|
31
|
|
|
|
|
230
|
my %position_table = ( |
|
21
|
|
|
|
|
|
|
'投' => 'p', |
|
22
|
|
|
|
|
|
|
'捕' => 'c', |
|
23
|
|
|
|
|
|
|
'一' => '1b', |
|
24
|
|
|
|
|
|
|
'二' => '2b', |
|
25
|
|
|
|
|
|
|
'三' => '3b', |
|
26
|
|
|
|
|
|
|
'遊' => 'ss', |
|
27
|
|
|
|
|
|
|
'左' => 'lf', |
|
28
|
|
|
|
|
|
|
'中' => 'cf', |
|
29
|
|
|
|
|
|
|
'右' => 'rf', |
|
30
|
|
|
|
|
|
|
'指' => 'dh', |
|
31
|
|
|
|
|
|
|
'打' => 'ph', |
|
32
|
|
|
|
|
|
|
'走' => 'pr' |
|
33
|
|
|
|
|
|
|
); |
|
34
|
31
|
|
|
|
|
42
|
my @positions = (); |
|
35
|
31
|
|
|
|
|
53
|
my $pos_rep = shift @$cells; |
|
36
|
31
|
100
|
|
|
|
119
|
if ($pos_rep =~ /^\((.*)\)$/) { |
|
37
|
20
|
|
|
|
|
23
|
my $i = 0; |
|
38
|
20
|
|
|
|
|
63
|
for my $p (split //, $1) { |
|
39
|
|
|
|
|
|
|
push(@positions, { |
|
40
|
72
|
100
|
|
|
|
265
|
position => $position_table{$p}, |
|
41
|
|
|
|
|
|
|
is_starting => $i++ > 0 ? 0 : 1 |
|
42
|
|
|
|
|
|
|
}); |
|
43
|
|
|
|
|
|
|
} |
|
44
|
|
|
|
|
|
|
} |
|
45
|
|
|
|
|
|
|
else { |
|
46
|
11
|
|
|
|
|
28
|
for my $p (split //, $pos_rep) { |
|
47
|
|
|
|
|
|
|
push(@positions, { |
|
48
|
33
|
|
|
|
|
90
|
position => $position_table{$p}, |
|
49
|
|
|
|
|
|
|
is_starting => 0 |
|
50
|
|
|
|
|
|
|
}); |
|
51
|
|
|
|
|
|
|
} |
|
52
|
|
|
|
|
|
|
} |
|
53
|
31
|
|
|
|
|
75
|
$stats->{positions} = \@positions; |
|
54
|
|
|
|
|
|
|
|
|
55
|
31
|
|
|
|
|
47
|
my $player_name = shift @$cells; |
|
56
|
|
|
|
|
|
|
|
|
57
|
31
|
|
|
|
|
88
|
my @indexes = qw(avg ab r h rbi k bbhbp shsf sb e hr); |
|
58
|
31
|
|
|
|
|
44
|
for my $i (@indexes) { |
|
59
|
341
|
|
|
|
|
623
|
$stats->{$i} = shift @$cells; |
|
60
|
|
|
|
|
|
|
} |
|
61
|
31
|
|
|
|
|
38
|
my $bi = 1; |
|
62
|
31
|
|
|
|
|
53
|
$stats->{innings} = {}; |
|
63
|
31
|
|
|
|
|
47
|
for my $bat (@$cells) { |
|
64
|
279
|
100
|
|
|
|
802
|
$stats->{innings}->{$bi++} = $bat ne '' ? [$bat] : []; |
|
65
|
|
|
|
|
|
|
} |
|
66
|
31
|
|
|
|
|
160
|
return $player_name, $stats; |
|
67
|
|
|
|
|
|
|
} |
|
68
|
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
sub parse_game_stats_page { |
|
70
|
1
|
|
|
1
|
0
|
5725
|
my %params = @_; |
|
71
|
|
|
|
|
|
|
my $stats_scraper = scraper { |
|
72
|
|
|
|
|
|
|
process '//*[@id="st_batth" or @id="st_battv"]//tr', 'lines[]' => scraper { |
|
73
|
35
|
|
|
|
|
1131445
|
process '//td', 'cells[]' => 'TEXT'; |
|
74
|
35
|
|
|
|
|
164643
|
process_first '//a[contains(@href, "/npb/player")]', 'player_uri' => '@href'; |
|
75
|
1
|
|
|
1
|
|
603578
|
}; |
|
76
|
1
|
|
|
|
|
33
|
}; |
|
77
|
1
|
50
|
|
|
|
59
|
my $res = $stats_scraper->scrape(defined $params{html} ? ($params{html}, $params{uri}) : $params{uri}); |
|
78
|
1
|
|
|
|
|
21172
|
my @players = (); |
|
79
|
1
|
|
|
|
|
3
|
for my $line (@{$res->{lines}}) { |
|
|
1
|
|
|
|
|
4
|
|
|
80
|
35
|
|
|
|
|
63
|
my $cells = $line->{cells}; |
|
81
|
35
|
100
|
66
|
|
|
290
|
unless ($cells and $line->{player_uri}) { |
|
82
|
4
|
|
|
|
|
7
|
next; |
|
83
|
|
|
|
|
|
|
} |
|
84
|
31
|
|
|
|
|
282
|
my ($player_name, $player_stats) = WWW::YahooJapan::Baseball::Parser::parse_game_player_row($cells); |
|
85
|
|
|
|
|
|
|
$player_stats->{player} = { |
|
86
|
|
|
|
|
|
|
name => $player_name, |
|
87
|
|
|
|
|
|
|
uri => $line->{player_uri}, |
|
88
|
|
|
|
|
|
|
$line->{player_uri}->query_form |
|
89
|
31
|
|
|
|
|
110
|
}; |
|
90
|
31
|
|
|
|
|
1067
|
push(@players, $player_stats); |
|
91
|
|
|
|
|
|
|
} |
|
92
|
1
|
|
|
|
|
54
|
@players; |
|
93
|
|
|
|
|
|
|
} |
|
94
|
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
1; |