line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Parse::MediaWikiDump; |
2
|
|
|
|
|
|
|
our $VERSION = '1.0.6'; |
3
|
|
|
|
|
|
|
|
4
|
8
|
|
|
8
|
|
254192
|
use Parse::MediaWikiDump::XML; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
use Parse::MediaWikiDump::Revisions; |
6
|
|
|
|
|
|
|
use Parse::MediaWikiDump::Pages; |
7
|
|
|
|
|
|
|
use Parse::MediaWikiDump::page; |
8
|
|
|
|
|
|
|
use Parse::MediaWikiDump::Links; |
9
|
|
|
|
|
|
|
use Parse::MediaWikiDump::link; |
10
|
|
|
|
|
|
|
use Parse::MediaWikiDump::CategoryLinks; |
11
|
|
|
|
|
|
|
use Parse::MediaWikiDump::category_link; |
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
#the POD is at the end of this file |
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
sub new { |
16
|
|
|
|
|
|
|
my ($class) = @_; |
17
|
|
|
|
|
|
|
return bless({}, $class); |
18
|
|
|
|
|
|
|
} |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
sub pages { |
21
|
|
|
|
|
|
|
shift(@_); |
22
|
|
|
|
|
|
|
return Parse::MediaWikiDump::Pages->new(@_); |
23
|
|
|
|
|
|
|
} |
24
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
sub revisions { |
26
|
|
|
|
|
|
|
shift(@_); |
27
|
|
|
|
|
|
|
return Parse::MediaWikiDump::Revisions->new(@_); |
28
|
|
|
|
|
|
|
} |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
sub links { |
31
|
|
|
|
|
|
|
shift(@_); |
32
|
|
|
|
|
|
|
return Parse::MediaWikiDump::Links->new(@_); |
33
|
|
|
|
|
|
|
} |
34
|
|
|
|
|
|
|
|
35
|
|
|
|
|
|
|
#just a place holder for something that might be used in the future |
36
|
|
|
|
|
|
|
#package Parse::MediaWikiDump::ExternalLinks; |
37
|
|
|
|
|
|
|
# |
38
|
|
|
|
|
|
|
#use strict; |
39
|
|
|
|
|
|
|
#use warnings; |
40
|
|
|
|
|
|
|
# |
41
|
|
|
|
|
|
|
#sub new { |
42
|
|
|
|
|
|
|
# my ($class, $source) = @_; |
43
|
|
|
|
|
|
|
# my $self = {}; |
44
|
|
|
|
|
|
|
# |
45
|
|
|
|
|
|
|
# $$self{BUFFER} = []; |
46
|
|
|
|
|
|
|
# $$self{BYTE} = 0; |
47
|
|
|
|
|
|
|
# |
48
|
|
|
|
|
|
|
# bless($self, $class); |
49
|
|
|
|
|
|
|
# |
50
|
|
|
|
|
|
|
# $self->open($source); |
51
|
|
|
|
|
|
|
# $self->init; |
52
|
|
|
|
|
|
|
# |
53
|
|
|
|
|
|
|
# return $self; |
54
|
|
|
|
|
|
|
#} |
55
|
|
|
|
|
|
|
# |
56
|
|
|
|
|
|
|
#sub next { |
57
|
|
|
|
|
|
|
# my ($self) = @_; |
58
|
|
|
|
|
|
|
# my $buffer = $$self{BUFFER}; |
59
|
|
|
|
|
|
|
# my $link; |
60
|
|
|
|
|
|
|
# |
61
|
|
|
|
|
|
|
# while(1) { |
62
|
|
|
|
|
|
|
# if (defined($link = pop(@$buffer))) { |
63
|
|
|
|
|
|
|
# last; |
64
|
|
|
|
|
|
|
# } |
65
|
|
|
|
|
|
|
# |
66
|
|
|
|
|
|
|
# #signals end of input |
67
|
|
|
|
|
|
|
# return undef unless $self->parse_more; |
68
|
|
|
|
|
|
|
# } |
69
|
|
|
|
|
|
|
# |
70
|
|
|
|
|
|
|
# return Parse::MediaWikiDump::external_link->new($link); |
71
|
|
|
|
|
|
|
#} |
72
|
|
|
|
|
|
|
# |
73
|
|
|
|
|
|
|
##private functions with OO interface |
74
|
|
|
|
|
|
|
#sub parse_more { |
75
|
|
|
|
|
|
|
# my ($self) = @_; |
76
|
|
|
|
|
|
|
# my $source = $$self{SOURCE}; |
77
|
|
|
|
|
|
|
# my $need_data = 1; |
78
|
|
|
|
|
|
|
# |
79
|
|
|
|
|
|
|
# while($need_data) { |
80
|
|
|
|
|
|
|
# my $line = <$source>; |
81
|
|
|
|
|
|
|
# |
82
|
|
|
|
|
|
|
# last unless defined($line); |
83
|
|
|
|
|
|
|
# |
84
|
|
|
|
|
|
|
# $$self{BYTE} += length($line); |
85
|
|
|
|
|
|
|
# |
86
|
|
|
|
|
|
|
# while($line =~ m/\((\d+),'(.*?)','(.*?)'\)[;,]/g) { |
87
|
|
|
|
|
|
|
# push(@{$$self{BUFFER}}, [$1, $2, $3]); |
88
|
|
|
|
|
|
|
# $need_data = 0; |
89
|
|
|
|
|
|
|
# } |
90
|
|
|
|
|
|
|
# } |
91
|
|
|
|
|
|
|
# |
92
|
|
|
|
|
|
|
# #if we still need data and we are here it means we ran out of input |
93
|
|
|
|
|
|
|
# if ($need_data) { |
94
|
|
|
|
|
|
|
# return 0; |
95
|
|
|
|
|
|
|
# } |
96
|
|
|
|
|
|
|
# |
97
|
|
|
|
|
|
|
# return 1; |
98
|
|
|
|
|
|
|
#} |
99
|
|
|
|
|
|
|
# |
100
|
|
|
|
|
|
|
#sub open { |
101
|
|
|
|
|
|
|
# my ($self, $source) = @_; |
102
|
|
|
|
|
|
|
# |
103
|
|
|
|
|
|
|
# if (ref($source) ne 'GLOB') { |
104
|
|
|
|
|
|
|
# die "could not open $source: $!" unless |
105
|
|
|
|
|
|
|
# open($$self{SOURCE}, $source); |
106
|
|
|
|
|
|
|
# |
107
|
|
|
|
|
|
|
# $$self{SOURCE_FILE} = $source; |
108
|
|
|
|
|
|
|
# } else { |
109
|
|
|
|
|
|
|
# $$self{SOURCE} = $source; |
110
|
|
|
|
|
|
|
# } |
111
|
|
|
|
|
|
|
# |
112
|
|
|
|
|
|
|
# binmode($$self{SOURCE}, ':utf8'); |
113
|
|
|
|
|
|
|
# |
114
|
|
|
|
|
|
|
# return 1; |
115
|
|
|
|
|
|
|
#} |
116
|
|
|
|
|
|
|
# |
117
|
|
|
|
|
|
|
#sub init { |
118
|
|
|
|
|
|
|
# my ($self) = @_; |
119
|
|
|
|
|
|
|
# my $source = $$self{SOURCE}; |
120
|
|
|
|
|
|
|
# my $found = 0; |
121
|
|
|
|
|
|
|
# |
122
|
|
|
|
|
|
|
# while(<$source>) { |
123
|
|
|
|
|
|
|
# if (m/^LOCK TABLES `externallinks` WRITE;/) { |
124
|
|
|
|
|
|
|
# $found = 1; |
125
|
|
|
|
|
|
|
# last; |
126
|
|
|
|
|
|
|
# } |
127
|
|
|
|
|
|
|
# } |
128
|
|
|
|
|
|
|
# |
129
|
|
|
|
|
|
|
# die "not a MediaWiki link dump file" unless $found; |
130
|
|
|
|
|
|
|
#} |
131
|
|
|
|
|
|
|
# |
132
|
|
|
|
|
|
|
#sub current_byte { |
133
|
|
|
|
|
|
|
# my ($self) = @_; |
134
|
|
|
|
|
|
|
# |
135
|
|
|
|
|
|
|
# return $$self{BYTE}; |
136
|
|
|
|
|
|
|
#} |
137
|
|
|
|
|
|
|
# |
138
|
|
|
|
|
|
|
#sub size { |
139
|
|
|
|
|
|
|
# my ($self) = @_; |
140
|
|
|
|
|
|
|
# |
141
|
|
|
|
|
|
|
# return undef unless defined $$self{SOURCE_FILE}; |
142
|
|
|
|
|
|
|
# |
143
|
|
|
|
|
|
|
# my @stat = stat($$self{SOURCE_FILE}); |
144
|
|
|
|
|
|
|
# |
145
|
|
|
|
|
|
|
# return $stat[7]; |
146
|
|
|
|
|
|
|
#} |
147
|
|
|
|
|
|
|
# |
148
|
|
|
|
|
|
|
#package Parse::MediaWikiDump::external_link; |
149
|
|
|
|
|
|
|
# |
150
|
|
|
|
|
|
|
##you must pass in a fully populated link array reference |
151
|
|
|
|
|
|
|
#sub new { |
152
|
|
|
|
|
|
|
# my ($class, $self) = @_; |
153
|
|
|
|
|
|
|
# |
154
|
|
|
|
|
|
|
# bless($self, $class); |
155
|
|
|
|
|
|
|
# |
156
|
|
|
|
|
|
|
# return $self; |
157
|
|
|
|
|
|
|
#} |
158
|
|
|
|
|
|
|
# |
159
|
|
|
|
|
|
|
#sub from { |
160
|
|
|
|
|
|
|
# my ($self) = @_; |
161
|
|
|
|
|
|
|
# return $$self[0]; |
162
|
|
|
|
|
|
|
#} |
163
|
|
|
|
|
|
|
# |
164
|
|
|
|
|
|
|
#sub to { |
165
|
|
|
|
|
|
|
# my ($self) = @_; |
166
|
|
|
|
|
|
|
# return $$self[1]; |
167
|
|
|
|
|
|
|
#} |
168
|
|
|
|
|
|
|
# |
169
|
|
|
|
|
|
|
#sub index { |
170
|
|
|
|
|
|
|
# my ($self) = @_; |
171
|
|
|
|
|
|
|
# return $$self[2]; |
172
|
|
|
|
|
|
|
#} |
173
|
|
|
|
|
|
|
# |
174
|
|
|
|
|
|
|
#sub timestamp { |
175
|
|
|
|
|
|
|
# my ($self) = @_; |
176
|
|
|
|
|
|
|
# return $$self[3]; |
177
|
|
|
|
|
|
|
# |
178
|
|
|
|
|
|
|
|
179
|
|
|
|
|
|
|
|
180
|
|
|
|
|
|
|
1; |
181
|
|
|
|
|
|
|
|
182
|
|
|
|
|
|
|
__END__ |