| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
# Copyright (c) 2004, 2005 Anders Ardö |
|
2
|
|
|
|
|
|
|
## $Id: DataBase.pm 326 2011-05-27 07:44:58Z it-aar $ |
|
3
|
|
|
|
|
|
|
|
|
4
|
|
|
|
|
|
|
# |
|
5
|
|
|
|
|
|
|
# See the file LICENCE included in the distribution. |
|
6
|
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
package Combine::DataBase; |
|
8
|
|
|
|
|
|
|
|
|
9
|
1
|
|
|
1
|
|
460
|
use strict; |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
36
|
|
|
10
|
1
|
|
|
1
|
|
482
|
use Combine::MySQLhdb; |
|
|
1
|
|
|
|
|
3
|
|
|
|
1
|
|
|
|
|
111
|
|
|
11
|
1
|
|
|
1
|
|
8
|
use Digest::MD5; |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
44
|
|
|
12
|
1
|
|
|
1
|
|
6
|
use Encode qw(encode_utf8); |
|
|
1
|
|
|
|
|
3
|
|
|
|
1
|
|
|
|
|
1323
|
|
|
13
|
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
sub new { |
|
15
|
0
|
|
|
0
|
0
|
|
my ($class, $xwi, $sv, $loghandle) = @_; |
|
16
|
0
|
0
|
|
|
|
|
$xwi = new Combine::XWI unless ref $xwi; |
|
17
|
0
|
|
|
|
|
|
my $self = {}; |
|
18
|
0
|
|
|
|
|
|
$self->{'xwi'} = $xwi; |
|
19
|
0
|
|
|
|
|
|
$self->{'databasehandle'} = $sv; |
|
20
|
0
|
|
|
|
|
|
$self->{'loghandle'} = $loghandle; |
|
21
|
0
|
|
|
|
|
|
bless $self, $class; |
|
22
|
0
|
|
|
|
|
|
return $self; |
|
23
|
|
|
|
|
|
|
} |
|
24
|
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
#uses table recordurl with columns: recordid, urlid, lastchecked PRIMARY KEY (urlid), key recordid |
|
26
|
|
|
|
|
|
|
# recordid and urlid starts at 1 !! |
|
27
|
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
sub delete { |
|
29
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
|
30
|
0
|
|
|
|
|
|
my $xwi = $self->{'xwi'}; |
|
31
|
0
|
0
|
|
|
|
|
return undef unless ref $xwi; |
|
32
|
|
|
|
|
|
|
|
|
33
|
0
|
|
|
|
|
|
my $urlid = $xwi->urlid; |
|
34
|
0
|
|
|
|
|
|
my ($recordid, $md5) = $self->{databasehandle}->selectrow_array( |
|
35
|
|
|
|
|
|
|
qq{SELECT recordid,md5 FROM recordurl WHERE urlid=$urlid;}); #Only one |
|
36
|
0
|
0
|
|
|
|
|
return if !defined($recordid); |
|
37
|
0
|
|
|
|
|
|
$self->{'loghandle'}->say("DataBase::delete $urlid, $recordid, $md5;"); |
|
38
|
|
|
|
|
|
|
#LOCK recordurl - needed? |
|
39
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{LOCK TABLES recordurl WRITE;})->execute(); |
|
40
|
|
|
|
|
|
|
#delete URL from recordurl |
|
41
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{DELETE FROM recordurl WHERE urlid=?;})->execute($urlid); |
|
42
|
0
|
|
|
|
|
|
my ($ant) = $self->{databasehandle}->selectrow_array( |
|
43
|
|
|
|
|
|
|
qq{SELECT recordid FROM recordurl WHERE recordid=$recordid LIMIT 1;}); |
|
44
|
|
|
|
|
|
|
#UNLOCK recordurl - needed? |
|
45
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{UNLOCK TABLES;})->execute(); |
|
46
|
0
|
0
|
0
|
|
|
|
if ( !defined($ant) || ($ant == 0) ) { Combine::MySQLhdb::DeleteKey($recordid,$md5); } #Should handle del's of non-existing recs |
|
|
0
|
|
|
|
|
|
|
|
47
|
|
|
|
|
|
|
|
|
48
|
0
|
|
|
|
|
|
$xwi->nofollow("true"); #?? check?? |
|
49
|
|
|
|
|
|
|
} |
|
50
|
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
sub insert { |
|
52
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
|
53
|
0
|
|
|
|
|
|
my $xwi = $self->{'xwi'}; |
|
54
|
0
|
0
|
|
|
|
|
return undef unless ref $xwi; |
|
55
|
|
|
|
|
|
|
|
|
56
|
0
|
|
|
|
|
|
my $urlid = $xwi->urlid; |
|
57
|
|
|
|
|
|
|
## my $md5 = $xwi->md5; |
|
58
|
|
|
|
|
|
|
|
|
59
|
0
|
|
|
|
|
|
my $md5D = new Digest::MD5; |
|
60
|
0
|
|
|
|
|
|
$md5D->reset; |
|
61
|
0
|
0
|
|
|
|
|
if ( length($xwi->text) > 0 ) { |
|
62
|
0
|
|
|
|
|
|
my $text = ${$xwi->text} . $xwi->title; |
|
|
0
|
|
|
|
|
|
|
|
63
|
0
|
|
|
|
|
|
$text =~ s/[\s\n\r]+//g; |
|
64
|
0
|
|
|
|
|
|
$md5D->add(encode_utf8($text)); #use only visible text without whitespace |
|
65
|
|
|
|
|
|
|
} else { |
|
66
|
0
|
|
|
|
|
|
$md5D->add($xwi->url); |
|
67
|
0
|
|
|
|
|
|
$md5D->add($xwi->type()); |
|
68
|
|
|
|
|
|
|
} |
|
69
|
0
|
|
|
|
|
|
$_ = $md5D->hexdigest; |
|
70
|
0
|
|
|
|
|
|
tr/a-z/A-Z/; |
|
71
|
0
|
|
|
|
|
|
$xwi->md5($_); |
|
72
|
0
|
|
|
|
|
|
my $md5 = $_; |
|
73
|
|
|
|
|
|
|
|
|
74
|
0
|
|
|
|
|
|
$self->{'loghandle'}->say("DataBase::insert $urlid, $md5;"); |
|
75
|
|
|
|
|
|
|
|
|
76
|
|
|
|
|
|
|
#actions according the following truth table based presence in recordurl |
|
77
|
|
|
|
|
|
|
#urlid: there is a document in the database for this url |
|
78
|
|
|
|
|
|
|
#recordid: there is as documenent in the database with the same MD5 as the new page |
|
79
|
|
|
|
|
|
|
# |
|
80
|
|
|
|
|
|
|
# recordid | ! recordid |
|
81
|
|
|
|
|
|
|
# urlid if same md5 | delete(urlid_recordid); |
|
82
|
|
|
|
|
|
|
# update(lastcheck) | update(urlid); insertRec |
|
83
|
|
|
|
|
|
|
# else delete(urlid_recordid);| |
|
84
|
|
|
|
|
|
|
# add(urlid) | |
|
85
|
|
|
|
|
|
|
# ----------------------------------------------------------------------------- |
|
86
|
|
|
|
|
|
|
# ! urlid add(urlid) | add(urlid); insertRec |
|
87
|
|
|
|
|
|
|
|
|
88
|
0
|
|
|
|
|
|
my $existurlid = 0; |
|
89
|
0
|
|
|
|
|
|
my $existrecordid = 0; |
|
90
|
0
|
|
|
|
|
|
my $oldmd5=''; |
|
91
|
|
|
|
|
|
|
#LOCK recordurl |
|
92
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{LOCK TABLES recordurl WRITE;})->execute(); |
|
93
|
0
|
|
|
|
|
|
($existurlid,$oldmd5) = $self->{databasehandle}->selectrow_array( |
|
94
|
|
|
|
|
|
|
qq{SELECT urlid,md5 FROM recordurl WHERE urlid=$urlid;}); |
|
95
|
0
|
|
|
|
|
|
($existrecordid) = $self->{databasehandle}->selectrow_array( |
|
96
|
|
|
|
|
|
|
qq{SELECT recordid FROM recordurl WHERE md5='$md5';}); |
|
97
|
|
|
|
|
|
|
|
|
98
|
0
|
0
|
|
|
|
|
if (!defined($existrecordid)) { $existrecordid = 0; } |
|
|
0
|
|
|
|
|
|
|
|
99
|
0
|
0
|
|
|
|
|
if (!defined($existurlid)) { $existurlid = 0; } |
|
|
0
|
|
|
|
|
|
|
|
100
|
0
|
0
|
|
|
|
|
if (!defined($oldmd5)) { $oldmd5 = ''; } |
|
|
0
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
#Log not locked $self->{'loghandle'}->say("DataBase:: $urlid, $md5; $existrecordid; $existurlid; $oldmd5;"); |
|
102
|
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
#CASE 1: There are documents for both the URL and the MD5 and they have the same md5 |
|
104
|
0
|
0
|
0
|
|
|
|
if ( ($existrecordid && $existurlid) && ($md5 eq $oldmd5) ) { |
|
|
|
0
|
0
|
|
|
|
|
|
|
|
0
|
0
|
|
|
|
|
|
|
|
0
|
0
|
|
|
|
|
|
|
|
0
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
105
|
|
|
|
|
|
|
# updateLastCheck |
|
106
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare( |
|
107
|
|
|
|
|
|
|
qq{UPDATE recordurl SET lastchecked=NOW() WHERE urlid=?;})->execute($urlid); |
|
108
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{UNLOCK TABLES;})->execute(); #UNLOCK recordurl |
|
109
|
0
|
|
|
|
|
|
$self->{'loghandle'}->say("DataBase:: case 1: $existrecordid; $existurlid; $oldmd5;"); |
|
110
|
|
|
|
|
|
|
|
|
111
|
|
|
|
|
|
|
#CASE 2 |
|
112
|
|
|
|
|
|
|
} elsif ( $existrecordid && $existurlid ) { |
|
113
|
|
|
|
|
|
|
#eg } elsif ( ($existrecordid && $existurlid) && ($md5 ne $oldmd5) ) { |
|
114
|
|
|
|
|
|
|
#There are documents for both the URL and the MD5 and they have different md5 |
|
115
|
|
|
|
|
|
|
# deleteOld |
|
116
|
0
|
|
|
|
|
|
my $oldrecordid = 0; |
|
117
|
0
|
|
|
|
|
|
($oldrecordid) = $self->{databasehandle}->selectrow_array( |
|
118
|
|
|
|
|
|
|
qq{SELECT recordid FROM recordurl WHERE urlid=$urlid;}); |
|
119
|
|
|
|
|
|
|
#delete URL from recordurl |
|
120
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{DELETE FROM recordurl WHERE urlid=?;})->execute($urlid); |
|
121
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare( |
|
122
|
|
|
|
|
|
|
qq{INSERT INTO recordurl SET urlid=?, recordid=?, md5=?, lastchecked=NOW();})->execute($urlid, $existrecordid, $md5); |
|
123
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{UNLOCK TABLES;})->execute(); #UNLOCK recordurl |
|
124
|
0
|
|
|
|
|
|
my ($ant) = $self->{databasehandle}->selectrow_array( |
|
125
|
|
|
|
|
|
|
qq{SELECT recordid FROM recordurl WHERE recordid=$oldrecordid LIMIT 1;}); #Outside LOCK? |
|
126
|
0
|
0
|
|
|
|
|
if ( ! defined($ant) ) { $ant = 0; } |
|
|
0
|
|
|
|
|
|
|
|
127
|
0
|
|
|
|
|
|
$self->{'loghandle'}->say("DataBase::DelURL case 2: $oldrecordid; $ant;; $existrecordid; $existurlid; $oldmd5;"); |
|
128
|
0
|
0
|
|
|
|
|
if ( $ant == 0 ) { Combine::MySQLhdb::DeleteKey($oldrecordid, $oldmd5); } |
|
|
0
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
#CASE 3 |
|
131
|
|
|
|
|
|
|
} elsif ( $existrecordid && ! $existurlid ) { |
|
132
|
|
|
|
|
|
|
# addUrlId |
|
133
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare( |
|
134
|
|
|
|
|
|
|
qq{INSERT INTO recordurl SET urlid=?, recordid=?, md5=?, lastchecked=NOW();})->execute($urlid, $existrecordid, $md5); |
|
135
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{UNLOCK TABLES;})->execute(); #UNLOCK recordurl |
|
136
|
0
|
|
|
|
|
|
$self->{'loghandle'}->say("DataBase:: case 3: $existrecordid; $existurlid; $oldmd5;"); |
|
137
|
|
|
|
|
|
|
|
|
138
|
|
|
|
|
|
|
#CASE 4 |
|
139
|
|
|
|
|
|
|
} elsif ( ! $existrecordid && $existurlid ) { |
|
140
|
|
|
|
|
|
|
# deleteOld |
|
141
|
0
|
|
|
|
|
|
my $oldrecordid = 0; |
|
142
|
0
|
|
|
|
|
|
($oldrecordid) = $self->{databasehandle}->selectrow_array( |
|
143
|
|
|
|
|
|
|
qq{SELECT recordid FROM recordurl WHERE urlid=$urlid;}); |
|
144
|
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
# delete($self, $urlid, $oldrecordid); #Problem med LOCK!!! -> ny subrutin |
|
146
|
|
|
|
|
|
|
#delete URL from recordurl |
|
147
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{DELETE FROM recordurl WHERE urlid=?;})->execute($urlid); |
|
148
|
|
|
|
|
|
|
#ASSIGN NEW RECORDID done with auto_increment in SQL |
|
149
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare( |
|
150
|
|
|
|
|
|
|
qq{INSERT INTO recordurl SET urlid=?, md5=?, lastchecked=NOW();})->execute($urlid,$md5); |
|
151
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{UNLOCK TABLES;})->execute(); #UNLOCK recordurl |
|
152
|
0
|
|
|
|
|
|
my ($ant) = $self->{databasehandle}->selectrow_array( |
|
153
|
|
|
|
|
|
|
qq{SELECT recordid FROM recordurl WHERE recordid=$oldrecordid LIMIT 1;}); |
|
154
|
0
|
0
|
|
|
|
|
if ( ! defined($ant) ) { $ant = 0; } |
|
|
0
|
|
|
|
|
|
|
|
155
|
0
|
|
|
|
|
|
$self->{'loghandle'}->say("DataBase::DelURL $oldrecordid; $ant;"); |
|
156
|
0
|
0
|
|
|
|
|
if ( $ant == 0 ) { Combine::MySQLhdb::DeleteKey($oldrecordid, $oldmd5); } |
|
|
0
|
|
|
|
|
|
|
|
157
|
0
|
|
|
|
|
|
my ($recordid) = $self->{databasehandle}->selectrow_array( |
|
158
|
|
|
|
|
|
|
qq{SELECT recordid FROM recordurl WHERE urlid=$urlid;}); |
|
159
|
0
|
|
|
|
|
|
$xwi->recordid($recordid); |
|
160
|
0
|
|
|
|
|
|
$self->{'loghandle'}->say("DataBase::Write $recordid case 4: $existrecordid; $existurlid; $oldmd5;"); |
|
161
|
0
|
|
|
|
|
|
Combine::MySQLhdb::Write($xwi); |
|
162
|
|
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
#CASE 5 |
|
164
|
|
|
|
|
|
|
} elsif ( ! $existrecordid && ! $existurlid ) { |
|
165
|
|
|
|
|
|
|
#ASSIGN NEW RECORDID done with auto_increment in SQL |
|
166
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare( |
|
167
|
|
|
|
|
|
|
qq{INSERT INTO recordurl SET urlid=?, md5=?, lastchecked=NOW();})->execute($urlid,$md5); |
|
168
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{UNLOCK TABLES;})->execute(); #UNLOCK recordurl |
|
169
|
0
|
|
|
|
|
|
my ($recordid) = $self->{databasehandle}->selectrow_array( |
|
170
|
|
|
|
|
|
|
qq{SELECT recordid FROM recordurl WHERE urlid=$urlid;}); |
|
171
|
0
|
|
|
|
|
|
$xwi->recordid($recordid); |
|
172
|
0
|
|
|
|
|
|
$self->{'loghandle'}->say("DataBase::Write $recordid case 5: $existrecordid; $existurlid; $oldmd5;"); |
|
173
|
0
|
|
|
|
|
|
Combine::MySQLhdb::Write($xwi); |
|
174
|
|
|
|
|
|
|
} |
|
175
|
|
|
|
|
|
|
|
|
176
|
|
|
|
|
|
|
#Should not happen |
|
177
|
|
|
|
|
|
|
else { |
|
178
|
0
|
|
|
|
|
|
$self->{'loghandle'}->say("DataBase::ERR $existrecordid; $existurlid; $oldmd5;"); |
|
179
|
0
|
|
|
|
|
|
print "ERR DataBase impossible case\n"; |
|
180
|
|
|
|
|
|
|
} |
|
181
|
|
|
|
|
|
|
|
|
182
|
0
|
|
|
|
|
|
$xwi->nofollow("false"); # was set to true by delete...??? |
|
183
|
|
|
|
|
|
|
# my ($follow,$add,$replaced) = &COMB::Policy::url_accept($url,@urls);??? |
|
184
|
|
|
|
|
|
|
} |
|
185
|
|
|
|
|
|
|
|
|
186
|
|
|
|
|
|
|
sub newLinks { |
|
187
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
|
188
|
0
|
|
|
|
|
|
my $xwi = $self->{'xwi'}; |
|
189
|
0
|
0
|
|
|
|
|
return undef unless ref $xwi; |
|
190
|
0
|
|
|
|
|
|
my $recordid = $xwi->recordid; #SANITY CHECK? |
|
191
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare( |
|
192
|
|
|
|
|
|
|
qq{INSERT IGNORE INTO newlinks SELECT urlid,netlocid FROM links WHERE recordid=?;})->execute($recordid); |
|
193
|
|
|
|
|
|
|
} |
|
194
|
|
|
|
|
|
|
|
|
195
|
|
|
|
|
|
|
sub newRedirect { |
|
196
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
|
197
|
0
|
|
|
|
|
|
my $xwi = $self->{'xwi'}; |
|
198
|
0
|
0
|
|
|
|
|
return undef unless ref $xwi; |
|
199
|
1
|
|
|
1
|
|
4
|
use Combine::selurl; |
|
|
1
|
|
|
|
|
3
|
|
|
|
1
|
|
|
|
|
219
|
|
|
200
|
0
|
|
|
|
|
|
my ($u, $netlocid, $urlid, $urlstr); |
|
201
|
|
|
|
|
|
|
# my $tl=$xwi->location; my $tb=$xwi->base; print "NL: $tl, $tb\n"; |
|
202
|
0
|
0
|
|
|
|
|
if ( $u = Combine::selurl->new_abs($xwi->location, $xwi->base) ) { |
|
203
|
0
|
|
|
|
|
|
$urlstr = $u->normalise(); |
|
204
|
|
|
|
|
|
|
# print "NL: $urlstr\n"; |
|
205
|
0
|
|
|
|
|
|
my $lsth = $self->{databasehandle}->prepare(qq{SELECT netlocid,urlid FROM urls WHERE urlstr=?;}); |
|
206
|
0
|
|
|
|
|
|
$lsth->execute($urlstr); |
|
207
|
0
|
|
|
|
|
|
($netlocid,$urlid) = $lsth->fetchrow_array; |
|
208
|
0
|
0
|
|
|
|
|
if ( !defined($urlid) ) { |
|
209
|
0
|
|
|
|
|
|
my $netlocstr = $u->authority; |
|
210
|
0
|
|
|
|
|
|
my $path_query = $u->path_query; |
|
211
|
|
|
|
|
|
|
# print "NL: $netlocstr, $path_query\n"; |
|
212
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{INSERT IGNORE INTO netlocs SET netlocstr=?;})->execute($netlocstr); |
|
213
|
0
|
|
|
|
|
|
($netlocid) = $self->{databasehandle}->selectrow_array(qq{SELECT netlocid FROM netlocs WHERE netlocstr='$netlocstr';}); |
|
214
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare(qq{INSERT IGNORE INTO urls SET urlstr=?, netlocid=?, path=?;})->execute($urlstr,$netlocid,$path_query); |
|
215
|
0
|
|
|
|
|
|
$lsth->execute($urlstr); |
|
216
|
0
|
|
|
|
|
|
($netlocid,$urlid) = $lsth->fetchrow_array; |
|
217
|
|
|
|
|
|
|
} |
|
218
|
|
|
|
|
|
|
# print "NL INS: $urlid,$netlocid\n"; |
|
219
|
|
|
|
|
|
|
#test if undefined |
|
220
|
0
|
|
|
|
|
|
$self->{databasehandle}->prepare( |
|
221
|
|
|
|
|
|
|
qq{INSERT IGNORE INTO newlinks SET urlid=?, netlocid=?;})->execute($urlid,$netlocid); |
|
222
|
|
|
|
|
|
|
} |
|
223
|
|
|
|
|
|
|
} |
|
224
|
|
|
|
|
|
|
|
|
225
|
|
|
|
|
|
|
1; |