?1 #
?2 #?Copyright?(C)?2007?羅明?<andyluo197(at)gmail.com>
?3 #
?4
?5 use ?strict? " vars " ;? # 嚴(yán)格遵循my?scope定義
?6
?7 use ?LWP :: Simple;
?8 use ?HTML :: LinkExtor;
?9 use ?HTTP :: Request :: Common;
10 use ?LWP :: UserAgent;
11 use ?URI :: URL;
12
13 # ?從命令行參數(shù)中取主網(wǎng)頁地址和保存文件夾
14 my ? $url ? = ? $ARGV [ 0 ];?
15 my ? $location ? = ? $ARGV [ 1 ];
16
17 # ?子程序，解析主網(wǎng)頁中的鏈接地址
18 my ? @urls ? = ?();
19 sub ?urlCallback
20 {
21 ???? my ( $tag , ? %attr )? = ? @_ ;
22 ???? return ? if ? $tag ?ne? ' a ' ;
23 ???? push ( @urls , ? values ? %attr );
24 }
25
26 # ?用于解析鏈接的LinkExtractor
27 my ? $p ? = ?HTML :: LinkExtor -> new( \& urlCallback);
28
29 # ?載入主網(wǎng)頁內(nèi)容，及解析鏈接地址
30 my ? $ua ? = ?LWP :: UserAgent -> new;
31 my ? $res ? = ? $ua -> request(HTTP :: Request -> new(GET? => ? $url ) , ? sub { $p -> parse( $_ [ 0 ])});
32
33 # ?將鏈接的相對路徑轉(zhuǎn)換為絕對路徑
34 my ? $base ? = ? $res -> base;
35 @urls ? = ? map ?{ $_ ? = ?url( $_ , ? $base )? -> ? abs }? @urls ;
36
37 # ?將網(wǎng)頁寫入文件
38 $ua -> request(GET( $url ) , ? $location ? . ? ' \\index.html ' );
39 my ? $index ? = ? 1 ;
40 foreach ( @urls )
41 {
42 ???? $ua -> request(GET( $_ ) , ? $location ? . ? ' \\ ' ? . ? $index ? . ? ' .html ' );
43 ???? $index += 1 ;
44 }