file processing, continued.

saving.

git-svn-id: svn://botcastle1b/yplom/facebug1@4 7dec801f-c475-4e67-ba99-809552d69c55

diff --git a/bot.1.pl b/bot.1.pl
index 3d00706ee279eb205f87b47e4b60403c1ce97768..7cc072944f1353712c40ff64287309f37909a8da 100644 (file)
--- a/bot.1.pl
+++ b/bot.1.pl
@@ -1,7 +1,7 @@
 ###PERL;
 
 use strict;
-
+use Fcntl;
 ###PROXY_LIB;
 use proxy_lib qw(url2path path2urldiv getcgi divideurl readconfigfile entitydecode urldecode);
 use POSIX qw(strftime);
@@ -10,6 +10,7 @@ use POSIX qw(strftime);
 ###GROUPSETTINGS_PATH;
 
 my $time = time();
+srand ($time-$$);
 
 print strftime("%d.%m.%Y %H:%M:%S", gmtime($time))."\n";
 if (scalar @ARGV) {
@@ -111,6 +112,7 @@ sub processfile {
        my $timenumber;
        
        my %thread;
+       my %thread2;
        
        if ($headerpath =~ /^((.+)\@h)$/) {
                $headerpath = $1;
@@ -189,6 +191,7 @@ sub processfile {
                my $ignoretext;
                my $link;
                my $hidename;
+               my $attnumber;
                
                local $/ = '<';
                unless (defined ($text = <$contentfile>)) {
@@ -217,7 +220,7 @@ sub processfile {
                                if ($tag{'<'} eq 'h3') {
                                        $mode = 'thread-author';
                                }
-                               elsif (($tag{'<'} eq 'div') and ($tag{'class'} =~ /^(bj|bk)$/)) {
+                               elsif (($tag{'<'} eq 'div') and ($tag{'class'} =~ /^(bj|bk|bm)$/)) { # These are very helpful names, facebug, thank you!
                                        $mode='thread-content';
                                        $level=0;
                                        $ignoretext=1;
@@ -226,18 +229,54 @@ sub processfile {
                                elsif ($tag{'<'} eq 'abbr') {
                                        $mode = 'thread-time';
                                }
+                               elsif (($tag{'<'} eq 'div') and ($tag{'class'} =~ /^(bn|bl)$/)) {
+                                       $mode='thread-attachment';
+                                       $level=0;
+                                       $attnumber=0;
+                               }
                                
                                elsif (($tag{'<'} eq 'div') and ($tag{'id'} =~ /^ufi_/)){
                                        $mode='posts';
                                        
-                                       #!!! {
+                                       my $threadfile;
+                                       my $threadpath = ARCH_PATH.$$settings{'id'}.'/'.'thread/'.$thread{'id'};
                                        
-                                       foreach my $ind (keys %thread) {
-                                               print "$ind: $thread{$ind}\n";
+                                       # if (open ($threadfile, "+<", $threadpath)) {
+                                       if (sysopen ($threadfile, $threadpath, O_RDWR | O_CREAT)) {
+                                               # In the future it might be necessary to use a different function.
+                                               local $/ = "\n";
+                                               %thread2 = readheaderfile($threadfile);
+                                               
+                                               if (($thread2{'timenumber'} ne '')and($thread2{'timenumber'}>$thread{'timenumber'})) {
+                                                       print ("Newer version already saved.\n\n");
+                                               }
+                                               else {
+                                                       foreach my $ind (keys %thread) {
+                                                               $thread2{$ind}=$thread{$ind};
+                                                       }
+                                                       if ($thread2{'key'} eq '') {
+                                                               $thread2{'key'} = key(256);
+                                                       }
+                                                       
+                                                       if (seek($threadfile, 0, 0)) {
+                                                               foreach my $ind (keys %thread2) {
+                                                                       print $threadfile "$ind: $thread2{$ind}\n";
+                                                                       print "$ind: $thread2{$ind}\n"; ####
+                                                               }
+                                                               print $threadfile "\n";
+                                                               truncate ($threadfile , tell($threadfile));
+                                                               print "saved.\n\n";
+                                                       }
+                                                       else {
+                                                               print "Failed seek $threadfile.\n\n";
+                                                       }
+                                               }
+                                               close ($threadfile);
+                                       }
+                                       else
+                                       {
+                                               print "Can't open $threadpath.\n\n";
                                        }
-                                       print "\n";
-                                       
-                                       #!!! }
                                }
                        }
                        elsif ($mode eq 'thread-author') {
@@ -295,9 +334,9 @@ sub processfile {
                                                        $hidename=0;
                                                }
                                        }
-                                       # else {
-                                       # $thread{'postcontent'}.='<<<'.$tag{'<'}.'>>>';
-                                       # }
+                                       else {
+                                       # $thread{'postcontent'}.='<!'.$tag{'<'}.'!>';
+                                       }
                                }
                                
                        }
@@ -307,6 +346,36 @@ sub processfile {
                                }
                        }
                        
+                       elsif ($mode eq 'thread-attachment') {
+                               ###############################################################
+                               if ($tag{'<'} eq 'div') {
+                                       ++$level;
+                               }
+                               elsif ($tag{'<'} eq '/div') {
+                                       if($level){
+                                               --$level;
+                                       }
+                                       else {
+                                               $mode = 'thread';
+                                       }
+                               }
+                               elsif (($tag{'<'} eq 'a') and ($tag{'href'} =~ /^\/photo\.php\?(.*&)?fbid=([0-9]+)(&.*)?$/)) {
+                                       ++$attnumber;
+                                       $thread{'img-'.$attnumber}=$2;
+                                       $mode = 'thread-attachment-img';
+                               }
+                       }
+                       
+                       elsif ($mode eq 'thread-attachment-img') {
+                               if ($tag{'<'} eq 'img') {
+                                       # SAVE THE IMG HERE
+                               }
+                               elsif ($tag{'<'} eq '/a') {
+                                       $mode = 'thread-attachment';
+                               }
+                       }
+                       
+                       
                        
                        if ($tag{"\\"} ne '') {
                                $closetag = 1;
@@ -350,34 +419,45 @@ sub readheaderfile {
        my $headerfile;
        my %header;
        
+       if(ref($headerpath)) {
+               $headerfile=$headerpath;
+       }
+       else {
+               unless (open ($headerfile, "<", $headerpath)) {
+                       return %header;
+               }
+       }
+       
        my $lastname='';
        
-       if (open ($headerfile, "<", $headerpath)) {
-               while (defined(my $line = <$headerfile>)) {
-                       $line =~ s/[\r\n]$//g;
-                       my $headname='';
-                       my $headval='';
+       while (defined(my $line = <$headerfile>)) {
+               $line =~ s/[\r\n]$//g;
+               my $headname='';
+               my $headval='';
+               
+               if($line =~ /^[ \t]+([^ \t].*)$/){
+                       if($lastname ne '') {
+                               $header{$lastname}.=$1;
+                       }
+               }
+               elsif ($line =~ /^([^:]*):[ \t]*([^ \t](.*[^ \t])?)[ \t]*$/) {
+                       $headname = lc($1);
+                       $headval = $2;
                        
-                       if($line =~ /^[ \t]+([^ \t].*)$/){
-                               if($lastname ne '') {
-                                       $header{$lastname}.=$1;
-                               }
+                       if ($header{$headname} ne '') {
+                               $header{$headname}.=', '.$headval;
                        }
-                       elsif ($line =~ /^([^:]*):[ \t]*([^ \t](.*[^ \t])?)[ \t]*$/) {
-                               $headname = lc($1);
-                               $headval = $2;
-                               
-                               if ($header{$headname} ne '') {
-                                       $header{$headname}.=', '.$headval;
-                               }
-                               else {
-                                       $header{$headname}=$headval;
-                               }
-                               $lastname = $headname;
+                       else {
+                               $header{$headname}=$headval;
                        }
+                       $lastname = $headname;
                }
+       }
+       
+       unless (ref($headerpath)) {
                close ($headerfile);
        }
+       
        return %header;
 }
 
@@ -497,4 +577,20 @@ sub taginfo {
                                        # print "\n";
        # }
        return %tag; 
+}
+
+sub key {
+       (my $bits) = @_;
+       my $p = int($bits / 16);
+       my $q = $bits % 16;
+       my $keytext;
+       
+       if ($q){
+        my $r = int(($q+3)/4);
+        $keytext=sprintf('%0'.$r.'x',int(rand(1<<$q)));
+       }
+       for(my $ind = 0; $ind < $p; ++$ind){
+               $keytext.=sprintf('%04x',int(rand(65536)));
+       }
+       return $keytext;
 }
\ No newline at end of file