From 41916d233bacd00c4c52d75714205feea5f068b0 Mon Sep 17 00:00:00 2001
From: b <b@7dec801f-c475-4e67-ba99-809552d69c55>
Date: Thu, 3 Dec 2015 21:48:27 +0000
Subject: [PATCH] Some more html processing

git-svn-id: svn://botcastle1b/yplom/facebug1@3 7dec801f-c475-4e67-ba99-809552d69c55
---
 bot.1.pl | 63 +++++++++++++++++++++++++++++++++++++++++++++++---------
 1 file changed, 53 insertions(+), 10 deletions(-)

diff --git a/bot.1.pl b/bot.1.pl
index 10ec131..3d00706 100644
--- a/bot.1.pl
+++ b/bot.1.pl
@@ -3,7 +3,7 @@
 use strict;
 
 ###PROXY_LIB;
-use proxy_lib qw(url2path path2urldiv getcgi divideurl readconfigfile entitydecode);
+use proxy_lib qw(url2path path2urldiv getcgi divideurl readconfigfile entitydecode urldecode);
 use POSIX qw(strftime);
 
 ###ARCH_PATH;
@@ -186,6 +186,9 @@ sub processfile {
 		my $mode = 'thread';
 		my $level = 0;
 		my $closetag=0;
+		my $ignoretext;
+		my $link;
+		my $hidename;
 		
 		local $/ = '<';
 		unless (defined ($text = <$contentfile>)) {
@@ -217,11 +220,14 @@ sub processfile {
 				elsif (($tag{'<'} eq 'div') and ($tag{'class'} =~ /^(bj|bk)$/)) {
 					$mode='thread-content';
 					$level=0;
+					$ignoretext=1;
+					$hidename=0;
+				}
+				elsif ($tag{'<'} eq 'abbr') {
+					$mode = 'thread-time';
 				}
-				
 				
 				elsif (($tag{'<'} eq 'div') and ($tag{'id'} =~ /^ufi_/)){
-					print "$tag{'id'} - $id\n";
 					$mode='posts';
 					
 					#!!! {
@@ -251,25 +257,57 @@ sub processfile {
 			elsif ($mode eq 'thread-content') {
 				if ($tag{'<'} eq 'div') { # There should not be any sub<div>s!
 					++$level;
-					$thread{'postcontent'}.="<br>\n<br>\n";
+					$thread{'postcontent'}.='<div>';
 				}
 				elsif ($tag{'<'} eq '/div') {
 					if($level){
 						--$level;
-						$thread{'postcontent'}.="<br>\n<br>\n";
+						$thread{'postcontent'}.='<div>';
 					}
 					else {
 						$mode = 'thread';
 					}
 				}
+				elsif ($tag{'<'} eq 'p') {
+					$thread{'postcontent'}.='<p>';
+					$ignoretext=0;
+				}
 				elsif ($tag{'<'} eq '/p') {
-					$thread{'postcontent'}.="<br>\n";
+					$thread{'postcontent'}.='</p>';
+					$ignoretext=1;
+				}
+				elsif (!$ignoretext) {
+					if ($tag{'<'} eq 'a') {
+						if ($tag{'href'} =~ /^https?:\/\/([a-z0-9]+\.)?facebook\.com\/l\.php\?(.*&)?u=([^&]+)(&.*)?$/) {
+							$thread{'postcontent'}.='<a href="'.urldecode($3).'">';
+							$link=1;
+						}
+						elsif ($tag{'href'} =~ /^\/([A-Za-z0-9\.]+)(\?.*)$/) {
+							$thread{'postcontent'}.='<a href="#">'.(($$names{$1} ne '')?$$names{$1}:$$names{'default'});
+							$link=1;
+							$hidename=1;
+						}
+					}
+					elsif ($tag{'<'} eq '/a') {
+						if($link) {
+							$thread{'postcontent'}.='</a>';
+							$link=0;
+							$hidename=0;
+						}
+					}
+					# else {
+					# $thread{'postcontent'}.='<<<'.$tag{'<'}.'>>>';
+					# }
+				}
+				
+			}
+			elsif ($mode eq 'thread-time') {
+				if ($tag{'<'} eq '/abbr') {
+					$mode = 'thread';
 				}
-				# else {
-					# $thread{'postcontent'}.='<'.$tag{'<'}.'>';
-				# }
 			}
 			
+			
 			if ($tag{"\\"} ne '') {
 				$closetag = 1;
 				next;
@@ -283,7 +321,12 @@ sub processfile {
 			$text =~ s/<$//;
 			
 			if($mode eq 'thread-content') {
-				$thread{'postcontent'}.=$text;
+				unless ($ignoretext or $hidename){
+					$thread{'postcontent'}.=$text;
+				}
+			}
+			elsif ($mode eq 'thread-time') {
+				$thread{'timetext'}.=$text;
 			}
 			# print "text: $text\n";
 		}
-- 
2.30.2