From: Luke Reeves <luke@neuro-tech.net>
Date: Sun, 12 Jul 2009 15:48:09 +0000 (-0400)
Subject: Fixed snapshot parsing and viewing
X-Git-Tag: 2.1-alpha-2019-0-29~153
X-Git-Url: http://91.132.146.200/gitweb/?a=commitdiff_plain;h=cad54fbbad645b876cce326e60bf2816144bc62d;p=insipid.git

Fixed snapshot parsing and viewing
---

diff --git a/lib/Insipid/Parser.pm b/lib/Insipid/Parser.pm
new file mode 100755
index 0000000..d9a41a3
--- /dev/null
+++ b/lib/Insipid/Parser.pm
@@ -0,0 +1,115 @@
+package Insipid::Parser;
+
+use HTML::Parser;
+use HTML::Entities ();
+use URI::URL;
+use Digest::MD5 qw(md5 md5_hex);
+use Insipid::Config;
+use Insipid::Database;
+
+use vars qw(@ISA);
+@ISA = qw(HTML::Parser);
+
+sub setSnapshotMap {
+	my($self, $ssMap) = (@_);
+	$self->{SSMAP} = $ssMap;
+}
+
+sub new {
+	my $pack = shift;
+	my $self = $pack->SUPER::new;
+	@{$self}{qw(__base __grabit)} = @_;
+	$self;
+}
+
+sub declaration {
+	my $self = shift;
+	my ($decl) = @_;
+}
+
+sub start {
+	my $self = shift;
+	my ($tag, $attr, $attrseq, $origtext) = @_;
+
+	if(!defined($self->{__grabit})) {
+		print("<$tag");
+	}
+
+	for (keys %$attr) {
+		my $val = $attr->{$_};
+		if(($_ eq "/") && ($val = "/")) { next; }
+
+		if(!defined($self->{__grabit})) { 
+			print(" $_=\""); 
+		}
+
+		if( "$tag $_" =~ /^(link href|img src)$/i) {
+			$val = url($val)->abs($self->{__base},1);
+
+			if(!defined($self->{__grabit})) {
+				if($val =~ /(\.gif|\.jpg|\.png|\.css)$/i) {
+					my $md5 = md5_hex("$val");
+					$val = $snapshot_url . $md5;
+				}
+			} else {
+				# JPG, GIF, PNG and CSS
+				if($val =~ /(\.gif|\.jpg|\.png|\.css)$/i) {
+					join_urls($self->{__base}, $val);
+					$val = $self->{__grabit}($val, $1);
+				}
+			}
+		}
+
+		if(!defined($self->{__grabit})) {
+			# Check against our snapshot map
+			if(($tag =~ /^a/i) && ($_ =~ /^href/i)) {
+				my $sst = $self->{SSMAP};
+
+				if(defined($sst->{$val})) {
+					print $snapshot_url . $sst->{$val};
+					print('"');
+				} else {
+					print("$val\"");
+				}
+			} else {
+				print("$val\"");
+			}
+		}
+	}
+
+	if(!defined($self->{__grabit})) { print(">"); }
+}
+
+sub end {
+	my $self = shift;
+	my ($tag) = @_;
+
+	if(!defined($self->{__grabit})) { print("</$tag>"); }
+}
+
+sub text {
+	my $self = shift;
+	my ($text) = @_;
+
+	if(!defined($self->{__grabit})) { print("$text"); }
+}
+
+sub comment {
+	my $self = shift;
+	my ($comment) = @_;
+
+	if(!defined($self->{__grabit})) { print("<!-- $comment -->"); }
+}
+
+sub join_urls {
+	my($parent, $child) = (@_);
+	my $sql = "insert into $tbl_pagecache_references(md5_parent, md5_child) values(?, ?)";
+	my $sth = $dbh->prepare($sql);
+	$sth->execute(md5_hex($parent), md5_hex($child));
+	if($sth->err) {
+		# ignore errors for now
+	}
+}
+
+1;
+__END__
diff --git a/lib/Insipid/Snapshots.pm b/lib/Insipid/Snapshots.pm
index 3e2f567..9b802b4 100755
--- a/lib/Insipid/Snapshots.pm
+++ b/lib/Insipid/Snapshots.pm
@@ -28,6 +28,7 @@ use Insipid::Config;
 use Insipid::Database;
 use Insipid::Util;
 use Insipid::LinkExtractor;
+use Insipid::Parser;
 use CGI qw/:standard/;
 use CGI::Carp qw(fatalsToBrowser);
 use Date::Format;
@@ -228,7 +229,7 @@ sub show_snapshot {
 		}
 
 		print "\r\n";
-		my $p = MyParser->new($row[2], undef);
+		my $p = Insipid::Parser->new($row[2], undef);
 		$p->setSnapshotMap(\%internalLinks);
 		
 		if($row[0] =~ /utf/i) {
@@ -476,7 +477,7 @@ sub do_snapshot {
 sub parsepage {
 	my ($url, $content, $content_type) = (@_);
 
-	my $p = MyParser->new($url, \&fetch_url);
+	my $p = Insipid::Parser->new($url, \&fetch_url);
 	if($content_type =~ /utf/i) { 
 		$p->utf8_mode(1);
 	}
@@ -484,122 +485,5 @@ sub parsepage {
 	$p->parse($content);
 }
 
-## "use MyParser;" ## TODO: Make this a separate file.
-BEGIN {
-	package MyParser;
-	use HTML::Parser;
-	use HTML::Entities ();
-	use URI::URL;
-	use Digest::MD5 qw(md5 md5_hex);
-	use Insipid::Config;
-	use Insipid::Database;
-
-	use vars qw(@ISA);
-	@ISA = qw(HTML::Parser);
-
-	sub setSnapshotMap {
-		my($self, $ssMap) = (@_);
-		$self->{SSMAP} = $ssMap;
-	}
-
-	sub new {
-		my $pack = shift;
-		my $self = $pack->SUPER::new;
-		@{$self}{qw(__base __grabit)} = @_;
-		$self;
-	}
-
-	sub declaration {
-		my $self = shift;
-		my ($decl) = @_;
-	}
-
-	sub start {
-		my $self = shift;
-		my ($tag, $attr, $attrseq, $origtext) = @_;
-
-		if(!defined($self->{__grabit})) {
-			print("<$tag");
-		}
-
-		for (keys %$attr) {
-			my $val = $attr->{$_};
-			if(($_ eq "/") && ($val = "/")) { next; }
-
-			if(!defined($self->{__grabit})) { 
-				print(" $_=\""); 
-			}
-
-			if( "$tag $_" =~ /^(link href|img src)$/i) {
-				$val = url($val)->abs($self->{__base},1);
-	
-				if(!defined($self->{__grabit})) {
-					if($val =~ /(\.gif|\.jpg|\.png|\.css)$/i) {
-						my $md5 = md5_hex("$val");
-						$val = $snapshot_url . $md5;
-					}
-				} else {
-					# JPG, GIF, PNG and CSS
-					if($val =~ /(\.gif|\.jpg|\.png|\.css)$/i) {
-						join_urls($self->{__base}, $val);
-						$val = $self->{__grabit}($val, $1);
-					}
-				}
-			}
-
-			if(!defined($self->{__grabit})) {
-				# Check against our snapshot map
-				if(($tag =~ /^a/i) && ($_ =~ /^href/i)) {
-					my $sst = $self->{SSMAP};
-
-					if(defined($sst->{$val})) {
-						print $snapshot_url . $sst->{$val};
-						print('"');
-					} else {
-						print("$val\"");
-					}
-				} else {
-					print("$val\"");
-				}
-			}
-		}
-
-		if(!defined($self->{__grabit})) { print(">"); }
-	}
-
-	sub end {
-		my $self = shift;
-		my ($tag) = @_;
-
-		if(!defined($self->{__grabit})) { print("</$tag>"); }
-	}
-
-	sub text {
-		my $self = shift;
-		my ($text) = @_;
-
-		if(!defined($self->{__grabit})) { print("$text"); }
-	}
-
-	sub comment {
-		my $self = shift;
-		my ($comment) = @_;
-
-		if(!defined($self->{__grabit})) { print("<!-- $comment -->"); }
-	}
-	
-	sub join_urls {
-		my($parent, $child) = (@_);
-		my $sql = "insert into $tbl_pagecache_references(md5_parent, md5_child) values(?, ?)";
-		my $sth = $dbh->prepare($sql);
-		$sth->execute(md5_hex($parent), md5_hex($child));
-		if($sth->err) {
-			# ignore errors for now
-		}
-	}
-
-}
-## end "use MyParser;" ##
-
 1;
 __END__