Benutzer:Jah/histInfo.js

#!/usr/bin/perl

use CGI qw(:standard);
use DB_File;
use Compress::Zlib;

$datadir = "data";
$lws = 5;

import_names;
$langProject = $Q::lang . $Q::project;

print header(-charset => 'utf-8');
if(defined $Q::page) {
	initDB();
	loadRevInfo();
	loadHashes();
	printAuthorInfo();
	printHistInfo();
	print "rhic2();\n";
}

sub initDB {
	my $page = $Q::page;
	$page =~ s/ /_/g;
	tie %idx, "DB_File", "$datadir/$langProject.idx", O_RDONLY;
	($offset, $len, $cm, $posSeq, $nSeq) = split / /, $idx{$page};
	untie %idx;
}

sub loadRevInfo {
	my $pid = open HIST, '-|';
	if(!$pid) {
		open REV, "$datadir/$langProject.rev";
		seek REV, $offset, 0;
		if($cm==0) {
			read REV, $buf, $len;
			print $buf;
		} elsif($cm==1) {
		        ($refinf, $status) = inflateInit(-WindowBits => 0 - MAX_WBITS);
			for(my $bufsize = 16*1024; $len>0; $len -= $bufsize) {
				read REV, $buf, $len<$bufsize?$len:$bufsize;
			        ($xml, $status) = $refinf->inflate($buf);
				print $xml;
			}
		}
		close REV;
		exit;
	}

	my $userNr = 0;
	my $revId = -1;
	while(<HIST>) {
		$len += do { use bytes; length };
		if($inRev) {
			if(/^\s*<\/revision/) {
				$user =~ s/'/\\'/g;
				$user{$revId} = $user;
				$userNr{$user} = $userNr++ unless defined $userNr{$user};
				$timestamp{$revId} = $timestamp;
				$length{$revId} = $length;
				# $diffId,$comment,$md5 are not used
				$comment = "";
				$inRev = 0;
			} elsif(/^\s*<timestamp>(.*?)<\/timestamp>/) {
				$timestamp = $1;
			} elsif(/\s*<id>(\d+)<\/id>/) {
				if(!$inContributor) {
					($revId, $diffId)  = ($1, $revId);
				}
			} elsif(/^\s*<comment>(.*?)<\/comment>/) {
				$comment = $1;
			} elsif(/^\s*<contributor>/) {
				$inContributor = 1;
			} elsif(/^\s*<\/contributor/) {
				$inContributor = 0;
			} elsif(/^\s*<(username|ip)>(.*?)<\/\1>/) {
				$user = $2;
			} elsif(/^\s*<text type="sectionlist" length="(\d+)" md5="(.*?)">(.*?)<\/text>/) {
					$length = $1;
					$md5    = $2;
			} elsif(/^\s*<text offset=\"(\d+?)\"(?: lengthGz=\"(\d+)\")? length=\"(\d+)\" md5=\"(.*?)\" \/>/) {
				$length = $3;
				$md5    = $4;
			}
		} elsif($inSgr) {
			if(/^\s*<\/sectiongroup>/) {
				$inSgr = 0;
			}
		} else {
			if(/^\s*<revision/) {
				$inRev = 1;
			} elsif(/^\s*<sectiongroup offset="(\d+)" length="(\d+)">/) {
				$inSgr = 1;
			}
		}
	}
	close HIST;
}

sub loadHashes {
	my $seqs;
	open SEQ, "$datadir/$langProject.seq";
	seek SEQ, $posSeq, 0;
	read SEQ, $seqs, 12*$nSeq;
	close SEQ;
	for(my $i=0; $i<$nSeq; $i++) {
		my ($hash, $first, $last) = unpack("N3", do { use bytes; substr($seqs, 12*$i, 12) });
		push @{$hashesFirst{$first}}, sprintf "%08x", $hash;
		push @{$hashesLast{$last}}, sprintf "%08x", $hash;
	}
}

sub printAuthorInfo {
	print "var histAuInfo = new Array(\n";
	foreach $user (sort { $userNr{$a} <=> $userNr{$b} } keys %userNr) {
		print ", " if $userNr{$user}>0;
		print "'$user'";
	}
	print ");\n";
}

sub printHistInfo {
	my $i=0;
	# The array constructor in Firefox accepts at most 2^16 arguments, so
	# we must work with nested arrays.
	print "var histInfo = new Array(\n";
	foreach $revId (sort { $a <=> $b } keys %user) {
		if($i%1000==0) {
			print ", " if $i>0;
			print "new Array(\n";
		}
		print "," if $i%1000>0;
		printf "%d,'%s',%d,'%s'\n", $revId, $timestamp{$revId}, $userNr{$user{$revId}},
			join("", @{$hashesFirst{$revId}});
		if($i%1000==999) {
			print ")\n";
		}
		$i++;
	}
	if($i%1000!=0) {
		print ")\n";
	}
	print ");\n";
}