#!/usr/bin/php
<?php

/*
	A script to download and save the "legacy" release notes of file
	releases on sourceforge.net

	Copyright 2009 Michiel 'elmuerte' Hendriks
	Released under the terms of the GNU Public License version 3
*/

$outputdir = "output";

if ($argc <= 1) die("Usage: php get-rls-notes.php <baseurl> [suffix]\n\n");

$baseurl = $argv[1];

if ($argc < 3) $suffix = "-notes.txt";
else $suffix = $argv[2];

echo "Processing SourceForge.net download overview: $baseurl\n";
flush();

$basedata = file_get_contents($baseurl);

if ($basedata === false) die("Failed to retrieve data");

$cnt = preg_match_all('/<a class="dload [^"]*"\s*href="([^"]*)"[^>]*>([^<]*)</is', $basedata, $files, PREG_OFFSET_CAPTURE);

for ($i = 0; $i < $cnt; ++$i)
{
	$name = $files[1][$i][0];
	$name = urldecode(preg_replace("#^/projects/[^/]*/files/(.*)/download$#", "\\1", $name));
	echo "Found file release: $name\n";

	flush();
	$offset = $files[1][$i][1];
	if (preg_match('#(http://sourceforge\.net/project/shownotes\.php\?release_id=[0-9]+)#', 
		$basedata, $rlsnote, PREG_OFFSET_CAPTURE, $offset))
	{
		if ($i < $cnt-1) {
			$nextoff = $files[1][$i+1][1];
			// validate offset
			if ($rlsnote[1][1] > $nextoff)
			{
				echo "    no release notes, skipping\n";
				continue;
			}
		}
		$notedata = file_get_contents($rlsnote[1][0]);
		if ($notedata === false)
		{
			echo "    failed to download release notes\n";
			continue;
		}
		if (preg_match('#<b>Notes:</b>\n(.*)\s+(</pre>\s*)?</p>\s+</p>\s+</div>#s', $notedata, $note))
		{
			$repl = array(
				'#^        #s',
				'#        <hr noshade>\n#s',
				'#        <b>Changes:</b>\n        #s',
			);
			$to = array(
				'',
				'',
				'',
			);
			$releasenote = preg_replace($repl, $to, $note[1]);
			if (empty($releasenote))
			{
				echo "    empty release notes\n";
				continue;
			}
			$dir = dirname($outputdir.'/'.$name);
			if (!file_exists($dir)) mkdir($dir, 0777, true);
			$rlsfilename = $outputdir.'/'.$name.$suffix;
			echo "    saving release notes to $rlsfilename\n";
			file_put_contents($rlsfilename, $releasenote);
		}
		else {
			echo "    no release note data found\n";
			continue;
		}
	}
	else {
		echo "    no legacy release notes attached\n";
	}
}

flush();