From 0a7ea3844b96abc66377773af4daef0c9fb8cf29 Mon Sep 17 00:00:00 2001
From: Paul J R <me@pjr.cc>
Date: Thu, 17 Jan 2013 06:41:15 +1100
Subject: [PATCH] createing a version file parser for the distro site of nodejs

---
 lxc/parse_dist.php |  158 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 158 insertions(+), 0 deletions(-)
 create mode 100644 lxc/parse_dist.php

diff --git a/lxc/parse_dist.php b/lxc/parse_dist.php
new file mode 100644
index 0000000..eb846a9
--- /dev/null
+++ b/lxc/parse_dist.php
@@ -0,0 +1,158 @@
+<?php
+
+$baseurl  = "http://nodejs.org/dist";
+
+// we scrape nodejs.org/dist site looking for versions that exist so we can construct a meta data database
+// so our lxc will be able to find versions in a more appropriate way
+
+// first, get the main site
+
+$fver = array();
+$fvern = 0;
+global $fver, $baseurl, $fvern;
+
+
+echo "Getting base list from $baseurl\n";
+$distbase = file_get_contents($baseurl);
+
+// get all links that are directories
+preg_match_all('/<a [^>]*href="(.+)"/', $distbase, $firstlist);
+
+$versions = array();
+
+foreach($firstlist[1] as $chk) {
+	//echo "Got: $chk\n";
+	if(preg_match('/v[0-9]+.[0-9]+.[0-9]+\//', $chk)) {
+		if(preg_match('/v([0-9]+.[0-9]+.[0-9]+)\//', $chk, $vlist)) {
+			$versions[$vlist[1]] = $chk;
+		}		
+	}
+}
+
+foreach($versions as $key => $val) {
+	// now we get each version and parse whats in its directory
+	echo "Checking $key => $baseurl/$val\n";
+	
+	//if(preg_match("/0\.8\.[1].*/", $key)) {
+		$urlnow = "$baseurl/$val";
+		$vers_dir = file_get_contents($urlnow);
+	
+		preg_match_all('/<a [^>]*href="(.+)"/', $vers_dir, $fileslist);
+		
+		checkFilesList(rtrim($val, "/"), $key, $fileslist);
+	//}
+}
+
+
+// done!
+//echo "and fver is now\n";
+//var_dump($fver);
+
+echo "dumping to versions file\n";
+$versfile = fopen("./versions_file", "w");
+if($versfile !== false) {
+	fwrite($versfile, "baseurl:$baseurl\n");
+	foreach($fver as $line) {
+		fwrite($versfile, "$line\n");
+	}
+	fclose($versfile);
+} else {
+	echo "Failed to open vers file!\n";	
+}
+
+function checkFilesList($url, $vers, $fileslist) {
+	global $fver, $baseurl, $fvern;
+			
+	echo "Checking $url, $vers\n";
+	
+	$shalist = array();
+	
+	foreach($fileslist[1] as $fname) {
+		$is_echo = false;
+		$flurl = "";
+		$arch = "";
+		$plat = "";
+		$thissha = "";
+		
+		if($fname == "SHASUMS.txt") {
+			//$shas = file_get_
+			$shas_r = file_get_contents("$baseurl/$url/SHASUMS.txt");
+			$shalist = explode("\n", $shas_r);
+			foreach($shalist as $ssha) {
+				$fsha = preg_split("/[ \t]+/", $ssha);
+				if(isset($fsha[1])) $shalist[$fsha[1]] = $fsha[0];
+			}
+			//echo "Shasums file: $fname\n";
+			//var_dump($shalist);
+		} else if($fname == "MD5SUMS.txt") {
+			// nothing yet
+		} else 	if(preg_match("/.*linux.*64\.tar\.gz/", $fname)) {
+			//echo "64 bit version for linux: $fname\n";
+			$thissha = "-";
+			if(isset($shalist[$fname])) $thissha = $shalist[$fname];
+			$is_echo = true;
+			$plat = "linux";
+			$arch = "x86_64";
+		} else if(preg_match("/.*linux.*86\.tar\.gz/", $fname)) {
+			//echo "32 bit version for linux: $fname\n";
+			$thissha = "-";
+			if(isset($shalist[$fname])) $thissha = $shalist[$fname];
+			//echo "$vers:$url/$fname:linux:i686:$thissha\n";
+			$is_echo = true;
+			$plat = "linux";
+			$arch = "i686";
+		} else 	if(preg_match("/.*darwin.*86\.tar\.gz/", $fname)) {
+			//echo "32 bit version for darwin: $fname\n";
+			$thissha = "-";
+			if(isset($shalist[$fname])) $thissha = $shalist[$fname];
+			//echo "$vers:$url/$fname:darwin:i686:$thissha\n";
+			$is_echo = true;
+			$plat = "darwin";
+			$arch = "i686";
+		} else if(preg_match("/.*darwin.*64\.tar\.gz/", $fname)) {
+			//echo "64 bit version for darwin: $fname\n";
+			$thissha = "-";
+			if(isset($shalist[$fname])) $thissha = $shalist[$fname];
+			//echo "$vers:$url/$fname:darwin:x86_64:$thissha\n";
+			$is_echo = true;
+			$plat = "darwin";
+			$arch = "x86_64";
+		} else if(preg_match("/.*sunos.*64\.tar\.gz/", $fname)) {
+			// echo "64 bit version for sunos: $fname\n";
+			$thissha = "-";
+			if(isset($shalist[$fname])) $thissha = $shalist[$fname];
+			//echo "$vers:$url/$fname:sunos:x86_64:$thissha\n";
+			$is_echo = true;
+			$plat = "sunos";
+			$arch = "x86_64";
+		} else if(preg_match("/.*sunos.*86\.tar\.gz/", $fname)) {
+			//echo "32 bit version for sunos: $fname\n";
+			$thissha = "-";
+			if(isset($shalist[$fname])) $thissha = $shalist[$fname];
+			//echo "$vers:$url/$fname:sunos:i686:$thissha\n";
+			$is_echo = true;
+			$plat = "sunos";
+			$arch = "i686";
+		} else if(preg_match("/.*linux-arm.*\.tar\.gz/", $fname)) {
+			// echo "arm for linux: $fname\n";
+			$thissha = "-";
+			if(isset($shalist[$fname])) $thissha = $shalist[$fname];
+			//echo "$vers:$url/$fname:linux:armhf:$thissha\n";
+			$is_echo = true;
+			$plat = "linux";
+			$arch = "armhf";
+		} else if(preg_match("/node-v[0-9]+\.[0-9]+\.[0-9]+\.tar\.gz/", $fname)) {
+			$thissha = "-";
+			if(isset($shalist[$fname])) $thissha = $shalist[$fname];
+			//echo "$vers:$url/$fname:linux:armhf:$thissha\n";
+			$is_echo = true;
+			$plat = "src";
+			$arch = "-";
+		}
+		
+		if($is_echo) {
+			$fver[$fvern++] = "$vers:$plat:$arch:$url/$fname:$thissha";
+		}
+	}
+}
+?>
\ No newline at end of file
-- 
1.7.0.4