createing a version file parser for the distro site of nodejs
[random_node_code.git] / nodejs_dist_list / parse_dist.php
1 <?php
2
3 $baseurl  = "http://nodejs.org/dist";
4
5 // we scrape nodejs.org/dist site looking for versions that exist so we can construct a meta data database
6 // so our lxc will be able to find versions in a more appropriate way
7
8 // first, get the main site
9
10 $fver = array();
11 $fvern = 0;
12 global $fver, $baseurl, $fvern;
13
14
15 echo "Getting base list from $baseurl\n";
16 $distbase = file_get_contents($baseurl);
17
18 // get all links that are directories
19 preg_match_all('/<a [^>]*href="(.+)"/', $distbase, $firstlist);
20
21 $versions = array();
22
23 foreach($firstlist[1] as $chk) {
24         //echo "Got: $chk\n";
25         if(preg_match('/v[0-9]+.[0-9]+.[0-9]+\//', $chk)) {
26                 if(preg_match('/v([0-9]+.[0-9]+.[0-9]+)\//', $chk, $vlist)) {
27                         $versions[$vlist[1]] = $chk;
28                 }               
29         }
30 }
31
32 foreach($versions as $key => $val) {
33         // now we get each version and parse whats in its directory
34         echo "Checking $key => $baseurl/$val\n";
35         
36         //if(preg_match("/0\.8\.[1].*/", $key)) {
37                 $urlnow = "$baseurl/$val";
38                 $vers_dir = file_get_contents($urlnow);
39         
40                 preg_match_all('/<a [^>]*href="(.+)"/', $vers_dir, $fileslist);
41                 
42                 checkFilesList(rtrim($val, "/"), $key, $fileslist);
43         //}
44 }
45
46
47 // done!
48 //echo "and fver is now\n";
49 //var_dump($fver);
50
51 echo "dumping to versions file\n";
52 $versfile = fopen("./versions_file", "w");
53 if($versfile !== false) {
54         fwrite($versfile, "baseurl:$baseurl\n");
55         foreach($fver as $line) {
56                 fwrite($versfile, "$line\n");
57         }
58         fclose($versfile);
59 } else {
60         echo "Failed to open vers file!\n";     
61 }
62
63 function checkFilesList($url, $vers, $fileslist) {
64         global $fver, $baseurl, $fvern;
65                         
66         echo "Checking $url, $vers\n";
67         
68         $shalist = array();
69         
70         foreach($fileslist[1] as $fname) {
71                 $is_echo = false;
72                 $flurl = "";
73                 $arch = "";
74                 $plat = "";
75                 $thissha = "";
76                 
77                 if($fname == "SHASUMS.txt") {
78                         //$shas = file_get_
79                         $shas_r = file_get_contents("$baseurl/$url/SHASUMS.txt");
80                         $shalist = explode("\n", $shas_r);
81                         foreach($shalist as $ssha) {
82                                 $fsha = preg_split("/[ \t]+/", $ssha);
83                                 if(isset($fsha[1])) $shalist[$fsha[1]] = $fsha[0];
84                         }
85                         //echo "Shasums file: $fname\n";
86                         //var_dump($shalist);
87                 } else if($fname == "MD5SUMS.txt") {
88                         // nothing yet
89                 } else  if(preg_match("/.*linux.*64\.tar\.gz/", $fname)) {
90                         //echo "64 bit version for linux: $fname\n";
91                         $thissha = "-";
92                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
93                         $is_echo = true;
94                         $plat = "linux";
95                         $arch = "x86_64";
96                 } else if(preg_match("/.*linux.*86\.tar\.gz/", $fname)) {
97                         //echo "32 bit version for linux: $fname\n";
98                         $thissha = "-";
99                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
100                         //echo "$vers:$url/$fname:linux:i686:$thissha\n";
101                         $is_echo = true;
102                         $plat = "linux";
103                         $arch = "i686";
104                 } else  if(preg_match("/.*darwin.*86\.tar\.gz/", $fname)) {
105                         //echo "32 bit version for darwin: $fname\n";
106                         $thissha = "-";
107                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
108                         //echo "$vers:$url/$fname:darwin:i686:$thissha\n";
109                         $is_echo = true;
110                         $plat = "darwin";
111                         $arch = "i686";
112                 } else if(preg_match("/.*darwin.*64\.tar\.gz/", $fname)) {
113                         //echo "64 bit version for darwin: $fname\n";
114                         $thissha = "-";
115                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
116                         //echo "$vers:$url/$fname:darwin:x86_64:$thissha\n";
117                         $is_echo = true;
118                         $plat = "darwin";
119                         $arch = "x86_64";
120                 } else if(preg_match("/.*sunos.*64\.tar\.gz/", $fname)) {
121                         // echo "64 bit version for sunos: $fname\n";
122                         $thissha = "-";
123                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
124                         //echo "$vers:$url/$fname:sunos:x86_64:$thissha\n";
125                         $is_echo = true;
126                         $plat = "sunos";
127                         $arch = "x86_64";
128                 } else if(preg_match("/.*sunos.*86\.tar\.gz/", $fname)) {
129                         //echo "32 bit version for sunos: $fname\n";
130                         $thissha = "-";
131                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
132                         //echo "$vers:$url/$fname:sunos:i686:$thissha\n";
133                         $is_echo = true;
134                         $plat = "sunos";
135                         $arch = "i686";
136                 } else if(preg_match("/.*linux-arm.*\.tar\.gz/", $fname)) {
137                         // echo "arm for linux: $fname\n";
138                         $thissha = "-";
139                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
140                         //echo "$vers:$url/$fname:linux:armhf:$thissha\n";
141                         $is_echo = true;
142                         $plat = "linux";
143                         $arch = "armhf";
144                 } else if(preg_match("/node-v[0-9]+\.[0-9]+\.[0-9]+\.tar\.gz/", $fname)) {
145                         $thissha = "-";
146                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
147                         //echo "$vers:$url/$fname:linux:armhf:$thissha\n";
148                         $is_echo = true;
149                         $plat = "src";
150                         $arch = "-";
151                 }
152                 
153                 if($is_echo) {
154                         $fver[$fvern++] = "$vers:$plat:$arch:$url/$fname:$thissha";
155                 }
156         }
157 }
158 ?>