new version data layout
[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 $versdate = date("Ymd");
53 $versfilename = "./versions_file.$versdate";
54 $versfile = fopen("$versfilename", "w");
55 if($versfile !== false) {
56         fwrite($versfile, "baseurl:$baseurl\n");
57         foreach($fver as $line) {
58                 fwrite($versfile, "$line\n");
59         }
60         fclose($versfile);
61
62         // compress teh file
63         system("gzip -c $versfilename > $versfilename.gz");
64         
65         // create a current versions file
66         $cv = fopen("current_version", "w");
67         if($cv !== false) {
68                 fwrite($cv, "version:$versdate\n");
69                 fclose($cv);
70         }
71 } else {
72         echo "Failed to open vers file!\n";     
73 }
74
75
76 function checkFilesList($url, $vers, $fileslist) {
77         global $fver, $baseurl, $fvern;
78                         
79         echo "Checking $url, $vers\n";
80         
81         $shalist = array();
82         
83         foreach($fileslist[1] as $fname) {
84                 $is_echo = false;
85                 $flurl = "";
86                 $arch = "";
87                 $plat = "";
88                 $thissha = "";
89                 
90                 if($fname == "SHASUMS.txt") {
91                         //$shas = file_get_
92                         $shas_r = file_get_contents("$baseurl/$url/SHASUMS.txt");
93                         $shalist = explode("\n", $shas_r);
94                         foreach($shalist as $ssha) {
95                                 $fsha = preg_split("/[ \t]+/", $ssha);
96                                 if(isset($fsha[1])) $shalist[$fsha[1]] = $fsha[0];
97                         }
98                         //echo "Shasums file: $fname\n";
99                         //var_dump($shalist);
100                 } else if($fname == "MD5SUMS.txt") {
101                         // nothing yet
102                 } else  if(preg_match("/.*linux.*64\.tar\.gz/", $fname)) {
103                         //echo "64 bit version for linux: $fname\n";
104                         $thissha = "-";
105                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
106                         $is_echo = true;
107                         $plat = "linux";
108                         $arch = "x86_64";
109                 } else if(preg_match("/.*linux.*86\.tar\.gz/", $fname)) {
110                         //echo "32 bit version for linux: $fname\n";
111                         $thissha = "-";
112                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
113                         //echo "$vers:$url/$fname:linux:i686:$thissha\n";
114                         $is_echo = true;
115                         $plat = "linux";
116                         $arch = "i686";
117                 } else  if(preg_match("/.*darwin.*86\.tar\.gz/", $fname)) {
118                         //echo "32 bit version for darwin: $fname\n";
119                         $thissha = "-";
120                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
121                         //echo "$vers:$url/$fname:darwin:i686:$thissha\n";
122                         $is_echo = true;
123                         $plat = "darwin";
124                         $arch = "i686";
125                 } else if(preg_match("/.*darwin.*64\.tar\.gz/", $fname)) {
126                         //echo "64 bit version for darwin: $fname\n";
127                         $thissha = "-";
128                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
129                         //echo "$vers:$url/$fname:darwin:x86_64:$thissha\n";
130                         $is_echo = true;
131                         $plat = "darwin";
132                         $arch = "x86_64";
133                 } else if(preg_match("/.*sunos.*64\.tar\.gz/", $fname)) {
134                         // echo "64 bit version for sunos: $fname\n";
135                         $thissha = "-";
136                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
137                         //echo "$vers:$url/$fname:sunos:x86_64:$thissha\n";
138                         $is_echo = true;
139                         $plat = "sunos";
140                         $arch = "x86_64";
141                 } else if(preg_match("/.*sunos.*86\.tar\.gz/", $fname)) {
142                         //echo "32 bit version for sunos: $fname\n";
143                         $thissha = "-";
144                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
145                         //echo "$vers:$url/$fname:sunos:i686:$thissha\n";
146                         $is_echo = true;
147                         $plat = "sunos";
148                         $arch = "i686";
149                 } else if(preg_match("/.*linux-arm.*\.tar\.gz/", $fname)) {
150                         // echo "arm for linux: $fname\n";
151                         $thissha = "-";
152                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
153                         //echo "$vers:$url/$fname:linux:armhf:$thissha\n";
154                         $is_echo = true;
155                         $plat = "linux";
156                         $arch = "armhf";
157                 } else if(preg_match("/node-v[0-9]+\.[0-9]+\.[0-9]+\.tar\.gz/", $fname)) {
158                         $thissha = "-";
159                         if(isset($shalist[$fname])) $thissha = $shalist[$fname];
160                         //echo "$vers:$url/$fname:linux:armhf:$thissha\n";
161                         $is_echo = true;
162                         $plat = "src";
163                         $arch = "-";
164                 }
165                 
166                 if($is_echo) {
167                         $fver[$fvern++] = "$vers:$plat:$arch:$url/$fname:$thissha";
168                 }
169         }
170 }
171 ?>