lib/cache.js

   1 var fs = require("fs");
   2 var http = require("http");
   3 var url = require("url");
   4 var path = require("path");
   5 var crypto = require("crypto");
   6 var log = require("./log.js");
   7
   8 function upstreamRequest(unify) {
   9         // first do a head request
  10         log.debug("upsteram as ", unify.requestFor);
  11
  12         var endData = false;
  13         var xpath = "";
  14         var filefd = null;
  15         if(unify.topPath !=null) if(unify.topPath != "") if(typeof global.repoproxy.repo[unify.topPath] != "undefined") {
  16                 var uplink = global.repoproxy.repo[unify.topPath].url;
  17                 xpath = uplink + unify.subPath;
  18         }
  19
  20         //unify.b.write("would send to '" + xpath + "'");
  21         //unify.b.end();
  22
  23         // not doing this properly yet...
  24         if(typeof global.repoproxy.downloads[unify.fullFilePath] != undefined && global.repoproxy.downloads[unify.fullFilePath] == 1) {
  25                 log.debug("request for file thats being downloaded already, doing inline request");
  26                 inlineService(unify);
  27                 return;
  28         }
  29
  30         log.debug("sending off to '%s'", xpath);
  31
  32         var headReq = url.parse(xpath);
  33         headReq["method"] = "HEAD";
  34
  35         getup = http.request(headReq, function(res) {
  36                 //res.setEncoding("utf8");
  37
  38                 if(!endData) {
  39                         log.debug("status code is ", typeof res.statusCode);
  40                         switch(res.statusCode) {
  41                         // TODO: this 301 directory redirect thing needs to work better
  42                         case 301:
  43                         case 302:
  44
  45                                 var loc = res.headers.location.substr(res.headers.location.length-4);
  46                                 var against_t = xpath + "/";
  47                                 var against = against_t.substr(against_t.length-4);
  48
  49                                 if(loc == against) {
  50                                         log.debug("got a redirect, upstream for loc => loc/ assuming its a directory");
  51                                         makeCacheDir(unify);
  52                                         unify.b.writeHead(302, { "Location": unify.originalReq + "/" });
  53                                 } else {
  54                                         log.debug("checked '%s' against '%s', was false, sending 404", loc, against);
  55                                         unify.b.writeHead(404, {"Content-Type": "text/plain"});
  56                                         unify.b.write("404 Not Found\n");
  57                                 }
  58                                 unify.b.end();
  59                                 endData = true;
  60                                 break;
  61
  62                         case 404:
  63                                 unify.b.writeHead(404, {"Content-Type": "text/plain"});
  64                                 unify.b.write("404 Not Found\n");
  65                                 unify.b.end();
  66                                 endData = true;
  67                                 break;
  68                         case 200:
  69                                 makeCacheDir(unify);
  70                                 if(unify.isDirectoryRequest) {
  71                                         serviceDirectory(unify);
  72                                         endData = true;
  73                                 } else {
  74                                         // this is where it gets ugly
  75                                         var filesize = res.headers["content-length"];
  76                                         log.debug("do ugly write: ", unify);
  77                                         //unify.b.write(data);
  78                                         var metafilename = unify.fullPathDirName + "/.meta."+ path.basename(unify.requestFor) +".filesize";
  79                                         var metafile = fs.createWriteStream(metafilename);
  80                                         metafile.write(filesize);
  81                                         metafile.end();
  82                                         getAndService(unify, xpath, filesize);
  83
  84                                 }
  85                                 break;
  86                         default:
  87                                 log.debug(".... data");
  88                                 //unify.b.write(data);
  89                         }
  90                 }
  91                 //log.debug("res is now ", res);
  92         });
  93
  94         getup.end();
  95
  96         //log.debug("getup: ", getup);
  97 }
  98
  99 exports.upstreamRequest = upstreamRequest;
 100
 101 function getAndService(unify, xpath, filesize) {
 102
 103         log.debug("calling in here with filesize, ", filesize)
 104         unify.b.writeHead(200, {'Content-Length' : filesize});
 105
 106         global.repoproxy.downloads[unify.fullFilePath] = 1;
 107
 108         http.get(xpath, function(res) {
 109
 110             var file = fs.createWriteStream(unify.fullFilePath);
 111
 112             //log.debug("res: ", res);
 113
 114             //res.setEncoding("utf8");
 115
 116             res.on("data", function(data) {
 117                     //log.debug("chunk");
 118                     file.write(data);
 119                     unify.b.write(data);
 120             });
 121
 122             res.on("end", function() {
 123                     log.debug("end...");
 124                     unify.b.end();
 125                     file.end();
 126                     global.repoproxy.downloads[unify.fullFilePath] = 0;
 127             });
 128
 129             res.on("error", function(err) {
 130                 log.debug("res threw error... ", err);
 131             });
 132         });
 133 }
 134
 135 // this is nasty nasty thing that can go horribly wrong in some ways, but currently works...
 136 function inlineService(unify) {
 137         // this method is called when we need to service a file thats being downloaded by something else
 138         var metafilename = unify.fullPathDirName + "/.meta."+ path.basename(unify.requestFor) +".filesize";
 139         var fsizef = fs.createReadStream(metafilename);
 140         var fsize = "";
 141         var lastchunk = 0;
 142         fsizef.on("data", function(data) {
 143                 fsize += data;
 144         });
 145
 146         fsizef.on("end", function() {
 147                 var sentSoFar = 0;
 148                 unify.b.writeHead(200, {"Content-Length" : fsize });
 149
 150                 // now we go into the file reading loop.
 151                 log.debug("start of inline services");
 152                 // we loop every 0.5s and do our thing
 153
 154                 function sendPieces() {
 155                         // this is going to be so fun i want to play real life frogger in real life traffic...
 156                         fs.stat(unify.fullFilePath, function(err, stats) {
 157                                 if(err == null) {
 158                                         if(stats["size"] > sentSoFar) {
 159                                                 // if file size changed between last chunk and this chunk, send the chunks
 160
 161                                                 lastChunk = 0;
 162                                                 // open the file, send the data
 163                                                 var rs = fs.createReadStream(unify.fullFilePath, {start: sentSoFar, end: stats["size"]});
 164
 165                                                 rs.on("data", function(thisdata) {
 166                                                         //log.debug("inline chunk: ", thisdata.length);
 167                                                         unify.b.write(thisdata);
 168                                                 });
 169
 170                                                 rs.on("end", function() {
 171                                                         sentSoFar = stats["size"];
 172                                                         // every second, we start again
 173                                                         if(sentSoFar != fsize) {
 174                                                                 setTimeout(sendPieces, 1000);
 175                                                         } else {
 176                                                                 // we're done!
 177                                                                 unify.b.end();
 178                                                         }
 179                                                 });
 180                                         } else {
 181                                                 // if file size did not change between last timeout and this one, incremement the chunk counter
 182                                                 // if we reach 60, we had a problem, and so we bomb out
 183
 184                                                 lastChunk++;
 185
 186                                                 // we bombed out somehow
 187                                                 if(lastChunk > 60) {
 188                                                         unify.b.end();
 189                                                 } else {
 190                                                         setTimeout(sendPieces, 1000);
 191                                                 }
 192                                         }
 193                                 } else {
 194                                         log.error("inline service - we're in a very bad place, how we ended up here we dont know, but we need to crash");
 195                                         process.exit(10);
 196                                 }
 197                         });
 198
 199                 }
 200
 201                 setTimeout(sendPieces, 100);
 202         });
 203 }
 204
 205 // the service file routine .... PLEASE KILL ME!
 206 function serviceFile(unify) {
 207
 208         // for now, ignore range.
 209         // however we need to check if a metadata file exists describing the filesize, check if its all correct
 210         // and if not, erase the file (and metafile) and forward the request back to upstream request
 211
 212
 213         checkFile(unify, function() {
 214
 215                 // file should already exist, so we just poop it out
 216                 var inp = fs.createReadStream(unify.fullFilePath);
 217                 //inp.setEncoding("utf8");
 218                 inp.on("data", function(data) {
 219                         unify.b.write(data);
 220                 });
 221
 222                 inp.on("end", function(closed) {
 223                         unify.b.end();
 224                 });
 225         });
 226 }
 227
 228 exports.serviceFile = serviceFile;
 229
 230
 231 function checkFile(unify, callback) {
 232         // in here we do the metadata checks
 233         var metafilename = unify.fullPathDirName + "/.meta."+ path.basename(unify.requestFor) +".filesize";
 234
 235         fs.exists(metafilename, function(existence) {
 236                 if(existence) {
 237                         var fsizef = fs.createReadStream(metafilename);
 238                         var fsize = "";
 239                         fsizef.on("data", function(data) {
 240                                 fsize += data;
 241                         });
 242
 243                         fsizef.on("end", function() {
 244                                 fs.stat(unify.fullFilePath, function(err, stats) {
 245                                         var rfsize = stats["size"];
 246                                         if(rfsize != fsize.trim()) {
 247                                                 // remove the file and start again
 248                                                 log.debug("reported filesizes dont match, '%s', '%s', removing file and starting again", rfsize, stats["size"]);
 249                                                 try {
 250                                                         fs.unlink(metafilename, function(){
 251                                                                 fs.unlink(unify.fullFilePath, function(){
 252                                                                         upstreamRequest(unify);
 253                                                                 })
 254                                                         });
 255                                                 } catch(e) {
 256                                                         upstreamRequest(unify);
 257                                                 }
 258                                         } else {
 259                                                 // we're good
 260                                                 unify.b.writeHead(200, {"Content-Length" : unify.fileSize})
 261                                                 callback();
 262                                         }
 263                                 });
 264                         });
 265                 } else {
 266                         log.debug("file, '%s' exists but has no filesize meta data, assuming it was put here manually and servicing", unify.fullFilePath);
 267                         unify.b.writeHead(200, {"Content-Length" : unify.fileSize})
 268                         callback();
 269                 }
 270         });
 271 }
 272
 273 function makeCacheDir(path) {
 274         log.debug("attempting to create... '%s' as '%s'", path.fullPathDirName, path.subPathDirName);
 275
 276         var startAt = path.topFullPath;
 277         var nextbits = path.subPathDirName.split("/");
 278         for(var i=0; i < nextbits.length; i++) {
 279                 startAt += "/" + nextbits[i];
 280                 log.debug("attempt mkdir on '%s'", startAt);
 281                 try {
 282                         fs.mkdirSync(startAt);
 283                 } catch(e) {
 284                         //log.debug("e in mkdir, ", e);
 285                 }
 286         }
 287         //process.exit(0);
 288 }
 289
 290 function serviceDirectory(unify) {
 291         var nfiles = 0;
 292         var res = unify.b;
 293
 294         res.write("<html><h1>Directory listing for " + unify.originalReq + "</h1><hr><pre>");
 295         if(unify.originalReq != "/") res.write("<a href=\"..\">Parent</a>\n\n");
 296         fs.readdir(unify.fullFilePath, function(err, files) {
 297                 log.debug("doing directory listing on: ", unify.fullFilePath);
 298                 if(err == null) {
 299
 300                         // TODO: make this work asynchronously...
 301                         for(var i=0; i<files.length; i++) {
 302                                 // avoiding statSync is too hard for now, will fix later TODO: fix this sync bit
 303                                 var stats = fs.statSync(unify.fullFilePath+"/"+files[i]);
 304
 305                                 if(files[i].match(/^\..*/) == null) {
 306                                         if(stats.isDirectory()) {
 307
 308                                                 res.write("Directory: <a href=\""+files[i]+"/\">"+files[i]+"/</a>\n");
 309                                                 nfiles++;
 310                                         } else if(stats.isFile()) {
 311                                                 var padlength = 80 - (files[i].length) - stats.size.toString().length;
 312                                                 var padding = "";
 313                                                 if(padlength > 0) {
 314                                                         padding = new Array(padlength).join(" ");
 315                                                 }
 316                                                 res.write("File:      <a href=\""+files[i]+"\">"+files[i]+"</a>"+padding+stats.size+" bytes\n");
 317                                                 nfiles++;
 318                                         }
 319                                 } else {
 320                                         log.debug("ignoring file, ", files[i]);
 321                                 }
 322                         }
 323
 324                         if(nfiles == 0) res.write("Empty directory....\n");
 325
 326                         res.write("<hr></pre>");
 327                         res.end();
 328                 } else {
 329                         res.write("we have entered bizaro world...\n");
 330                         res.write("</pre>");
 331                         res.end();
 332                 }
 333         });
 334 }
 335
 336 function moveToCleanup(file_or_dir) {
 337         // err..?
 338         var cleanup = global.repoproxy.cacheDir + "/.cleanup";
 339         var ctime = new Date().getTime();
 340         var encoded = (++global.repoproxy.fileid).toString();
 341         var toloc = cleanup + "/" + ctime.toString() + "." + encoded;
 342
 343         //log.debug("Moving %s to %s for cleanup", file_or_dir.replace(/\/$/, ""), toloc);
 344
 345         fs.renameSync(file_or_dir.replace(/\/$/, ""), toloc);
 346 }
 347
 348
 349 exports.serviceDirectory = serviceDirectory;
 350 exports.moveToCleanup = moveToCleanup;