Merge branch 'master' of /home/magne/remoteglot
[remoteglot] / server / serve-analysis.js
1 // node.js version of analysis.pl; hopefully scales a bit better
2 // for this specific kind of task.
3
4 // Modules.
5 var http = require('http');
6 var fs = require('fs');
7 var url = require('url');
8 var querystring = require('querystring');
9 var path = require('path');
10 var zlib = require('zlib');
11 var readline = require('readline');
12 var child_process = require('child_process');
13 var delta = require('../www/js/json_delta.js');
14 var hash_lookup = require('./hash-lookup.js');
15
16 // Constants.
17 var HISTORY_TO_KEEP = 5;
18 var MINIMUM_VERSION = null;
19 var COUNT_FROM_VARNISH_LOG = true;
20
21 // Filename to serve.
22 var json_filename = '/srv/analysis.sesse.net/www/analysis.json';
23 if (process.argv.length >= 3) {
24         json_filename = process.argv[2];
25 }
26
27 // Expected destination filenames.
28 var serve_url = '/analysis.pl';
29 var hash_serve_url = '/hash';
30 if (process.argv.length >= 4) {
31         serve_url = process.argv[3];
32 }
33 if (process.argv.length >= 5) {
34         hash_serve_url = process.argv[4];
35 }
36
37 // TCP port to listen on.
38 var port = 5000;
39 if (process.argv.length >= 6) {
40         port = parseInt(process.argv[5]);
41 }
42
43 // gRPC backends.
44 var grpc_backends = ["localhost:50051", "localhost:50052"];
45 if (process.argv.length >= 7) {
46         grpc_backends = process.argv[6].split(",");
47 }
48 hash_lookup.init(grpc_backends);
49
50 // If set to 1, we are already processing a JSON update and should not
51 // start a new one. If set to 2, we are _also_ having one in the queue.
52 var json_lock = 0;
53
54 // The current contents of the file to hand out, and its last modified time.
55 var json = undefined;
56
57 // The last five timestamps, and diffs from them to the latest version.
58 var historic_json = [];
59 var diff_json = {};
60
61 // The list of clients that are waiting for new data to show up.
62 // Uniquely keyed by request_id so that we can take them out of
63 // the queue if they close the socket.
64 var sleeping_clients = {};
65 var request_id = 0;
66
67 // List of when clients were last seen, keyed by their unique ID.
68 // Used to show a viewer count to the user.
69 var last_seen_clients = {};
70
71 // The timer used to touch the file every 30 seconds if nobody
72 // else does it for us. This makes sure we don't have clients
73 // hanging indefinitely (which might have them return errors).
74 var touch_timer = undefined;
75
76 // If we are behind Varnish, we can't count the number of clients
77 // ourselves, so we need to get it from parsing varnishncsa.
78 var viewer_count_override = undefined;
79
80 var replace_json = function(new_json_contents, mtime) {
81         // Generate the list of diffs from the last five versions.
82         if (json !== undefined) {
83                 // If two versions have the same mtime, clients could have either.
84                 // Note the fact, so that we never insert it.
85                 if (json.last_modified == mtime) {
86                         json.invalid_base = true;
87                 }
88                 if (!json.invalid_base) {
89                         historic_json.push(json);
90                         if (historic_json.length > HISTORY_TO_KEEP) {
91                                 historic_json.shift();
92                         }
93                 }
94         }
95
96         var parsed = JSON.parse(new_json_contents);
97
98         if (parsed['internal']) {
99                 if (parsed['internal']['grpc_backends'] &&
100                     hash_lookup.need_reinit(parsed['internal']['grpc_backends'])) {
101                         hash_lookup.init(parsed['internal']['grpc_backends']);
102                 }
103                 delete parsed['internal'];
104         }
105
106         var new_json = {
107                 parsed: parsed,
108                 plain: new_json_contents,
109                 last_modified: mtime
110         };
111         create_json_historic_diff(new_json, historic_json.slice(0), {}, function(new_diff_json) {
112                 // gzip the new version (non-delta), and put it into place.
113                 zlib.gzip(new_json_contents, function(err, buffer) {
114                         if (err) throw err;
115
116                         new_json.gzip = buffer;
117                         json = new_json;
118                         diff_json = new_diff_json;
119                         json_lock = 0;
120
121                         // Finally, wake up any sleeping clients.
122                         possibly_wakeup_clients();
123                 });
124         });
125 }
126
127 var create_json_historic_diff = function(new_json, history_left, new_diff_json, cb) {
128         if (history_left.length == 0) {
129                 cb(new_diff_json);
130                 return;
131         }
132
133         var histobj = history_left.shift();
134         var diff = delta.JSON_delta.diff(histobj.parsed, new_json.parsed);
135         var diff_text = JSON.stringify(diff);
136         zlib.gzip(diff_text, function(err, buffer) {
137                 if (err) throw err;
138                 new_diff_json[histobj.last_modified] = {
139                         parsed: diff,
140                         plain: diff_text,
141                         gzip: buffer,
142                         last_modified: new_json.last_modified,
143                 };
144                 create_json_historic_diff(new_json, history_left, new_diff_json, cb);
145         });
146 }
147
148 var reread_file = function(event, filename) {
149         if (filename != path.basename(json_filename)) {
150                 return;
151         }
152         if (json_lock >= 2) {
153                 return;
154         }
155         if (json_lock == 1) {
156                 // Already processing; wait a bit.
157                 json_lock = 2;
158                 setTimeout(function() { if (json_lock == 2) json_lock = 1; reread_file(event, filename); }, 100);
159                 return;
160         }
161         json_lock = 1;
162
163         console.log("Rereading " + json_filename);
164         fs.open(json_filename, 'r', function(err, fd) {
165                 if (err) throw err;
166                 fs.fstat(fd, function(err, st) {
167                         if (err) throw err;
168                         var buffer = new Buffer(1048576);
169                         fs.read(fd, buffer, 0, 1048576, 0, function(err, bytesRead, buffer) {
170                                 if (err) throw err;
171                                 fs.close(fd, function() {
172                                         var new_json_contents = buffer.toString('utf8', 0, bytesRead);
173                                         replace_json(new_json_contents, st.mtime.getTime());
174                                 });
175                         });
176                 });
177         });
178
179         if (touch_timer !== undefined) {
180                 clearTimeout(touch_timer);
181         }
182         touch_timer = setTimeout(function() {
183                 console.log("Touching analysis.json due to no other activity");
184                 var now = Date.now() / 1000;
185                 fs.utimes(json_filename, now, now);
186         }, 30000);
187 }
188 var possibly_wakeup_clients = function() {
189         var num_viewers = count_viewers();
190         for (var i in sleeping_clients) {
191                 mark_recently_seen(sleeping_clients[i].unique);
192                 send_json(sleeping_clients[i].response,
193                           sleeping_clients[i].ims,
194                           sleeping_clients[i].accept_gzip,
195                           num_viewers);
196         }
197         sleeping_clients = {};
198 }
199 var send_404 = function(response) {
200         response.writeHead(404, {
201                 'Content-Type': 'text/plain',
202         });
203         response.write('Something went wrong. Sorry.');
204         response.end();
205 }
206 var send_json = function(response, ims, accept_gzip, num_viewers) {
207         var this_json = diff_json[ims] || json;
208
209         var headers = {
210                 'Content-Type': 'text/json',
211                 'X-RGLM': this_json.last_modified,
212                 'X-RGNV': num_viewers,
213                 'Access-Control-Expose-Headers': 'X-RGLM, X-RGNV, X-RGMV',
214                 'Vary': 'Accept-Encoding',
215         };
216
217         if (MINIMUM_VERSION) {
218                 headers['X-RGMV'] = MINIMUM_VERSION;
219         }
220
221         if (accept_gzip) {
222                 headers['Content-Length'] = this_json.gzip.length;
223                 headers['Content-Encoding'] = 'gzip';
224                 response.writeHead(200, headers);
225                 response.write(this_json.gzip);
226         } else {
227                 headers['Content-Length'] = this_json.plain.length;
228                 response.writeHead(200, headers);
229                 response.write(this_json.plain);
230         }
231         response.end();
232 }
233 var mark_recently_seen = function(unique) {
234         if (unique) {
235                 last_seen_clients[unique] = (new Date).getTime();
236         }
237 }
238 var count_viewers = function() {
239         if (viewer_count_override !== undefined) {
240                 return viewer_count_override;
241         }
242
243         var now = (new Date).getTime();
244
245         // Go through and remove old viewers, and count them at the same time.
246         var new_last_seen_clients = {};
247         var num_viewers = 0;
248         for (var unique in last_seen_clients) {
249                 if (now - last_seen_clients[unique] < 5000) {
250                         ++num_viewers;
251                         new_last_seen_clients[unique] = last_seen_clients[unique];
252                 }
253         }
254
255         // Also add sleeping clients that we would otherwise assume timed out.
256         for (var request_id in sleeping_clients) {
257                 var unique = sleeping_clients[request_id].unique;
258                 if (unique && !(unique in new_last_seen_clients)) {
259                         ++num_viewers;
260                 }
261         }
262
263         last_seen_clients = new_last_seen_clients;
264         return num_viewers;
265 }
266 var log = function(str) {
267         console.log("[" + ((new Date).getTime()*1e-3).toFixed(3) + "] " + str);
268 }
269
270 // Set up a watcher to catch changes to the file, then do an initial read
271 // to make sure we have a copy.
272 fs.watch(path.dirname(json_filename), reread_file);
273 reread_file(null, path.basename(json_filename));
274
275 if (COUNT_FROM_VARNISH_LOG) {
276         // Note: We abuse serve_url as a regex.
277         var varnishncsa = child_process.spawn(
278                 'varnishncsa', ['-F', '%{%s}t %U %q tffb=%{Varnish:time_firstbyte}x',
279                 '-q', 'ReqURL ~ "^' + serve_url + '"']);
280         var rl = readline.createInterface({
281                 input: varnishncsa.stdout,
282                 output: varnishncsa.stdin,
283                 terminal: false
284         });
285
286         var uniques = [];
287         rl.on('line', function(line) {
288                 var v = line.match(/(\d+) .*\?ims=\d+&unique=(.*) tffb=(.*)/);
289                 if (v) {
290                         uniques[v[2]] = {
291                                 last_seen: (parseInt(v[1]) + parseFloat(v[3])) * 1e3,
292                                 grace: null,
293                         };
294                         log(v[1] + " " + v[2] + " " + v[3]);
295                 } else {
296                         log("VARNISHNCSA UNPARSEABLE LINE: " + line);
297                 }
298         });
299         setInterval(function() {
300                 var mtime = json.last_modified - 1000;  // Compensate for subsecond issues.
301                 var now = (new Date).getTime();
302                 var num_viewers = 0;
303
304                 for (var unique in uniques) {
305                         ++num_viewers;
306                         var last_seen = uniques[unique].last_seen;
307                         if (now - last_seen <= 5000) {
308                                 // We've seen this user in the last five seconds;
309                                 // it's okay.
310                                 continue;
311                         }
312                         if (last_seen >= mtime) {
313                                 // This user has the latest version;
314                                 // they are probably just hanging.
315                                 continue;
316                         }
317                         if (uniques[unique].grace === null) {
318                                 // They have five seconds after a new JSON has been
319                                 // provided to get get it, or they're out.
320                                 // We don't simply use mtime, since we don't want to
321                                 // reset the grace timer just because a new JSON is
322                                 // published.
323                                 uniques[unique].grace = mtime;
324                         }
325                         if (now - uniques[unique].grace > 5000) {
326                                 log("Timing out " + unique + " (last_seen=" + last_seen + ", now=" + now +
327                                         ", mtime=" + mtime, ", grace=" + uniques[unique].grace + ")");
328                                 delete uniques[unique];
329                                 --num_viewers;
330                         }
331                 }
332
333                 log(num_viewers + " entries in hash, mtime=" + mtime);
334                 viewer_count_override = num_viewers;
335         }, 1000);
336 }
337
338 var server = http.createServer();
339 server.on('request', function(request, response) {
340         var u = url.parse(request.url, true);
341         var ims = (u.query)['ims'];
342         var unique = (u.query)['unique'];
343
344         log(request.url);
345         if (u.pathname === hash_serve_url) {
346                 var fen = (u.query)['fen'];
347                 hash_lookup.handle_request(fen, response);
348                 return;
349         }
350         if (u.pathname !== serve_url) {
351                 // This is not the request you are looking for.
352                 send_404(response);
353                 return;
354         }
355
356         mark_recently_seen(unique);
357
358         var accept_encoding = request.headers['accept-encoding'];
359         var accept_gzip;
360         if (accept_encoding !== undefined && accept_encoding.match(/\bgzip\b/)) {
361                 accept_gzip = true;
362         } else {
363                 accept_gzip = false;
364         }
365
366         // If we already have something newer than what the user has,
367         // just send it out and be done with it.
368         if (json !== undefined && (!ims || json.last_modified > ims)) {
369                 send_json(response, ims, accept_gzip, count_viewers());
370                 return;
371         }
372
373         // OK, so we need to hang until we have something newer.
374         // Put the user on the wait list.
375         var client = {};
376         client.response = response;
377         client.request_id = request_id;
378         client.accept_gzip = accept_gzip;
379         client.unique = unique;
380         client.ims = ims;
381         sleeping_clients[request_id++] = client;
382
383         request.socket.client = client;
384 });
385 server.on('connection', function(socket) {
386         socket.on('close', function() {
387                 var client = socket.client;
388                 if (client) {
389                         mark_recently_seen(client.unique);
390                         delete sleeping_clients[client.request_id];
391                 }
392         });
393 });
394
395 server.listen(port);