ec419162b033c18106e23b34675e42859618f1ab
[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                 new_json_contents = JSON.stringify(parsed);
105         }
106
107         var new_json = {
108                 parsed: parsed,
109                 plain: new_json_contents,
110                 last_modified: mtime
111         };
112         create_json_historic_diff(new_json, historic_json.slice(0), {}, function(new_diff_json) {
113                 // gzip the new version (non-delta), and put it into place.
114                 zlib.gzip(new_json_contents, function(err, buffer) {
115                         if (err) throw err;
116
117                         new_json.gzip = buffer;
118                         json = new_json;
119                         diff_json = new_diff_json;
120                         json_lock = 0;
121
122                         // Finally, wake up any sleeping clients.
123                         possibly_wakeup_clients();
124                 });
125         });
126 }
127
128 var create_json_historic_diff = function(new_json, history_left, new_diff_json, cb) {
129         if (history_left.length == 0) {
130                 cb(new_diff_json);
131                 return;
132         }
133
134         var histobj = history_left.shift();
135         var diff = delta.JSON_delta.diff(histobj.parsed, new_json.parsed, true);
136         var diff_text = JSON.stringify(diff);
137         zlib.gzip(diff_text, function(err, buffer) {
138                 if (err) throw err;
139                 new_diff_json[histobj.last_modified] = {
140                         parsed: diff,
141                         plain: diff_text,
142                         gzip: buffer,
143                         last_modified: new_json.last_modified,
144                 };
145                 create_json_historic_diff(new_json, history_left, new_diff_json, cb);
146         });
147 }
148
149 var reread_file = function(event, filename) {
150         if (filename != path.basename(json_filename)) {
151                 return;
152         }
153         if (json_lock >= 2) {
154                 return;
155         }
156         if (json_lock == 1) {
157                 // Already processing; wait a bit.
158                 json_lock = 2;
159                 setTimeout(function() { if (json_lock == 2) json_lock = 1; reread_file(event, filename); }, 100);
160                 return;
161         }
162         json_lock = 1;
163
164         console.log("Rereading " + json_filename);
165         fs.open(json_filename, 'r', function(err, fd) {
166                 if (err) throw err;
167                 fs.fstat(fd, function(err, st) {
168                         if (err) throw err;
169                         var buffer = new Buffer(1048576);
170                         fs.read(fd, buffer, 0, 1048576, 0, function(err, bytesRead, buffer) {
171                                 if (err) throw err;
172                                 fs.close(fd, function() {
173                                         var new_json_contents = buffer.toString('utf8', 0, bytesRead);
174                                         replace_json(new_json_contents, st.mtime.getTime());
175                                 });
176                         });
177                 });
178         });
179
180         if (touch_timer !== undefined) {
181                 clearTimeout(touch_timer);
182         }
183         touch_timer = setTimeout(function() {
184                 console.log("Touching analysis.json due to no other activity");
185                 var now = Date.now() / 1000;
186                 fs.utimes(json_filename, now, now);
187         }, 30000);
188 }
189 var possibly_wakeup_clients = function() {
190         var num_viewers = count_viewers();
191         for (var i in sleeping_clients) {
192                 mark_recently_seen(sleeping_clients[i].unique);
193                 send_json(sleeping_clients[i].response,
194                           sleeping_clients[i].ims,
195                           sleeping_clients[i].accept_gzip,
196                           num_viewers);
197         }
198         sleeping_clients = {};
199 }
200 var send_404 = function(response) {
201         response.writeHead(404, {
202                 'Content-Type': 'text/plain',
203         });
204         response.write('Something went wrong. Sorry.');
205         response.end();
206 }
207 var send_json = function(response, ims, accept_gzip, num_viewers) {
208         var this_json = diff_json[ims] || json;
209
210         var headers = {
211                 'Content-Type': 'text/json',
212                 'X-RGLM': this_json.last_modified,
213                 'X-RGNV': num_viewers,
214                 'Access-Control-Expose-Headers': 'X-RGLM, X-RGNV, X-RGMV',
215                 'Vary': 'Accept-Encoding',
216         };
217
218         if (MINIMUM_VERSION) {
219                 headers['X-RGMV'] = MINIMUM_VERSION;
220         }
221
222         if (accept_gzip) {
223                 headers['Content-Length'] = this_json.gzip.length;
224                 headers['Content-Encoding'] = 'gzip';
225                 response.writeHead(200, headers);
226                 response.write(this_json.gzip);
227         } else {
228                 headers['Content-Length'] = this_json.plain.length;
229                 response.writeHead(200, headers);
230                 response.write(this_json.plain);
231         }
232         response.end();
233 }
234 var mark_recently_seen = function(unique) {
235         if (unique) {
236                 last_seen_clients[unique] = (new Date).getTime();
237         }
238 }
239 var count_viewers = function() {
240         if (viewer_count_override !== undefined) {
241                 return viewer_count_override;
242         }
243
244         var now = (new Date).getTime();
245
246         // Go through and remove old viewers, and count them at the same time.
247         var new_last_seen_clients = {};
248         var num_viewers = 0;
249         for (var unique in last_seen_clients) {
250                 if (now - last_seen_clients[unique] < 5000) {
251                         ++num_viewers;
252                         new_last_seen_clients[unique] = last_seen_clients[unique];
253                 }
254         }
255
256         // Also add sleeping clients that we would otherwise assume timed out.
257         for (var request_id in sleeping_clients) {
258                 var unique = sleeping_clients[request_id].unique;
259                 if (unique && !(unique in new_last_seen_clients)) {
260                         ++num_viewers;
261                 }
262         }
263
264         last_seen_clients = new_last_seen_clients;
265         return num_viewers;
266 }
267 var log = function(str) {
268         console.log("[" + ((new Date).getTime()*1e-3).toFixed(3) + "] " + str);
269 }
270
271 // Set up a watcher to catch changes to the file, then do an initial read
272 // to make sure we have a copy.
273 fs.watch(path.dirname(json_filename), reread_file);
274 reread_file(null, path.basename(json_filename));
275
276 if (COUNT_FROM_VARNISH_LOG) {
277         // Note: We abuse serve_url as a regex.
278         var varnishncsa = child_process.spawn(
279                 'varnishncsa', ['-F', '%{%s}t %U %q tffb=%{Varnish:time_firstbyte}x',
280                 '-q', 'ReqURL ~ "^' + serve_url + '"']);
281         var rl = readline.createInterface({
282                 input: varnishncsa.stdout,
283                 output: varnishncsa.stdin,
284                 terminal: false
285         });
286
287         var uniques = [];
288         rl.on('line', function(line) {
289                 var v = line.match(/(\d+) .*\?ims=\d+&unique=(.*) tffb=(.*)/);
290                 if (v) {
291                         uniques[v[2]] = {
292                                 last_seen: (parseInt(v[1]) + parseFloat(v[3])) * 1e3,
293                                 grace: null,
294                         };
295                         log(v[1] + " " + v[2] + " " + v[3]);
296                 } else {
297                         log("VARNISHNCSA UNPARSEABLE LINE: " + line);
298                 }
299         });
300         setInterval(function() {
301                 var mtime = json.last_modified - 1000;  // Compensate for subsecond issues.
302                 var now = (new Date).getTime();
303                 var num_viewers = 0;
304
305                 for (var unique in uniques) {
306                         ++num_viewers;
307                         var last_seen = uniques[unique].last_seen;
308                         if (now - last_seen <= 5000) {
309                                 // We've seen this user in the last five seconds;
310                                 // it's okay.
311                                 continue;
312                         }
313                         if (last_seen >= mtime) {
314                                 // This user has the latest version;
315                                 // they are probably just hanging.
316                                 continue;
317                         }
318                         if (uniques[unique].grace === null) {
319                                 // They have five seconds after a new JSON has been
320                                 // provided to get get it, or they're out.
321                                 // We don't simply use mtime, since we don't want to
322                                 // reset the grace timer just because a new JSON is
323                                 // published.
324                                 uniques[unique].grace = mtime;
325                         }
326                         if (now - uniques[unique].grace > 5000) {
327                                 log("Timing out " + unique + " (last_seen=" + last_seen + ", now=" + now +
328                                         ", mtime=" + mtime, ", grace=" + uniques[unique].grace + ")");
329                                 delete uniques[unique];
330                                 --num_viewers;
331                         }
332                 }
333
334                 log(num_viewers + " entries in hash, mtime=" + mtime);
335                 viewer_count_override = num_viewers;
336         }, 1000);
337 }
338
339 var server = http.createServer();
340 server.on('request', function(request, response) {
341         var u = url.parse(request.url, true);
342         var ims = (u.query)['ims'];
343         var unique = (u.query)['unique'];
344
345         log(request.url);
346         if (u.pathname === hash_serve_url) {
347                 var fen = (u.query)['fen'];
348                 hash_lookup.handle_request(fen, response);
349                 return;
350         }
351         if (u.pathname !== serve_url) {
352                 // This is not the request you are looking for.
353                 send_404(response);
354                 return;
355         }
356
357         mark_recently_seen(unique);
358
359         var accept_encoding = request.headers['accept-encoding'];
360         var accept_gzip;
361         if (accept_encoding !== undefined && accept_encoding.match(/\bgzip\b/)) {
362                 accept_gzip = true;
363         } else {
364                 accept_gzip = false;
365         }
366
367         // If we already have something newer than what the user has,
368         // just send it out and be done with it.
369         if (json !== undefined && (!ims || json.last_modified > ims)) {
370                 send_json(response, ims, accept_gzip, count_viewers());
371                 return;
372         }
373
374         // OK, so we need to hang until we have something newer.
375         // Put the user on the wait list.
376         var client = {};
377         client.response = response;
378         client.request_id = request_id;
379         client.accept_gzip = accept_gzip;
380         client.unique = unique;
381         client.ims = ims;
382         sleeping_clients[request_id++] = client;
383
384         request.socket.client = client;
385 });
386 server.on('connection', function(socket) {
387         socket.on('close', function() {
388                 var client = socket.client;
389                 if (client) {
390                         mark_recently_seen(client.unique);
391                         delete sleeping_clients[client.request_id];
392                 }
393         });
394 });
395
396 server.listen(port);