git ssb

0+

cel / ssb-wikimedia



Tree: caf3d30d63d6aeaa044de2339d4f9a0bd26a80e2

Files: caf3d30d63d6aeaa044de2339d4f9a0bd26a80e2 / bin.js

14767 bytesRaw
1#!/usr/bin/env node
2
3var fs = require('fs')
4var path = require('path')
5var URL = require('url')
6var http = require('http')
7var https = require('https')
8var crypto = require('crypto')
9var readline = require('readline')
10var os = require('os')
11
12var ssbClient = require('ssb-client')
13var pull = require('pull-stream')
14
15var pkg = require('./package')
16
17var userAgentBase = pkg.name + '/' + pkg.version
18var userAgentContact
19var userAgentBot = false
20
21function estimateMessageSize(content) {
22 var draftMsg = {
23 key: '%0000000000000000000000000000000000000000000=.sha256',
24 value: {
25 previous: '%0000000000000000000000000000000000000000000=.sha256',
26 author: '@0000000000000000000000000000000000000000000=.ed25519',
27 sequence: 100000,
28 timestamp: 1000000000000.0001,
29 hash: 'sha256',
30 content: content,
31 signature: '00000000000000000000000000000000000000000000000000000000000000000000000000000000000000==.sig.ed25519'
32 }
33 }
34 return JSON.stringify(draftMsg, null, 2).length
35}
36
37function mapCollect(fn) {
38 var aborted
39 return function (read) {
40 var queue = []
41 return function (abort, cb) {
42 if (aborted = abort) return read(abort, cb)
43 read(null, function next(end, data) {
44 if (end) return cb(end)
45 queue.push(data)
46 var result = fn(queue)
47 if (result) read(null, next)
48 })
49 }
50 }
51}
52
53function getJson(url, cb) {
54 var opts = URL.parse(url)
55 opts.headers = {
56 'User-Agent': userAgentBase
57 + (userAgentContact ? ' (' + userAgentContact + ')' : '')
58 + (userAgentBot ? ' bot' : '')
59 }
60 var h = opts.protocol === 'https:' ? https : http
61 h.get(opts, function (res) {
62 if (res.statusCode !== 200) return cb(new Error('HTTP ' + res.statusCode + ' ' + res.statusMessage))
63 var bufs = []
64 res.on('data', function (buf) {
65 bufs.push(buf)
66 })
67 res.on('end', function () {
68 res.removeListener('error', cb)
69 var buf = Buffer.concat(bufs)
70 bufs = null
71 var data
72 try {
73 data = JSON.parse(buf.toString('utf8'))
74 } catch(e) {
75 return cb(e)
76 }
77 cb(null, data)
78 })
79 res.on('error', cb)
80 })
81}
82
83function publishDrafts(sbot, drafts, cb) {
84 var draftIdIndex = {}
85 drafts.forEach(function (draft, i) {
86 draftIdIndex[draft.draftId] = i
87 })
88 var ids = []
89
90 function replaceDraftIds(obj) {
91 if (typeof obj === 'string') {
92 var i = draftIdIndex[obj]
93 if (typeof i === 'number') {
94 var id = ids[i]
95 if (!id) throw new ReferenceError('draft referernces unknown message')
96 return id
97 }
98 } else if (Array.isArray(obj)) {
99 return obj.map(replaceDraftIds)
100 } else if (obj !== null && typeof obj === 'object') {
101 var o = {}
102 for (var k in obj) o[k] = replaceDraftIds(obj[k])
103 return o
104 }
105 return obj
106 }
107
108 pull(
109 pull.values(drafts),
110 pull.asyncMap(function (draft, cb) {
111 var content = replaceDraftIds(draft.content)
112 sbot.publish(content, function (err, msg) {
113 if (err) return cb(err)
114 ids.push(msg.key)
115 cb(null, msg)
116 })
117 }),
118 pull.collect(cb)
119 )
120}
121
122var args = process.argv.slice(2)
123var yes = false
124var dry = false
125var help = false
126var urls = []
127args.forEach(function (arg) {
128 if (arg[0] === '-') switch (arg) {
129 case '-n': return dry = true
130 case '-y': return yes = true
131 case '-h': return help = true
132 default: throw 'Unknown argument: ' + arg
133 } else urls.push(arg)
134})
135
136if (help) {
137 process.stdout.write(fs.readFileSync(path.join(__dirname, 'usage.txt')))
138 process.exit(0)
139}
140
141ssbClient(function (err, sbot, config) {
142 if (err) throw err
143 var conf = config.wikimedia || {}
144 userAgentContact = conf.contact
145 userAgentBot = conf.bot
146
147 if (urls.length === 0) {
148 var pagesFile = path.join(config.path, 'wikimedia-pages.txt')
149 var pagesData = fs.readFileSync(pagesFile, 'utf8')
150 urls = pagesData.split('\n').filter(RegExp.prototype.test.bind(/[^#]/))
151 if (!urls.length) {
152 console.log('No pages to sync.')
153 return sbot.close()
154 }
155 }
156
157 var pagesInfo = urls.map(function (page) {
158 var m = /^(.*?)\/wiki\/(.*)$/.exec(page)
159 if (!m) throw 'Unable to parse page URL ' + page
160 return {
161 site: m[1] + '/',
162 api: m[1] + '/w/api.php',
163 title: m[2]
164 }
165 })
166 var pagesInfoByApi = {}
167 pagesInfo.forEach(function (pageInfo) {
168 var infos = pagesInfoByApi[pageInfo.api] || (pagesInfoByApi[pageInfo.api] = [])
169 infos.push(pageInfo)
170 })
171 console.log('Normalizing titles...')
172 var waiting = 0
173 for (var api in pagesInfoByApi) (function (api) {
174 var pagesInfoForApi = pagesInfoByApi[api]
175 var pagesInfoForApiByTitle = {}
176 var titles = pagesInfoForApi.map(function (info) {
177 pagesInfoForApiByTitle[info.title] = info
178 return info.title
179 })
180 var url = api + '?format=json&action=query' +
181 '&titles=' + encodeURIComponent('\x1f' + titles.join('\x1f'))
182 waiting++
183 getJson(url, function (err, data) {
184 if (err) throw err
185 if (data.warnings) console.trace('Warnings:', data.warnings)
186 if (data.query.normalized) data.query.normalized.forEach(function (norm) {
187 var info = pagesInfoForApiByTitle[norm.from]
188 if (!info) {
189 console.error(JSON.stringify({titles: titles, response: data}, 0, 2))
190 throw new Error('Unexpected title in server response')
191 }
192 // console.log('Normalized title', norm.from, norm.to)
193 info.title = norm.to
194 })
195 if (!--waiting) next()
196 })
197 }(api))
198
199 function next() {
200 console.log('Getting revisions...')
201 var userHashes = {}
202 pull(
203 pull.values(pagesInfo),
204 pull.asyncMap(function (pageInfo, cb) {
205 // Calculate blob id for page URL + title, for linking
206 pull(
207 pull.once(pageInfo.site + '\t' + pageInfo.title),
208 sbot.blobs.add(function (err, hash) {
209 pageInfo.hash = hash
210 cb(null, pageInfo)
211 })
212 )
213 }),
214 pull.asyncMap(function (pageInfo, cb) {
215 // Get previous messages for this page.
216 // Simple solution: find the revision with latest timestamp.
217 var maxRevTs = ''
218 var maxRevMsgId
219 pull(
220 sbot.links({
221 dest: pageInfo.hash,
222 rel: 'pageId',
223 values: true,
224 meta: false
225 }),
226 pull.filter(function (msg) {
227 var c = msg && msg.value && msg.value.content
228 return c
229 && c.type === 'wikimedia/revisions'
230 && c.site === pageInfo.site
231 && c.title === pageInfo.title
232 }),
233 pull.drain(function (msg) {
234 var c = msg && msg.value && msg.value.content
235 var revs = Array.isArray(c.revisions) && c.revisions
236 if (revs) revs.forEach(function (rev) {
237 if (rev && rev.timestamp > maxRevTs) {
238 maxRevTs = rev.timestamp
239 maxRevMsgId == msg.key
240 }
241 })
242 }, function (err) {
243 if (err) return cb(err)
244 pageInfo.latestMsgId = maxRevMsgId
245 pageInfo.latestRevTs = maxRevTs
246 cb(null, pageInfo)
247 })
248 )
249 }),
250 pull.map(function (pageInfo) {
251 // Get new revisions.
252 var rvcontinue, rvdone
253 var rvstart = pageInfo.latestRevTs
254 var prevId = pageInfo.latestMsgId
255 var aborted
256 var revisions = pull(
257 function (abort, cb) {
258 if (aborted = abort) return cb(abort)
259 if (rvdone) return cb(true)
260 console.log('Getting revisions for', pageInfo.title + '...',
261 rvstart || '', rvcontinue || '')
262 var url = api + '?format=json&action=query&prop=revisions&rvslots=*'
263 + '&titles=' + encodeURIComponent(pageInfo.title)
264 + '&rvprop=ids|timestamp|comment|user|sha1|size|slotsha1|slotsize|content|roles|flags|tags'
265 + '&rvdir=newer'
266 + (rvcontinue ? '&rvcontinue=' + rvcontinue : '')
267 + (rvstart ? '&rvstart=' + rvstart : '')
268 + '&rvlimit=50'
269 getJson(url, function (err, data) {
270 if (aborted) return err && console.trace(err)
271 if (err) return cb(err)
272
273 var warnings = data.warnings
274 if (warnings) {
275 if (warnings.main) {
276 if (warnings.main['*'] === 'Unrecognized parameter: rvslots.') {
277 delete warnings.main['*']
278 if (Object.keys(warnings.main).length === 0) {
279 delete warnings.main
280 }
281 }
282 }
283 if (warnings.revisions) {
284 if (warnings.revisions['*'] === 'Unrecognized values for parameter "rvprop": slotsha1, slotsize, roles.') {
285 delete warnings.revisions['*']
286 if (Object.keys(warnings.revisions).length === 0) {
287 delete warnings.revisions
288 }
289 }
290 }
291 if (Object.keys(warnings).length > 0) {
292 console.trace('Warnings:', warnings)
293 }
294 }
295
296 rvcontinue = data.continue && data.continue.rvcontinue
297 if (!rvcontinue) rvdone = true
298 var page
299 if (data.query) for (var pageid in data.query.pages) {
300 page = data.query.pages[pageid]
301 if (page.title === pageInfo.title) break
302 else page = null
303 }
304 if (!page) {
305 console.trace(data.query.pages, pageInfo)
306 return cb(new Error('Unable to find page'))
307 }
308 var revs = page.revisions
309 if (!revs) {
310 console.trace(page, pageInfo)
311 return cb(new Error('Unable to get revisions'))
312 }
313 console.log('Got ' + page.revisions.length + ' revisions')
314 cb(null, page.revisions)
315 })
316 },
317 pull.flatten(),
318
319 pull.through(function (rev) {
320 if (!rev.slots) {
321 // old API does not use slots.
322 // Transform result to be forward-compatible.
323 rev.slots = {
324 main: {
325 size: rev.size,
326 sha1: rev.sha1,
327 contentmodel: rev.contentmodel,
328 contentformat: rev.contentformat,
329 '*': rev['*']
330 }
331 }
332 delete rev.contentmodel
333 delete rev.contentformat
334 delete rev['*']
335 }
336 // duplicate values supplied in new API in slotsize and slotsha1
337 delete rev.sha1
338 delete rev.size
339 }),
340
341 pull.asyncMap(function (rev, cb) {
342 // Calculate blob id for user page URL + title, for linking
343 var hash = userHashes[rev.user]
344 if (hash) {
345 rev.userId = hash
346 return cb(null, rev)
347 }
348 pull(
349 pull.once(pageInfo.site + '\tUser:' + rev.user),
350 sbot.blobs.add(function (err, hash) {
351 rev.userId = userHashes[rev.user] = hash
352 cb(null, rev)
353 })
354 )
355 }),
356 pull.asyncMap(function (rev, cb) {
357 var waiting = 0
358 for (var slot in rev.slots) (function (slot) {
359 waiting++
360 var slotInfo = rev.slots[slot]
361 var content = slotInfo['*']
362 if (!content) {
363 console.trace(slotInfo)
364 return cb(new Error('Missing content'))
365 }
366 var sha1 = crypto.createHash('sha1').update(content).digest('hex')
367 if (sha1 !== slotInfo.sha1) {
368 console.trace(slotInfo, sha1)
369 return cb(new Error('Mismatched content sha1'))
370 }
371 pull(
372 pull.once(content),
373 sbot.blobs.add(function (err, hash) {
374 if (err) return cb(err)
375 slotInfo.link = hash
376 delete slotInfo['*']
377 if (!--waiting) cb(null, rev)
378 })
379 )
380 }(slot))
381 })
382 )
383
384 var queuedRevisions = []
385 var ended
386 function cbDraft(content, cb) {
387 if (!content.revisions.length) {
388 console.log('No revisions for', pageInfo.title)
389 return cb(true)
390 }
391 console.log('Prepared a message',
392 'with', content.revisions.length, 'revisions',
393 'for', pageInfo.title)
394 prevId = '%' + crypto.createHash('sha256').update(JSON.stringify(content)).digest('base64') + '.draft6'
395 cb(null, {
396 draftId: prevId,
397 content: content
398 })
399 }
400 return function (abort, cb) {
401 if (abort) return revisions(abort, cb)
402 if (ended) return cb(true)
403 var content = {
404 type: 'wikimedia/revisions',
405 site: pageInfo.site,
406 title: pageInfo.title,
407 pageId: pageInfo.hash,
408 parents: prevId ? [prevId] : undefined,
409 revisions: queuedRevisions.splice(0)
410 }
411 revisions(null, function next(end, revision) {
412 if (ended = end) return cbDraft(content, cb)
413 content.revisions.push(revision)
414 if (estimateMessageSize(content) > 8192) {
415 queuedRevisions.push(content.revisions.pop())
416 // console.log('filled msg for ', pageInfo.title, ' with ', content.revisions.length, 'revisions')
417 return cbDraft(content, cb)
418 }
419 revisions(null, next)
420 })
421 }
422 }),
423 pull.flatten(),
424 pull.collect(function (err, drafts) {
425 if (err) throw err
426 if (dry) {
427 console.log(JSON.stringify(drafts, 0, 2))
428 return sbot.close()
429 }
430 if (yes) return confirmed(true)
431 var rl = readline.createInterface({
432 input: process.stdin,
433 output: process.stdout
434 })
435 rl.question('Publish ' + drafts.length + ' messages? [Y/n] ', function (answer) {
436 rl.close()
437 confirmed(!/^n/i.test(answer))
438 })
439 function confirmed(yes) {
440 if (!yes) return sbot.close()
441 publishDrafts(sbot, drafts, function (err, msgs) {
442 if (err) throw err
443 console.log('Published:\n' + msgs.map(function (msg) {
444 return msg.key
445 }.join('\n')))
446 sbot.close()
447 })
448 }
449 })
450 )
451 }
452})
453

Built with git-ssb-web