git ssb

1
var buffered = require('pull-buffered')
2
var pull = require('pull-stream')
3
var toPull = require('stream-to-pull-stream')
4
var pako = require('pako')
5
var createHash = require('./util').createHash
6
var cat = require('pull-cat')
7
8
exports.decode = decodePack
9
exports.encode = encodePack
10
11
var PACK_VERSION = 2
12
13
var objectTypes = [
14
  'none', 'commit', 'tree', 'blob',
15
  'tag', 'unused', 'ofs-delta', 'ref-delta'
16
]
17
var objectTypeNums = {
18
  commit: 1,
19
  tree: 2,
20
  blob: 3,
21
  tag: 4,
22
  'ofs-delta': 6,
23
  'ref-delta': 7
24
}
25
26
function error(cb) {
27
  return function (err) {
28
    cb(err || true)
29
  }
30
}
31
32
function inflateBytes(read) {
33
  var inflate = new pako.Inflate()
34
  var ended, dataOut
35
36
  inflate.onData = function (data) {
37
    dataOut = new Buffer(data)
38
    // console.error('inflated data', data.length)
39
  }
40
41
  inflate.onEnd = function (status) {
42
    ended = (status === 0) ? true : new Error(inflate.strm.msg)
43
    // console.error('inflated end', status, ended)
44
  }
45
46
  return function (abort, cb) {
47
    if (ended) return cb(ended)
48
    read(abort, function next(end, data) {
49
      if (end === true) {
50
        end = null
51
        data = []
52
      }
53
      if (ended = end) return cb(end)
54
      if (data.length > 1) return cb(new Error('got more than one byte'))
55
      dataOut = null
56
      inflate.push(data, end === true)
57
      if (dataOut)
58
        cb(null, dataOut)
59
      else if (ended)
60
        cb(ended)
61
      else
62
        // let the stack unwind
63
        setImmediate(function () {
64
          read(null, next)
65
        })
66
    })
67
  }
68
}
69
70
function deflate(read) {
71
  var def = new pako.Deflate()
72
  var queue = []
73
  var ended
74
75
  def.onData = function (data) {
76
    queue.push([null, new Buffer(data)])
77
  }
78
79
  def.onEnd = function (status) {
80
    queue.push([(status === 0) ? true : new Error(def.strm.msg)])
81
  }
82
83
  return function readOut(abort, cb) {
84
    if (ended)
85
      cb(ended)
86
    else if (queue.length)
87
      cb.apply(this, queue.shift())
88
    else
89
      read(abort, function next(end, data) {
90
        if (end === true) def.push([], true)
91
        else if (end) return cb(end)
92
        else def.push(data)
93
        readOut(null, cb)
94
      })
95
  }
96
}
97
98
function decodePack(opts, repo, onEnd, read) {
99
  if (read === undefined)
100
    return decodePack.bind(this, opts, repo, onEnd)
101
102
  var ended
103
  var numObjects = -1
104
  var checksum = createHash('sha1')
105
  var b = buffered(read)
106
  // TODO: optimize to pass through buffers to checksum
107
  var readByte = checksum(b.chunks(1))
108
  var readWord = checksum(b.chunks(4))
109
  var readHash = checksum(b.chunks(20))
110
  var readChecksum = b.chunks(20)
111
  var expectChecksum = true
112
113
  function readHeader(cb) {
114
    readWord(null, function (end, header) {
115
      if (ended = end) return cb(end)
116
      if (!header.equals(header, new Buffer('PACK')))
117
        read(new Error('Invalid packfile header'), error(cb))
118
      else
119
        readVersion(cb)
120
    })
121
  }
122
123
  function readVersion(cb) {
124
    readWord(null, function (end, word) {
125
      if (ended = end) return cb(end)
126
      var version = word.readUInt32BE()
127
      if (version < 2 || version > 3)
128
        read(new Error('Invalid packfile version ' + version), error(cb))
129
      else
130
        readNumObjects(cb)
131
    })
132
  }
133
134
  function readNumObjects(cb) {
135
    readWord(null, function (end, word) {
136
      if (ended = end) return cb(end)
137
      numObjects = word.readUInt32BE()
138
      if (opts.verbosity >= 2)
139
        console.error(numObjects + ' objects')
140
      if (opts.onHeader)
141
        opts.onHeader(numObjects)
142
      readObject(null, cb)
143
    })
144
  }
145
146
  function readTypedVarInt(cb) {
147
    var type, value, shift
148
    // https://codewords.recurse.com/images/three/varint.svg
149
    readByte(null, function (end, buf) {
150
      if (ended = end) return cb(end)
151
      var firstByte = buf[0]
152
      type = objectTypes[(firstByte >> 4) & 7]
153
      value = firstByte & 15
154
      // console.error('byte1', firstByte, firstByte.toString(2))
155
      shift = 4
156
      checkByte(firstByte)
157
    })
158
159
    function checkByte(byte) {
160
      if (byte & 0x80)
161
        readByte(null, gotByte)
162
      else
163
        cb(null, type, value)
164
    }
165
166
    function gotByte(end, buf) {
167
      if (ended = end) return cb(end)
168
      var byte = buf[0]
169
      value += (byte & 0x7f) << shift
170
      shift += 7
171
      // console.error('byte', byte, byte.toString(2))
172
      checkByte(byte)
173
    }
174
  }
175
176
  function getObject(cb) {
177
    readTypedVarInt(function (end, type, length) {
178
      if (opts.verbosity >= 2)
179
        console.error('read object header', end, type, length)
180
      numObjects--
181
      if (end === true && expectChecksum)
182
        onEnd(new Error('Missing checksum'))
183
      if (ended = end) return cb(end)
184
      // TODO: verify that the inflated data is the correct length
185
      if (type == 'ref-delta')
186
        getObjectFromRefDelta(length, cb)
187
      else
188
        cb(null, {
189
          type: type,
190
          length: length,
191
          read: inflateBytes(readByte)
192
        })
193
    })
194
  }
195
196
  // TODO: test with ref-delta objects in pack
197
  function getObjectFromRefDelta(length, cb) {
198
    readHash(null, function (end, sourceHash) {
199
      if (end) return cb(end)
200
      sourceHash = sourceHash.toString('hex')
201
      var b = buffered(inflateBytes(readByte))
202
      var readInflatedByte = b.chunks(1)
203
      readVarInt(readInflatedByte, function (err, expectedSourceLength) {
204
        if (err) return cb(err)
205
        readVarInt(readInflatedByte, function (err, expectedTargetLength) {
206
          if (err) return cb(err)
207
          if (opts.verbosity >= 3)
208
            console.error('getting object', sourceHash)
209
          repo.getObject(sourceHash, function (err, sourceObject) {
210
            if (opts.verbosity >= 3)
211
              console.error('got object', sourceHash, sourceObject, err)
212
            if (err) return cb(err)
213
            if (sourceObject.length != expectedSourceLength)
214
              cb(new Error('Incorrect source object size in ref delta'))
215
            else
216
              patchObject(opts, b, length, sourceObject,
217
                expectedTargetLength, cb)
218
          })
219
        })
220
      })
221
    })
222
  }
223
224
  function readTrailer(cb) {
225
    // read the checksum before it updates to include the trailer
226
    var expected = checksum.digest()
227
    readChecksum(null, function (end, value) {
228
      cb(true)
229
      if (end === true && expectChecksum)
230
        onEnd(new Error('Missing checksum'))
231
      if (!value.equals(expected)) {
232
        onEnd(new Error('Checksum mismatch: ' +
233
          expected.hexSlice() + ' != ' + value.hexSlice()))
234
      } else {
235
        if (opts.verbosity >= 3)
236
          console.error('checksum ok', expected.hexSlice())
237
        onEnd(null)
238
      }
239
    })
240
  }
241
242
  function readObject(abort, cb) {
243
    if (ended) cb(ended)
244
    else if (abort) read(abort, function (err) { cb(ended = err || abort) })
245
    else if (numObjects < 0) readHeader(cb)
246
    else if (numObjects > 0) getObject(cb)
247
    else if (expectChecksum) readTrailer(cb)
248
  }
249
250
  return readObject
251
}
252
253
function readVarInt(readByte, cb) {
254
  var value = 0, shift = 0
255
  readByte(null, function gotByte(end, buf) {
256
    if (ended = end) return cb(end)
257
    var byte = buf[0]
258
    value += (byte & 0x7f) << shift
259
    shift += 7
260
    if (byte & 0x80)
261
      readByte(null, gotByte)
262
    else
263
      cb(null, value)
264
  })
265
}
266
267
function patchObject(opts, deltaB, deltaLength, srcObject, targetLength, cb) {
268
  var readByte = deltaB.chunks(1)
269
  var srcBuf
270
  var ended
271
272
  if (opts.verbosity >= 2)
273
    console.error('patching', srcObject.type, targetLength)
274
  pull(
275
    srcObject.read,
276
    pull.collect(function (err, bufs) {
277
      srcBuf = Buffer.concat(bufs, srcObject.length)
278
      cb(null, {
279
        type: srcObject.type,
280
        length: targetLength,
281
        read: read
282
      })
283
    })
284
  )
285
286
  function read(abort, cb) {
287
    if (ended) return cb(ended)
288
    readByte(null, function (end, dBuf) {
289
      if (ended = end) return cb(end)
290
      var cmd = dBuf[0]
291
      if (cmd & 0x80)
292
        // skip a variable amount and then pass through a variable amount
293
        readOffsetSize(cmd, deltaB, function (err, offset, size) {
294
          if (err) return earlyEnd(err)
295
          var buf = srcBuf.slice(offset, offset + size)
296
          cb(end, buf)
297
        })
298
      else if (cmd)
299
        // insert `cmd` bytes from delta
300
        deltaB.chunks(cmd)(null, cb)
301
      else
302
        cb(new Error("unexpected delta opcode 0"))
303
    })
304
305
    function earlyEnd(err) {
306
      cb(err === true ? new Error('stream ended early') : err)
307
    }
308
  }
309
}
310
311
function readOffsetSize(cmd, b, readCb) {
312
  var readByte = b.chunks(1)
313
  var offset = 0, size = 0
314
315
  function addByte(bit, outPos, cb) {
316
    if (cmd & (1 << bit))
317
      readByte(null, function (err, buf) {
318
        if (err) readCb(err)
319
        else cb(buf[0] << (outPos << 3))
320
      })
321
    else
322
      cb(0)
323
  }
324
325
  addByte(0, 0, function (val) {
326
    offset = val
327
    addByte(1, 1, function (val) {
328
      offset |= val
329
      addByte(2, 2, function (val) {
330
        offset |= val
331
        addByte(3, 3, function (val) {
332
          offset |= val
333
          addSize()
334
        })
335
      })
336
    })
337
  })
338
  function addSize() {
339
    addByte(4, 0, function (val) {
340
      size = val
341
      addByte(5, 1, function (val) {
342
        size |= val
343
        addByte(6, 2, function (val) {
344
          size |= val
345
          readCb(null, offset, size || 0x10000)
346
        })
347
      })
348
    })
349
  }
350
}
351
352
function encodeTypedVarInt(typeStr, length, cb) {
353
  var type = objectTypeNums[typeStr]
354
  // console.error('TYPE', type, typeStr, 'len', length, typeof cb)
355
  if (!type)
356
    return cb(new Error("Bad object type " + typeStr))
357
358
  var vals = []
359
  var b = (type << 4) | (length & 15)
360
  for (length >>= 4; length; length >>= 7) {
361
    vals.push(b | 0x80)
362
    b = length & 0x7f
363
  }
364
  vals.push(b)
365
  /*
366
  console.error('sending var int', vals, vals.map(function (n) {
367
    return ('00000000' + Number(n).toString(2)).substr(-8)
368
  }))
369
  */
370
  cb(null, new Buffer(vals))
371
}
372
373
function encodePack(opts, numObjects, readObject) {
374
  if (readObject === undefined)
375
    return encodePack.bind(this, numObjects)
376
377
  var header = new Buffer(12)
378
  header.write('PACK')
379
  header.writeUInt32BE(PACK_VERSION, 4)
380
  header.writeUInt32BE(numObjects, 8)
381
  var checksum = createHash('sha1')
382
  var readData
383
384
  return cat([
385
    checksum(cat([
386
      pull.once(header),
387
      encodeObject
388
    ])),
389
    checksum.readDigest
390
  ])
391
392
  function encodeObject(abort, cb) {
393
    if (readData)
394
      readData(abort, function (end, data) {
395
        if (end === true)
396
          readObject(abort, nextObject)
397
        else
398
          cb(end, data)
399
      })
400
    else
401
      readObject(abort, nextObject)
402
403
    function nextObject(end, object) {
404
      if (end) return cb(end)
405
      readData = deflate(object.read)
406
      encodeTypedVarInt(object.type, object.length, cb)
407
    }
408
  }
409
}
410