wip: more git

7 files changed, 571 insertions, 10 deletions

diff --git a/lib/app.js b/lib/app.js
index 4219151..1daf064 100644
--- a/lib/app.js
+++ b/lib/app.js
@@ -11,6 +11,7 @@ var Contacts = require('ssb-contact')
 var About = require('./about')
 var Serve = require('./serve')
 var Render = require('./render')
+var Git = require('./git')
 
 module.exports = App
 
@@ -41,6 +42,7 @@ function App(sbot, config) {
   this.unboxMsg = this.unboxMsg.bind(this)
 
   this.render = new Render(this, this.opts)
+  this.git = new Git(this)
 }
 
 App.prototype.go = function () {
@@ -186,6 +188,36 @@ App.prototype.pushBlob = function (id, cb) {
   this.sbot.blobs.push(id, cb)
 }
 
+App.prototype.readBlob = function (link, opts) {
+  link = u.toLink(link)
+  opts = opts || {}
+  return this.sbot.blobs.get({
+    hash: link.link,
+    size: link.size,
+    start: opts.start,
+    end: opts.end,
+  })
+}
+
+App.prototype.ensureHasBlobs = function (links, cb) {
+  var self = this
+  var done = multicb({pluck: 1})
+  links.forEach(function (link) {
+    var cb = done()
+    self.sbot.blobs.size(link.link, function (err, size) {
+      if (err) cb(err)
+      else if (size == null) cb(null, link)
+      else cb()
+    })
+  })
+  done(function (err, missingLinks) {
+    if (err) console.trace(err)
+    missingLinks = missingLinks.filter(Boolean)
+    if (missingLinks.length == 0) return cb()
+    return cb({name: 'BlobNotFoundError', links: missingLinks})
+  })
+}
+
 App.prototype.getReverseNameSync = function (name) {
   var id = this.reverseNameCache.get(name)
   return id
diff --git a/lib/git.js b/lib/git.js
new file mode 100644
index 0000000..36e4587
--- /dev/null
+++ b/lib/git.js
@@ -0,0 +1,380 @@
+var pull = require('pull-stream')
+var paramap = require('pull-paramap')
+var lru = require('hashlru')
+var memo = require('asyncmemo')
+var u = require('./util')
+var packidx = require('pull-git-packidx-parser')
+var Reader = require('pull-reader')
+var toPull = require('stream-to-pull-stream')
+var zlib = require('zlib')
+
+module.exports = Git
+
+function Git(app) {
+  this.app = app
+  this.findObject = memo({
+    cache: false,
+    asString: function (opts) {
+      return opts.id + opts.headMsgId
+    }
+  }, this._findObject.bind(this))
+}
+
+Git.prototype.getObject = function (opts, cb) {
+  pull(
+    this.readObject(opts),
+    u.pullConcat(cb)
+  )
+}
+
+// get a message that pushed an object
+Git.prototype.getObjectMsg = function (opts) {
+  this.findObject(opts, function (err, loc) {
+    if (err) return cb(err)
+    cb(null, loc.msg)
+  })
+}
+
+Git.prototype.readObject = function (opts) {
+  var self = this
+  return u.readNext(function (cb) {
+    self.findObject(opts, function (err, loc) {
+      if (err) return cb(err)
+      self.app.ensureHasBlobs([loc.packLink], function (err) {
+        if (err) return cb(err)
+        cb(null, pull(
+          self.app.readBlob(loc.packLink, {start: loc.offset, end: loc.next}),
+          self.decodeObject({
+            type: opts.type,
+            length: opts.length,
+            packLink: loc.packLink,
+            idx: loc.idx,
+          })
+        ))
+      })
+    })
+  })
+}
+
+// find which packfile contains a git object, and where in the packfile it is
+// located
+Git.prototype._findObject = function (opts, cb) {
+  var self = this
+  var objId = opts.id
+  var objIdBuf = new Buffer(objId, 'hex')
+  self.findObjectMsgs(opts, function (err, msgs) {
+    if (err) return cb(err)
+    if (msgs.length === 0)
+      return cb(new Error('unable to find git object ' + objId))
+    // if blobs may need to be fetched, try to ask the user about as many of them
+    // at one time as possible
+    var packidxs = [].concat.apply([], msgs.map(function (msg) {
+      var c = msg.value.content
+      var idxs = u.toArray(c.indexes).map(u.toLink)
+      return u.toArray(c.packs).map(u.toLink).map(function (pack, i) {
+        var idx = idxs[i]
+        if (pack && idx) return {
+          msg: msg,
+          packLink: pack,
+          idxLink: idx,
+        }
+      })
+    })).filter(Boolean)
+    var blobLinks = packidxs.length === 1
+      ? [packidxs[0].idxLink, packidxs[0].packLink]
+      : packidxs.map(function (packidx) {
+        return packidx.idxLink
+      })
+    self.app.ensureHasBlobs(blobLinks, function (err) {
+      if (err) return cb(err)
+      pull(
+        pull.values(packidxs),
+        paramap(function (pack, cb) {
+          console.error('get idx', pack.idxLink)
+          self.getPackIndex(pack.idxLink, function (err, idx) {
+            if (err) return cb(err)
+            var offset = idx.find(objIdBuf)
+            // console.error('got idx', err, pack.idxId, offset)
+            if (!offset) return cb()
+            cb(null, {
+              offset: offset.offset,
+              next: offset.next,
+              packLink: pack.packLink,
+              idx: idx,
+              msg: pack.msg,
+            })
+          })
+        }, 4),
+        pull.filter(),
+        pull.take(1),
+        pull.collect(function (err, offsets) {
+          if (err) return cb(err)
+          if (offsets.length === 0)
+            return cb(new Error('unable to find git object ' + objId +
+              ' in ' + msgs.length + ' messages'))
+          cb(null, offsets[0])
+        })
+      )
+    })
+  })
+}
+
+// given an object id and ssb msg id, get a set of messages of which at least one pushed the object.
+Git.prototype.findObjectMsgs = function (opts, cb) {
+  var self = this
+  var id = opts.id
+  var headMsgId = opts.headMsgId
+  var ended = false
+  var waiting = 0
+  var maybeMsgs = []
+
+  function cbOnce(err, msgs) {
+    if (ended) return
+    ended = true
+    cb(err, msgs)
+  }
+
+  function objectMatches(commit) {
+    return commit && (commit === id || commit.sha1 === id)
+  }
+
+  if (!headMsgId) return cb(new TypeError('missing head message id'))
+  if (!u.isRef(headMsgId))
+    return cb(new TypeError('bad head message id \'' + headMsgId + '\''))
+
+  ;(function getMsg(id) {
+    waiting++
+    console.error('get msg', id)
+    self.app.getMsgDecrypted(id, function (err, msg) {
+      waiting--
+      if (ended) return
+      if (err && err.name == 'NotFoundError')
+        return cbOnce(new Error('missing message ' + headMsgId))
+      if (err) return cbOnce(err)
+      var c = msg.value.content
+      if (typeof c === 'string')
+        return cbOnce(new Error('unable to decrypt message ' + msg.key))
+      if ((u.toArray(c.object_ids).some(objectMatches))
+      || (u.toArray(c.tags).some(objectMatches))
+      || (u.toArray(c.commits).some(objectMatches))) {
+        // found the object
+        return cbOnce(null, [msg])
+        // console.error('found', msg.key)
+      } else if (!c.object_ids) {
+        // the object might be here
+        maybeMsgs.push(msg)
+      }
+      // traverse the DAG to keep looking for the object
+      u.toArray(c.repoBranch).filter(u.isRef).forEach(getMsg)
+      if (waiting === 0) {
+        // console.error('trying messages', maybeMsgs.map(function (msg) { return msg.key}))
+        cbOnce(null, maybeMsgs)
+      }
+    })
+  })(headMsgId)
+}
+
+Git.prototype.getPackIndex = function (idxBlobLink, cb) {
+  pull(this.app.readBlob(idxBlobLink), packidx(cb))
+}
+
+var objectTypes = [
+  'none', 'commit', 'tree', 'blob',
+  'tag', 'unused', 'ofs-delta', 'ref-delta'
+]
+
+function readTypedVarInt(reader, cb) {
+  var type, value, shift
+  reader.read(1, function (end, buf) {
+    if (ended = end) return cb(end)
+    var firstByte = buf[0]
+    type = objectTypes[(firstByte >> 4) & 7]
+    value = firstByte & 15
+    shift = 4
+    checkByte(firstByte)
+  })
+
+  function checkByte(byte) {
+    if (byte & 0x80)
+      reader.read(1, gotByte)
+    else
+      cb(null, type, value)
+  }
+
+  function gotByte(end, buf) {
+    if (ended = end) return cb(end)
+    var byte = buf[0]
+    value += (byte & 0x7f) << shift
+    shift += 7
+    checkByte(byte)
+  }
+}
+
+function readVarInt(reader, cb) {
+  var value = 0, shift = 0
+  reader.read(1, function gotByte(end, buf) {
+    if (ended = end) return cb(end)
+    var byte = buf[0]
+    value += (byte & 0x7f) << shift
+    shift += 7
+    if (byte & 0x80)
+      reader.read(1, gotByte)
+    else
+      cb(null, value)
+  })
+}
+
+function inflate(read) {
+  return toPull(zlib.createInflate())(read)
+}
+
+Git.prototype.decodeObject = function (opts) {
+  var self = this
+  var packLink = opts.packLink
+  return function (read) {
+    var reader = Reader()
+    reader(read)
+    return u.readNext(function (cb) {
+      readTypedVarInt(reader, function (end, type, length) {
+        if (end === true) cb(new Error('Missing object type'))
+        else if (end) cb(end)
+        else if (type === 'ref-delta') getObjectFromRefDelta(length, cb)
+        else if (opts.type && type !== opts.type)
+          cb(new Error('expected type \'' + opts.type + '\' ' +
+            'but found \'' + type + '\''))
+        else if (opts.length && length !== opts.length)
+          cb(new Error('expected length ' + opts.length + ' ' +
+            'but found ' + length))
+          else cb(null, inflate(reader.read()))
+      })
+    })
+
+    function getObjectFromRefDelta(length, cb) {
+      console.error('read from ref delta')
+      reader.read(20, function (end, sourceHash) {
+        if (end) return cb(end)
+        var inflatedReader = Reader()
+        pull(reader.read(), inflate, inflatedReader)
+        readVarInt(inflatedReader, function (err, expectedSourceLength) {
+          if (err) return cb(err)
+          readVarInt(inflatedReader, function (err, expectedTargetLength) {
+            if (err) return cb(err)
+            // console.error('getting object', sourceHash)
+            var offset = opts.idx.find(sourceHash)
+            if (!offset) return cb(null, 'missing source object ' +
+              sourcehash.toString('hex'))
+            console.error('get pack', opts.packLink, offset.offset, offset.next)
+            var readSource = pull(
+              self.app.readBlob(opts.packLink, {
+                start: offset.offset,
+                end: offset.next
+              }),
+              self.decodeObject({
+                type: opts.type,
+                length: expectedSourceLength,
+                packLink: opts.packLink,
+                idx: opts.idx
+              })
+            )
+            // console.error('patching', length, expectedTargetLength)
+            cb(null, patchObject(inflatedReader, length, readSource, expectedTargetLength))
+          })
+        })
+      })
+    }
+  }
+}
+
+function readOffsetSize(cmd, reader, readCb) {
+  var offset = 0, size = 0
+
+  function addByte(bit, outPos, cb) {
+    if (cmd & (1 << bit))
+      reader.read(1, function (err, buf) {
+        if (err) readCb(err)
+        else cb(buf[0] << (outPos << 3))
+      })
+    else
+      cb(0)
+  }
+
+  addByte(0, 0, function (val) {
+    offset = val
+    addByte(1, 1, function (val) {
+      offset |= val
+      addByte(2, 2, function (val) {
+        offset |= val
+        addByte(3, 3, function (val) {
+          offset |= val
+          addSize()
+        })
+      })
+    })
+  })
+  function addSize() {
+    addByte(4, 0, function (val) {
+      size = val
+      addByte(5, 1, function (val) {
+        size |= val
+        addByte(6, 2, function (val) {
+          size |= val
+          readCb(null, offset, size || 0x10000)
+        })
+      })
+    })
+  }
+}
+
+function patchObject(deltaReader, deltaLength, readSource, targetLength) {
+  var srcBuf
+  var ended
+  // console.error('patching', deltaLength, targetLength)
+
+  return u.readNext(function (cb) {
+    pull(readSource, u.pullConcat(function (err, buf) {
+      if (err) return cb(err)
+      srcBuf = buf
+      cb(null, read)
+    }))
+  })
+
+  function read(abort, cb) {
+    // console.error('pa', abort, ended)
+    if (ended) return cb(ended)
+    deltaReader.read(1, function (end, dBuf) {
+      // console.error("read", end, dBuf)
+      // if (ended = end) return console.error('patched', deltaLength, targetLength, end),  cb(end)
+      if (ended = end) return cb(end)
+      var cmd = dBuf[0]
+      // console.error('cmd', cmd & 0x80, cmd)
+      if (cmd & 0x80)
+        // skip a variable amount and then pass through a variable amount
+        readOffsetSize(cmd, deltaReader, function (err, offset, size) {
+        // console.error('offset', err, offset, size)
+          if (err) return earlyEnd(err)
+          var buf = srcBuf.slice(offset, offset + size)
+        // console.error('buf', buf)
+          cb(end, buf)
+        })
+      else if (cmd)
+        // insert `cmd` bytes from delta
+        deltaReader.read(cmd, cb)
+      else
+        cb(new Error("unexpected delta opcode 0"))
+    })
+
+    function earlyEnd(err) {
+      cb(err === true ? new Error('stream ended early') : err)
+    }
+  }
+}
+
+Git.prototype.getCommit = function (opts, cb) {
+  this.getObject(opts, function (err, buf) {
+    if (err) return cb(err)
+    var commit = {
+      body: buf.toString('ascii')
+    }
+    cb(null, commit)
+  })
+}
diff --git a/lib/render-msg.js b/lib/render-msg.js
index c0001fc..ef23b3f 100644
--- a/lib/render-msg.js
+++ b/lib/render-msg.js
@@ -25,11 +25,7 @@ RenderMsg.prototype.toUrl = function (href) {
 }
 
 RenderMsg.prototype.linkify = function (text) {
-  var arr = text.split(u.ssbRefRegex)
-  for (var i = 1; i < arr.length; i += 2) {
-    arr[i] = h('a', {href: this.toUrl(arr[i])}, arr[i])
-  }
-  return arr
+  return this.render.linkify(text)
 }
 
 function token() {
@@ -482,14 +478,20 @@ RenderMsg.prototype.gitUpdate = function (cb) {
         !isNaN(size) ? [self.render.formatSize(size), ' '] : '',
         self.c.refs ? h('ul', Object.keys(self.c.refs).map(function (ref) {
           var id = self.c.refs[ref]
-        return h('li',
-          ref.replace(/^refs\/(heads|tags)\//, ''), ': ',
-          id ? h('code', id) : h('em', 'deleted'))
+          var type = /^refs\/tags/.test(ref) ? 'tag' : 'commit'
+          var path = id && ('/git/' + type + '/' + encodeURIComponent(id)
+            + '?head=' + encodeURIComponent(self.msg.key))
+          return h('li',
+            ref.replace(/^refs\/(heads|tags)\//, ''), ': ',
+            id ? h('a', {href: self.render.toUrl(path)}, h('code', id))
+               : h('em', 'deleted'))
       })) : '',
       Array.isArray(self.c.commits) ?
         h('ul', self.c.commits.map(function (commit) {
-          return h('li',
-            h('code', String(commit.sha1).substr(0, 8)), ' ',
+          var path = '/git/commit/' + encodeURIComponent(commit.sha1)
+            + '?head=' + encodeURIComponent(self.msg.key)
+          return h('li', h('a', {href: self.render.toUrl(path)},
+            h('code', String(commit.sha1).substr(0, 8))), ' ',
             self.linkify(String(commit.title)),
             self.gitCommitBody(commit.body)
           )
diff --git a/lib/render.js b/lib/render.js
index dd6d55e..02d7285 100644
--- a/lib/render.js
+++ b/lib/render.js
@@ -148,6 +148,14 @@ Render.prototype.formatSize = function (size) {
   return size.toFixed(2) + ' MB'
 }
 
+Render.prototype.linkify = function (text) {
+  var arr = text.split(u.ssbRefRegex)
+  for (var i = 1; i < arr.length; i += 2) {
+    arr[i] = h('a', {href: this.toUrl(arr[i])}, arr[i])
+  }
+  return arr
+}
+
 Render.prototype.toUrl = function (href) {
   if (!href) return href
   var mentions = this._mentions
diff --git a/lib/serve.js b/lib/serve.js
index 8d3eeba..4e8c15d 100644
--- a/lib/serve.js
+++ b/lib/serve.js
@@ -119,6 +119,7 @@ Serve.prototype.go = function () {
     else if (data.action === 'publish') self.publishJSON(next)
     else if (data.action === 'vote') self.publishVote(next)
     else if (data.action === 'contact') self.publishContact(next)
+    else if (data.action === 'want-blobs') self.wantBlobs(next)
     else next()
   }
 
@@ -172,6 +173,22 @@ Serve.prototype.publishContact = function (cb) {
   this.publish(content, cb)
 }
 
+Serve.prototype.wantBlobs = function (cb) {
+  var self = this
+  if (!self.data.blob_ids) return cb()
+  var ids = self.data.blob_ids.split(',')
+  if (!ids.every(u.isRef)) return cb(new Error('bad blob ids ' + ids.join(',')))
+  var done = multicb({pluck: 1})
+  ids.forEach(function (id) {
+    self.app.sbot.blobs.want(id, done())
+  })
+  done(function (err) {
+    if (err) return cb(err)
+    // self.note = h('div', 'wanted blobs: ' + ids.join(', ') + '.')
+    cb()
+  })
+}
+
 Serve.prototype.publish = function (content, cb) {
   var self = this
   var done = multicb({pluck: 1, spread: true})
@@ -257,6 +274,7 @@ Serve.prototype.path = function (url) {
     case '/emoji': return this.emoji(m[2])
     case '/contacts': return this.contacts(m[2])
     case '/about': return this.about(m[2])
+    case '/git': return this.git(m[2])
   }
   return this.respond(404, 'Not found')
 }
@@ -1063,6 +1081,20 @@ function catchHTMLError() {
   }
 }
 
+function catchTextError() {
+  return function (read) {
+    var ended
+    return function (abort, cb) {
+      if (ended) return cb(ended)
+      read(abort, function (end, data) {
+        if (!end || end === true) return cb(end, data)
+        ended = true
+        cb(null, end.stack + '\n')
+      })
+    }
+  }
+}
+
 function styles() {
   return fs.readFileSync(path.join(__dirname, '../static/styles.css'), 'utf8')
 }
@@ -1117,6 +1149,7 @@ Serve.prototype.wrapPage = function (title, searchQ) {
           'published ',
           self.app.render.msgLink(self.publishedMsg, done())
         ) : '',
+        // self.note,
         content
       )))
       done(cb)
@@ -1245,6 +1278,65 @@ Serve.prototype.wrapUserFeed = function (isScrolled, id) {
   })
 }
 
+Serve.prototype.git = function (url) {
+  var m = /^\/?([^\/]*)\/?(.*)?$/.exec(url)
+  switch (m[1]) {
+    case 'commit': return this.gitCommit(m[2])
+    case 'tag': return this.gitCommit(m[2])
+    case 'raw': return this.gitRaw(m[2])
+    default: return this.respond(404, 'Not found')
+  }
+}
+
+Serve.prototype.gitRaw = function (rev) {
+  var self = this
+  if (!/[0-9a-f]{24}/.test(rev)) {
+    return pull(
+      pull.once('\'' + rev + '\' is not a git object id'),
+      self.respondSink(400, {'Content-Type': 'text/plain'})
+    )
+  }
+
+  var headMsgId = self.query.head
+  pull(
+    self.app.git.readObject({
+      id: rev,
+      headMsgId: headMsgId,
+    }),
+    catchTextError(),
+    self.respondSink(200, {'Content-type': 'text/plain'})
+  )
+}
+
+Serve.prototype.gitCommit = function (rev) {
+  var self = this
+  if (!/[0-9a-f]{24}/.test(rev)) {
+    return pull(
+      ph('div.error', 'rev is not a git object id'),
+      self.wrapPage('git'),
+      self.respondSink(400)
+    )
+  }
+
+  self.app.git.getCommit({
+    id: rev,
+    headMsgId: self.query.head,
+  }, function (err, commit) {
+    if (err && err.name === 'BlobNotFoundError')
+      return self.askWantBlobs(err.links)
+    if (err) return pull(
+      pull.once(u.renderError(err).outerHTML),
+      self.wrapPage('git object ' + rev),
+      self.respondSink(400)
+    )
+    pull(
+      pull.once(h('pre', self.app.render.linkify(commit.body)).outerHTML),
+      self.wrapPage('git object ' + rev),
+      self.respondSink(200)
+    )
+  })
+}
+
 Serve.prototype.wrapPublic = function (opts) {
   var self = this
   return u.hyperwrap(function (thread, cb) {
@@ -1260,6 +1352,31 @@ Serve.prototype.wrapPublic = function (opts) {
   })
 }
 
+Serve.prototype.askWantBlobs = function (links) {
+  var self = this
+  pull(
+    ph('form', {action: '', method: 'post'}, [
+      ph('section', [
+        ph('h3', 'Missing blobs'),
+        ph('p', 'The application needs these blobs to continue:'),
+        ph('table', links.map(u.toLink).map(function (link) {
+          if (!u.isRef(link.link)) return
+          return ph('tr', [
+            ph('td', ph('code', link.link)),
+            ph('td', self.app.render.formatSize(link.size)),
+          ])
+        })),
+        ph('input', {type: 'hidden', name: 'action', value: 'want-blobs'}),
+        ph('input', {type: 'hidden', name: 'blob_ids',
+          value: links.map(u.linkDest).join(',')}),
+        ph('p', ph('input', {type: 'submit', value: 'Want Blobs'}))
+      ])
+    ]),
+    self.wrapPage('missing blobs'),
+    self.respondSink(409)
+  )
+}
+
 Serve.prototype.wrapPrivate = function (opts) {
   var self = this
   return u.hyperwrap(function (thread, cb) {
diff --git a/lib/util.js b/lib/util.js
index 5546716..dbefa14 100644
--- a/lib/util.js
+++ b/lib/util.js
@@ -62,6 +62,10 @@ u.hyperwrap = function (fn) {
   }
 }
 
+u.toLink = function (link) {
+  return typeof link === 'string' ? {link: link} : link
+}
+
 u.linkDest = function (link) {
   return typeof link === 'string' ? link : link && link.link || link
 }
@@ -114,3 +118,19 @@ u.isMsgEncrypted = function (msg) {
   var c = msg && msg.value.content
   return typeof c === 'string'
 }
+
+u.pullConcat = function (cb) {
+  return pull.collect(function (err, bufs) {
+    if (err) return cb(err)
+    cb(null, Buffer.concat(bufs))
+  })
+}
+
+u.customError = function (name) {
+  return function (message) {
+    var error = new Error(message)
+    error.name = name
+    error.stack = error.stack.replace(/^    at .*\n/m, '')
+    return error
+  }
+}
diff --git a/package.json b/package.json
index 7b77ca5..2a78964 100644
--- a/package.json
+++ b/package.json
@@ -13,11 +13,13 @@
     "mime-types": "^2.1.12",
     "multicb": "^1.2.1",
     "pull-cat": "^1.1.11",
+    "pull-git-packidx-parser": "^1.0.0",
     "pull-hash": "^1.0.0",
     "pull-hyperscript": "^0.2.2",
     "pull-identify-filetype": "^1.1.0",
     "pull-paginate": "^1.0.0",
     "pull-paramap": "^1.2.1",
+    "pull-reader": "^1.2.9",
     "pull-stream": "^3.5.0",
     "ssb-contact": "^1.0.0",
     "ssb-marked": "^0.7.1",