refactor git stuff

- parse commits
- link to git objects with specific message id
- open object before reading it
- cleanup debug things

1 files changed, 170 insertions, 92 deletions

diff --git a/lib/git.js b/lib/git.js
index 36e4587..048e2b3 100644
--- a/lib/git.js
+++ b/lib/git.js
@@ -8,121 +8,166 @@ var Reader = require('pull-reader')
 var toPull = require('stream-to-pull-stream')
 var zlib = require('zlib')
 
+var ObjectNotFoundError = u.customError('ObjectNotFoundError')
+
 module.exports = Git
 
 function Git(app) {
   this.app = app
+
   this.findObject = memo({
-    cache: false,
+    cache: lru(5),
     asString: function (opts) {
-      return opts.id + opts.headMsgId
+      return opts.obj + opts.headMsgId
     }
   }, this._findObject.bind(this))
+
+  this.findObjectInMsg = memo({
+    cache: lru(5),
+    asString: function (opts) {
+      return opts.obj + opts.msg
+    }
+  }, this._findObjectInMsg.bind(this))
+
+  this.getPackIndex = memo({
+    cache: lru(4),
+    asString: JSON.stringify
+  }, this._getPackIndex.bind(this))
 }
 
+// open, read, buffer and callback an object
 Git.prototype.getObject = function (opts, cb) {
-  pull(
-    this.readObject(opts),
-    u.pullConcat(cb)
-  )
+  var self = this
+  self.openObject(opts, function (err, obj) {
+    if (err) return cb(err)
+    pull(
+      self.readObject(obj),
+      u.pullConcat(cb)
+    )
+  })
 }
 
 // get a message that pushed an object
-Git.prototype.getObjectMsg = function (opts) {
+Git.prototype.getObjectMsg = function (opts, cb) {
   this.findObject(opts, function (err, loc) {
     if (err) return cb(err)
     cb(null, loc.msg)
   })
 }
 
-Git.prototype.readObject = function (opts) {
+Git.prototype.openObject = function (opts, cb) {
   var self = this
-  return u.readNext(function (cb) {
-    self.findObject(opts, function (err, loc) {
+  self.findObjectInMsg(opts, function (err, loc) {
+    if (err) return cb(err)
+    self.app.ensureHasBlobs([loc.packLink], function (err) {
       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,
-          })
-        ))
+      cb(null, {
+        type: opts.type,
+        length: opts.length,
+        offset: loc.offset,
+        next: loc.next,
+        packLink: loc.packLink,
+        idx: loc.idx,
+        msg: loc.msg,
       })
     })
   })
 }
 
+Git.prototype.readObject = function (obj) {
+  return pull(
+    this.app.readBlob(obj.packLink, {start: obj.offset, end: obj.next}),
+    this.decodeObject({
+      type: obj.type,
+      length: obj.length,
+      packLink: obj.packLink,
+      idx: obj.idx,
+    })
+  )
+}
+
 // find which packfile contains a git object, and where in the packfile it is
 // located
 Git.prototype._findObject = function (opts, cb) {
+  if (!opts.headMsgId) return cb(new TypeError('missing head message id'))
+  if (!opts.obj) return cb(new TypeError('missing object id'))
   var self = this
-  var objId = opts.id
-  var objIdBuf = new Buffer(objId, 'hex')
+  var objId = opts.obj
   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) {
+      return cb(new ObjectNotFoundError('unable to find git object ' + objId))
+    self.findObjectInMsgs(objId, msgs, cb)
+  })
+}
+
+Git.prototype._findObjectInMsg = function (opts, cb) {
+  if (!opts.msg) return cb(new TypeError('missing message id'))
+  if (!opts.obj) return cb(new TypeError('missing object id'))
+  var self = this
+  self.app.getMsgDecrypted(opts.msg, function (err, msg) {
+    if (err) return cb(err)
+    self.findObjectInMsgs(opts.obj, [msg], cb)
+  })
+}
+
+Git.prototype.findObjectInMsgs = function (objId, msgs, cb) {
+  var self = this
+  var objIdBuf = new Buffer(objId, 'hex')
+  // 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) {
+        self.getPackIndex(pack.idxLink, function (err, idx) {
           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])
+          var offset = idx.find(objIdBuf)
+          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 ObjectNotFoundError('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 id = opts.obj
   var headMsgId = opts.headMsgId
   var ended = false
   var waiting = 0
@@ -144,7 +189,6 @@ Git.prototype.findObjectMsgs = function (opts, cb) {
 
   ;(function getMsg(id) {
     waiting++
-    console.error('get msg', id)
     self.app.getMsgDecrypted(id, function (err, msg) {
       waiting--
       if (ended) return
@@ -159,7 +203,6 @@ Git.prototype.findObjectMsgs = function (opts, cb) {
       || (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)
@@ -167,14 +210,13 @@ Git.prototype.findObjectMsgs = function (opts, cb) {
       // 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) {
+Git.prototype._getPackIndex = function (idxBlobLink, cb) {
   pull(this.app.readBlob(idxBlobLink), packidx(cb))
 }
 
@@ -250,7 +292,6 @@ Git.prototype.decodeObject = function (opts) {
     })
 
     function getObjectFromRefDelta(length, cb) {
-      console.error('read from ref delta')
       reader.read(20, function (end, sourceHash) {
         if (end) return cb(end)
         var inflatedReader = Reader()
@@ -259,11 +300,9 @@ Git.prototype.decodeObject = function (opts) {
           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,
@@ -276,7 +315,6 @@ Git.prototype.decodeObject = function (opts) {
                 idx: opts.idx
               })
             )
-            // console.error('patching', length, expectedTargetLength)
             cb(null, patchObject(inflatedReader, length, readSource, expectedTargetLength))
           })
         })
@@ -328,7 +366,6 @@ function readOffsetSize(cmd, reader, readCb) {
 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) {
@@ -339,21 +376,15 @@ function patchObject(deltaReader, deltaLength, readSource, targetLength) {
   })
 
   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)
@@ -369,12 +400,59 @@ function patchObject(deltaReader, deltaLength, readSource, targetLength) {
   }
 }
 
-Git.prototype.getCommit = function (opts, cb) {
-  this.getObject(opts, function (err, buf) {
+var gitNameRegex = /^(.*) <(([^>@]*)(@[^>]*)?)> (.*) (.*)$/
+function parseName(line) {
+  var m = gitNameRegex.exec(line)
+  if (!m) return null
+  return {
+    name: m[1],
+    email: m[2],
+    localpart: m[3],
+    feed: u.isRef(m[4]) && m[4] || undefined,
+    date: new Date(m[5] * 1000),
+    tz: m[6],
+  }
+}
+
+Git.prototype.getCommit = function (obj, cb) {
+  pull(this.readObject(obj), u.pullConcat(function (err, buf) {
     if (err) return cb(err)
     var commit = {
-      body: buf.toString('ascii')
+      msg: obj.msg,
+      parents: [],
     }
+    var authorLine, committerLine
+    var lines = buf.toString('utf8').split('\n')
+    for (var line; (line = lines.shift()); ) {
+      var parts = line.split(' ')
+      var prop = parts.shift()
+      var value = parts.join(' ')
+      switch (prop) {
+        case 'tree':
+          commit.tree = value
+          break
+        case 'parent':
+          commit.parents.push(value)
+          break
+        case 'author':
+          authorLine = value
+          break
+        case 'committer':
+          committerLine = value
+          break
+        case 'gpgsig':
+          var sigLines = [value]
+          while (lines[0] && lines[0][0] == ' ')
+            sigLines.push(lines.shift().slice(1))
+          commit.gpgsig = sigLines.join('\n')
+          break
+        default:
+          return cb(new TypeError('unknown git object property ' + prop))
+      }
+    }
+    commit.committer = parseName(committerLine)
+    if (authorLine !== committerLine) commit.author = parseName(authorLine)
+    commit.body = lines.join('\n')
     cb(null, commit)
-  })
+  }))
 }