Merge pull request #7 from cubehero/streaming

Added data streaming for blob files

.gitignore

@@ -1,2 +1,13 @@
-lib
-node_modules
+# ignore emacs and other editor files
+*~
+\#*\#
+.\#*
+*.swp
+*.swo
+
+src/*.js
+test/*.js
+
+lib/
+node_modules/
+

README.md

@@ -302,6 +302,25 @@ Each file has the following properties:
 
   * `callback` - `(err, data)`
 
+Warning: this method only returns the complete file up to 200k, which is the default
+buffer size for running child_process.exec(). If the file you're reading is bigger than
+that, or if you're not sure, you need to use dataStream()
+
+### `Blob#dataStream()`
+
+  * returns - [dataStream, errorStream]
+
+Returns streams for you to use to get the data.
+
+Usage:
+
+    data = ""
+    [dataStream, _] = blob.dataStream()
+    dataStream.on 'data', (buf) ->
+      data += buf.toString()
+    .on 'end', ->
+      callback(data)
+
 ## Submodule
 ### `Submodule#id`
 `String`

src/blob.coffee

@@ -8,10 +8,26 @@ module.exports = class Blob
   # 
   # callback - Receives `(err, data)`.
   # 
+  # Warning, this only returns files less than 200k, the standard buffer size for 
+  # node's exec(). If you need to get bigger files, you should use dataStream() to
+  # get a stream for the file's data
+  #
   data: (callback) ->
     @repo.git "cat-file", {p: true}, @id
     , (err, stdout, stderr) ->
       return callback err, stdout
 
+  # Public: Get the blob contents as a stream
+  #
+  # returns - [dataStream, errstream]
+  #
+  # Usage:
+  #   [blobstream, _] = blob.dataStream()
+  #   blobstream.pipe(res)
+  #
+  dataStream: () ->
+    streams = @repo.git.streamCmd "cat-file", {p: true}, [@id]
+    return streams
+
   toString: ->
     "#<Blob '#{@id}'>"

src/git.coffee

@@ -1,5 +1,5 @@
 fs     = require 'fs'
-{exec} = require 'child_process'
+{exec, spawn} = require 'child_process'
 
 module.exports = Git = (git_dir, dot_git) ->
   dot_git ||= "#{git_dir}/.git"
@@ -21,6 +21,19 @@ module.exports = Git = (git_dir, dot_git) ->
   git.cmd  = (command, options, args, callback) ->
     git command, options, args, callback
 
+  # Public: stream results of git command
+  #
+  # This is used for large files that you'd need to stream.
+  #
+  # returns [outstream, errstream]
+  #
+  git.streamCmd = (command, options, args) ->
+    options ?= {}
+    options  = options_to_argv options
+    args    ?= []
+    allargs = [command].concat(options).concat(args)
+    process  = spawn Git.bin, allargs, {cwd: git_dir, encoding: 'binary'}
+    return [process.stdout, process.stderr]
 
   # Public: Get a list of the remote names.
   #

test/blob.test.coffee

@@ -43,3 +43,36 @@ describe "Blob", ->
         data.should.be.type "string"
         data.should.include "!!!"
 
+  describe "#dataStream", ->
+    describe "of a file off the root", ->
+      repo = git "#{__dirname}/fixtures/branched"
+      data = ""
+      before (done) ->
+        repo.tree().blobs (err, blobs) ->
+          [dataStream, _] = blobs[0].dataStream()
+          dataStream.on 'data', (buf) ->
+            data += buf.toString()
+          .on 'end', ->
+            done()
+
+      it "is a string", ->
+        data.should.be.type "string"
+        data.should.include "Bla"
+
+    describe "of a file in a subdir", ->
+      repo = git "#{__dirname}/fixtures/branched"
+      data = ""
+      before (done) ->
+        repo.tree().trees (err, trees) ->
+          trees[0].blobs (err, blobs) ->
+            [dataStream, _] = blobs[0].dataStream()
+            dataStream.on 'data', (buf) ->
+              data += buf.toString()
+            .on 'end', ->
+              done()
+
+      it "is a string", ->
+        data.should.be.type "string"
+        data.should.include "!!!"
+
+