schacon-grit 0.9.1

data/lib/grit/git-ruby/internal/mmap.rb

@@ -0,0 +1,59 @@
+#
+# converted from the gitrb project
+#
+# authors: 
+#    Matthias Lederhofer <matled@gmx.net>
+#    Simon 'corecode' Schubert <corecode@fs.ei.tum.de>
+#    Scott Chacon <schacon@gmail.com>
+#
+# provides native ruby access to git objects and pack files
+#
+
+begin
+  require 'mmap'
+rescue LoadError
+
+module Grit 
+  module GitRuby 
+    module Internal
+      class Mmap
+        def initialize(file)
+          @file = file
+          @offset = nil
+        end
+
+        def unmap
+          @file = nil
+        end
+
+        def [](*idx)
+          idx = idx[0] if idx.length == 1
+          case idx
+          when Range
+            offset = idx.first
+            len = idx.last - idx.first + idx.exclude_end? ? 0 : 1
+          when Fixnum
+            offset = idx
+            len = nil
+          when Array
+            offset, len = idx
+          else
+            raise RuntimeError, "invalid index param: #{idx.class}"
+          end
+          if @offset != offset
+            @file.seek(offset)
+          end
+          @offset = offset + len ? len : 1
+          if not len
+            @file.read(1)[0]
+          else
+            @file.read(len)
+          end
+        end
+      end
+    end
+  end 
+end
+
+end     # rescue LoadError
+

data/lib/grit/git-ruby/internal/pack.rb

@@ -0,0 +1,332 @@
+#
+# converted from the gitrb project
+#
+# authors: 
+#    Matthias Lederhofer <matled@gmx.net>
+#    Simon 'corecode' Schubert <corecode@fs.ei.tum.de>
+#    Scott Chacon <schacon@gmail.com>
+#
+# provides native ruby access to git objects and pack files
+#
+
+require 'zlib'
+require 'grit/git-ruby/internal/raw_object'
+require 'grit/git-ruby/internal/mmap'
+
+module Grit 
+  module GitRuby 
+    module Internal
+      class PackFormatError < StandardError
+      end
+
+      class PackStorage
+        OBJ_OFS_DELTA = 6
+        OBJ_REF_DELTA = 7
+
+        FanOutCount = 256
+        SHA1Size = 20
+        IdxOffsetSize = 4
+        OffsetSize = 4
+        OffsetStart = FanOutCount * IdxOffsetSize
+        SHA1Start = OffsetStart + OffsetSize
+        EntrySize = OffsetSize + SHA1Size
+
+        def initialize(file)
+          if file =~ /\.idx$/
+            file = file[0...-3] + 'pack'
+          end
+          
+          @cache = {}
+          @name = file
+          
+          with_idx do |idx|
+            @offsets = [0]
+            FanOutCount.times do |i|
+              pos = idx[i * IdxOffsetSize,IdxOffsetSize].unpack('N')[0]
+              if pos < @offsets[i]
+                raise PackFormatError, "pack #@name has discontinuous index #{i}"
+              end
+              @offsets << pos
+            end
+
+            @size = @offsets[-1]
+          end
+        end
+        
+        def with_idx(index_file = nil)
+          if !index_file
+            index_file = @name
+            idxfile = File.open(@name[0...-4]+'idx')
+          else
+            idxfile = File.open(index_file)
+          end
+          idx = Mmap.new(idxfile)
+          yield idx
+          idx.unmap
+          idxfile.close
+        end
+        
+        def with_packfile
+          packfile = File.open(@name)
+          yield packfile
+          packfile.close
+        end
+        
+        # given an index file, list out the shas that it's packfile contains
+        def self.get_shas(index_file)
+          with_idx(index_file) do |idx|
+            @offsets = [0]
+            FanOutCount.times do |i|
+              pos = idx[i * IdxOffsetSize,IdxOffsetSize].unpack('N')[0]
+              if pos < @offsets[i]
+                raise PackFormatError, "pack #@name has discontinuous index #{i}"
+              end
+              @offsets << pos
+            end
+        
+            @size = @offsets[-1]
+            shas = []
+          
+            pos = SHA1Start
+            @size.times do
+              sha1 = idx[pos,SHA1Size]
+              pos += EntrySize
+              shas << sha1.unpack("H*").first
+            end
+            shas
+          end
+        end
+        
+        def cache_objects
+          @cache = {}
+          with_packfile do |packfile|          
+            each_entry do |sha, offset|
+              data, type = unpack_object(packfile, offset, {:caching => true})
+              if data
+                @cache[sha] = RawObject.new(OBJ_TYPES[type], data)
+              end
+            end
+          end
+        end
+
+        def name
+          @name
+        end
+        
+        def close
+          # shouldnt be anything open now
+        end
+
+        def [](sha1)
+          if obj = @cache[sha1]
+            return obj 
+          end
+          
+          offset = find_object(sha1)
+          return nil if !offset
+          @cache[sha1] = obj = parse_object(offset)
+          return obj
+        end
+
+        def each_entry
+          with_idx do |idx|
+            pos = OffsetStart
+            @size.times do
+              offset = idx[pos,OffsetSize].unpack('N')[0]
+              sha1 = idx[pos+OffsetSize,SHA1Size]
+              pos += EntrySize
+              yield sha1, offset
+            end
+          end
+        end
+
+        def each_sha1
+          with_idx do |idx|
+            # unpacking the offset is quite expensive, so
+            # we avoid using #each
+            pos = SHA1Start
+            @size.times do
+              sha1 = idx[pos,SHA1Size]
+              pos += EntrySize
+              yield sha1
+            end
+          end
+        end
+
+        def find_object(sha1)
+          obj = nil
+          with_idx do |idx|
+            obj = find_object_in_index(idx, sha1)
+          end
+          obj
+        end    
+        private :find_object
+
+        def find_object_in_index(idx, sha1)
+          slot = sha1[0]
+          return nil if !slot
+          first, last = @offsets[slot,2] 
+          while first < last
+            mid = (first + last) / 2
+            midsha1 = idx[SHA1Start + mid * EntrySize,SHA1Size]
+            cmp = midsha1 <=> sha1
+
+            if cmp < 0
+              first = mid + 1
+            elsif cmp > 0
+              last = mid
+            else
+              pos = OffsetStart + mid * EntrySize
+              offset = idx[pos,OffsetSize].unpack('N')[0]
+              return offset
+            end
+          end
+          nil
+        end
+
+        def parse_object(offset)
+          obj = nil
+          with_packfile do |packfile|
+            data, type = unpack_object(packfile, offset)
+            obj = RawObject.new(OBJ_TYPES[type], data)
+          end
+          obj
+        end
+        protected :parse_object
+
+        def unpack_object(packfile, offset, options = {})
+          obj_offset = offset
+          packfile.seek(offset)
+
+          c = packfile.read(1)[0]
+          size = c & 0xf
+          type = (c >> 4) & 7
+          shift = 4
+          offset += 1
+          while c & 0x80 != 0
+            c = packfile.read(1)[0]
+            size |= ((c & 0x7f) << shift)
+            shift += 7
+            offset += 1
+          end
+        
+          return [false, false] if !(type == OBJ_COMMIT || type == OBJ_TREE) && options[:caching]
+          
+          case type
+          when OBJ_OFS_DELTA, OBJ_REF_DELTA
+            data, type = unpack_deltified(packfile, type, offset, obj_offset, size, options)
+            #puts type
+          when OBJ_COMMIT, OBJ_TREE, OBJ_BLOB, OBJ_TAG
+            data = unpack_compressed(offset, size)
+          else
+            raise PackFormatError, "invalid type #{type}"
+          end
+          [data, type]
+        end
+        private :unpack_object
+
+        def unpack_deltified(packfile, type, offset, obj_offset, size, options = {})
+          packfile.seek(offset)
+          data = packfile.read(SHA1Size)
+
+          if type == OBJ_OFS_DELTA
+            i = 0
+            c = data[i]
+            base_offset = c & 0x7f
+            while c & 0x80 != 0
+              c = data[i += 1]
+              base_offset += 1
+              base_offset <<= 7
+              base_offset |= c & 0x7f
+            end
+            base_offset = obj_offset - base_offset
+            offset += i + 1
+          else
+            base_offset = find_object(data)
+            offset += SHA1Size
+          end
+
+          base, type = unpack_object(packfile, base_offset)
+          
+          return [false, false] if !(type == OBJ_COMMIT || type == OBJ_TREE) && options[:caching]
+          
+          delta = unpack_compressed(offset, size)
+          [patch_delta(base, delta), type]
+        end
+        private :unpack_deltified
+
+        def unpack_compressed(offset, destsize)
+          outdata = ""
+          with_packfile do |packfile|
+            packfile.seek(offset)
+            zstr = Zlib::Inflate.new
+            while outdata.size < destsize
+              indata = packfile.read(4096)
+              if indata.size == 0
+                raise PackFormatError, 'error reading pack data'
+              end
+              outdata += zstr.inflate(indata)
+            end
+            if outdata.size > destsize
+              raise PackFormatError, 'error reading pack data'
+            end
+            zstr.close
+          end
+          outdata
+        end
+        private :unpack_compressed
+
+        def patch_delta(base, delta)
+          src_size, pos = patch_delta_header_size(delta, 0)
+          if src_size != base.size
+            raise PackFormatError, 'invalid delta data'
+          end
+
+          dest_size, pos = patch_delta_header_size(delta, pos)
+          dest = ""
+          while pos < delta.size
+            c = delta[pos]
+            pos += 1
+            if c & 0x80 != 0
+              pos -= 1
+              cp_off = cp_size = 0
+              cp_off = delta[pos += 1] if c & 0x01 != 0
+              cp_off |= delta[pos += 1] << 8 if c & 0x02 != 0
+              cp_off |= delta[pos += 1] << 16 if c & 0x04 != 0
+              cp_off |= delta[pos += 1] << 24 if c & 0x08 != 0
+              cp_size = delta[pos += 1] if c & 0x10 != 0
+              cp_size |= delta[pos += 1] << 8 if c & 0x20 != 0
+              cp_size |= delta[pos += 1] << 16 if c & 0x40 != 0
+              cp_size = 0x10000 if cp_size == 0
+              pos += 1
+              dest += base[cp_off,cp_size]
+            elsif c != 0
+              dest += delta[pos,c]
+              pos += c
+            else
+              raise PackFormatError, 'invalid delta data'
+            end
+          end
+          dest
+        end
+        private :patch_delta
+
+        def patch_delta_header_size(delta, pos)
+          size = 0
+          shift = 0
+          begin
+            c = delta[pos]
+            if c == nil
+              raise PackFormatError, 'invalid delta header'
+            end
+            pos += 1
+            size |= (c & 0x7f) << shift
+            shift += 7
+          end while c & 0x80 != 0
+          [size, pos]
+        end
+        private :patch_delta_header_size
+      end
+    end
+  end 
+end

data/lib/grit/git-ruby/internal/raw_object.rb

@@ -0,0 +1,37 @@
+#
+# converted from the gitrb project
+#
+# authors: 
+#    Matthias Lederhofer <matled@gmx.net>
+#    Simon 'corecode' Schubert <corecode@fs.ei.tum.de>
+#
+# provides native ruby access to git objects and pack files
+#
+
+require 'digest/sha1'
+
+module Grit 
+  module GitRuby 
+    module Internal
+      OBJ_NONE = 0
+      OBJ_COMMIT = 1
+      OBJ_TREE = 2
+      OBJ_BLOB = 3
+      OBJ_TAG = 4
+
+      OBJ_TYPES = [nil, :commit, :tree, :blob, :tag].freeze
+
+      class RawObject
+        attr_accessor :type, :content
+        def initialize(type, content)
+          @type = type
+          @content = content
+        end
+
+        def sha1
+          Digest::SHA1.digest("%s %d\0" % [@type, @content.length] + @content)
+        end
+      end
+    end 
+  end
+end

data/lib/grit/git-ruby/object.rb

@@ -0,0 +1,319 @@
+#
+# converted from the gitrb project
+#
+# authors: 
+#    Matthias Lederhofer <matled@gmx.net>
+#    Simon 'corecode' Schubert <corecode@fs.ei.tum.de>
+#    Scott Chacon <schacon@gmail.com>
+#
+# provides native ruby access to git objects and pack files
+#
+
+# These classes translate the raw binary data kept in the sha encoded files
+# into parsed data that can then be used in another fashion
+require 'stringio'
+
+module Grit
+  module GitRuby
+
+  # class for author/committer/tagger lines
+  class UserInfo
+    attr_accessor :name, :email, :date, :offset
+
+    def initialize(str)
+      m = /^(.*?) <(.*)> (\d+) ([+-])0*(\d+?)$/.match(str)
+      if !m
+        raise RuntimeError, "invalid header '%s' in commit" % str
+      end
+      @name = m[1]
+      @email = m[2]
+      @date = Time.at(Integer(m[3]))
+      @offset = (m[4] == "-" ? -1 : 1)*Integer(m[5])
+    end
+
+    def to_s
+      "%s <%s> %s %+05d" % [@name, @email, @date.to_i, @offset]
+    end
+  end
+
+  # base class for all git objects (blob, tree, commit, tag)
+  class Object
+    attr_accessor :repository
+
+    def Object.from_raw(rawobject, repository = nil)      
+      case rawobject.type
+      when :blob
+        return Blob.from_raw(rawobject, repository)
+      when :tree
+        return Tree.from_raw(rawobject, repository)
+      when :commit
+        return Commit.from_raw(rawobject, repository)
+      when :tag
+        return Tag.from_raw(rawobject, repository)
+      else
+        raise RuntimeError, "got invalid object-type"
+      end
+    end
+
+    def initialize
+      raise NotImplemented, "abstract class"
+    end
+
+    def type
+      raise NotImplemented, "abstract class"
+    end
+
+    def raw_content
+      raise NotImplemented, "abstract class"
+    end
+
+    def sha1
+      Digest::SHA1.hexdigest("%s %d\0" % \
+                             [self.type, self.raw_content.length] + \
+                             self.raw_content)
+    end
+  end
+
+  class Blob < Object
+    attr_accessor :content
+
+    def self.from_raw(rawobject, repository)
+      new(rawobject.content)
+    end
+
+    def initialize(content, repository=nil)
+      @content = content
+      @repository = repository
+    end
+
+    def type
+      :blob
+    end
+
+    def raw_content
+      @content
+    end
+  end
+
+  class DirectoryEntry
+    S_IFMT  = 00170000
+    S_IFLNK =  0120000
+    S_IFREG =  0100000
+    S_IFDIR =  0040000
+
+    attr_accessor :mode, :name, :sha1
+    def initialize(mode, filename, sha1o)
+      @mode = 0
+      mode.each_byte do |i|
+        @mode = (@mode << 3) | (i-'0'[0])
+      end
+      @name = filename
+      @sha1 = sha1o
+      if ![S_IFLNK, S_IFDIR, S_IFREG].include?(@mode & S_IFMT)
+        raise RuntimeError, "unknown type for directory entry"
+      end
+    end
+
+    def type
+      case @mode & S_IFMT
+      when S_IFLNK
+        @type = :link
+      when S_IFDIR
+        @type = :directory
+      when S_IFREG
+        @type = :file
+      else
+        raise RuntimeError, "unknown type for directory entry"
+      end
+    end
+
+    def type=(type)
+      case @type
+      when :link
+        @mode = (@mode & ~S_IFMT) | S_IFLNK
+      when :directory
+        @mode = (@mode & ~S_IFMT) | S_IFDIR
+      when :file
+        @mode = (@mode & ~S_IFMT) | S_IFREG
+      else
+        raise RuntimeError, "invalid type"
+      end
+    end
+
+    def format_type
+      case type
+      when :link
+        'link'
+      when :directory
+        'tree'
+      when :file
+        'blob'
+      end
+    end
+
+    def format_mode
+      "%06o" % @mode
+    end
+    
+    def raw
+      "%o %s\0%s" % [@mode, @name, [@sha1].pack("H*")]
+    end
+  end
+
+
+  def self.read_bytes_until(io, char)
+    string = ''
+    while ((next_char = io.getc.chr) != char) && !io.eof
+      string += next_char
+    end
+    string
+  end
+
+  
+  class Tree < Object
+    attr_accessor :entry
+
+    def self.from_raw(rawobject, repository=nil)
+      raw = StringIO.new(rawobject.content)
+  
+      entries = []
+      while !raw.eof?
+        mode      = Grit::GitRuby.read_bytes_until(raw, ' ')
+        file_name = Grit::GitRuby.read_bytes_until(raw, "\0")
+        raw_sha   = raw.read(20)
+        sha = raw_sha.unpack("H*").first
+        
+        entries << DirectoryEntry.new(mode, file_name, sha)
+      end
+      new(entries, repository)
+    end
+
+    def initialize(entries=[], repository = nil)
+      @entry = entries
+      @repository = repository
+    end
+
+    def type
+      :tree
+    end
+
+    def raw_content
+      # TODO: sort correctly
+      #@entry.sort { |a,b| a.name <=> b.name }.
+      @entry.collect { |e| [[e.format_mode, e.format_type, e.sha1].join(' '), e.name].join("\t") }.join("\n")
+    end
+    
+    def actual_raw
+      #@entry.collect { |e| e.raw.join(' '), e.name].join("\t") }.join("\n")
+    end
+  end
+
+  class Commit < Object
+    attr_accessor :author, :committer, :tree, :parent, :message, :headers
+
+    def self.from_raw(rawobject, repository=nil)
+      parent = []
+      tree = author = committer = nil
+
+      headers, message = rawobject.content.split(/\n\n/, 2)
+      all_headers = headers.split(/\n/).map { |header| header.split(/ /, 2) }
+      all_headers.each do |key, value|
+        case key
+        when "tree"
+          tree = value
+        when "parent"
+          parent.push(value)
+        when "author"
+          author = UserInfo.new(value)
+        when "committer"
+          committer = UserInfo.new(value)
+        else
+          warn "unknown header '%s' in commit %s" % \
+            [key, rawobject.sha1.unpack("H*")[0]]
+        end
+      end
+      if not tree && author && committer
+        raise RuntimeError, "incomplete raw commit object"
+      end
+      new(tree, parent, author, committer, message, headers, repository)
+    end
+
+    def initialize(tree, parent, author, committer, message, headers, repository=nil)
+      @tree = tree
+      @author = author
+      @parent = parent
+      @committer = committer
+      @message = message
+      @headers = headers
+      @repository = repository
+    end
+
+    def type
+      :commit
+    end
+
+    def raw_content
+      "tree %s\n%sauthor %s\ncommitter %s\n\n" % [
+        @tree,
+        @parent.collect { |i| "parent %s\n" % i }.join,
+        @author, @committer] + @message
+    end
+    
+    def raw_log(sha)
+      output = "commit #{sha}\n"
+      output += @headers + "\n\n"
+      output += @message.split("\n").map { |l| '    ' + l }.join("\n") + "\n\n"
+    end
+    
+  end
+
+  class Tag < Object
+    attr_accessor :object, :type, :tag, :tagger, :message
+
+    def self.from_raw(rawobject, repository=nil)
+      headers, message = rawobject.content.split(/\n\n/, 2)
+      headers = headers.split(/\n/).map { |header| header.split(/ /, 2) }
+      headers.each do |key, value|
+        case key
+        when "object"
+          object = value
+        when "type"
+          if !["blob", "tree", "commit", "tag"].include?(value)
+            raise RuntimeError, "invalid type in tag"
+          end
+          type = value.to_sym
+        when "tag"
+          tag = value
+        when "tagger"
+          tagger = UserInfo.new(value)
+        else
+          warn "unknown header '%s' in tag" % \
+            [key, rawobject.sha1.unpack("H*")[0]]
+        end
+        if not object && type && tag && tagger
+          raise RuntimeError, "incomplete raw tag object"
+        end
+      end
+      new(object, type, tag, tagger, repository)
+    end
+
+    def initialize(object, type, tag, tagger, repository=nil)
+      @object = object
+      @type = type
+      @tag = tag
+      @tagger = tagger
+      @repository = repository
+    end
+
+    def raw_content
+      "object %s\ntype %s\ntag %s\ntagger %s\n\n" % \
+        [@object, @type, @tag, @tagger] + @message
+    end
+
+    def type
+      :tag
+    end
+  end
+
+  end
+end