summaryrefslogtreecommitdiffstats
path: root/lib/git/raw/internal
diff options
context:
space:
mode:
authorscott Chacon <schacon@agadorsparticus.corp.reactrix.com>2007-11-20 12:07:46 -0800
committerscott Chacon <schacon@agadorsparticus.corp.reactrix.com>2007-11-20 12:07:46 -0800
commitd07a2c04c1599fe707831afdd29397cc36e02fa6 (patch)
tree8bde8310874c92c4c2a51998ff53166a8e8e2d94 /lib/git/raw/internal
parent2cef1e66c395296620225c7ffd77b772c6ad4215 (diff)
downloadthird_party-ruby-git-d07a2c04c1599fe707831afdd29397cc36e02fa6.tar.gz
third_party-ruby-git-d07a2c04c1599fe707831afdd29397cc36e02fa6.tar.xz
third_party-ruby-git-d07a2c04c1599fe707831afdd29397cc36e02fa6.zip
added files from the gitrb project, which seems abandoned, but which is great code
Diffstat (limited to 'lib/git/raw/internal')
-rw-r--r--lib/git/raw/internal/loose.rb96
-rw-r--r--lib/git/raw/internal/mmap.rb44
-rw-r--r--lib/git/raw/internal/object.rb23
-rw-r--r--lib/git/raw/internal/pack.rb240
4 files changed, 403 insertions, 0 deletions
diff --git a/lib/git/raw/internal/loose.rb b/lib/git/raw/internal/loose.rb
new file mode 100644
index 0000000..0e4020c
--- /dev/null
+++ b/lib/git/raw/internal/loose.rb
@@ -0,0 +1,96 @@
+require 'zlib'
+require 'digest/sha1'
+
+require 'git/raw/internal/object'
+
+module Git module Raw module Internal
+ class LooseObjectError < StandardError
+ end
+
+ class LooseStorage
+ def initialize(directory)
+ @directory = directory
+ end
+
+ def [](sha1)
+ sha1 = sha1.unpack("H*")[0]
+
+ path = @directory+'/'+sha1[0...2]+'/'+sha1[2..40]
+ begin
+ get_raw_object(File.read(path))
+ rescue Errno::ENOENT
+ nil
+ end
+ end
+
+ def get_raw_object(buf)
+ if buf.length < 2
+ raise LooseObjectError, "object file too small"
+ end
+
+ if legacy_loose_object?(buf)
+ content = Zlib::Inflate.inflate(buf)
+ header, content = content.split(/\0/, 2)
+ if !header || !content
+ raise LooseObjectError, "invalid object header"
+ end
+ type, size = header.split(/ /, 2)
+ if !%w(blob tree commit tag).include?(type) || size !~ /^\d+$/
+ raise LooseObjectError, "invalid object header"
+ end
+ type = type.to_sym
+ size = size.to_i
+ else
+ type, size, used = unpack_object_header_gently(buf)
+ content = Zlib::Inflate.inflate(buf[used..-1])
+ end
+ raise LooseObjectError, "size mismatch" if content.length != size
+ return RawObject.new(type, content)
+ end
+
+ # private
+ def unpack_object_header_gently(buf)
+ used = 0
+ c = buf[used]
+ used += 1
+
+ type = (c >> 4) & 7;
+ size = c & 15;
+ shift = 4;
+ while c & 0x80 != 0
+ if buf.length <= used
+ raise LooseObjectError, "object file too short"
+ end
+ c = buf[used]
+ used += 1
+
+ size += (c & 0x7f) << shift
+ shift += 7
+ end
+ type = OBJ_TYPES[type]
+ if ![:blob, :tree, :commit, :tag].include?(type)
+ raise LooseObjectError, "invalid loose object type"
+ end
+ return [type, size, used]
+ end
+ private :unpack_object_header_gently
+
+ def legacy_loose_object?(buf)
+ word = (buf[0] << 8) + buf[1]
+ buf[0] == 0x78 && word % 31 == 0
+ end
+ private :legacy_loose_object?
+ end
+end end
+
+if $0 == __FILE__
+ require 'find'
+ ARGV.each do |path|
+ storage = Git::Internal::LooseStorage.new(path)
+ Find.find(path) do |p|
+ next if !/\/([0-9a-f]{2})\/([0-9a-f]{38})$/.match(p)
+ obj = storage[[$1+$2].pack("H*")]
+ puts "%s %s" % [obj.sha1.unpack("H*")[0], obj.type]
+ end
+ end
+end
diff --git a/lib/git/raw/internal/mmap.rb b/lib/git/raw/internal/mmap.rb
new file mode 100644
index 0000000..d7390b1
--- /dev/null
+++ b/lib/git/raw/internal/mmap.rb
@@ -0,0 +1,44 @@
+begin
+ require 'mmap'
+rescue LoadError
+
+module Git module Raw 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 # rescue LoadError
+
diff --git a/lib/git/raw/internal/object.rb b/lib/git/raw/internal/object.rb
new file mode 100644
index 0000000..b81df2b
--- /dev/null
+++ b/lib/git/raw/internal/object.rb
@@ -0,0 +1,23 @@
+require 'digest/sha1'
+
+module Git module Raw 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
diff --git a/lib/git/raw/internal/pack.rb b/lib/git/raw/internal/pack.rb
new file mode 100644
index 0000000..edfeada
--- /dev/null
+++ b/lib/git/raw/internal/pack.rb
@@ -0,0 +1,240 @@
+require 'zlib'
+require 'git/raw/internal/object'
+require 'git/raw/internal/mmap'
+
+module Git module Raw 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
+
+ @name = file
+ @packfile = File.open(file)
+ @idxfile = File.open(file[0...-4]+'idx')
+ @idx = Mmap.new(@idxfile)
+
+ @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
+
+ def close
+ @packfile.close
+ @idx.unmap
+ @idxfile.close
+ end
+
+ def [](sha1)
+ offset = find_object(sha1)
+ return nil if !offset
+ return parse_object(offset)
+ end
+
+ def each_entry
+ pos = OffsetStart
+ @size.times do
+ offset = @idx[pos,OffsetSize].unpack('N')[0]
+ sha1 = @idx[pos+OffsetSize,SHA1Size]
+ pos += EntrySize
+ yield sha1, offset
+ end
+ end
+
+ def each_sha1
+ # 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
+
+ def find_object(sha1)
+ slot = sha1[0]
+ 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
+ private :find_object
+
+ def parse_object(offset)
+ data, type = unpack_object(offset)
+ RawObject.new(OBJ_TYPES[type], data)
+ end
+ protected :parse_object
+
+ def unpack_object(offset)
+ 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
+
+ case type
+ when OBJ_OFS_DELTA, OBJ_REF_DELTA
+ data, type = unpack_deltified(type, offset, obj_offset, size)
+ 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(type, offset, obj_offset, size)
+ @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(base_offset)
+ delta = unpack_compressed(offset, size)
+ [patch_delta(base, delta), type]
+ end
+ private :unpack_deltified
+
+ def unpack_compressed(offset, destsize)
+ outdata = ""
+ @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
+ 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
+
+if $0 == __FILE__
+ ARGV.each do |path|
+ storage = Git::Internal::PackStorage.new(path)
+ storage.each_sha1 do |sha1|
+ obj = storage[sha1]
+ puts "%s %s" % [obj.sha1.unpack('H*'), obj.type]
+ end
+ end
+end