omnizip 0.3.8 → 0.3.10

data/lib/omnizip/formats/xz.rb

@@ -17,10 +17,6 @@ module Omnizip
         # @option options [Integer] :dict_size Dictionary size (default: 8MB to match XZ Utils preset 6)
         # @option options [Integer] :check Check type (default: CRC64)
         def create(input, output = nil, options = {})
-          require "omnizip/formats/xz_impl"
-          require "omnizip/formats/xz_const"
-          require "omnizip/formats/xz_impl/stream_encoder"
-
           encoder = ::Omnizip::Formats::XzImpl::StreamEncoder.new(
             check_type: options[:check] || ::Omnizip::Formats::XzConst::CHECK_CRC64,
             dict_size: options[:dict_size] || (64 * 1024 * 1024),
@@ -29,11 +25,7 @@ module Omnizip
           compressed = encoder.encode(input)
 
           if output
-            if output.respond_to?(:write)
-              output.write(compressed)
-            else
-              File.binwrite(output, compressed)
-            end
+            Omnizip::IO::Sink.for(output).write(compressed)
           else
             compressed
           end
@@ -72,42 +64,13 @@ module Omnizip
         # @param options [Hash] Options (reserved for future use)
         # @return [String, nil] Decompressed data (if output is nil)
         def decompress(input, output = nil, _options = {})
-          # Handle raw data string vs file path
-          data = if input.respond_to?(:read)
-                   # Already an IO object - read content
-                   if input.respond_to?(:size)
-                     # Seekable IO (File, etc.) - read without consuming
-                     original_pos = input.pos
-                     content = input.read
-                     input.seek(original_pos)
-                   else
-                     # Non-seekable IO - read and consume
-                     content = input.read
-                   end
-                   content
-                 elsif input.is_a?(String)
-                   # Could be file path or raw data
-                   # If string contains null byte, it's definitely data (not a path)
-                   # Also check if it's a valid file path first
-                   if !input.include?("\0") && File.exist?(input)
-                     File.binread(input)
-                   else
-                     input.b
-                   end
-                 else
-                   raise ArgumentError,
-                         "Input must be a String or IO object"
-                 end
+          data = Omnizip::IO::Source.for(input).read
 
           # Detect format and decode
           decompressed = decode_lzma_data(data)
 
           if output
-            if output.respond_to?(:write)
-              output.write(decompressed)
-            else
-              File.binwrite(output, decompressed)
-            end
+            Omnizip::IO::Sink.for(output).write(decompressed)
             nil
           else
             decompressed

data/lib/omnizip/formats/xz_impl/block_decoder.rb

@@ -567,11 +567,8 @@ module Omnizip
 
           # Create LZMA2 decoder with raw_mode for XZ format
           decoder = Omnizip::Implementations::XZUtils::LZMA2::Decoder.new(input_buffer,
-                                                                          raw_mode: true)
-
-          # Set dict_size directly since we skipped property byte reading
-          decoder.instance_variable_set(:@dict_size, dict_size)
-          decoder.instance_variable_set(:@properties, Omnizip::Algorithms::LZMA2::Properties.new(dict_size))
+                                                                          raw_mode: true,
+                                                                          dict_size: dict_size)
 
           # Decode stream
           uncompressed_data = decoder.decode_stream
@@ -623,11 +620,8 @@ module Omnizip
 
           # Create LZMA2 decoder with raw_mode for XZ format
           decoder = Omnizip::Implementations::XZUtils::LZMA2::Decoder.new(input_buffer,
-                                                                          raw_mode: true)
-
-          # Set dict_size directly since we skipped property byte reading
-          decoder.instance_variable_set(:@dict_size, dict_size)
-          decoder.instance_variable_set(:@properties, Omnizip::Algorithms::LZMA2::Properties.new(dict_size))
+                                                                          raw_mode: true,
+                                                                          dict_size: dict_size)
 
           # Decode stream
           decoder.decode_stream

data/lib/omnizip/formats/xz_impl/block_encoder.rb

@@ -1,8 +1,6 @@
 # frozen_string_literal: true
 
 require "stringio"
-require "omnizip/algorithms/xz_lzma2"
-require "omnizip/checksums/crc64"
 require "zlib"
 
 module Omnizip

data/lib/omnizip/formats/xz_impl/block_header_parser.rb

@@ -22,8 +22,6 @@
 
 require "stringio"
 require "zlib"
-require "omnizip/error"
-require "omnizip/checksums/verifier"
 
 module Omnizip
   module Formats

data/lib/omnizip/formats/xz_impl/index_decoder.rb

@@ -22,7 +22,6 @@
 
 require "stringio"
 require "zlib"
-require "omnizip/error"
 
 module Omnizip
   module Formats

data/lib/omnizip/formats/xz_impl/index_encoder.rb

@@ -2,7 +2,6 @@
 
 require "stringio"
 require "zlib"
-require "omnizip/error"
 
 module Omnizip
   module Formats

data/lib/omnizip/formats/xz_impl/stream_encoder.rb

@@ -25,7 +25,7 @@ module Omnizip
           output.set_encoding(Encoding::BINARY)
 
           # Read input data
-          input_data = input.respond_to?(:read) ? input.read : input.to_s
+          input_data = Omnizip::IO::Source.for(input).read
           input_data = input_data.dup.force_encoding(Encoding::BINARY)
 
           # 1. Write Stream Header

data/lib/omnizip/formats/xz_impl/stream_footer.rb

@@ -1,7 +1,6 @@
 # frozen_string_literal: true
 
 require "zlib"
-require "omnizip/error"
 
 module Omnizip
   module Formats

data/lib/omnizip/formats/xz_impl/stream_footer_parser.rb

@@ -21,7 +21,6 @@
 # DEALINGS IN THE SOFTWARE.
 
 require "zlib"
-require "omnizip/error"
 
 module Omnizip
   module Formats

data/lib/omnizip/formats/xz_impl/stream_header.rb

@@ -1,7 +1,6 @@
 # frozen_string_literal: true
 
 require "zlib"
-require "omnizip/error"
 
 module Omnizip
   module Formats

data/lib/omnizip/formats/xz_impl/stream_header_parser.rb

@@ -21,8 +21,6 @@
 # DEALINGS IN THE SOFTWARE.
 
 require "zlib"
-require "omnizip/error"
-require "omnizip/checksums/verifier"
 
 module Omnizip
   module Formats

data/lib/omnizip/formats/zip/central_directory_header.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require "omnizip/formats/zip"
-
 module Omnizip
   module Formats
     module Zip
@@ -14,7 +12,8 @@ module Omnizip
                       :crc32, :compressed_size, :uncompressed_size,
                       :filename_length, :extra_field_length, :comment_length,
                       :disk_number_start, :internal_attributes, :external_attributes,
-                      :local_header_offset, :filename, :extra_field, :comment
+                      :local_header_offset, :filename, :extra_field, :comment,
+                      :cached_original_data
 
         def initialize(
           signature: CENTRAL_DIRECTORY_SIGNATURE,

data/lib/omnizip/formats/zip/end_of_central_directory.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require "omnizip/formats/zip"
-
 module Omnizip
   module Formats
     module Zip

data/lib/omnizip/formats/zip/local_file_header.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require "omnizip/formats/zip"
-
 module Omnizip
   module Formats
     module Zip

data/lib/omnizip/formats/zip/reader.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require "omnizip/formats/zip"
-
 module Omnizip
   module Formats
     module Zip
@@ -149,6 +147,18 @@ dereference_links: false)
           end
         end
 
+        # Decompress +compressed_data+ according to +method+. Public API
+        # for callers (e.g., ParallelExtractor) that hold raw compressed
+        # payload outside an entry stream.
+        #
+        # @param compressed_data [String] Compressed bytes
+        # @param method [Integer] ZIP compression method code
+        # @param uncompressed_size [Integer] Expected uncompressed size
+        # @return [String] Decompressed bytes
+        def decompress(compressed_data, method, uncompressed_size)
+          decompress_data(compressed_data, method, uncompressed_size)
+        end
+
         private
 
         # Read central directory entries

data/lib/omnizip/formats/zip/writer.rb

@@ -1,7 +1,6 @@
 # frozen_string_literal: true
 
 require "fileutils"
-require "omnizip/formats/zip"
 
 module Omnizip
   module Formats
@@ -112,6 +111,79 @@ module Omnizip
           end
         end
 
+        # Add an entry whose payload is already compressed. Bypasses
+        # per-entry compression in +#write+, so +#write_precompressed+
+        # must be used instead of +#write+ to emit the archive.
+        #
+        # @param filename [String] Entry name in the archive
+        # @param uncompressed_size [Integer] Size before compression
+        # @param compressed_size [Integer] Size of +compressed_data+
+        # @param crc32 [Integer] CRC32 of the uncompressed data
+        # @param compressed_data [String] Pre-compressed payload
+        # @param stat [File::Stat, nil] Optional stat for the source file
+        def add_precompressed_entry(filename:, uncompressed_size:,
+                                    compressed_size:, crc32:,
+                                    compressed_data:, stat: nil)
+          entry = create_entry(
+            filename: filename,
+            uncompressed_data: "",
+            stat: stat,
+          )
+          entry[:compressed_size] = compressed_size
+          entry[:uncompressed_size] = uncompressed_size
+          entry[:crc32] = crc32
+          entry[:compressed_data] = compressed_data
+          @entries << entry
+          entry
+        end
+
+        # Write the archive using pre-compressed entry payloads (set
+        # via +#add_precompressed_entry+).
+        #
+        # @param compression_method [Integer] ZIP compression method
+        #   code recorded in the headers
+        def write_precompressed(compression_method: COMPRESSION_DEFLATE)
+          File.open(file_path, "wb") do |io|
+            write_precompressed_to_io(io, compression_method: compression_method)
+          end
+        end
+
+        def write_precompressed_to_io(io, compression_method: COMPRESSION_DEFLATE)
+          local_header_offsets = []
+
+          entries.each do |entry|
+            offset = io.pos
+            local_header_offsets << offset
+
+            local_header = create_local_header(entry, compression_method)
+            local_header.compressed_size = entry[:compressed_size]
+            local_header.uncompressed_size = entry[:uncompressed_size]
+            local_header.crc32 = entry[:crc32]
+            io.write(local_header.to_binary)
+            io.write(entry[:compressed_data]) unless entry[:directory]
+          end
+
+          central_directory_offset = io.pos
+
+          entries.each_with_index do |entry, index|
+            central_header = create_central_header(
+              entry,
+              compression_method,
+              local_header_offsets[index],
+            )
+            io.write(central_header.to_binary)
+          end
+
+          central_directory_size = io.pos - central_directory_offset
+
+          eocd = create_eocd(
+            total_entries: entries.size,
+            central_directory_size: central_directory_size,
+            central_directory_offset: central_directory_offset,
+          )
+          io.write(eocd.to_binary)
+        end
+
         # Write to an IO object
         def write_to_io(io, compression_method: COMPRESSION_DEFLATE, level: 6)
           local_header_offsets = []

data/lib/omnizip/formats/zip/zip64_end_of_central_directory.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require "omnizip/formats/zip"
-
 module Omnizip
   module Formats
     module Zip

data/lib/omnizip/formats/zip/zip64_end_of_central_directory_locator.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require "omnizip/formats/zip"
-
 module Omnizip
   module Formats
     module Zip

data/lib/omnizip/formats/zip/zip64_extra_field.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require "omnizip/formats/zip"
-
 module Omnizip
   module Formats
     module Zip

data/lib/omnizip/formats/zip.rb

@@ -47,7 +47,6 @@ module Omnizip
 
         # Auto-register .zip format when loaded
         def register!
-          require "omnizip/format_registry"
           FormatRegistry.register(".zip", Omnizip::Formats::Zip)
         end
       end

data/lib/omnizip/formats.rb

@@ -2,39 +2,26 @@
 
 module Omnizip
   module Formats
-    # Archive format support
-    #
-    # This module contains format-specific autoload declarations.
-    # Less commonly used formats are autoloaded for lazy loading.
-
-    # Format specification loader
     autoload :FormatSpecLoader, "omnizip/formats/format_spec_loader"
 
-    # CPIO archive format
-    autoload :Cpio, "omnizip/formats/cpio"
+    autoload :SevenZip, "omnizip/formats/seven_zip"
+    autoload :Zip, "omnizip/formats/zip"
+    autoload :Rar, "omnizip/formats/rar"
+    autoload :Tar, "omnizip/formats/tar"
+    autoload :Gzip, "omnizip/formats/gzip"
+    autoload :Bzip2File, "omnizip/formats/bzip2_file"
+    autoload :Xz, "omnizip/formats/xz"
 
-    # RPM package format
+    autoload :Cpio, "omnizip/formats/cpio"
     autoload :Rpm, "omnizip/formats/rpm"
-
-    # OLE compound documents (MSI, DOC, XLS, PPT)
     autoload :Ole, "omnizip/formats/ole"
-
-    # XAR archive format
+    autoload :Msi, "omnizip/formats/msi"
     autoload :Xar, "omnizip/formats/xar"
-
-    # ISO 9660 CD-ROM format
     autoload :Iso, "omnizip/formats/iso"
-
-    # LZMA alone format
     autoload :LzmaAlone, "omnizip/formats/lzma_alone"
-
-    # LZIP format
     autoload :Lzip, "omnizip/formats/lzip"
 
-    # XZ format constants (at Formats level for cross-module access)
     autoload :XzConst, "omnizip/formats/xz_const"
-
-    # XZ implementation details
     autoload :XzImpl, "omnizip/formats/xz_impl"
   end
 end

data/lib/omnizip/implementations/seven_zip/lzma/range_decoder.rb

@@ -74,14 +74,14 @@ module Omnizip
               # UPDATE_0: bit is 0
               @range = bound & 0xFFFFFFFF
               new_prob = prob + ((BIT_MODEL_TOTAL - prob) >> MOVE_BITS)
-              model.instance_variable_set(:@probability, new_prob)
+              model.probability = new_prob
               0
             else
               # UPDATE_1: bit is 1
               @range = (@range - bound) & 0xFFFFFFFF
               @code = (@code - bound) & 0xFFFFFFFF
               new_prob = prob - (prob >> MOVE_BITS)
-              model.instance_variable_set(:@probability, new_prob)
+              model.probability = new_prob
               1
             end
           end

data/lib/omnizip/implementations/seven_zip/lzma/range_encoder.rb

@@ -91,7 +91,7 @@ module Omnizip
                          # Update probability: ttt -= ttt >> kNumMoveBits
                          prob - (prob >> MOVE_BITS)
                        end
-            model.instance_variable_set(:@probability, new_prob)
+            model.probability = new_prob
 
             # RC_NORM: Normalize AFTER encoding (key SDK difference!)
             normalize

data/lib/omnizip/implementations/seven_zip/lzma2/encoder.rb

@@ -71,12 +71,6 @@ module Omnizip
 
             # Initialize shared state across all chunks
             # Using XZ Utils components (tested and working)
-            require "omnizip/algorithms/lzma/dictionary"
-            require "omnizip/algorithms/lzma/lzma_state"
-            require "omnizip/algorithms/lzma/xz_probability_models"
-            require "omnizip/algorithms/lzma/match_finder"
-            require "omnizip/algorithms/lzma/optimal_encoder"
-            require "omnizip/algorithms/lzma/xz_range_encoder_exact"
 
             @dictionary = Omnizip::Algorithms::LZMA::Dictionary.new(dict_size)
             @state = Omnizip::Algorithms::LZMA::LZMAState.new(0)