omnizip 0.3.52 → 0.3.53

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8002bf03f99715929f352dfb92a91a10d02f6dac7152006a6cb87ca20b3647ae
-  data.tar.gz: b095258b0b0af358ae9c0a21482a6f8b29de1620bb2162d64dd8794f29eb8f59
+  metadata.gz: a7e72a57eaeb6e597e40020de67aa0242da65c6e6e8ceaa933d779d926e55666
+  data.tar.gz: 105cf4f238610f518462c115d398403fef5daca9b2994788361b027c7c5ec648
 SHA512:
-  metadata.gz: 527be0108ee0fa63b6e7ebafdeaafd05430a17c08e0d59b0f731db7a3b76401da96daec205947c7c3b5baa07aba886515959513ef3d4b31438889761219e958d
-  data.tar.gz: e1f8665f545fc81a4459cbdffbb9e5d5aca219a177a5b2d2ac3bd47c589a2ceb11b02d8cfdc748ac39381417f12682bf53cf14105dcbcd3d04bd59b77d7f9e87
+  metadata.gz: a5301b14047d1150885624527d03176c61a979464b7a16329a47449e13fe05679c1e5ce11a650bfd29101aaaf502b5ab81513558fc080dca85554266ae30f8dc
+  data.tar.gz: a44ffd04cc02fa03e2b23a330da4f6612d268bfd350d9b9672e42d7a107dcad65201ed9707651b3533fdd790a5841a5ff9297d6f3a8e937b51e0f1614d756c55

data/CHANGELOG.md

@@ -7,6 +7,35 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.3.52] - 2026-09-01
+
+### Changed
+- The extract command's pattern path opens archives through the
+  Archive facade instead of a hand-rolled extension `case` that
+  misrouted `.cpio`/`.iso`/`.rpm` files into the 7z reader — pattern
+  extraction now works for every routed format. The metadata
+  command's 7z view lists through the handler registry (the detail
+  contract gains `:crc`), and SafeExtract extracts through
+  `extract_archive`, so every routed format gets rollback and
+  verification instead of ZIP-only with a `NotImplementedError`
+  for 7z. SelectiveExtractor understands both entry contracts
+  (legacy `entry_name` and reader-session `name`) and skips
+  directory entries when extracting.
+- `CliOutputFormatter.format_size` is now the single byte formatter;
+  the nine private `format_bytes`/`format_size` copies (four
+  different shapes, two unit ceilings) become delegators.
+
+### Fixed
+- The long-deferred parity spec flake: `Par2Verifier` crashed five
+  frames deep on a nil PAR2 path (`File.exist?(nil)` → TypeError)
+  where the repairer already raised a clean ArgumentError; the
+  verifier now mirrors it, and the spec's `.verify` group gains the
+  nil-index fallback `.repair` already had. The dead shadowed
+  `analyze_file` definition in `Par2Creator` (silently overridden
+  by a second definition with a divergent file_id format) is
+  deleted. The tautological "LZMA2 is implicit" xz assertion became
+  a real block-header parse asserting the explicit 0x21 filter ID.
+
 ## [0.3.51] - 2026-09-01
 
 ### Changed

data/lib/omnizip/archive_handlers/zip_handler.rb

@@ -38,7 +38,7 @@ module Omnizip
           extracted = []
           reader.entries.each do |entry|
             dest_path = ::File.join(output_dir, entry.filename)
-            raise "File exists: #{dest_path}" if ::File.exist?(dest_path)
+            raise Errno::EEXIST, "File exists: #{dest_path}" if ::File.exist?(dest_path)
 
             reader.extract_entry(entry, output_dir)
             extracted << dest_path

data/lib/omnizip/buffer/memory_extractor.rb

@@ -151,24 +151,12 @@ module Omnizip
 
       private
 
-      # Detect archive format from magic bytes
+      # Delegate to the shared buffer-side sniffer
       #
       # @return [Symbol] Detected format
       # @raise [Omnizip::FormatError] If format cannot be detected
       def detect_format
-        magic = @buffer.read(4).to_s.b
-        @buffer.rewind
-
-        case magic
-        when "PK\x03\x04".b, "PK\x05\x06".b, "PK\x07\x08".b
-          # ZIP signatures: local file header, EOCD, data descriptor
-          :zip
-        when "7z\xBC\xAF".b
-          :seven_zip
-        else
-          raise Omnizip::FormatError,
-                "Unknown archive format (magic: #{magic.inspect})"
-        end
+        Omnizip::Buffer.detect_format(@buffer)
       end
 
       # Extract all entries from ZIP
@@ -207,8 +195,7 @@ module Omnizip
       def extract_entry_seven_zip(name)
         @buffer.rewind
         SevenZipBridge.open(@buffer) do |archive|
-          entry = archive.raw_entries.find { |e| e.name == name }
-          archive.extract_all_to_memory[name] if entry
+          archive.read_entry(name)
         end
       end
 

data/lib/omnizip/buffer/seven_zip_bridge.rb

@@ -112,6 +112,19 @@ module Omnizip
           @reader.list_files
         end
 
+        # Content of a single named file entry, or nil when absent
+        # (directories carry no content). Extracts just that entry —
+        # not the whole archive.
+        #
+        # @param name [String] Entry name
+        # @return [String, nil]
+        def read_entry(name)
+          entry = raw_entries.find { |e| e.name == name && !e.is_dir }
+          return nil unless entry
+
+          Entry.new(entry, @reader, @tmp).read
+        end
+
         # Entry wrapper matching MemoryArchive::Entry's surface
         class Entry
           attr_reader :name, :size

data/lib/omnizip/buffer.rb

@@ -107,27 +107,9 @@ module Omnizip
         extractor.extract_all
       end
 
-      # Create archive from Hash of filename => content
-      #
-      # @param hash [Hash<String, String>] Filename => content mapping
-      # @param format [Symbol] Archive format
-      # @param options [Hash] Format-specific options
-      # @return [StringIO] Complete archive in memory
-      #
-      # @example Create from Hash
-      #   data = {'file1.txt' => 'content1', 'file2.txt' => 'content2'}
-      #   zip = Omnizip::Buffer.create_from_hash(data, :zip)
-      def create_from_hash(hash, format = :zip, **options)
-        create(format, **options) do |archive|
-          hash.each do |name, content|
-            archive.add(name, content)
-          end
-        end
-      end
-
-      private
-
-      # Detect archive format from magic bytes
+      # Detect archive format from magic bytes. THE buffer-side
+      # sniffer — in-memory data has no extension to route by, so
+      # every consumer delegates here.
       #
       # @param buffer [StringIO] Buffer containing archive data
       # @return [Symbol] Detected format
@@ -148,6 +130,26 @@ module Omnizip
         end
       end
 
+      # Create archive from Hash of filename => content
+      #
+      # @param hash [Hash<String, String>] Filename => content mapping
+      # @param format [Symbol] Archive format
+      # @param options [Hash] Format-specific options
+      # @return [StringIO] Complete archive in memory
+      #
+      # @example Create from Hash
+      #   data = {'file1.txt' => 'content1', 'file2.txt' => 'content2'}
+      #   zip = Omnizip::Buffer.create_from_hash(data, :zip)
+      def create_from_hash(hash, format = :zip, **options)
+        create(format, **options) do |archive|
+          hash.each do |name, content|
+            archive.add(name, content)
+          end
+        end
+      end
+
+      private
+
       # Create ZIP archive in buffer
       #
       # @param buffer [StringIO] Buffer to write to

data/lib/omnizip/checksums/verifier.rb

@@ -94,7 +94,7 @@ module Omnizip
         when CHECK_SHA256
           verify_sha256(data, expected)
         else
-          raise "Unknown check type: #{check_type}"
+          raise Omnizip::UnknownChecksumError, "Unknown check type: #{check_type}"
         end
       end
 
@@ -122,7 +122,7 @@ module Omnizip
         when CHECK_SHA256
           Digest::SHA256.digest(data)
         else
-          raise "Unknown check type: #{check_type}"
+          raise Omnizip::UnknownChecksumError, "Unknown check type: #{check_type}"
         end
       end
     end

data/lib/omnizip/chunked.rb

@@ -95,30 +95,23 @@ module Omnizip
                 "Input archive not found: #{input}"
         end
 
-        writer = Writer.new(output, chunk_size: chunk_size)
-        processed = 0
-
-        Omnizip::Zip::File.open(input) do |zip|
-          entry = zip.entries.first
-          total_size = entry.size
-
-          # Read the full entry content
-          content = zip.get_input_stream(entry)
+        reader = Omnizip::Formats::Zip::Reader.new(input)
+        entry = reader.entries.first
+        raise Omnizip::Error, "No entries to decompress in #{input}" unless entry
 
-          # Write in chunks
-          offset = 0
-          while offset < content.bytesize
-            chunk = content.byteslice(offset, chunk_size) || ""
-            break if chunk.empty?
+        total_size = entry.uncompressed_size.to_i
+        processed = 0
+        writer = Writer.new(output, chunk_size: chunk_size)
 
-            writer.write_chunk(chunk)
-            processed += chunk.bytesize
-            offset += chunk.bytesize
+        # Streams entry-by-entry in bounded memory — the whole entry
+        # is never held at once
+        reader.read_entry_stream(entry.filename, chunk_size: chunk_size) do |chunk|
+          writer.write_chunk(chunk)
+          processed += chunk.bytesize
 
-            if progress
-              percentage = (processed.to_f / total_size * 100).round(2)
-              progress.call(processed, total_size, percentage)
-            end
+          if progress
+            percentage = (processed.to_f / total_size * 100).round(2)
+            progress.call(processed, total_size, percentage)
           end
         end
 

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

@@ -64,7 +64,7 @@ module Omnizip
         # @raise [RuntimeError] if entry not found
         def extract_entry(entry_name, output_path)
           entry = @entries.find { |e| e.name == entry_name }
-          raise "Entry not found: #{entry_name}" unless entry
+          raise Errno::ENOENT, "Entry not found: #{entry_name}" unless entry
 
           extract_single_entry(entry, output_path)
         end

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

@@ -67,7 +67,7 @@ module Omnizip
         # @param output_path [String] Destination path
         def extract_entry(entry_path, output_path)
           entry = find_entry(entry_path)
-          raise "Entry not found: #{entry_path}" unless entry
+          raise Errno::ENOENT, "Entry not found: #{entry_path}" unless entry
 
           if entry.directory?
             FileUtils.mkdir_p(output_path)
@@ -111,7 +111,7 @@ module Omnizip
           loop do
             @io.seek(sector * Iso::SECTOR_SIZE)
             data = @io.read(Iso::SECTOR_SIZE)
-            raise "Failed to read volume descriptor" unless data
+            raise Omnizip::InvalidArchiveError, "Failed to read volume descriptor" unless data
 
             vd = VolumeDescriptor.parse(data)
 
@@ -127,7 +127,7 @@ module Omnizip
 
           return if @primary_volume_descriptor
 
-          raise "No primary volume descriptor found"
+          raise Omnizip::InvalidArchiveError, "No primary volume descriptor found"
         end
 
         # Parse directory structure

data/lib/omnizip/formats/iso/volume_descriptor.rb

@@ -29,7 +29,7 @@ module Omnizip
         #
         # @param data [String] Binary data
         def parse(data)
-          raise "Invalid volume descriptor size" unless data.bytesize >= 2048
+          raise Omnizip::InvalidArchiveError, "Invalid volume descriptor size" unless data.bytesize >= 2048
 
           # Byte 0: Volume descriptor type
           @type = data.getbyte(0)
@@ -37,7 +37,7 @@ module Omnizip
           # Bytes 1-5: Standard identifier "CD001"
           @identifier = data[1, 5]
           unless @identifier == ISO_IDENTIFIER
-            raise "Invalid ISO identifier: expected #{ISO_IDENTIFIER}, got #{@identifier}"
+            raise Omnizip::InvalidArchiveError, "Invalid ISO identifier: expected #{ISO_IDENTIFIER}, got #{@identifier}"
           end
 
           # Byte 6: Version (should be 1)

data/lib/omnizip/formats/rar/compression/ppmd/decoder.rb

@@ -230,26 +230,6 @@ module Omnizip
               # Decode as 8 direct bits
               @range_decoder.decode_direct_bits(8)
             end
-
-            # Decode RAR-specific escape code
-            #
-            # RAR variant H uses different escape code values
-            # and handling compared to standard PPMd7.
-            #
-            # Escape codes in RAR:
-            # - 0: New symbol follows
-            # - 1: Same as last symbol (run-length)
-            # - 2-255: Reserved for future use
-            #
-            # @return [Integer, nil] Escape code or nil
-            def decode_escape_code
-              # RAR escape codes differ from PPMd7
-              # This is a placeholder for the proper implementation
-
-              # For now, return 0 (new symbol follows)
-              # Real implementation would decode from range coder
-              0
-            end
           end
         end
       end

data/lib/omnizip/formats/rar/decompressor.rb

@@ -61,7 +61,7 @@ module Omnizip
           # @param password [String, nil] Optional password
           # @raise [RuntimeError] if extraction fails
           def extract(archive_path, output_dir, password: nil)
-            raise "RAR extraction not available" unless available?
+            raise Omnizip::RarNotAvailableError, "RAR extraction not available" unless available?
 
             if gem_available?
               extract_with_gem(archive_path, output_dir, password)
@@ -78,7 +78,7 @@ module Omnizip
           # @return [Array<Hash>] Entry information
           # @raise [RuntimeError] if listing fails
           def list(archive_path)
-            raise "RAR extraction not available" unless available?
+            raise Omnizip::RarNotAvailableError, "RAR extraction not available" unless available?
 
             if gem_available?
               list_with_gem(archive_path)
@@ -97,7 +97,7 @@ module Omnizip
           # @param password [String, nil] Optional password
           def extract_entry(archive_path, entry_name, output_path,
                             password: nil)
-            raise "RAR extraction not available" unless available?
+            raise Omnizip::RarNotAvailableError, "RAR extraction not available" unless available?
 
             if gem_available?
               extract_entry_with_gem(archive_path, entry_name,
@@ -185,7 +185,7 @@ module Omnizip
             require "unrar"
             Unrar.extract(archive_path, output_dir, password: password)
           rescue StandardError => e
-            raise "Gem extraction failed: #{e.message}"
+            raise Omnizip::DecompressionError, "Gem extraction failed: #{e.message}"
           end
 
           # Extract with system command
@@ -193,7 +193,7 @@ module Omnizip
             cmd = build_extract_command(archive_path, output_dir, password)
             return if system(*cmd)
 
-            raise "Command extraction failed: #{archive_path}"
+            raise Omnizip::DecompressionError, "Command extraction failed: #{archive_path}"
           end
 
           # List with unrar gem
@@ -210,13 +210,13 @@ module Omnizip
               }
             end
           rescue StandardError => e
-            raise "Gem listing failed: #{e.message}"
+            raise Omnizip::DecompressionError, "Gem listing failed: #{e.message}"
           end
 
           # List with system command
           def list_with_command(archive_path)
             output = `"#{command_path}" vb "#{archive_path}" 2>&1`
-            raise "Command listing failed" unless $CHILD_STATUS.success?
+            raise Omnizip::DecompressionError, "Command listing failed" unless $CHILD_STATUS.success?
 
             output.split("\n").map do |line|
               { name: line.strip, size: 0, compressed_size: 0,

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

@@ -64,7 +64,7 @@ module Omnizip
         def extract_entry(entry_name, output_path, password: nil)
           ensure_open
           entry = @entries.find { |e| e.name == entry_name }
-          raise "Entry not found: #{entry_name}" unless entry
+          raise Errno::ENOENT, "Entry not found: #{entry_name}" unless entry
 
           # Create directory if needed
           FileUtils.mkdir_p(File.dirname(output_path))
@@ -150,7 +150,7 @@ module Omnizip
         def parse_archive(io)
           # Read and validate header
           @header = Header.read(io)
-          raise "Invalid RAR archive" unless @header.valid?
+          raise Omnizip::InvalidArchiveError, "Invalid RAR archive" unless @header.valid?
 
           # Update archive info
           @archive_info.version = @header.version

data/lib/omnizip/formats/rpm.rb

@@ -301,7 +301,7 @@ module Omnizip
         #
         # @param output_dir [String] Output directory
         def extract(output_dir)
-          raise "RPM not opened" unless @file
+          raise Omnizip::IOError, "RPM not opened" unless @file
 
           FileUtils.mkdir_p(output_dir)
 
@@ -322,7 +322,7 @@ module Omnizip
         #
         # @return [String] Raw compressed payload data
         def raw_payload
-          raise "RPM not opened" unless @file
+          raise Omnizip::IOError, "RPM not opened" unless @file
 
           payload_io = payload
           payload_io.read
@@ -370,7 +370,7 @@ module Omnizip
         end
 
         def payload
-          raise "RPM not opened" unless @file
+          raise Omnizip::IOError, "RPM not opened" unless @file
 
           # Calculate payload offset
           offset = @lead.length
@@ -434,7 +434,7 @@ module Omnizip
 
           output
         rescue StandardError => e
-          raise "Failed to decompress with #{cmd}: #{e.message}"
+          raise Omnizip::DecompressionError, "Failed to decompress with #{cmd}: #{e.message}"
         end
 
         def extract_cpio(source, output_dir)

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

@@ -103,7 +103,7 @@ module Omnizip
           end
 
           entry = @entries.find { |e| e.name == entry_name }
-          raise "Entry not found: #{entry_name}" unless entry
+          raise Errno::ENOENT, "Entry not found: #{entry_name}" unless entry
 
           # Create directory if needed
           FileUtils.mkdir_p(File.dirname(output_path))

data/lib/omnizip/formats/seven_zip/split_archive_reader.rb

@@ -70,7 +70,7 @@ module Omnizip
         # @raise [RuntimeError] if entry not found or extraction fails
         def extract_entry(entry_name, output_path)
           entry = @entries.find { |e| e.name == entry_name }
-          raise "Entry not found: #{entry_name}" unless entry
+          raise Errno::ENOENT, "Entry not found: #{entry_name}" unless entry
 
           # Create directory if needed
           FileUtils.mkdir_p(File.dirname(output_path))

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

@@ -203,7 +203,7 @@ module Omnizip
           @file.seek(@header.header_size)
           compressed_toc = @file.read(@header.toc_compressed_size)
 
-          raise "Failed to read TOC" unless compressed_toc
+          raise Omnizip::InvalidArchiveError, "Failed to read TOC" unless compressed_toc
 
           # Parse TOC
           @toc = Toc.parse(compressed_toc, @header.toc_uncompressed_size)
@@ -272,7 +272,7 @@ module Omnizip
             inf.close
             result
           rescue Zlib::Error => e
-            raise "Failed to decompress data: #{e.message}"
+            raise Omnizip::DecompressionError, "Failed to decompress data: #{e.message}"
           end
         end
 

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

@@ -151,19 +151,94 @@ dereference_links: false)
           read_entry_data(entry)
         end
 
+        # Stream a single entry's decompressed content in +chunk_size+
+        # chunks, yielding each. STORE copies raw bytes and DEFLATE
+        # feeds an incremental inflater, so memory stays bounded by
+        # the chunk size; methods without an incremental decoder fall
+        # back to the whole-entry read (yielded as one chunk). The
+        # streamed path verifies the entry CRC after the last chunk.
+        #
+        # @param entry_name [String] Name of the entry to stream
+        # @param chunk_size [Integer] Read granularity in bytes
+        # @yield [chunk] Decompressed chunk
+        # @return [void]
+        def read_entry_stream(entry_name, chunk_size: 64 * 1024, &block)
+          ensure_read
+          entry = @entries.find { |e| e.filename == entry_name }
+          raise Errno::ENOENT, "Entry not found: #{entry_name}" unless entry
+          return if entry.directory? || entry.symlink?
+
+          crc = Omnizip::Checksums::Crc32.new
+          streamed =
+            File.open(file_path, "rb") do |io|
+              io.seek(entry_data_offset(io, entry), ::IO::SEEK_SET)
+
+              # Bound reads by the entry payload — what follows it in
+              # the file is the central directory, not entry data
+              remaining = entry.compressed_size
+
+              case entry.compression_method
+              when COMPRESSION_STORE
+                stream_raw(io, chunk_size, crc, remaining, &block)
+              when COMPRESSION_DEFLATE
+                stream_deflate(io, chunk_size, crc, remaining, &block)
+              else
+                yield(read_entry_data(entry))
+                false
+              end
+            end
+
+          return unless streamed
+
+          if crc.finalize != entry.crc32
+            raise Omnizip::ChecksumError,
+                  "CRC mismatch for #{entry.filename}"
+          end
+        end
+
         private
 
+        def stream_raw(io, chunk_size, crc, remaining)
+          while remaining.positive? &&
+              (chunk = io.read([chunk_size, remaining].min))
+            break if chunk.empty?
+
+            remaining -= chunk.bytesize
+            crc.update(chunk)
+            yield chunk
+          end
+          true
+        end
+
+        def stream_deflate(io, chunk_size, crc, remaining, &block)
+          require "zlib"
+          inflater = Zlib::Inflate.new(-Zlib::MAX_WBITS)
+          begin
+            while remaining.positive? &&
+                (chunk = io.read([chunk_size, remaining].min))
+              break if chunk.empty?
+
+              remaining -= chunk.bytesize
+              emit_inflated(inflater.inflate(chunk), crc, &block)
+            end
+            emit_inflated(inflater.finish, crc, &block)
+          ensure
+            inflater.close
+          end
+          true
+        end
+
+        def emit_inflated(out, crc)
+          return if out.nil? || out.empty?
+
+          crc.update(out)
+          yield out
+        end
+
         # Decompress and CRC-verify a single entry's data
         def read_entry_data(entry)
           File.open(file_path, "rb") do |io|
-            io.seek(entry.local_header_offset, ::IO::SEEK_SET)
-
-            fixed_header = io.read(30)
-            _signature, _version, _flags, _method, _time, _date, _crc32,
-            _comp_size, _uncomp_size, filename_length, extra_length = fixed_header.unpack("VvvvvvVVVvv")
-
-            variable_data = io.read(filename_length + extra_length)
-            LocalFileHeader.from_binary(fixed_header + variable_data)
+            io.seek(entry_data_offset(io, entry), ::IO::SEEK_SET)
 
             compressed_data = io.read(entry.compressed_size)
             decompressed_data = decompress_data(
@@ -184,6 +259,19 @@ dereference_links: false)
           end
         end
 
+        # File offset of an entry's compressed payload: the local
+        # header's fixed 30 bytes plus its filename/extra fields.
+        def entry_data_offset(io, entry)
+          io.seek(entry.local_header_offset, ::IO::SEEK_SET)
+
+          fixed_header = io.read(30)
+          _signature, _version, _flags, _method, _time, _date, _crc32,
+          _comp_size, _uncomp_size, filename_length, extra_length = fixed_header.unpack("VvvvvvVVVvv")
+
+          io.read(filename_length + extra_length)
+          io.pos
+        end
+
         # Parse the central directory once, on demand
         def ensure_read
           read if @entries.empty? && @central_directory.empty?