ziperu 0.1.1-aarch64-linux

data/lib/ziperu/inflater.rb

@@ -0,0 +1,60 @@
+# frozen_string_literal: true
+
+# Ported verbatim from rubyzip (Copyright (C) 2002-2026 the rubyzip authors).
+# Not on any ziperu code path: the Rust core compresses and decompresses.
+# Kept so that code which references these classes keeps loading.
+module Ziperu
+  class Inflater < Decompressor # :nodoc:all
+    def initialize(*args)
+      super
+
+      @buffer = +''.b
+      @zlib_inflater = ::Zlib::Inflate.new(-Zlib::MAX_WBITS)
+    end
+
+    def read(maxlen = nil)
+      return (maxlen.nil? || maxlen.zero? ? '' : nil) if eof?
+
+      while maxlen.nil? || (@buffer.bytesize < maxlen)
+        break if input_finished?
+
+        @buffer << produce_input
+      end
+
+      @buffer.slice!(0...(maxlen || @buffer.bytesize))
+    end
+
+    def eof?
+      @buffer.empty? && input_finished?
+    end
+
+    # Alias for compatibility. Remove for version 4.
+    alias eof eof?
+
+    private
+
+    def produce_input
+      retried = 0
+      begin
+        @zlib_inflater.inflate(input_stream.read(Decompressor::CHUNK_SIZE))
+      rescue Zlib::BufError
+        raise if retried >= 5 # how many times should we retry?
+
+        retried += 1
+        retry
+      end
+    rescue Zlib::Error => e
+      raise ::Ziperu::DecompressionError, e
+    end
+
+    def input_finished?
+      @zlib_inflater.finished?
+    end
+  end
+
+  ::Ziperu::Decompressor.register(::Ziperu::COMPRESSION_METHOD_DEFLATE, ::Ziperu::Inflater)
+end
+
+# Copyright (C) 2002, 2003 Thomas Sondergaard
+# rubyzip is free software; you can redistribute it and/or
+# modify it under the terms of the ruby license.

data/lib/ziperu/input_stream.rb

@@ -0,0 +1,296 @@
+# frozen_string_literal: true
+
+module Ziperu
+  # Reads the entries of an archive one after another, exposing each as an
+  # IO-like stream, exactly as rubyzip's `Ziperu::InputStream` does:
+  #
+  #   Ziperu::InputStream.open("my.zip") do |io|
+  #     while (entry = io.get_next_entry)
+  #       puts "Contents of #{entry.name}: '#{io.read}'"
+  #     end
+  #   end
+  #
+  # Unlike rubyzip, entries are located through the central directory rather
+  # than by walking local headers, so archives whose local headers lack sizes
+  # (general-purpose bit 3, as written by OS X Archive Utility and most
+  # streaming writers) read fine instead of raising `Ziperu::StreamingError`.
+  class InputStream
+    CHUNK_SIZE = IOExtras::CHUNK_SIZE # :nodoc:
+
+    include ::Ziperu::IOExtras::AbstractInputStream
+
+    # Opens the indicated zip file.
+    #
+    # @param context [String||IO||StringIO] file path or IO/StringIO object
+    # @param offset [Integer] byte offset of the local header to start at
+    # This method also accepts the standard IO encoding options:
+    # `external_encoding:`, `internal_encoding:` and `encoding:`.
+    def initialize(context, offset: 0, decrypter: nil, enforce_size: ::Ziperu.validate_entry_sizes, **opts)
+      super(**opts)
+      @split = !context.kind_of?(Native::Archive) && self.class.split_archive?(context)
+      # An archive handed in (by Entry#get_input_stream) stays open for its
+      # owner; one opened here is closed with the stream. A split segment
+      # has no central directory to open; rubyzip fails on the first entry,
+      # so defer the error to `get_next_entry`.
+      @owns_archive = !context.kind_of?(Native::Archive)
+      @archive =
+        if !@owns_archive
+          context
+        elsif !@split
+          begin
+            self.class.open_archive(context)
+          rescue LimitError
+            raise
+          rescue Error => e
+            # Not an archive: rubyzip finds no entries rather than raising,
+            # and `while (entry = zis.get_next_entry)` over untrusted input
+            # relies on that.
+            @not_an_archive = e
+            nil
+          end
+        end
+      @decrypter = decrypter
+      @enforce_size = enforce_size
+      # An archive with several records under one name is walked from its
+      # listing, since the native core addresses only the last of each name
+      # by index; every record is yielded, as rubyzip's walk of the local
+      # headers would.
+      @records = @archive.entries if @archive&.duplicate_names?
+      @next_index = offset.zero? || @archive.nil? ? 0 : index_at_offset(offset)
+      @reader = nil
+      @current_entry = nil
+    end
+
+    # Close this InputStream. All further IO will raise an IOError.
+    def close
+      @reader&.close
+      @reader = nil
+      @archive&.close if @owns_archive
+      @archive = nil
+      @not_an_archive = nil
+    end
+
+    # Returns an Entry object and positions the stream at the beginning of
+    # the entry data. Returns nil when there are no more entries.
+    def get_next_entry
+      raise SplitArchiveError if @split
+      raise IOError, 'closed stream' if closed?
+      return @current_entry = nil if @archive.nil? || @next_index.nil?
+      if @next_index >= (@records ? @records.size : @archive.entry_count)
+        # Past the last entry there is nothing to read, not the previous
+        # entry again.
+        @reader&.close
+        @reader = nil
+        flush
+        return @current_entry = nil
+      end
+
+      meta = @records ? @records[@next_index] : @archive.entry_meta(@next_index)
+      @current_entry = Entry.from_meta(@archive, meta)
+      @next_index += 1
+      open_entry
+      @current_entry
+    end
+
+    # Position the stream at an entry that already exists as an object
+    # (used by Entry#get_input_stream to avoid rebuilding it).
+    #
+    # @api private
+    def get_next_stream_entry(entry)
+      raise IOError, 'closed stream' if closed?
+
+      @current_entry = entry
+      @next_index =
+        if @records
+          at = @records.index { |meta| meta[:header_offset] == entry.local_header_offset }
+          at && at + 1
+        else
+          entry.index + 1
+        end
+      open_entry
+      @current_entry
+    end
+
+    # Rewinds the stream to the beginning of the current entry.
+    def rewind
+      raise IOError, 'closed stream' if closed?
+      return if @current_entry.nil?
+
+      @lineno = 0
+      @pos    = 0
+      wrap_errors { @reader.rewind }
+      flush
+    end
+
+    # Modelled after IO#sysread. Reads up to maxlen bytes from the stream,
+    # bypassing the line buffer; returns a string (either a new string or the
+    # given out_string).
+    def sysread(maxlen = nil, out_string = nil)
+      return (maxlen.nil? || maxlen.zero? ? '' : nil) if eof?
+
+      output = produce_input(maxlen) || +''.b
+
+      if out_string.nil?
+        output.force_encoding(@internal_encoding || @external_encoding)
+      else
+        encoding = out_string.encoding
+        out_string.replace(output).force_encoding(encoding)
+      end
+    end
+
+    # Returns the size of the current entry, or `nil` if there isn't one.
+    def size
+      return if @current_entry.nil?
+
+      @current_entry.size
+    end
+
+    class << self
+      # Same as #initialize but if a block is passed the opened stream is
+      # passed to the block and closed when the block returns.
+      def open(filename_or_io, offset: 0, decrypter: nil, **opts)
+        zio = new(filename_or_io, offset: offset, decrypter: decrypter, **opts)
+        return zio unless block_given?
+
+        begin
+          yield zio
+        ensure
+          zio.close if zio
+        end
+      end
+
+      # @api private
+      def open_archive(context)
+        check_limits(open_native(context))
+      end
+
+      # Ziperu.max_entries / max_total_size, checked before any entry is
+      # read. An archive that fails is closed again.
+      def check_limits(native)
+        if (limit = ::Ziperu.max_entries) && native.entry_count > limit
+          raise LimitError.new('an entry count', native.entry_count, limit)
+        end
+        if (limit = ::Ziperu.max_total_size) && (total = native.total_size) > limit
+          raise LimitError.new('a declared total size', total, limit)
+        end
+
+        native
+      rescue LimitError
+        native.close
+        raise
+      end
+
+      def open_native(context)
+        return Native::Archive.open_path(context.to_s) unless context.respond_to?(:read)
+
+        # A StringIO's bytes are already in memory: snapshot them once rather
+        # than paying a method call per chunk (12x faster per entry).
+        return Native::Archive.open_bytes(context.string) if context.kind_of?(::StringIO)
+
+        context.binmode if context.respond_to?(:binmode)
+        if IOExtras.seekable?(context)
+          # Driven lazily through seek/read; nothing is copied up front.
+          Native::Archive.open_io(context)
+        else
+          # A pipe or socket: there is no way back to the central directory
+          # once the data has gone past, so take it all.
+          bytes = context.read.to_s.b
+          raise SplitArchiveError if split_signature?(bytes)
+
+          Native::Archive.open_bytes(bytes)
+        end
+      end
+
+      # The first part of a multi-part archive starts with a split-file
+      # signature.
+      #
+      # @api private
+      def split_archive?(context)
+        head =
+          if context.respond_to?(:read)
+            # A pipe cannot be peeked at; `open_archive` checks what it reads.
+            return false unless IOExtras.seekable?(context)
+
+            context.seek(0, ::IO::SEEK_SET)
+            context.read(4)
+          elsif ::File.file?(context.to_s)
+            ::File.binread(context.to_s, 4)
+          end
+        split_signature?(head)
+      end
+
+      def split_signature?(head)
+        !head.nil? && head.bytesize >= 4 && head.unpack1('V') == SPLIT_FILE_SIGNATURE
+      end
+    end
+
+    protected
+
+    def open_entry # :nodoc:
+      entry = @current_entry
+      unless READABLE_COMPRESSION_METHODS.include?(entry.compression_method)
+        raise CompressionMethodError, entry.compression_method
+      end
+
+      if entry.encrypted? && @decrypter.nil?
+        raise Error, PASSWORD_REQUIRED_MESSAGE
+      end
+
+      password = @decrypter&.password&.b
+      @reader = wrap_errors do
+        if entry.index
+          @archive.open_entry(entry.index, password, @enforce_size)
+        else
+          @archive.open_entry_at(entry.local_header_offset, entry.size, entry.compressed_size,
+                                 entry.crc, password, @enforce_size)
+        end
+      end
+      flush
+    end
+
+    def produce_input(maxlen = CHUNK_SIZE) # :nodoc:
+      raise IOError, 'closed stream' if closed?
+      raise ArgumentError, "negative length #{maxlen} given" if maxlen&.negative?
+      return nil if @reader.nil?
+
+      wrap_errors { @reader.read(maxlen) }
+    end
+
+    def input_finished? # :nodoc:
+      # A closed stream is not "at EOF": reading it raises, as with an IO.
+      return false if closed?
+
+      @reader.nil? || @reader.eof?
+    end
+
+    private
+
+    # Nil when no entry starts there, which reads as "no more entries".
+    def index_at_offset(offset)
+      return @records.index { |meta| meta[:header_offset] == offset } if @records
+
+      found = @archive.entries.find { |meta| meta[:header_offset] == offset }
+      found && found[:index]
+    end
+
+    # Closed by `close`, as opposed to never opened on an archive at all.
+    def closed?
+      @archive.nil? && @not_an_archive.nil?
+    end
+
+    # Re-raise the native core's errors as the rubyzip-shaped ones.
+    def wrap_errors
+      yield
+    rescue Native::SizeLimitError
+      raise EntrySizeError, @current_entry
+    rescue Native::PasswordError => e
+      # rubyzip verifies AES passwords up front but just inflates garbage for
+      # a wrong ZipCrypto one, so the exception classes differ.
+      raise Error, 'Bad password' if @decrypter.kind_of?(AESDecrypter)
+
+      raise DecompressionError, e
+    rescue Native::UnsupportedError => e
+      raise Error, e.message
+    end
+  end
+end

data/lib/ziperu/ioextras/abstract_input_stream.rb

@@ -0,0 +1,137 @@
+# frozen_string_literal: true
+
+# Ported from rubyzip (Copyright (C) 2002-2026 the rubyzip authors).
+
+require_relative 'fake_io'
+
+module Ziperu
+  module IOExtras # :nodoc:
+    # Implements many of the convenience methods of IO such as gets, getc,
+    # read, readline and readlines. Depends on: input_finished?,
+    # produce_input and read.
+    module AbstractInputStream
+      include Enumerable
+      include FakeIO
+
+      def initialize(**opts)
+        super
+        @lineno        = 0
+        @pos           = 0
+        @output_buffer = +''.b
+      end
+
+      attr_accessor :lineno
+      attr_reader :pos
+
+      def read(maxlen = nil, out_string = nil) # rubocop:disable Metrics/PerceivedComplexity,Metrics/CyclomaticComplexity
+        return (maxlen.nil? || maxlen.zero? ? '' : nil) if eof?
+
+        tbuf = if @output_buffer.bytesize > 0
+                 if maxlen && maxlen <= @output_buffer.bytesize
+                   @output_buffer.slice!(0, maxlen)
+                 else
+                   maxlen -= @output_buffer.bytesize if maxlen
+                   rbuf = produce_input(maxlen)
+                   out  = @output_buffer
+                   out << rbuf if rbuf
+                   @output_buffer = +''.b
+                   out
+                 end
+               else
+                 produce_input(maxlen)
+               end
+
+        if tbuf.nil? || tbuf.empty?
+          return nil if maxlen&.positive?
+
+          return ''
+        end
+
+        @pos += tbuf.length
+
+        if out_string.nil?
+          tbuf.force_encoding(@internal_encoding || @external_encoding)
+        else
+          encoding = out_string.encoding
+          out_string.replace(tbuf).force_encoding(encoding)
+        end
+      end
+
+      def readlines(sep = $INPUT_RECORD_SEPARATOR, limit = nil, chomp: false)
+        each(sep, limit, chomp: chomp).to_a
+      end
+
+      def gets(sep = $INPUT_RECORD_SEPARATOR, limit = nil, chomp: false) # rubocop:disable Metrics/AbcSize,Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
+        if sep.nil?
+          return nil if eof?
+
+          @lineno = @lineno.next
+          return read(limit)
+        end
+
+        if sep.respond_to?(:to_int)
+          limit = sep.to_int
+          sep   = $INPUT_RECORD_SEPARATOR
+        elsif sep&.empty?
+          sep = "#{$INPUT_RECORD_SEPARATOR}#{$INPUT_RECORD_SEPARATOR}"
+        end
+
+        encoding = @internal_encoding || @external_encoding
+
+        buffer_index = 0
+        while (sep_index = @output_buffer.index(sep, buffer_index)).nil?
+          break if limit && @output_buffer.bytesize >= limit
+
+          if input_finished?
+            return nil if @output_buffer.empty?
+
+            @lineno = @lineno.next
+            @pos += @output_buffer.bytesize
+            return @output_buffer.slice!(0..).force_encoding(encoding)
+          end
+
+          buffer_index = [buffer_index, @output_buffer.bytesize - sep.bytesize].max
+          @output_buffer << produce_input
+        end
+
+        limit ||= @output_buffer.bytesize
+        cut_index = sep_index ? [sep_index + sep.bytesize, limit].min : limit
+        @lineno = @lineno.next
+        @pos += cut_index
+        data = @output_buffer.slice!(0, cut_index)
+        data&.chomp!(sep) if chomp
+        data&.force_encoding(encoding)
+      end
+
+      def ungetc(byte) # :nodoc:
+        @output_buffer = byte.chr + @output_buffer
+      end
+
+      def flush # :nodoc:
+        @output_buffer.slice!(0..)
+      end
+
+      def readline(sep = $INPUT_RECORD_SEPARATOR, limit = nil, chomp: false)
+        raise EOFError if eof?
+
+        gets(sep, limit, chomp: chomp)
+      end
+
+      def each(sep = $INPUT_RECORD_SEPARATOR, limit = nil, chomp: false)
+        return to_enum(:each, sep, limit, chomp: chomp) unless block_given?
+
+        while (line = gets(sep, limit, chomp: chomp))
+          yield line
+        end
+      end
+
+      alias each_line each
+
+      def eof?
+        @output_buffer.empty? && input_finished?
+      end
+
+      alias eof eof? # :nodoc:
+    end
+  end
+end

data/lib/ziperu/ioextras/abstract_output_stream.rb

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+# Ported from rubyzip (Copyright (C) 2002-2026 the rubyzip authors).
+
+require_relative 'fake_io'
+
+module Ziperu
+  module IOExtras # :nodoc:
+    # Implements many of the output convenience methods of IO.
+    # relies on <<
+    module AbstractOutputStream # :nodoc:
+      include FakeIO
+
+      def write(data)
+        self << data
+        data.to_s.bytesize
+      end
+
+      def print(*params)
+        self << params.join
+
+        # Deflate doesn't like `nil`s or empty strings!
+        unless $OUTPUT_RECORD_SEPARATOR.nil? || $OUTPUT_RECORD_SEPARATOR.empty?
+          self << $OUTPUT_RECORD_SEPARATOR
+        end
+
+        nil
+      end
+
+      def printf(a_format_string, *params)
+        self << format(a_format_string, *params)
+        nil
+      end
+
+      def putc(an_object)
+        self << case an_object
+                when Integer
+                  an_object.chr
+                when String
+                  an_object
+                else
+                  raise TypeError, 'putc: Only Integer and String supported'
+                end
+        an_object
+      end
+
+      def puts(*params)
+        params << "\n" if params.empty?
+        params.flatten.each do |element|
+          val = element.to_s
+          self << val
+          self << "\n" unless val[-1, 1] == "\n"
+        end
+      end
+    end
+  end
+end

data/lib/ziperu/ioextras/fake_io.rb

@@ -0,0 +1,41 @@
+# frozen_string_literal: true
+
+# Ported from rubyzip (Copyright (C) 2002-2026 the rubyzip authors).
+
+module Ziperu
+  module IOExtras # :nodoc:
+    module FakeIO # :nodoc:
+      attr_reader :external_encoding, :internal_encoding
+
+      def initialize(**opts)
+        @internal_encoding = Encoding.find(opts[:internal_encoding]) if opts[:internal_encoding]
+
+        if opts[:encoding].kind_of?(String)
+          _ext, int = opts[:encoding].split(':', 2)
+          @internal_encoding = Encoding.find(int) unless @internal_encoding || int.nil?
+        end
+
+        @external_encoding = Encoding::ASCII_8BIT
+      end
+
+      def set_encoding(ext_enc, int_enc = nil)
+        if ext_enc.kind_of?(String) && ext_enc.include?(':')
+          _ext_enc, int_enc = ext_enc.split(':', 2)
+        end
+
+        begin
+          @internal_encoding = Encoding.find(int_enc) if int_enc
+        rescue ArgumentError
+          warn "warning: Unsupported encoding #{int_enc} ignored"
+        end
+
+        @external_encoding = Encoding::ASCII_8BIT
+      end
+
+      # Implement kind_of? in order to pretend to be an IO object.
+      def kind_of?(object)
+        object == IO || super
+      end
+    end
+  end
+end

data/lib/ziperu/ioextras.rb

@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+
+# For $INPUT_RECORD_SEPARATOR, the line-separator default in gets and friends.
+require 'English'
+
+require_relative 'ioextras/fake_io'
+require_relative 'ioextras/abstract_input_stream'
+require_relative 'ioextras/abstract_output_stream'
+
+module Ziperu
+  # IO-like conveniences shared by the stream classes. Ported from rubyzip so
+  # that `gets`, `readline`, `each_line` and friends behave identically.
+  module IOExtras # :nodoc:
+    CHUNK_SIZE = 32_768
+
+    class << self
+      # Whether `io` can be positioned: it answers `seek` and `tell` and a
+      # no-op seek succeeds (a pipe or socket raises).
+      def seekable?(io)
+        return false unless io.respond_to?(:seek) && io.respond_to?(:tell)
+
+        io.seek(0, ::IO::SEEK_CUR)
+        true
+      rescue SystemCallError, IOError
+        false
+      end
+
+      def copy_stream(ostream, istream)
+        ostream.write(istream.read(CHUNK_SIZE, +'')) until istream.eof?
+      end
+
+      def copy_stream_n(ostream, istream, nbytes)
+        toread = nbytes
+        while toread > 0 && !istream.eof?
+          tr = [toread, CHUNK_SIZE].min
+          ostream.write(istream.read(tr, +''))
+          toread -= tr
+        end
+      end
+    end
+  end
+end

data/lib/ziperu/null_compressor.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+# Ported verbatim from rubyzip (Copyright (C) 2002-2026 the rubyzip authors).
+# Not on any ziperu code path: the Rust core compresses and decompresses.
+# Kept so that code which references these classes keeps loading.
+module Ziperu
+  class NullCompressor < Compressor # :nodoc:all
+    include Singleton
+
+    def <<(_data)
+      raise IOError, 'closed stream'
+    end
+
+    attr_reader :size, :compressed_size
+  end
+end
+
+# Copyright (C) 2002, 2003 Thomas Sondergaard
+# rubyzip is free software; you can redistribute it and/or
+# modify it under the terms of the ruby license.

data/lib/ziperu/null_decompressor.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+# Ported verbatim from rubyzip (Copyright (C) 2002-2026 the rubyzip authors).
+# Not on any ziperu code path: the Rust core compresses and decompresses.
+# Kept so that code which references these classes keeps loading.
+module Ziperu
+  module NullDecompressor # :nodoc:all
+    module_function
+
+    def read(_length = nil, _outbuf = nil)
+      nil
+    end
+
+    def eof?
+      true
+    end
+
+    # Alias for compatibility. Remove for version 4.
+    alias eof eof?
+  end
+end
+
+# Copyright (C) 2002, 2003 Thomas Sondergaard
+# rubyzip is free software; you can redistribute it and/or
+# modify it under the terms of the ruby license.