ruby-lzws 1.0.0 → 1.1.4

data/ext/lzws_ext/stream/decompressor.h

@@ -5,22 +5,25 @@
 #define LZWS_EXT_STREAM_DECOMPRESSOR_H
 
 #include <lzws/decompressor/state.h>
+#include <stdlib.h>
 
+#include "lzws_ext/common.h"
 #include "ruby.h"
 
 typedef struct {
   lzws_decompressor_state_t* state_ptr;
-  uint8_t*                   destination_buffer;
+  lzws_ext_byte_t*           destination_buffer;
   size_t                     destination_buffer_length;
-  uint8_t*                   remaining_destination_buffer;
+  lzws_ext_byte_t*           remaining_destination_buffer;
   size_t                     remaining_destination_buffer_length;
 } lzws_ext_decompressor_t;
 
 VALUE lzws_ext_allocate_decompressor(VALUE klass);
 VALUE lzws_ext_initialize_decompressor(VALUE self, VALUE options);
-VALUE lzws_ext_decompressor_read_magic_header(VALUE self, VALUE source);
 VALUE lzws_ext_decompress(VALUE self, VALUE source);
 VALUE lzws_ext_decompressor_read_result(VALUE self);
 VALUE lzws_ext_decompressor_close(VALUE self);
 
+void lzws_ext_decompressor_exports(VALUE root_module);
+
 #endif // LZWS_EXT_STREAM_DECOMPRESSOR_H

data/ext/lzws_ext/string.c

@@ -1,78 +1,280 @@
 // Ruby bindings for lzws library.
 // Copyright (c) 2019 AUTHORS, MIT License.
 
-#include <lzws/string.h>
+#include "lzws_ext/string.h"
 
-#include "ruby.h"
+#include <lzws/buffer.h>
+#include <lzws/common.h>
+#include <lzws/compressor/common.h>
+#include <lzws/compressor/main.h>
+#include <lzws/compressor/state.h>
+#include <lzws/decompressor/common.h>
+#include <lzws/decompressor/main.h>
+#include <lzws/decompressor/state.h>
+#include <stdlib.h>
 
+#include "lzws_ext/buffer.h"
 #include "lzws_ext/error.h"
 #include "lzws_ext/macro.h"
 #include "lzws_ext/option.h"
-#include "lzws_ext/string.h"
+#include "ruby.h"
+
+// -- buffer --
+
+static inline lzws_ext_result_t increase_destination_buffer(
+  VALUE destination_value, size_t destination_length,
+  size_t* remaining_destination_buffer_length_ptr, size_t destination_buffer_length)
+{
+  if (*remaining_destination_buffer_length_ptr == destination_buffer_length) {
+    // We want to write more data at once, than buffer has.
+    return LZWS_EXT_ERROR_NOT_ENOUGH_DESTINATION_BUFFER;
+  }
+
+  int exception;
+
+  LZWS_EXT_RESIZE_STRING_BUFFER(destination_value, destination_length + destination_buffer_length, exception);
+  if (exception != 0) {
+    return LZWS_EXT_ERROR_ALLOCATE_FAILED;
+  }
+
+  *remaining_destination_buffer_length_ptr = destination_buffer_length;
+
+  return 0;
+}
+
+// -- utils --
+
+#define GET_SOURCE_DATA(source_value)                    \
+  Check_Type(source_value, T_STRING);                    \
+                                                         \
+  const char* source        = RSTRING_PTR(source_value); \
+  size_t      source_length = RSTRING_LEN(source_value);
 
-#define GET_STRING(source)                         \
-  Check_Type(source, T_STRING);                    \
-                                                   \
-  const char* source_data   = RSTRING_PTR(source); \
-  size_t      source_length = RSTRING_LEN(source);
+// -- compress --
 
-VALUE lzws_ext_compress_string(VALUE LZWS_EXT_UNUSED(self), VALUE source, VALUE options)
+#define BUFFERED_COMPRESS(function, ...)                                                                                               \
+  while (true) {                                                                                                                       \
+    lzws_ext_byte_t* remaining_destination_buffer             = (lzws_ext_byte_t*)RSTRING_PTR(destination_value) + destination_length; \
+    size_t           prev_remaining_destination_buffer_length = remaining_destination_buffer_length;                                   \
+                                                                                                                                       \
+    result = function(__VA_ARGS__, &remaining_destination_buffer, &remaining_destination_buffer_length);                               \
+                                                                                                                                       \
+    if (                                                                                                                               \
+      result != 0 &&                                                                                                                   \
+      result != LZWS_COMPRESSOR_NEEDS_MORE_DESTINATION) {                                                                              \
+      return LZWS_EXT_ERROR_UNEXPECTED;                                                                                                \
+    }                                                                                                                                  \
+                                                                                                                                       \
+    destination_length += prev_remaining_destination_buffer_length - remaining_destination_buffer_length;                              \
+                                                                                                                                       \
+    if (result == LZWS_COMPRESSOR_NEEDS_MORE_DESTINATION) {                                                                            \
+      ext_result = increase_destination_buffer(                                                                                        \
+        destination_value, destination_length,                                                                                         \
+        &remaining_destination_buffer_length, destination_buffer_length);                                                              \
+                                                                                                                                       \
+      if (ext_result != 0) {                                                                                                           \
+        return ext_result;                                                                                                             \
+      }                                                                                                                                \
+                                                                                                                                       \
+      continue;                                                                                                                        \
+    }                                                                                                                                  \
+                                                                                                                                       \
+    break;                                                                                                                             \
+  }
+
+static inline lzws_ext_result_t compress(
+  lzws_compressor_state_t* state_ptr,
+  const char* source, size_t source_length,
+  VALUE destination_value, size_t destination_buffer_length)
+{
+  lzws_result_t     result;
+  lzws_ext_result_t ext_result;
+
+  lzws_ext_byte_t* remaining_source        = (lzws_ext_byte_t*)source;
+  size_t           remaining_source_length = source_length;
+
+  size_t destination_length                  = 0;
+  size_t remaining_destination_buffer_length = destination_buffer_length;
+
+  BUFFERED_COMPRESS(lzws_compress, state_ptr, &remaining_source, &remaining_source_length);
+  BUFFERED_COMPRESS(lzws_compressor_finish, state_ptr);
+
+  int exception;
+
+  LZWS_EXT_RESIZE_STRING_BUFFER(destination_value, destination_length, exception);
+  if (exception != 0) {
+    return LZWS_EXT_ERROR_ALLOCATE_FAILED;
+  }
+
+  return 0;
+}
+
+VALUE lzws_ext_compress_string(VALUE LZWS_EXT_UNUSED(self), VALUE source_value, VALUE options)
 {
-  GET_STRING(source);
+  GET_SOURCE_DATA(source_value);
+  Check_Type(options, T_HASH);
   LZWS_EXT_GET_COMPRESSOR_OPTIONS(options);
+  LZWS_EXT_GET_BUFFER_LENGTH_OPTION(options, destination_buffer_length);
 
-  char*  destination;
-  size_t destination_length;
+  lzws_compressor_state_t* state_ptr;
 
-  lzws_result_t result = lzws_compress_string(
-    (uint8_t*)source_data, source_length,
-    (uint8_t**)&destination, &destination_length, buffer_length,
+  lzws_result_t result = lzws_compressor_get_initial_state(
+    &state_ptr,
     without_magic_header, max_code_bit_length, block_mode, msb, unaligned_bit_groups, quiet);
 
-  if (result == LZWS_STRING_ALLOCATE_FAILED) {
-    lzws_ext_raise_error("AllocateError", "allocate error");
+  if (result != 0) {
+    switch (result) {
+      case LZWS_COMPRESSOR_ALLOCATE_FAILED:
+        lzws_ext_raise_error(LZWS_EXT_ERROR_ALLOCATE_FAILED);
+      case LZWS_COMPRESSOR_INVALID_MAX_CODE_BIT_LENGTH:
+        lzws_ext_raise_error(LZWS_EXT_ERROR_VALIDATE_FAILED);
+      default:
+        lzws_ext_raise_error(LZWS_EXT_ERROR_UNEXPECTED);
+    }
   }
-  else if (result == LZWS_STRING_VALIDATE_FAILED) {
-    lzws_ext_raise_error("ValidateError", "validate error");
+
+  if (destination_buffer_length == 0) {
+    destination_buffer_length = LZWS_DEFAULT_DESTINATION_BUFFER_LENGTH_FOR_COMPRESSOR;
   }
-  else if (result != 0) {
-    lzws_ext_raise_error("UnexpectedError", "unexpected error");
+
+  int exception;
+
+  LZWS_EXT_CREATE_STRING_BUFFER(destination_value, destination_buffer_length, exception);
+  if (exception != 0) {
+    lzws_compressor_free_state(state_ptr);
+    lzws_ext_raise_error(LZWS_EXT_ERROR_ALLOCATE_FAILED);
   }
 
-  // Ruby copies string on initialization.
-  VALUE result_string = rb_str_new(destination, destination_length);
-  free(destination);
-  return result_string;
+  lzws_ext_result_t ext_result = compress(
+    state_ptr,
+    source, source_length,
+    destination_value, destination_buffer_length);
+
+  lzws_compressor_free_state(state_ptr);
+
+  if (ext_result != 0) {
+    lzws_ext_raise_error(ext_result);
+  }
+
+  return destination_value;
 }
 
-VALUE lzws_ext_decompress_string(VALUE LZWS_EXT_UNUSED(self), VALUE source, VALUE options)
+// -- decompress --
+
+static inline lzws_ext_result_t decompress(
+  lzws_decompressor_state_t* state_ptr,
+  const char* source, size_t source_length,
+  VALUE destination_value, size_t destination_buffer_length)
 {
-  GET_STRING(source);
+  lzws_result_t     result;
+  lzws_ext_result_t ext_result;
+
+  lzws_ext_byte_t* remaining_source        = (lzws_ext_byte_t*)source;
+  size_t           remaining_source_length = source_length;
+
+  size_t destination_length                  = 0;
+  size_t remaining_destination_buffer_length = destination_buffer_length;
+
+  while (true) {
+    lzws_ext_byte_t* remaining_destination_buffer             = (lzws_ext_byte_t*)RSTRING_PTR(destination_value) + destination_length;
+    size_t           prev_remaining_destination_buffer_length = remaining_destination_buffer_length;
+
+    result = lzws_decompress(
+      state_ptr,
+      &remaining_source, &remaining_source_length,
+      &remaining_destination_buffer, &remaining_destination_buffer_length);
+
+    if (
+      result != 0 &&
+      result != LZWS_DECOMPRESSOR_NEEDS_MORE_DESTINATION) {
+      switch (result) {
+        case LZWS_DECOMPRESSOR_INVALID_MAGIC_HEADER:
+        case LZWS_DECOMPRESSOR_INVALID_MAX_CODE_BIT_LENGTH:
+          return LZWS_EXT_ERROR_VALIDATE_FAILED;
+        case LZWS_DECOMPRESSOR_CORRUPTED_SOURCE:
+          return LZWS_EXT_ERROR_DECOMPRESSOR_CORRUPTED_SOURCE;
+        default:
+          return LZWS_EXT_ERROR_UNEXPECTED;
+      }
+    }
+
+    destination_length += prev_remaining_destination_buffer_length - remaining_destination_buffer_length;
+
+    if (result == LZWS_DECOMPRESSOR_NEEDS_MORE_DESTINATION) {
+      ext_result = increase_destination_buffer(
+        destination_value, destination_length,
+        &remaining_destination_buffer_length, destination_buffer_length);
+
+      if (ext_result != 0) {
+        return ext_result;
+      }
+
+      continue;
+    }
+
+    break;
+  }
+
+  int exception;
+
+  LZWS_EXT_RESIZE_STRING_BUFFER(destination_value, destination_length, exception);
+  if (exception != 0) {
+    return LZWS_EXT_ERROR_ALLOCATE_FAILED;
+  }
+
+  return 0;
+}
+
+VALUE lzws_ext_decompress_string(VALUE LZWS_EXT_UNUSED(self), VALUE source_value, VALUE options)
+{
+  GET_SOURCE_DATA(source_value);
+  Check_Type(options, T_HASH);
   LZWS_EXT_GET_DECOMPRESSOR_OPTIONS(options);
+  LZWS_EXT_GET_BUFFER_LENGTH_OPTION(options, destination_buffer_length);
 
-  char*  destination;
-  size_t destination_length;
+  lzws_decompressor_state_t* state_ptr;
 
-  lzws_result_t result = lzws_decompress_string(
-    (uint8_t*)source_data, source_length,
-    (uint8_t**)&destination, &destination_length, buffer_length,
+  lzws_result_t result = lzws_decompressor_get_initial_state(
+    &state_ptr,
     without_magic_header, msb, unaligned_bit_groups, quiet);
 
-  if (result == LZWS_STRING_ALLOCATE_FAILED) {
-    lzws_ext_raise_error("AllocateError", "allocate error");
+  if (result != 0) {
+    switch (result) {
+      case LZWS_DECOMPRESSOR_ALLOCATE_FAILED:
+        lzws_ext_raise_error(LZWS_EXT_ERROR_ALLOCATE_FAILED);
+      default:
+        lzws_ext_raise_error(LZWS_EXT_ERROR_UNEXPECTED);
+    }
   }
-  else if (result == LZWS_STRING_VALIDATE_FAILED) {
-    lzws_ext_raise_error("ValidateError", "validate error");
+
+  if (destination_buffer_length == 0) {
+    destination_buffer_length = LZWS_DEFAULT_DESTINATION_BUFFER_LENGTH_FOR_DECOMPRESSOR;
   }
-  else if (result == LZWS_STRING_DECOMPRESSOR_CORRUPTED_SOURCE) {
-    lzws_ext_raise_error("DecompressorCorruptedSourceError", "decompressor received corrupted source");
+
+  int exception;
+
+  LZWS_EXT_CREATE_STRING_BUFFER(destination_value, destination_buffer_length, exception);
+  if (exception != 0) {
+    lzws_decompressor_free_state(state_ptr);
+    lzws_ext_raise_error(LZWS_EXT_ERROR_ALLOCATE_FAILED);
   }
-  else if (result != 0) {
-    lzws_ext_raise_error("UnexpectedError", "unexpected error");
+
+  lzws_ext_result_t ext_result = decompress(
+    state_ptr,
+    source, source_length,
+    destination_value, destination_buffer_length);
+
+  lzws_decompressor_free_state(state_ptr);
+
+  if (ext_result != 0) {
+    lzws_ext_raise_error(ext_result);
   }
 
-  // Ruby copies string on initialization.
-  VALUE result_string = rb_str_new(destination, destination_length);
-  free(destination);
-  return result_string;
+  return destination_value;
+}
+
+void lzws_ext_string_exports(VALUE root_module)
+{
+  rb_define_module_function(root_module, "_native_compress_string", RUBY_METHOD_FUNC(lzws_ext_compress_string), 2);
+  rb_define_module_function(root_module, "_native_decompress_string", RUBY_METHOD_FUNC(lzws_ext_decompress_string), 2);
 }

data/ext/lzws_ext/string.h

@@ -9,4 +9,6 @@
 VALUE lzws_ext_compress_string(VALUE self, VALUE source, VALUE options);
 VALUE lzws_ext_decompress_string(VALUE self, VALUE source, VALUE options);
 
+void lzws_ext_string_exports(VALUE root_module);
+
 #endif // LZWS_EXT_STRING_H

data/lib/lzws.rb

@@ -1,4 +1,8 @@
-require_relative "lzws/file"
+# Ruby bindings for lzws library.
+# Copyright (c) 2019 AUTHORS, MIT License.
+
 require_relative "lzws/stream/reader"
 require_relative "lzws/stream/writer"
+require_relative "lzws/file"
 require_relative "lzws/string"
+require_relative "lzws/version"

data/lib/lzws/error.rb

@@ -4,17 +4,18 @@
 module LZWS
   class BaseError < ::StandardError; end
 
-  class ValidateError   < BaseError; end
-  class AllocateError   < BaseError; end
-  class UnexpectedError < BaseError; end
-
-  class NotEnoughDestinationError < BaseError; end
-  class UsedAfterCloseError       < BaseError; end
+  class AllocateError < BaseError; end
+  class ValidateError < BaseError; end
 
+  class UsedAfterCloseError              < BaseError; end
+  class NotEnoughSourceBufferError       < BaseError; end
+  class NotEnoughDestinationBufferError  < BaseError; end
+  class NotEnoughDestinationError        < BaseError; end
   class DecompressorCorruptedSourceError < BaseError; end
 
-  class OpenFileError < BaseError; end
   class AccessIOError < BaseError; end
   class ReadIOError   < BaseError; end
   class WriteIOError  < BaseError; end
+
+  class UnexpectedError < BaseError; end
 end

data/lib/lzws/file.rb

@@ -9,11 +9,13 @@ require_relative "validation"
 
 module LZWS
   module File
+    BUFFER_LENGTH_NAMES = %i[source_buffer_length destination_buffer_length].freeze
+
     def self.compress(source, destination, options = {})
       Validation.validate_string source
       Validation.validate_string destination
 
-      options = Option.get_compressor_options options
+      options = Option.get_compressor_options options, BUFFER_LENGTH_NAMES
 
       open_files(source, destination) do |source_io, destination_io|
         LZWS._native_compress_io source_io, destination_io, options
@@ -24,7 +26,7 @@ module LZWS
       Validation.validate_string source
       Validation.validate_string destination
 
-      options = Option.get_decompressor_options options
+      options = Option.get_decompressor_options options, BUFFER_LENGTH_NAMES
 
       open_files(source, destination) do |source_io, destination_io|
         LZWS._native_decompress_io source_io, destination_io, options
@@ -32,25 +34,11 @@ module LZWS
     end
 
     private_class_method def self.open_files(source, destination, &_block)
-      open_file(source, "rb") do |source_io|
-        open_file(destination, "wb") do |destination_io|
+      ::File.open source, "rb" do |source_io|
+        ::File.open destination, "wb" do |destination_io|
           yield source_io, destination_io
         end
       end
     end
-
-    private_class_method def self.open_file(path, mode, &_block)
-      begin
-        io = ::File.open path, mode
-      rescue ::StandardError
-        raise OpenFileError, "open file failed"
-      end
-
-      begin
-        yield io
-      ensure
-        io.close
-      end
-    end
   end
 end