[GR-19220] Support digest plugins (#3822)

PullRequest: truffleruby/4507

Author: andrykonchin

Diff for: CHANGELOG.md

@@ -15,6 +15,7 @@ Compatibility:
 * Implement `ObjectSpace::WeakKeyMap` (#3681, @andrykonchin).
 * Fix `String#slice` called with negative offset (@andrykonchin).
 * Fix explicitly inherited `Struct` subclasses and don't provide `#members` method (#3802, @andrykonchin).
+* Support Digest plugins (#1390, @nirvdrum).
 
 Performance:
 

Diff for: lib/cext/include/ruby/digest.h

@@ -0,0 +1,73 @@
+/************************************************
+
+  digest.h - header file for ruby digest modules
+
+  $Author$
+  created at: Fri May 25 08:54:56 JST 2001
+
+
+  Copyright (C) 2001-2006 Akinori MUSHA
+
+  $RoughId: digest.h,v 1.3 2001/07/13 15:38:27 knu Exp $
+  $Id$
+
+************************************************/
+
+#include "ruby.h"
+
+#define RUBY_DIGEST_API_VERSION	3
+
+typedef int (*rb_digest_hash_init_func_t)(void *);
+typedef void (*rb_digest_hash_update_func_t)(void *, unsigned char *, size_t);
+typedef int (*rb_digest_hash_finish_func_t)(void *, unsigned char *);
+
+typedef struct {
+    int api_version;
+    size_t digest_len;
+    size_t block_len;
+    size_t ctx_size;
+    rb_digest_hash_init_func_t init_func;
+    rb_digest_hash_update_func_t update_func;
+    rb_digest_hash_finish_func_t finish_func;
+} rb_digest_metadata_t;
+
+#define DEFINE_UPDATE_FUNC_FOR_UINT(name) \
+void \
+rb_digest_##name##_update(void *ctx, unsigned char *ptr, size_t size) \
+{ \
+    const unsigned int stride = 16384; \
+ \
+    for (; size > stride; size -= stride, ptr += stride) { \
+        name##_Update(ctx, ptr, stride); \
+    } \
+    /* Since size <= stride, size should fit into an unsigned int */ \
+    if (size > 0) name##_Update(ctx, ptr, (unsigned int)size); \
+}
+
+#define DEFINE_FINISH_FUNC_FROM_FINAL(name) \
+int \
+rb_digest_##name##_finish(void *ctx, unsigned char *ptr) \
+{ \
+    return name##_Final(ptr, ctx); \
+}
+
+static inline VALUE
+rb_digest_namespace(void)
+{
+    rb_require("digest");
+    return rb_path2class("Digest");
+}
+
+static inline ID
+rb_id_metadata(void)
+{
+    return rb_intern_const("metadata");
+}
+
+static inline VALUE
+rb_digest_make_metadata(const rb_digest_metadata_t *meta)
+{
+#undef RUBY_UNTYPED_DATA_WARNING
+#define RUBY_UNTYPED_DATA_WARNING 0
+    return rb_obj_freeze(Data_Wrap_Struct(0, 0, 0, (void *)meta));
+}

Diff for: lib/cext/include/truffleruby/truffleruby-abi-version.h

@@ -20,6 +20,6 @@
 // $RUBY_VERSION must be the same as TruffleRuby.LANGUAGE_VERSION.
 // $ABI_NUMBER starts at 1 and is incremented for every ABI-incompatible change.
 
-#define TRUFFLERUBY_ABI_VERSION "3.3.7.2"
+#define TRUFFLERUBY_ABI_VERSION "3.3.7.3"
 
 #endif

Diff for: lib/truffle/digest.rb

@@ -13,6 +13,24 @@
 # under LICENSE.RUBY as it is derived from lib/ruby/stdlib/digest.rb.
 
 require_relative 'digest/version'
+require_relative 'ffi'
+
+module Truffle
+  class Digest
+    module Foreign
+      class RbDigestMetadata < ::FFI::Struct
+        layout :api_version, :int,
+               :digest_len, :size_t,
+               :block_len, :size_t,
+               :ctx_size, :size_t,
+               :init_func, ::FFI::FunctionType.new(:int, [:pointer]),
+               :update_func, ::FFI::FunctionType.new(:void, [:pointer, :pointer, :size_t]),
+               :finish_func, ::FFI::FunctionType.new(:int, [:pointer, :pointer])
+      end
+    end
+  end
+end
+
 
 module Digest
 
@@ -170,6 +188,12 @@ def self.hexdigest(str, *args)
   end
 
   class Base < ::Digest::Class
+    def self.inherited(klass)
+      unless %w[Digest::MD5 Digest::SHA1 Digest::SHA2 Digest::SHA256 Digest::SHA384 Digest::SHA512].include?(klass.name)
+        klass.include(Digest::Plugin)
+      end
+    end
+
     def block_length
       Truffle::Digest.digest_block_length @digest
     end
@@ -186,9 +210,65 @@ def reset
     def update(str)
       str = StringValue(str)
       Truffle::Digest.update(@digest, str)
+
+      self
+    end
+    alias_method :<<, :update
+  end
+
+  module Plugin
+    attr_reader :metadata
+
+    def initialize
+      super
+      metadata_wrapped = Primitive.object_ivar_get(Primitive.class(self), :metadata)
+      metadata_pointer = Primitive.object_hidden_var_get(metadata_wrapped, Truffle::CExt::DATA_STRUCT).data
+      @metadata = Truffle::Digest::Foreign::RbDigestMetadata.new(FFI::Pointer.new(metadata_pointer))
+
+      reset
+    end
+
+    def initialize_copy(from)
+      @metadata = from.metadata
+      @context = from.context.clone
+
+      self
+    end
+
+    def context
+      @context ||= Truffle::FFI::MemoryPointer.new(:uchar, metadata[:ctx_size])
+    end
+
+    def block_length
+      metadata[:block_len]
+    end
+
+    def digest_length
+      metadata[:digest_len]
+    end
+
+    def update(str)
+      metadata[:update_func].call(context, Truffle::CExt.string_to_ffi_pointer_inplace(str), str.bytesize)
+
       self
     end
     alias_method :<<, :update
+
+    def finish
+      str = Truffle::FFI::MemoryPointer.new(:uchar, metadata[:digest_len], false)
+      metadata[:finish_func].call(context, str)
+
+      reset
+
+      Primitive.pointer_read_bytes str.address, metadata[:digest_len]
+    end
+
+    def reset
+      @context = nil
+      if metadata[:init_func].call(context) != 1
+        raise RuntimeError, 'Digest initialization failed'
+      end
+    end
   end
 
   class MD5 < Base

Diff for: lib/truffle/truffle/cext_ruby.rb

@@ -13,7 +13,7 @@ module Truffle::CExt
   # Methods defined in this file are not considered as Ruby code implementing MRI C parts,
   # see org.truffleruby.cext.CExtNodes.BlockProcNode
 
-  # methods defined with rb_define_method are normal Ruby methods therefore they cannot be defined in the cext.rb file
+  # methods defined with rb_define_method are normal Ruby methods therefore they cannot be defined in the cext.rb
   # file because blocks passed as arguments would be skipped by org.truffleruby.cext.CExtNodes.BlockProcNode
   def rb_define_method(mod, name, function, argc)
     if argc < -2 or 15 < argc

Diff for: spec/ruby/optional/capi/digest_spec.rb

@@ -0,0 +1,99 @@
+require_relative 'spec_helper'
+
+require 'fiddle'
+
+load_extension('digest')
+
+describe "C-API Digest functions" do
+  before :each do
+    @s = CApiDigestSpecs.new
+  end
+
+  describe "rb_digest_make_metadata" do
+    before :each do
+      @metadata = @s.rb_digest_make_metadata
+    end
+
+    it "should store the block length" do
+      @s.block_length(@metadata).should == 40
+    end
+
+    it "should store the digest length" do
+      @s.digest_length(@metadata).should == 20
+    end
+
+    it "should store the context size" do
+      @s.context_size(@metadata).should == 129
+    end
+  end
+
+  describe "digest plugin" do
+    before :each do
+      @s = CApiDigestSpecs.new
+      @digest = Digest::TestDigest.new
+
+      # A pointer to the CTX type defined in the extension for this spec. Digest does not make the context directly
+      # accessible as part of its API. However, to ensure we are properly loading the plugin, it's useful to have
+      # direct access to the context pointer to verify its contents.
+      @context = Fiddle::Pointer.new(@s.context(@digest))
+    end
+
+    it "should report the block length" do
+      @digest.block_length.should == 40
+    end
+
+    it "should report the digest length" do
+      @digest.digest_length.should == 20
+    end
+
+    it "should initialize the context" do
+      # Our test plugin always writes the string "Initialized\n" when its init function is called.
+      verify_context("Initialized\n")
+    end
+
+    it "should update the digest" do
+      @digest.update("hello world")
+
+      # Our test plugin always writes the string "Updated: <data>\n" when its update function is called.
+      current = "Initialized\nUpdated: hello world"
+      verify_context(current)
+
+      @digest << "blah"
+
+      current = "Initialized\nUpdated: hello worldUpdated: blah"
+      verify_context(current)
+    end
+
+    it "should finalize the digest" do
+      @digest.update("")
+
+      finish_string = @digest.instance_eval { finish }
+
+      # We expect the plugin to write out the last `@digest.digest_length` bytes, followed by the string "Finished\n".
+      #
+      finish_string.should == "d\nUpdated: Finished\n"
+      finish_string.encoding.should == Encoding::ASCII_8BIT
+    end
+
+    it "should reset the context" do
+      @digest.update("foo")
+      verify_context("Initialized\nUpdated: foo")
+
+      @digest.reset
+
+      # The context will be recreated as a result of the `reset` so we must fetch the latest context pointer.
+      @context = Fiddle::Pointer.new(@s.context(@digest))
+
+      verify_context("Initialized\n")
+    end
+
+    def verify_context(current_body)
+      # In the CTX type, the length of the current context contents is stored in the first byte.
+      byte_count = @context[0]
+      byte_count.should == current_body.bytesize
+
+      # After the size byte follows a string.
+      @context[1, byte_count].should == current_body
+    end
+  end
+end