[GR-58266] Prepare strscan.rb to be upstreamed

PullRequest: truffleruby/4485

Author: eregon

lib/mri/monitor.rb

@@ -214,7 +214,7 @@ def mon_owned?
   def mon_synchronize(&block)
     Primitive.monitor_synchronize(@mon_mutex, block)
   end
-  Truffle::Graal.always_split instance_method(:mon_synchronize)
+  Primitive.always_split self, :mon_synchronize
   alias synchronize mon_synchronize
 
   #

lib/truffle/strscan.rb

@@ -99,10 +99,12 @@ def charpos
   def check(pattern)
     scan_internal pattern, false, true, true
   end
+  Primitive.always_split self, :check
 
   def check_until(pattern)
     scan_internal pattern, false, true, false
   end
+  Primitive.always_split self, :check_until
 
   def clear
     warn 'StringScanner#clear is obsolete; use #terminate instead' if $VERBOSE
@@ -128,6 +130,7 @@ def eos?
   def exist?(pattern)
     scan_internal pattern, false, false, false
   end
+  Primitive.always_split self, :exist?
 
   def fixed_anchor?
     @fixed_anchor
@@ -192,6 +195,7 @@ def inspect
   def match?(pattern)
     scan_internal pattern, false, false, true
   end
+  Primitive.always_split self, :match?
 
   def matched
     @match&.to_s
@@ -247,6 +251,7 @@ def restsize
   def scan(pattern)
     scan_internal pattern, true, true, true
   end
+  Primitive.always_split self, :scan
 
   def scan_byte
     if eos?
@@ -277,21 +282,24 @@ def scan_integer(base: 10)
     end
 
     if substr
-      Primitive.string_to_inum(substr, base, true, true)
+      substr.to_i(base)
     end
   end
 
   def scan_until(pattern)
     scan_internal pattern, true, true, false
   end
+  Primitive.always_split self, :scan_until
 
   def scan_full(pattern, advance_pos, getstr)
     scan_internal pattern, advance_pos, getstr, true
   end
+  Primitive.always_split self, :scan_full
 
   def search_full(pattern, advance_pos, getstr)
     scan_internal pattern, advance_pos, getstr, false
   end
+  Primitive.always_split self, :search_full
 
   def self.must_C_version
     self
@@ -304,10 +312,12 @@ def size
   def skip(pattern)
     scan_internal pattern, true, false, true
   end
+  Primitive.always_split self, :skip
 
   def skip_until(pattern)
     scan_internal pattern, true, false, false
   end
+  Primitive.always_split self, :skip_until
 
   def string
     @original
@@ -332,7 +342,7 @@ def terminate
   end
 
   def unscan
-    raise ScanError if Primitive.nil?(@match)
+    raise ScanError unless @match
     @pos = @prev_pos
     @prev_pos = nil
     @match = nil
@@ -358,7 +368,7 @@ def peep(len)
     peek len
   end
 
-  private def scan_check_args(pattern, headonly)
+  private def scan_check_args(pattern)
     unless Primitive.is_a?(pattern, Regexp) || Primitive.is_a?(pattern, String)
       raise TypeError, "bad pattern argument: #{pattern.inspect}"
     end
@@ -369,15 +379,18 @@ def peep(len)
   # This method is kept very small so that it should fit within 100
   # AST nodes and can be split. This is done to avoid indirect calls
   # to TRegex.
-  private def scan_internal(pattern, advance_pos, getstr, headonly)
-    scan_check_args(pattern, headonly)
+  private def scan_internal(pattern, advance_pos, getstr, only_match_at_start)
+    scan_check_args(pattern)
 
     if Primitive.is_a?(pattern, String)
-      md = scan_internal_string_pattern(pattern, headonly)
+      md = scan_internal_string_pattern(pattern, only_match_at_start)
     else
       start = @fixed_anchor ? 0 : @pos
-      md = Truffle::RegexpOperations.match_in_region pattern, @string, @pos, @string.bytesize, headonly, start
-      Primitive.matchdata_fixup_positions(md, start) if md
+      if only_match_at_start
+        md = Primitive.regexp_match_at_start(pattern, @string, @pos, start)
+      else
+        md = Primitive.regexp_search_with_start(pattern, @string, @pos, start)
+      end
     end
 
     if md
@@ -387,11 +400,12 @@ def peep(len)
       @match = nil
     end
   end
+  Primitive.always_split self, :scan_internal
 
-  private def scan_internal_string_pattern(pattern, headonly)
+  private def scan_internal_string_pattern(pattern, only_match_at_start)
     pos = @pos
 
-    if headonly
+    if only_match_at_start
       if @string.byteslice(pos..).start_with?(pattern)
         Primitive.matchdata_create_single_group(pattern, @string.dup, pos, pos + pattern.bytesize)
       else

lib/truffle/truffle/cext.rb

@@ -1193,7 +1193,7 @@ def rb_hash_foreach(hash, func, farg)
       end
     end
   end
-  Truffle::Graal.always_split instance_method(:rb_hash_foreach)
+  Primitive.always_split self, :rb_hash_foreach
 
   def rb_path_to_class(path)
     begin
@@ -1264,7 +1264,7 @@ def rb_protect(function, arg, write_status, status)
     Truffle::Interop.execute_without_conversion(write_status, status, pos)
     res
   end
-  Truffle::Graal.always_split instance_method(:rb_protect)
+  Primitive.always_split self, :rb_protect
 
   def rb_jump_tag(pos)
     if pos > 0
@@ -1279,12 +1279,12 @@ def rb_jump_tag(pos)
   def rb_yield(value)
     rb_block_proc.call(value)
   end
-  Truffle::Graal.always_split instance_method(:rb_yield)
+  Primitive.always_split self, :rb_yield
 
   def rb_yield_splat(values)
     rb_block_proc.call(*values)
   end
-  Truffle::Graal.always_split instance_method(:rb_yield_splat)
+  Primitive.always_split self, :rb_yield_splat
 
   def rb_ivar_lookup(object, name, default_value)
     # TODO CS 24-Jul-16 races - needs a new primitive or be defined in Java?
@@ -1649,7 +1649,7 @@ def rb_mutex_synchronize(mutex, func, arg)
       Primitive.interop_execute(POINTER_TO_POINTER_WRAPPER, [func, arg])
     end
   end
-  Truffle::Graal.always_split instance_method(:rb_mutex_synchronize)
+  Primitive.always_split self, :rb_mutex_synchronize
 
   def rb_gc_enable
     GC.enable
@@ -1879,7 +1879,7 @@ def rb_ensure(b_proc, data1, e_proc, data2)
       end
     end
   end
-  Truffle::Graal.always_split instance_method(:rb_ensure)
+  Primitive.always_split self, :rb_ensure
 
   def rb_rescue(b_proc, data1, r_proc, data2)
     begin
@@ -1898,7 +1898,7 @@ def rb_rescue(b_proc, data1, r_proc, data2)
       end
     end
   end
-  Truffle::Graal.always_split instance_method(:rb_rescue)
+  Primitive.always_split self, :rb_rescue
 
   def rb_rescue2(b_proc, data1, r_proc, data2, rescued)
     begin
@@ -1913,7 +1913,7 @@ def rb_rescue2(b_proc, data1, r_proc, data2, rescued)
       end
     end
   end
-  Truffle::Graal.always_split instance_method(:rb_rescue2)
+  Primitive.always_split self, :rb_rescue2
 
   def rb_exec_recursive(func, obj, arg)
     result = nil
@@ -1932,7 +1932,7 @@ def rb_exec_recursive(func, obj, arg)
       result
     end
   end
-  Truffle::Graal.always_split instance_method(:rb_exec_recursive)
+  Primitive.always_split self, :rb_exec_recursive
 
   def rb_catch_obj(tag, func, data)
     use_cext_lock = Primitive.use_cext_lock?

spec/truffle/array/steal_array_storage.rb renamed to spec/truffle/array/steal_array_storage_spec.rb

@@ -16,31 +16,25 @@ def storage(ary)
     Truffle::Debug.array_storage(ary)
   end
 
-  before :each do
-    @array = %i[first second third]
-  end
-
   guard -> { !Truffle::Boot.get_option('chaos-data') } do
     it "should no-op when called on itself" do
-      copy = @array.dup
+      array = %i[first second third]
+      Primitive.steal_array_storage(array, array)
 
-      Primitive.steal_array_storage(@array, @array)
-
-      storage(@array).should == "Object[]"
-      @array.should == copy
+      storage(array).should == "Object[]"
+      array.should == %i[first second third]
     end
 
     it "should take ownership of the store" do
+      array = %i[first second third]
       other = [1, 2, 3, 4, 5]
-      other_copy = other.dup
-
-      Primitive.steal_array_storage(@array, other)
+      Primitive.steal_array_storage(array, other)
 
-      storage(@array).should == "int[]"
-      @array.should == other_copy
+      storage(array).should == "int[]"
+      array.should == [1, 2, 3, 4, 5]
 
-      storage(other).should == "null"
-      other.empty?.should == true
+      storage(other).should == "empty"
+      other.should.empty?
     end
   end
 end

spec/truffle/splitting_spec.rb

@@ -35,8 +35,6 @@
   Regexp.instance_method(:match?),
   Truffle::RegexpOperations.method(:match),
   Truffle::RegexpOperations.method(:match?),
-  Truffle::RegexpOperations.method(:search_region),
-  Truffle::RegexpOperations.method(:match_in_region),
 
   String.instance_method(:[]),
   Truffle::StringOperations.method(:subpattern),

spec/truffleruby.mspec

@@ -8,6 +8,8 @@ load "#{__dir__}/ruby/default.mspec"
 # Don't run ruby/spec as root on TruffleRuby
 raise 'ruby/spec is not designed to be run as root on TruffleRuby' if Process.uid == 0
 
+ENV['TRUFFLERUBY_ALLOW_PRIVATE_PRIMITIVES_IN'] = "#{__dir__}/truffle/"
+
 class MSpecScript
   def self.windows?
     ENV.key?('WINDIR') || ENV.key?('windir')

src/annotations/java/org/truffleruby/annotations/Primitive.java

@@ -33,4 +33,7 @@
 
     /** Use these names in Ruby core methods stubs, ignore argument names in Java specializations. */
     String[] argumentNames() default {};
+
+    /** Whether this Primitive can be used outside the TruffleRuby repository. */
+    boolean isPublic() default false;
 }

src/main/java/org/truffleruby/RubyLanguage.java

@@ -208,6 +208,8 @@ public abstract static class RubySourceOptions {
      * {@link TranslatorEnvironment#newFrameDescriptorBuilderForBlock(BlockDescriptorInfo)}. */
     public static final FrameDescriptor EMPTY_FRAME_DESCRIPTOR = new FrameDescriptor(nil);
 
+    public static final String INTERNAL_CORE_PREFIX = "<internal:core> ";
+
     /** Global cache of call targets that {@code RBSprintfCompiler.compile} returns */
     public static final Map<TStringWithEncoding, RootCallTarget> sprintfCompilerCallTargets = new ConcurrentHashMap<>();
 
@@ -284,6 +286,7 @@ private RubyThread getOrCreateForeignThread(RubyContext context, Thread thread)
     @CompilationFinal public LanguageOptions options;
     @CompilationFinal private String rubyHome;
     @CompilationFinal public String cextPath;
+    public String[] allowPrivatePrimitivesPrefixes;
 
     private TruffleFile rubyHomeTruffleFile;
 
@@ -777,13 +780,36 @@ private void setRubyHome(TruffleFile home) {
         rubyHomeTruffleFile = home;
         rubyHome = home.getPath();
         cextPath = rubyHome + "/lib/truffle/truffle/cext_ruby.rb";
+
+        String allowIn = System.getenv("TRUFFLERUBY_ALLOW_PRIVATE_PRIMITIVES_IN");
+        if (allowIn != null) {
+            if (!allowIn.endsWith("/spec/truffle/") && !allowIn.endsWith("/test/truffle/compiler/") &&
+                    !allowIn.endsWith("/bench/metrics/")) {
+                throw CompilerDirectives
+                        .shouldNotReachHere("Invalid value for TRUFFLERUBY_ALLOW_PRIMITIVES_IN: " + allowIn);
+            }
+
+            allowPrivatePrimitivesPrefixes = new String[]{
+                    rubyHome + "/lib/truffle/",
+                    rubyHome + "/lib/patches/",
+                    rubyHome + "/lib/mri/",
+                    allowIn,
+            };
+        } else {
+            allowPrivatePrimitivesPrefixes = new String[]{
+                    rubyHome + "/lib/truffle/",
+                    rubyHome + "/lib/patches/",
+                    rubyHome + "/lib/mri/",
+            };
+        }
     }
 
     private void resetRubyHome() {
         assert Thread.holdsLock(this);
         rubyHomeTruffleFile = null;
         rubyHome = null;
         cextPath = null;
+        allowPrivatePrimitivesPrefixes = null;
     }
 
     private TruffleFile findRubyHome(Env env) {
@@ -954,7 +980,7 @@ public static String getPath(Source source) {
     public String getSourcePath(Source source) {
         final String path = getPath(source);
         if (path.startsWith(coreLoadPath)) {
-            return "<internal:core> " + path.substring(coreLoadPath.length() + 1);
+            return INTERNAL_CORE_PREFIX + path.substring(coreLoadPath.length() + 1);
         } else {
             return path;
         }

src/main/java/org/truffleruby/builtins/PrimitiveManager.java

@@ -43,7 +43,7 @@ public PrimitiveNodeConstructor getPrimitive(String name) {
             }
         }
 
-        throw new Error("Primitive :" + name + " not found");
+        throw new Error("Primitive." + name + " not found");
     }
 
     public void addLazyPrimitive(String primitive, String nodeFactoryClass) {

src/main/java/org/truffleruby/builtins/PrimitiveNodeConstructor.java

@@ -28,6 +28,10 @@ public PrimitiveNodeConstructor(Primitive annotation, NodeFactory<? extends Ruby
         this.factory = factory;
     }
 
+    public boolean isPublic() {
+        return annotation.isPublic();
+    }
+
     public int getPrimitiveArity() {
         return factory.getExecutionSignature().size();
     }