[macruby-changes] [3622] MacRuby/branches/icu

25 Feb 2010

Revision: 3622
          http://trac.macosforge.org/projects/ruby/changeset/3622
Author:   lsansonetti@apple.com
Date:     2010-02-25 13:53:42 -0800 (Thu, 25 Feb 2010)
Log Message:
-----------
regexps are now AOT compilable

Modified Paths:
--------------
    MacRuby/branches/icu/compiler.cpp
    MacRuby/branches/icu/compiler.h
    MacRuby/branches/icu/encoding.h
    MacRuby/branches/icu/include/ruby/ruby.h
    MacRuby/branches/icu/re.cpp
    MacRuby/branches/icu/re.h
    MacRuby/branches/icu/string.c

Modified: MacRuby/branches/icu/compiler.cpp
===================================================================

--- MacRuby/branches/icu/compiler.cpp	2010-02-25 21:13:29 UTC (rev 3621)
+++ MacRuby/branches/icu/compiler.cpp	2010-02-25 21:53:42 UTC (rev 3622)
@@ -18,12 +18,13 @@
 #include "ruby/ruby.h"
 #include "ruby/encoding.h"
 #include "ruby/node.h"
-#include "ruby/re.h"
 #include "id.h"
 #include "vm.h"
 #include "compiler.h"
 #include "objc.h"
 #include "version.h"
+#include "encoding.h"
+#include "re.h"
 
 extern "C" const char *ruby_node_name(int node);
 
@@ -498,10 +499,12 @@
 
 GlobalVariable *
 RoxorCompiler::compile_const_global_ustring(const UniChar *str,
-	const size_t len, CFHashCode hash)
+	const size_t len)
 {
     assert(len > 0);
 
+    const unsigned long hash = rb_str_hash_uchars(str, len);
+
     std::map<CFHashCode, GlobalVariable *>::iterator iter =
 	static_ustrings.find(hash);
 
@@ -2797,22 +2800,18 @@
 	    return CallInst::Create(newString3Func, "", bb);
 	}
 	else {
-	    UniChar *buf = (UniChar *)CFStringGetCharactersPtr(
-		    (CFStringRef)val);
-	    bool free_buf = false;
-	    if (buf == NULL) {
-		buf = (UniChar *)malloc(sizeof(UniChar) * str_len);
-		CFStringGetCharacters((CFStringRef)val,
-			CFRangeMake(0, str_len), buf);
-		free_buf = true;
-	    }
+	    UChar *chars = NULL;
+	    long chars_len = 0;
+	    bool need_free = false;
 
-	    GlobalVariable *str_gvar = compile_const_global_ustring(buf,
-		    str_len, CFHash((CFTypeRef)val));
+	    rb_str_get_uchars(val, &chars, &chars_len, &need_free);
 
-	    if (free_buf) {
-		free(buf);
-		buf = NULL;
+	    GlobalVariable *str_gvar = compile_const_global_ustring(chars,
+		    chars_len);
+
+	    if (need_free) {
+		free(chars);
+		chars = NULL;
 	    }
 
 	    std::vector<Value *> idxs;
@@ -5621,8 +5620,10 @@
 		    RubyObjTy, PtrTy, NULL));
 
     Function *newRegexp2Func =
-	cast<Function>(module->getOrInsertFunction("rb_reg_new_retained",
-		    RubyObjTy, PtrTy, Int32Ty, Int32Ty, NULL));
+	cast<Function>(module->getOrInsertFunction(
+		    "rb_unicode_regex_new_retained",
+		    RubyObjTy, PointerType::getUnqual(Int16Ty), Int32Ty,
+		    Int32Ty, NULL));
 
     Function *newBignumFunc =
 	cast<Function>(module->getOrInsertFunction("rb_bignum_new_retained",
@@ -5669,28 +5670,33 @@
 
 	    case T_REGEXP:
 		{
-		    struct RRegexp *re = (struct RRegexp *)val;
+		    const UChar *chars = NULL;
+		    long chars_len = 0;
 
+		    regexp_get_uchars(val, &chars, &chars_len);
+
 		    Value *re_str;
-		    if (re->len == 0) {
+		    if (chars_len == 0) {
+			re_str = ConstantPointerNull::get(
+				PointerType::getUnqual(Int16Ty));
 			re_str = compile_const_pointer(NULL, NULL);
 		    }
 		    else {
-			GlobalVariable *rename_gvar =
-			    compile_const_global_string(re->str, re->len);
+			GlobalVariable *re_name_gvar =
+			    compile_const_global_ustring(chars, chars_len);
 
 			std::vector<Value *> idxs;
 			idxs.push_back(ConstantInt::get(Int32Ty, 0));
 			idxs.push_back(ConstantInt::get(Int32Ty, 0));
-			re_str = GetElementPtrInst::Create(rename_gvar,
+			re_str = GetElementPtrInst::Create(re_name_gvar,
 				idxs.begin(), idxs.end(), "");
 		    }
 
 		    std::vector<Value *> params;
 		    params.push_back(re_str);
-		    params.push_back(ConstantInt::get(Int32Ty, re->len));
+		    params.push_back(ConstantInt::get(Int32Ty, chars_len));
 		    params.push_back(ConstantInt::get(Int32Ty,
-				re->ptr->options));
+				rb_reg_options(val)));
 
 		    Instruction *call = CallInst::Create(newRegexp2Func,
 			    params.begin(), params.end(), "");

Modified: MacRuby/branches/icu/compiler.h
===================================================================
--- MacRuby/branches/icu/compiler.h	2010-02-25 21:13:29 UTC (rev 3621)
+++ MacRuby/branches/icu/compiler.h	2010-02-25 21:53:42 UTC (rev 3622)
@@ -314,7 +314,7 @@
 	    return compile_const_global_string(str, strlen(str));
 	}
 	GlobalVariable *compile_const_global_ustring(const UniChar *str,
-		const size_t str_len, CFHashCode hash);
+		const size_t str_len);
 
 	Value *compile_arity(rb_vm_arity_t &arity);
 	Instruction *compile_range(Value *beg, Value *end, bool exclude_end,

Modified: MacRuby/branches/icu/encoding.h
===================================================================
--- MacRuby/branches/icu/encoding.h	2010-02-25 21:13:29 UTC (rev 3621)
+++ MacRuby/branches/icu/encoding.h	2010-02-25 21:53:42 UTC (rev 3622)
@@ -301,7 +301,7 @@
 long rb_str_chars_len(VALUE str);
 UChar rb_str_get_uchar(VALUE str, long pos);
 void rb_str_append_uchar(VALUE str, UChar c);
-unsigned long rb_str_hash_uchars(UChar *chars, long chars_len);
+unsigned long rb_str_hash_uchars(const UChar *chars, long chars_len);
 
 VALUE mr_enc_s_is_compatible(VALUE klass, SEL sel, VALUE str1, VALUE str2);
 

Modified: MacRuby/branches/icu/include/ruby/ruby.h
===================================================================
--- MacRuby/branches/icu/include/ruby/ruby.h	2010-02-25 21:13:29 UTC (rev 3621)
+++ MacRuby/branches/icu/include/ruby/ruby.h	2010-02-25 21:53:42 UTC (rev 3622)
@@ -641,13 +641,6 @@
 # define RARRAY_PTR(a) (rb_ary_ptr((VALUE)a)) 
 #endif
 
-struct RRegexp {
-    struct RBasic basic;
-    struct re_pattern_buffer *ptr;
-    long len;
-    char *str;
-};
-
 #if !WITH_OBJC
 struct RHash {
     struct RBasic basic;

Modified: MacRuby/branches/icu/re.cpp
===================================================================
--- MacRuby/branches/icu/re.cpp	2010-02-25 21:13:29 UTC (rev 3621)
+++ MacRuby/branches/icu/re.cpp	2010-02-25 21:53:42 UTC (rev 3622)
@@ -1336,6 +1336,29 @@
 	    "finalize", (IMP)match_finalize_imp);
 }
 
+// Compiler primitives.
+
+void
+regexp_get_uchars(VALUE re, const UChar **chars_p, long *chars_len_p)
+{
+    assert(chars_p != NULL && chars_len_p != NULL);
+
+    UnicodeString *unistr = RREGEXP(re)->unistr;
+    assert(unistr != NULL);
+
+    *chars_p = unistr->getBuffer();
+    *chars_len_p = unistr->length();
+}
+
+VALUE
+rb_unicode_regex_new_retained(UChar *chars, int chars_len, int options)
+{
+    VALUE str = rb_unicode_str_new(chars, chars_len);
+    VALUE re = rb_reg_new_str(str, options);
+    GC_RETAIN(re);
+    return re;
+}
+
 // MRI compatibility.
 
 VALUE

Modified: MacRuby/branches/icu/re.h
===================================================================
--- MacRuby/branches/icu/re.h	2010-02-25 21:13:29 UTC (rev 3621)
+++ MacRuby/branches/icu/re.h	2010-02-25 21:53:42 UTC (rev 3622)
@@ -23,6 +23,8 @@
 VALUE rb_reg_regcomp(VALUE str);
 int rb_reg_search(VALUE re, VALUE str, int pos, bool reverse);
 
+void regexp_get_uchars(VALUE re, const UChar **chars_p, long *chars_len_p);
+
 typedef struct rb_match_result {
     unsigned int beg;
     unsigned int end;

Modified: MacRuby/branches/icu/string.c
===================================================================
--- MacRuby/branches/icu/string.c	2010-02-25 21:13:29 UTC (rev 3621)
+++ MacRuby/branches/icu/string.c	2010-02-25 21:53:42 UTC (rev 3622)
@@ -3731,7 +3731,7 @@
 // Unicode characters hashing function, copied from CoreFoundation.
 // This function might have some performance issues on large strings.
 unsigned long
-rb_str_hash_uchars(UChar *chars, long len)
+rb_str_hash_uchars(const UChar *chars, long len)
 {
 #define HashNextFourUniChars(accessStart, accessEnd, pointer) \
     {result = result * 67503105 + (accessStart 0 accessEnd) * 16974593  + (accessStart 1 accessEnd) * 66049  + (accessStart 2 accessEnd) * 257 + (accessStart 3 accessEnd); pointer += 4;}

    

[macruby-changes] [3622] MacRuby/branches/icu

source_changes＠macosforge.org