kanayago 0.1.1 → 0.2.0

data/ext/kanayago/parse.h

@@ -1,4 +1,4 @@
-/* A Bison parser, made by Lrama 0.6.9.  */
+/* A Bison parser, made by Lrama 0.7.0.  */
 
 /* Bison interface for Yacc-like parsers in C
 
@@ -34,8 +34,8 @@
 /* DO NOT RELY ON FEATURES THAT ARE NOT DOCUMENTED in the manual,
    especially those whose name start with YY_ or yy_.  They are
    private implementation details that can be changed or removed.  */
-#ifndef YY_YY_EXT_KANAYAGO_PARSE_H_INCLUDED
-# define YY_YY_EXT_KANAYAGO_PARSE_H_INCLUDED
+#ifndef YY_YY_PARSE_H_INCLUDED
+# define YY_YY_PARSE_H_INCLUDED
 /* Debug traces.  */
 #ifndef YYDEBUG
 # define YYDEBUG 0
@@ -149,39 +149,38 @@ extern int yydebug;
     tASSOC = 327,                  /* "=>"  */
     tLPAREN = 328,                 /* "("  */
     tLPAREN_ARG = 329,             /* "( arg"  */
-    tRPAREN = 330,                 /* ")"  */
-    tLBRACK = 331,                 /* "["  */
-    tLBRACE = 332,                 /* "{"  */
-    tLBRACE_ARG = 333,             /* "{ arg"  */
-    tSTAR = 334,                   /* "*"  */
-    tDSTAR = 335,                  /* "**arg"  */
-    tAMPER = 336,                  /* "&"  */
-    tLAMBDA = 337,                 /* "->"  */
-    tSYMBEG = 338,                 /* "symbol literal"  */
-    tSTRING_BEG = 339,             /* "string literal"  */
-    tXSTRING_BEG = 340,            /* "backtick literal"  */
-    tREGEXP_BEG = 341,             /* "regexp literal"  */
-    tWORDS_BEG = 342,              /* "word list"  */
-    tQWORDS_BEG = 343,             /* "verbatim word list"  */
-    tSYMBOLS_BEG = 344,            /* "symbol list"  */
-    tQSYMBOLS_BEG = 345,           /* "verbatim symbol list"  */
-    tSTRING_END = 346,             /* "terminator"  */
-    tSTRING_DEND = 347,            /* "'}'"  */
-    tSTRING_DBEG = 348,            /* "'#{'"  */
-    tSTRING_DVAR = 349,            /* tSTRING_DVAR  */
-    tLAMBEG = 350,                 /* tLAMBEG  */
-    tLABEL_END = 351,              /* tLABEL_END  */
-    tIGNORED_NL = 352,             /* tIGNORED_NL  */
-    tCOMMENT = 353,                /* tCOMMENT  */
-    tEMBDOC_BEG = 354,             /* tEMBDOC_BEG  */
-    tEMBDOC = 355,                 /* tEMBDOC  */
-    tEMBDOC_END = 356,             /* tEMBDOC_END  */
-    tHEREDOC_BEG = 357,            /* tHEREDOC_BEG  */
-    tHEREDOC_END = 358,            /* tHEREDOC_END  */
-    k__END__ = 359,                /* k__END__  */
-    tLOWEST = 360,                 /* tLOWEST  */
-    tUMINUS_NUM = 361,             /* tUMINUS_NUM  */
-    tLAST_TOKEN = 362              /* tLAST_TOKEN  */
+    tLBRACK = 330,                 /* "["  */
+    tLBRACE = 331,                 /* "{"  */
+    tLBRACE_ARG = 332,             /* "{ arg"  */
+    tSTAR = 333,                   /* "*"  */
+    tDSTAR = 334,                  /* "**arg"  */
+    tAMPER = 335,                  /* "&"  */
+    tLAMBDA = 336,                 /* "->"  */
+    tSYMBEG = 337,                 /* "symbol literal"  */
+    tSTRING_BEG = 338,             /* "string literal"  */
+    tXSTRING_BEG = 339,            /* "backtick literal"  */
+    tREGEXP_BEG = 340,             /* "regexp literal"  */
+    tWORDS_BEG = 341,              /* "word list"  */
+    tQWORDS_BEG = 342,             /* "verbatim word list"  */
+    tSYMBOLS_BEG = 343,            /* "symbol list"  */
+    tQSYMBOLS_BEG = 344,           /* "verbatim symbol list"  */
+    tSTRING_END = 345,             /* "terminator"  */
+    tSTRING_DEND = 346,            /* "'}'"  */
+    tSTRING_DBEG = 347,            /* "'#{'"  */
+    tSTRING_DVAR = 348,            /* tSTRING_DVAR  */
+    tLAMBEG = 349,                 /* tLAMBEG  */
+    tLABEL_END = 350,              /* tLABEL_END  */
+    tIGNORED_NL = 351,             /* tIGNORED_NL  */
+    tCOMMENT = 352,                /* tCOMMENT  */
+    tEMBDOC_BEG = 353,             /* tEMBDOC_BEG  */
+    tEMBDOC = 354,                 /* tEMBDOC  */
+    tEMBDOC_END = 355,             /* tEMBDOC_END  */
+    tHEREDOC_BEG = 356,            /* tHEREDOC_BEG  */
+    tHEREDOC_END = 357,            /* tHEREDOC_END  */
+    k__END__ = 358,                /* k__END__  */
+    tLOWEST = 359,                 /* tLOWEST  */
+    tUMINUS_NUM = 360,             /* tUMINUS_NUM  */
+    tLAST_TOKEN = 361              /* tLAST_TOKEN  */
   };
   typedef enum yytokentype yytoken_kind_t;
 #endif
@@ -190,7 +189,7 @@ extern int yydebug;
 #if ! defined YYSTYPE && ! defined YYSTYPE_IS_DECLARED
 union YYSTYPE
 {
-#line 2696 "ext/kanayago/parse.tmp.y"
+#line 2670 "parse.y"
 
     NODE *node;
     rb_node_fcall_t *node_fcall;
@@ -202,6 +201,7 @@ union YYSTYPE
     rb_node_masgn_t *node_masgn;
     rb_node_def_temp_t *node_def_temp;
     rb_node_exits_t *node_exits;
+    struct rb_locations_lambda_body_t *locations_lambda_body;
     ID id;
     int num;
     st_table *tbl;
@@ -211,7 +211,7 @@ union YYSTYPE
     struct lex_context ctxt;
     enum lex_state_e state;
 
-#line 215 "ext/kanayago/parse.h"
+#line 215 "parse.h"
 
 };
 typedef union YYSTYPE YYSTYPE;
@@ -240,5 +240,5 @@ int yyparse (struct parser_params *p);
 
 
 
-#endif /* !YY_YY_EXT_KANAYAGO_PARSE_H_INCLUDED  */
+#endif /* !YY_YY_PARSE_H_INCLUDED  */
 

data/ext/kanayago/parser_st.c

@@ -154,7 +154,8 @@ nonempty_memcpy(void *dest, const void *src, size_t n)
     _Pragma("GCC diagnostic push") \
     _Pragma("GCC diagnostic ignored \"-Wattributes\"") \
     __attribute__((__no_sanitize__(x))) y; \
-    _Pragma("GCC diagnostic pop")
+    _Pragma("GCC diagnostic pop") \
+    y
 #endif
 
 #ifndef NO_SANITIZE

data/ext/kanayago/pattern_node.c

@@ -0,0 +1,78 @@
+#include "pattern_node.h"
+#include "internal/ruby_parser.h"
+#include "kanayago.h"
+
+VALUE rb_cInNode;
+VALUE rb_cArrayPatternNode;
+VALUE rb_cHashPatternNode;
+VALUE rb_cFindPatternNode;
+
+VALUE
+in_node_new(const NODE *node)
+{
+    VALUE obj = rb_class_new_instance(0, 0, rb_cInNode);
+
+    rb_ivar_set(obj, rb_intern("@head"), ast_to_node_instance(RNODE_IN(node)->nd_head));
+    rb_ivar_set(obj, rb_intern("@body"), ast_to_node_instance(RNODE_IN(node)->nd_body));
+    rb_ivar_set(obj, rb_intern("@next"), ast_to_node_instance(RNODE_IN(node)->nd_next));
+
+    return obj;
+}
+
+VALUE
+array_pattern_node_new(const NODE *node)
+{
+    VALUE obj = rb_class_new_instance(0, 0, rb_cArrayPatternNode);
+    NODE *rest_arg = RNODE_ARYPTN(node)->rest_arg;
+
+    rb_ivar_set(obj, rb_intern("@pconst"), ast_to_node_instance(RNODE_ARYPTN(node)->nd_pconst));
+    rb_ivar_set(obj, rb_intern("@pre_args"), ast_to_node_instance(RNODE_ARYPTN(node)->pre_args));
+    // rest_arg can be NODE_SPECIAL_NO_NAME_REST (-1), which should be preserved as a special marker
+    rb_ivar_set(obj, rb_intern("@rest_arg"),
+                (rest_arg == (NODE *)-1) ? INT2FIX(-1) : ast_to_node_instance(rest_arg));
+    rb_ivar_set(obj, rb_intern("@post_args"), ast_to_node_instance(RNODE_ARYPTN(node)->post_args));
+
+    return obj;
+}
+
+VALUE
+hash_pattern_node_new(const NODE *node)
+{
+    VALUE obj = rb_class_new_instance(0, 0, rb_cHashPatternNode);
+
+    rb_ivar_set(obj, rb_intern("@pconst"), ast_to_node_instance(RNODE_HSHPTN(node)->nd_pconst));
+    rb_ivar_set(obj, rb_intern("@pkwargs"), ast_to_node_instance(RNODE_HSHPTN(node)->nd_pkwargs));
+    rb_ivar_set(obj, rb_intern("@pkwrestarg"), ast_to_node_instance(RNODE_HSHPTN(node)->nd_pkwrestarg));
+
+    return obj;
+}
+
+VALUE
+find_pattern_node_new(const NODE *node)
+{
+    VALUE obj = rb_class_new_instance(0, 0, rb_cFindPatternNode);
+    NODE *pre_rest_arg = RNODE_FNDPTN(node)->pre_rest_arg;
+    NODE *post_rest_arg = RNODE_FNDPTN(node)->post_rest_arg;
+
+    rb_ivar_set(obj, rb_intern("@pconst"), ast_to_node_instance(RNODE_FNDPTN(node)->nd_pconst));
+    // rest args can be NODE_SPECIAL_NO_NAME_REST (-1)
+    rb_ivar_set(obj, rb_intern("@pre_rest_arg"),
+                (pre_rest_arg == (NODE *)-1) ? INT2FIX(-1) : ast_to_node_instance(pre_rest_arg));
+    rb_ivar_set(obj, rb_intern("@args"), ast_to_node_instance(RNODE_FNDPTN(node)->args));
+    rb_ivar_set(obj, rb_intern("@post_rest_arg"),
+                (post_rest_arg == (NODE *)-1) ? INT2FIX(-1) : ast_to_node_instance(post_rest_arg));
+
+    return obj;
+}
+
+void
+Init_PatternNode(VALUE module)
+{
+    rb_cInNode = rb_define_class_under(module, "InNode", rb_cObject);
+
+    rb_cArrayPatternNode = rb_define_class_under(module, "ArrayPatternNode", rb_cObject);
+
+    rb_cHashPatternNode = rb_define_class_under(module, "HashPatternNode", rb_cObject);
+
+    rb_cFindPatternNode = rb_define_class_under(module, "FindPatternNode", rb_cObject);
+}

data/ext/kanayago/pattern_node.h

@@ -0,0 +1,13 @@
+#ifndef KANAYAGO_PATTERN_NODE_H
+#define KANAYAGO_PATTERN_NODE_H
+
+#include "kanayago.h"
+
+VALUE in_node_new(const NODE *);
+VALUE array_pattern_node_new(const NODE *);
+VALUE hash_pattern_node_new(const NODE *);
+VALUE find_pattern_node_new(const NODE *);
+
+void Init_PatternNode(VALUE);
+
+#endif

data/ext/kanayago/ruby_atomic.h

@@ -1,5 +1,10 @@
+#ifndef INTERNAL_ATOMIC_H
+#define INTERNAL_ATOMIC_H
+
 #include "ruby/atomic.h"
 
+#define RUBY_ATOMIC_VALUE_LOAD(x) rbimpl_atomic_value_load(&(x), RBIMPL_ATOMIC_SEQ_CST)
+
 /* shim macros only */
 #define ATOMIC_ADD(var, val) RUBY_ATOMIC_ADD(var, val)
 #define ATOMIC_CAS(var, oldval, newval) RUBY_ATOMIC_CAS(var, oldval, newval)
@@ -21,3 +26,41 @@
 #define ATOMIC_SUB(var, val) RUBY_ATOMIC_SUB(var, val)
 #define ATOMIC_VALUE_CAS(var, oldval, val) RUBY_ATOMIC_VALUE_CAS(var, oldval, val)
 #define ATOMIC_VALUE_EXCHANGE(var, val) RUBY_ATOMIC_VALUE_EXCHANGE(var, val)
+
+#define ATOMIC_LOAD_RELAXED(var) rbimpl_atomic_load(&(var), RBIMPL_ATOMIC_RELAXED)
+
+typedef RBIMPL_ALIGNAS(8) uint64_t rbimpl_atomic_uint64_t;
+
+static inline uint64_t
+rbimpl_atomic_u64_load_relaxed(const volatile rbimpl_atomic_uint64_t *value)
+{
+#if defined(HAVE_GCC_ATOMIC_BUILTINS_64)
+    return __atomic_load_n(value, __ATOMIC_RELAXED);
+#elif defined(_WIN32)
+    uint64_t val = *value;
+    return InterlockedCompareExchange64(RBIMPL_CAST((uint64_t *)value), val, val);
+#elif defined(__sun) && defined(HAVE_ATOMIC_H) && (defined(_LP64) || defined(_I32LPx))
+    uint64_t val = *value;
+    return atomic_cas_64(value, val, val);
+#else
+    return *value;
+#endif
+}
+#define ATOMIC_U64_LOAD_RELAXED(var) rbimpl_atomic_u64_load_relaxed(&(var))
+
+static inline void
+rbimpl_atomic_u64_set_relaxed(volatile rbimpl_atomic_uint64_t *address, uint64_t value)
+{
+#if defined(HAVE_GCC_ATOMIC_BUILTINS_64)
+    __atomic_store_n(address, value, __ATOMIC_RELAXED);
+#elif defined(_WIN32)
+    InterlockedExchange64(address, value);
+#elif defined(__sun) && defined(HAVE_ATOMIC_H) && (defined(_LP64) || defined(_I32LPx))
+    atomic_swap_64(address, value);
+#else
+    *address = value;
+#endif
+}
+#define ATOMIC_U64_SET_RELAXED(var, val) rbimpl_atomic_u64_set_relaxed(&(var), val)
+
+#endif

data/ext/kanayago/ruby_parser.c

@@ -34,12 +34,6 @@
 
 #define parser_encoding const void
 
-static int
-is_ascii_string2(VALUE str)
-{
-    return is_ascii_string(str);
-}
-
 RBIMPL_ATTR_FORMAT(RBIMPL_PRINTF_FORMAT, 6, 0)
 static VALUE
 syntax_error_append(VALUE exc, VALUE file, int line, int column,
@@ -144,12 +138,6 @@ utf8_encoding(void)
     return rb_utf8_encoding();
 }
 
-static VALUE
-enc_associate(VALUE obj, parser_encoding *enc)
-{
-    return rb_enc_associate(obj, enc);
-}
-
 static parser_encoding *
 ascii8bit_encoding(void)
 {
@@ -186,12 +174,6 @@ intern3(const char *name, long len, parser_encoding *enc)
     return rb_intern3(name, len, enc);
 }
 
-static parser_encoding *
-usascii_encoding(void)
-{
-    return rb_usascii_encoding();
-}
-
 static int
 enc_symname_type(const char *name, long len, parser_encoding *enc, unsigned int allowed_attrset)
 {
@@ -203,6 +185,7 @@ typedef struct {
     rb_encoding *enc;
     NODE *succ_block;
     const rb_code_location_t *loc;
+    rb_parser_assignable_func assignable;
 } reg_named_capture_assign_t;
 
 static int
@@ -216,11 +199,12 @@ reg_named_capture_assign_iter(const OnigUChar *name, const OnigUChar *name_end,
     long len = name_end - name;
     const char *s = (const char *)name;
 
-    return rb_reg_named_capture_assign_iter_impl(p, s, len, enc, &arg->succ_block, loc);
+    return rb_reg_named_capture_assign_iter_impl(p, s, len, enc, &arg->succ_block, loc, arg->assignable);
 }
 
 static NODE *
-reg_named_capture_assign(struct parser_params* p, VALUE regexp, const rb_code_location_t *loc)
+reg_named_capture_assign(struct parser_params* p, VALUE regexp, const rb_code_location_t *loc,
+                         rb_parser_assignable_func assignable)
 {
     reg_named_capture_assign_t arg;
 
@@ -228,6 +212,7 @@ reg_named_capture_assign(struct parser_params* p, VALUE regexp, const rb_code_lo
     arg.enc = rb_enc_get(regexp);
     arg.succ_block = 0;
     arg.loc = loc;
+    arg.assignable = assignable;
     onig_foreach_name(RREGEXP_PTR(regexp), reg_named_capture_assign_iter, &arg);
 
     if (!arg.succ_block) return 0;
@@ -339,7 +324,7 @@ xmalloc_mul_add(size_t x, size_t y, size_t z)
 static VALUE
 suppress_tracing(VALUE (*func)(VALUE), VALUE arg)
 {
-    return rb_suppress_tracing(func, arg);
+    return func(arg);
 }
 
 static ID
@@ -385,8 +370,6 @@ static const rb_parser_config_t rb_global_parser_config = {
 
     .attr_get = rb_attr_get,
 
-    .ary_new = rb_ary_new,
-    .ary_push = rb_ary_push,
     .ary_new_from_args = rb_ary_new_from_args,
     .ary_unshift = rb_ary_unshift,
 
@@ -405,25 +388,18 @@ static const rb_parser_config_t rb_global_parser_config = {
     .id2name = rb_id2name,
     .id2str = rb_id2str,
     .id2sym = rb_id2sym,
-    .sym2id = rb_sym2id,
 
     .str_catf = rb_str_catf,
     .str_cat_cstr = rb_str_cat_cstr,
-    .str_modify = rb_str_modify,
-    .str_set_len = rb_str_set_len,
-    .str_cat = rb_str_cat,
     .str_resize = rb_str_resize,
     .str_new = rb_str_new,
     .str_new_cstr = rb_str_new_cstr,
     .str_to_interned_str = rb_str_to_interned_str,
-    .is_ascii_string = is_ascii_string2,
     .enc_str_new = enc_str_new,
     .str_vcatf = rb_str_vcatf,
     .rb_sprintf = rb_sprintf,
     .rstring_ptr = RSTRING_PTR,
-    .rstring_end = RSTRING_END,
     .rstring_len = RSTRING_LEN,
-    .obj_as_string = rb_obj_as_string,
 
     .int2num = rb_int2num_inline,
 
@@ -446,7 +422,6 @@ static const rb_parser_config_t rb_global_parser_config = {
     .enc_get = enc_get,
     .enc_asciicompat = enc_asciicompat,
     .utf8_encoding = utf8_encoding,
-    .enc_associate = enc_associate,
     .ascii8bit_encoding = ascii8bit_encoding,
     .enc_codelen = enc_codelen,
     .enc_mbcput = enc_mbcput,
@@ -455,7 +430,6 @@ static const rb_parser_config_t rb_global_parser_config = {
     .enc_isspace = enc_isspace,
     .enc_coderange_7bit = ENC_CODERANGE_7BIT,
     .enc_coderange_unknown = ENC_CODERANGE_UNKNOWN,
-    .usascii_encoding = usascii_encoding,
     .enc_mbminlen = enc_mbminlen,
     .enc_isascii = enc_isascii,
     .enc_mbc_to_codepoint = enc_mbc_to_codepoint,
@@ -469,7 +443,6 @@ static const rb_parser_config_t rb_global_parser_config = {
 
     .errinfo = rb_errinfo,
     .set_errinfo = rb_set_errinfo,
-    .exc_raise = rb_exc_raise,
     .make_exception = rb_make_exception,
 
     .sized_xfree = ruby_sized_xfree,
@@ -774,8 +747,7 @@ ast_free(void *ptr)
     rb_ast_free(ast);
 }
 
-// Not static const for Kanayago
-const rb_data_type_t ast_data_type = {
+static const rb_data_type_t ast_data_type = {
     "AST",
     {
         NULL,