RegExp refactoring and improvements

Move RegExp bytecode functions to a separate file.
Optimize bytecode lenght on character matching.
Implement a basic RegExp cache to optimize memory
usage on duplicated RegExp in JS files. Also fix
minor style issues and add missing comments. Improve
existing comments.

JerryScript-DCO-1.0-Signed-off-by: László Langó llango.u-szeged@partner.samsung.com

Author: LaszloLango

jerry-core/ecma/base/ecma-gc.c

@@ -32,6 +32,7 @@
 #include "jrt.h"
 #include "jrt-libc-includes.h"
 #include "jrt-bit-fields.h"
+#include "re-compiler.h"
 #include "vm-defines.h"
 #include "vm-stack.h"
 
@@ -549,6 +550,11 @@ ecma_gc_run (void)
   ecma_gc_objects_lists[ECMA_GC_COLOR_BLACK] = NULL;
 
   ecma_gc_visited_flip_flag = !ecma_gc_visited_flip_flag;
+
+#ifndef CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN
+  /* Free RegExp bytecodes stored in cache */
+  re_cache_gc_run ();
+#endif /* !CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */
 } /* ecma_gc_run */
 
 /**

jerry-core/ecma/builtin-objects/ecma-builtin-regexp-prototype.c

@@ -53,7 +53,9 @@
  * See also:
  *          ECMA-262 v5, B.2.5.1
  *
- * @return ecma value
+ * @return undefined        - if compiled successfully
+ *         error ecma value - otherwise
+ *
  *         Returned value must be freed with ecma_free_value.
  */
 static ecma_value_t
@@ -240,8 +242,10 @@ ecma_builtin_regexp_prototype_compile (ecma_value_t this_arg, /**< this argument
  * See also:
  *          ECMA-262 v5, 15.10.6.2
  *
- * @return ecma value
- *         Returned value must be freed with ecma_free_value.
+ * @return array object containing the results - if the matched
+ *         null                                - otherwise
+ *
+ *         May raise error, so returned value must be freed with ecma_free_value.
  */
 static ecma_value_t
 ecma_builtin_regexp_prototype_exec (ecma_value_t this_arg, /**< this argument */
@@ -314,8 +318,10 @@ ecma_builtin_regexp_prototype_exec (ecma_value_t this_arg, /**< this argument */
  * See also:
  *          ECMA-262 v5, 15.10.6.3
  *
- * @return ecma value
- *         Returned value must be freed with ecma_free_value.
+ * @return true  - if match is not null
+ *         false - otherwise
+ *
+ *         May raise error, so returned value must be freed with ecma_free_value.
  */
 static ecma_value_t
 ecma_builtin_regexp_prototype_test (ecma_value_t this_arg, /**< this argument */
@@ -439,4 +445,4 @@ ecma_builtin_regexp_prototype_to_string (ecma_value_t this_arg) /**< this argume
  * @}
  */
 
-#endif /* CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */
+#endif /* !CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */

jerry-core/ecma/builtin-objects/ecma-builtin-regexp.c

@@ -46,6 +46,7 @@
  * Handle calling [[Call]] of built-in RegExp object
  *
  * @return ecma value
+ *         Returned value must be freed with ecma_free_value.
  */
 ecma_value_t
 ecma_builtin_regexp_dispatch_call (const ecma_value_t *arguments_list_p, /**< arguments list */
@@ -58,6 +59,7 @@ ecma_builtin_regexp_dispatch_call (const ecma_value_t *arguments_list_p, /**< ar
  * Handle calling [[Construct]] of built-in RegExp object
  *
  * @return ecma value
+ *         Returned value must be freed with ecma_free_value.
  */
 ecma_value_t
 ecma_builtin_regexp_dispatch_construct (const ecma_value_t *arguments_list_p, /**< arguments list */
@@ -152,4 +154,4 @@ ecma_builtin_regexp_dispatch_construct (const ecma_value_t *arguments_list_p, /*
  * @}
  */
 
-#endif /* CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */
+#endif /* !CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */

jerry-core/ecma/builtin-objects/ecma-builtin-string-prototype.inc.h

@@ -73,7 +73,7 @@ ROUTINE (LIT_MAGIC_STRING_MATCH, ecma_builtin_string_prototype_object_match, 1,
 ROUTINE (LIT_MAGIC_STRING_REPLACE, ecma_builtin_string_prototype_object_replace, 2, 2)
 ROUTINE (LIT_MAGIC_STRING_SEARCH, ecma_builtin_string_prototype_object_search, 1, 1)
 ROUTINE (LIT_MAGIC_STRING_SPLIT, ecma_builtin_string_prototype_object_split, 2, 2)
-#endif /* CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */
+#endif /* !CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */
 
 ROUTINE (LIT_MAGIC_STRING_SUBSTRING, ecma_builtin_string_prototype_object_substring, 2, 2)
 ROUTINE (LIT_MAGIC_STRING_TO_LOWER_CASE_UL, ecma_builtin_string_prototype_object_to_lower_case, 0, 0)
@@ -84,7 +84,7 @@ ROUTINE (LIT_MAGIC_STRING_TRIM, ecma_builtin_string_prototype_object_trim, 0, 0)
 
 #ifndef CONFIG_ECMA_COMPACT_PROFILE_DISABLE_ANNEXB_BUILTIN
 ROUTINE (LIT_MAGIC_STRING_SUBSTR, ecma_builtin_string_prototype_object_substr, 2, 2)
-#endif /* CONFIG_ECMA_COMPACT_PROFILE_DISABLE_ANNEXB_BUILTIN */
+#endif /* !CONFIG_ECMA_COMPACT_PROFILE_DISABLE_ANNEXB_BUILTIN */
 
 #undef OBJECT_ID
 #undef SIMPLE_VALUE

jerry-core/ecma/operations/ecma-objects.c

@@ -958,13 +958,13 @@ ecma_object_get_class_name (ecma_object_t *obj_p) /**< object */
           {
             return LIT_MAGIC_STRING_DATE_UL;
           }
-#endif /* CONFIG_ECMA_COMPACT_PROFILE_DISABLE_DATE_BUILTIN */
+#endif /* !CONFIG_ECMA_COMPACT_PROFILE_DISABLE_DATE_BUILTIN */
 #ifndef CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN
           case ECMA_BUILTIN_ID_REGEXP_PROTOTYPE:
           {
             return LIT_MAGIC_STRING_REGEXP_UL;
           }
-#endif /* CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */
+#endif /* !CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */
           default:
           {
             JERRY_ASSERT (ecma_builtin_is (obj_p, ECMA_BUILTIN_ID_GLOBAL));

jerry-core/ecma/operations/ecma-regexp-object.c

@@ -55,12 +55,19 @@
 #define RE_GLOBAL_START_IDX 0
 #define RE_GLOBAL_END_IDX   1
 
+/**
+ * Check if a RegExp opcode is a capture group or not
+ */
+#define RE_IS_CAPTURE_GROUP(x) (((x) < RE_OP_NON_CAPTURE_GROUP_START) ? 1 : 0)
+
 /**
  * Parse RegExp flags (global, ignoreCase, multiline)
  *
  * See also: ECMA-262 v5, 15.10.4.1
  *
- * @return ecma value
+ * @return empty ecma value - if parsed successfully
+ *         error ecma value - otherwise
+ *
  *         Returned value must be freed with ecma_free_value
  */
 ecma_value_t
@@ -123,7 +130,7 @@ re_parse_regexp_flags (ecma_string_t *flags_str_p, /**< Input string with flags
   return ret_value;
 } /* re_parse_regexp_flags  */
 
-/*
+/**
  * Initializes the source, global, ignoreCase, multiline, and lastIndex properties of RegExp instance.
  */
 void
@@ -223,11 +230,11 @@ re_initialize_props (ecma_object_t *re_obj_p, /**< RegExp obejct */
  *
  * See also: ECMA-262 v5, 15.10.4.1
  *
- * @return ecma value
+ * @return constructed RegExp object
  *         Returned value must be freed with ecma_free_value
  */
 ecma_value_t
-ecma_op_create_regexp_object_from_bytecode (re_compiled_code_t *bytecode_p) /**< input pattern */
+ecma_op_create_regexp_object_from_bytecode (re_compiled_code_t *bytecode_p) /**< RegExp bytecode */
 {
   JERRY_ASSERT (bytecode_p != NULL);
 
@@ -259,7 +266,9 @@ ecma_op_create_regexp_object_from_bytecode (re_compiled_code_t *bytecode_p) /**<
  *
  * See also: ECMA-262 v5, 15.10.4.1
  *
- * @return ecma value
+ * @return constructed RegExp object - if pattern and flags were parsed successfully
+ *         error ecma value          - otherwise
+ *
  *         Returned value must be freed with ecma_free_value
  */
 ecma_value_t
@@ -367,8 +376,10 @@ re_canonicalize (ecma_char_t ch, /**< character */
  * See also:
  *          ECMA-262 v5, 15.10.2.1
  *
- * @return ecma value
- *         Returned value must be freed with ecma_free_value
+ * @return true  - if matched
+ *         false - otherwise
+ *
+ *         May raise error, so returned value must be freed with ecma_free_value
  */
 static ecma_value_t
 re_match_regexp (re_matcher_ctx_t *re_ctx_p, /**< RegExp matcher context */
@@ -400,7 +411,7 @@ re_match_regexp (re_matcher_ctx_t *re_ctx_p, /**< RegExp matcher context */
         }
 
         bool is_ignorecase = re_ctx_p->flags & RE_FLAG_IGNORE_CASE;
-        ecma_char_t ch1 = (ecma_char_t) re_get_value (&bc_p); /* Already canonicalized. */
+        ecma_char_t ch1 = (ecma_char_t) re_get_char (&bc_p); /* Already canonicalized. */
         ecma_char_t ch2 = re_canonicalize (lit_utf8_read_next (&str_curr_p), is_ignorecase);
         JERRY_DDLOG ("Character matching %d to %d: ", ch1, ch2);
 
@@ -613,8 +624,8 @@ re_match_regexp (re_matcher_ctx_t *re_ctx_p, /**< RegExp matcher context */
 
         while (num_of_ranges)
         {
-          ecma_char_t ch1 = re_canonicalize ((ecma_char_t) re_get_value (&bc_p), is_ignorecase);
-          ecma_char_t ch2 = re_canonicalize ((ecma_char_t) re_get_value (&bc_p), is_ignorecase);
+          ecma_char_t ch1 = re_canonicalize (re_get_char (&bc_p), is_ignorecase);
+          ecma_char_t ch2 = re_canonicalize (re_get_char (&bc_p), is_ignorecase);
           JERRY_DDLOG ("num_of_ranges=%d, ch1=%d, ch2=%d, curr_ch=%d; ",
                        num_of_ranges, ch1, ch2, curr_ch);
 
@@ -698,6 +709,7 @@ re_match_regexp (re_matcher_ctx_t *re_ctx_p, /**< RegExp matcher context */
           uint32_t offset = re_get_value (&bc_p);
           lit_utf8_byte_t *sub_str_p = NULL;
           ecma_value_t match_value = re_match_regexp (re_ctx_p, bc_p, str_curr_p, &sub_str_p);
+
           if (ecma_is_value_true (match_value))
           {
             *out_str_p = sub_str_p;
@@ -707,6 +719,7 @@ re_match_regexp (re_matcher_ctx_t *re_ctx_p, /**< RegExp matcher context */
           {
             return match_value;
           }
+
           bc_p += offset;
           old_bc_p = bc_p;
         }
@@ -839,6 +852,7 @@ re_match_regexp (re_matcher_ctx_t *re_ctx_p, /**< RegExp matcher context */
         {
           offset = re_get_value (&bc_p);
           ecma_value_t match_value = re_match_regexp (re_ctx_p, bc_p, str_curr_p, &sub_str_p);
+
           if (ecma_is_value_true (match_value))
           {
             *out_str_p = sub_str_p;
@@ -848,6 +862,7 @@ re_match_regexp (re_matcher_ctx_t *re_ctx_p, /**< RegExp matcher context */
           {
             return match_value;
           }
+
           bc_p += offset;
           old_bc_p = bc_p;
         }
@@ -915,6 +930,7 @@ re_match_regexp (re_matcher_ctx_t *re_ctx_p, /**< RegExp matcher context */
 
           lit_utf8_byte_t *sub_str_p = NULL;
           ecma_value_t match_value = re_match_regexp (re_ctx_p, bc_p, str_curr_p, &sub_str_p);
+
           if (ecma_is_value_true (match_value))
           {
             *out_str_p = sub_str_p;
@@ -1225,7 +1241,13 @@ re_set_result_array_properties (ecma_object_t *array_obj_p, /**< result array */
  * RegExp helper function to start the recursive matching algorithm
  * and create the result Array object
  *
- * @return ecma value
+ * See also:
+ *          ECMA-262 v5, 15.10.6.2
+ *
+ * @return array object - if matched
+ *         null         - otherwise
+ *
+ *         May raise error.
  *         Returned value must be freed with ecma_free_value
  */
 ecma_value_t
@@ -1475,4 +1497,4 @@ ecma_regexp_exec_helper (ecma_value_t regexp_value, /**< RegExp object */
  * @}
  */
 
-#endif /* CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */
+#endif /* !CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */

jerry-core/ecma/operations/ecma-regexp-object.h

@@ -32,48 +32,39 @@
 /**
  * RegExp flags
  */
-#define RE_FLAG_GLOBAL              (1u << 1) /* ECMA-262 v5, 15.10.7.2 */
-#define RE_FLAG_IGNORE_CASE         (1u << 2) /* ECMA-262 v5, 15.10.7.3 */
-#define RE_FLAG_MULTILINE           (1u << 3) /* ECMA-262 v5, 15.10.7.4 */
+typedef enum
+{
+  RE_FLAG_GLOBAL = (1u << 1),      /**< ECMA-262 v5, 15.10.7.2 */
+  RE_FLAG_IGNORE_CASE = (1u << 2), /**< ECMA-262 v5, 15.10.7.3 */
+  RE_FLAG_MULTILINE = (1u << 3)    /**< ECMA-262 v5, 15.10.7.4 */
+} re_flags_t;
 
 /**
  * RegExp executor context
  */
 typedef struct
 {
-  lit_utf8_byte_t **saved_p; /**< saved result string pointers, ECMA 262 v5, 15.10.2.1, State */
+  lit_utf8_byte_t **saved_p;            /**< saved result string pointers, ECMA 262 v5, 15.10.2.1, State */
   const lit_utf8_byte_t *input_start_p; /**< start of input pattern string */
-  const lit_utf8_byte_t *input_end_p; /**< end of input pattern string */
-  uint32_t num_of_captures; /**< number of capture groups */
-  uint32_t num_of_non_captures; /**< number of non-capture groups */
-  uint32_t *num_of_iterations_p; /**< number of iterations */
-  uint16_t flags; /**< RegExp flags */
+  const lit_utf8_byte_t *input_end_p;   /**< end of input pattern string */
+  uint32_t num_of_captures;             /**< number of capture groups */
+  uint32_t num_of_non_captures;         /**< number of non-capture groups */
+  uint32_t *num_of_iterations_p;        /**< number of iterations */
+  uint16_t flags;                       /**< RegExp flags */
 } re_matcher_ctx_t;
 
-extern ecma_value_t
-ecma_op_create_regexp_object_from_bytecode (re_compiled_code_t *);
-
-extern ecma_value_t
-ecma_op_create_regexp_object (ecma_string_t *, ecma_string_t *);
-
-extern ecma_value_t
-ecma_regexp_exec_helper (ecma_value_t, ecma_value_t, bool);
-
-extern ecma_char_t
-re_canonicalize (ecma_char_t, bool);
-extern void
-re_set_result_array_properties (ecma_object_t *, ecma_string_t *, uint32_t, int32_t);
-
-extern ecma_value_t
-re_parse_regexp_flags (ecma_string_t *, uint16_t *);
-
-extern void
-re_initialize_props (ecma_object_t *, ecma_string_t *, uint16_t);
+ecma_value_t ecma_op_create_regexp_object_from_bytecode (re_compiled_code_t *);
+ecma_value_t ecma_op_create_regexp_object (ecma_string_t *, ecma_string_t *);
+ecma_value_t ecma_regexp_exec_helper (ecma_value_t, ecma_value_t, bool);
+ecma_char_t re_canonicalize (ecma_char_t, bool);
+void re_set_result_array_properties (ecma_object_t *, ecma_string_t *, uint32_t, int32_t);
+ecma_value_t re_parse_regexp_flags (ecma_string_t *, uint16_t *);
+void re_initialize_props (ecma_object_t *, ecma_string_t *, uint16_t);
 
 /**
  * @}
  * @}
  */
 
-#endif /* CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */
+#endif /* !CONFIG_ECMA_COMPACT_PROFILE_DISABLE_REGEXP_BUILTIN */
 #endif /* !ECMA_REGEXP_OBJECT_H */

jerry-core/jerry.c

@@ -1655,9 +1655,9 @@ jerry_cleanup (void)
 
   bool is_show_mem_stats = ((jerry_flags & JERRY_FLAG_MEM_STATS) != 0);
 
+  vm_finalize ();
   ecma_finalize ();
   lit_finalize ();
-  vm_finalize ();
   mem_finalize (is_show_mem_stats);
 } /* jerry_cleanup */
 

jerry-core/parser/js/js-lexer.c

@@ -1812,7 +1812,6 @@ lexer_construct_regexp_object (parser_context_t *context_p, /**< context */
   ecma_value_t completion_value;
 
   ecma_string_t *pattern_str_p = ecma_new_ecma_string_from_utf8 (regex_start_p, length);
-  // FIXME: check return value of 're_compile_bytecode' and throw an error
   completion_value = re_compile_bytecode (&re_bytecode_p,
                                           pattern_str_p,
                                           current_flags);