Refactor builtins to handle CESU-8 encoded strings.

JerryScript-DCO-1.0-Signed-off-by: Zsolt Borbély zsborbely.u-szeged@partner.samsung.com
JerryScript-DCO-1.0-Signed-off-by: Dániel Bátyai dbatyai.u-szeged@partner.samsung.com

Author: dbatyai

jerry-core/ecma/base/ecma-helpers-conversion.cpp

@@ -335,7 +335,7 @@
  * @return ecma-number
  */
 ecma_number_t
-ecma_utf8_string_to_number (const lit_utf8_byte_t *str_p, /**< utf-8 string */
+ecma_cesu8_string_to_number (const lit_utf8_byte_t *str_p, /**< utf-8 string */
                             lit_utf8_size_t str_size) /**< string size */
 {
   TODO (Check license issues);
@@ -354,42 +354,40 @@ ecma_utf8_string_to_number (const lit_utf8_byte_t *str_p, /**< utf-8 string */
     return ECMA_NUMBER_ZERO;
   }
 
-  lit_utf8_iterator_t iter = lit_utf8_iterator_create (str_p, str_size);
+  lit_utf8_byte_t *str_curr_p = (lit_utf8_byte_t *) str_p;
+  const lit_utf8_byte_t *str_end_p = str_p + str_size;
   ecma_char_t code_unit;
 
-  while (!lit_utf8_iterator_is_eos (&iter))
+  while (str_curr_p < str_end_p)
   {
-    code_unit = lit_utf8_iterator_peek_next (&iter);
+    code_unit = lit_cesu8_peek_next (&str_curr_p);
     if (lit_char_is_white_space (code_unit) || lit_char_is_line_terminator (code_unit))
     {
-      lit_utf8_iterator_incr (&iter);
+      lit_cesu8_incr (&str_curr_p);
     }
     else
     {
       break;
     }
   }
 
-  JERRY_ASSERT (!iter.buf_pos.is_non_bmp_middle);
-  const lit_utf8_byte_t *begin_p = iter.buf_p + iter.buf_pos.offset;
+  const lit_utf8_byte_t *begin_p = str_curr_p;
+  str_curr_p = (lit_utf8_byte_t *) str_end_p;
 
-  iter = lit_utf8_iterator_create (iter.buf_p + iter.buf_pos.offset, str_size - iter.buf_pos.offset);
-  lit_utf8_iterator_seek_eos (&iter);
-  while (!lit_utf8_iterator_is_bos (&iter))
+  while (str_curr_p > str_p)
   {
-    code_unit = lit_utf8_iterator_peek_prev (&iter);
+    code_unit = lit_cesu8_peek_prev (&str_curr_p);
     if (lit_char_is_white_space (code_unit) || lit_char_is_line_terminator (code_unit))
     {
-      lit_utf8_iterator_decr (&iter);
+      lit_cesu8_decr (&str_curr_p);
     }
     else
     {
       break;
     }
   }
 
-  JERRY_ASSERT (!iter.buf_pos.is_non_bmp_middle);
-  const lit_utf8_byte_t *end_p = iter.buf_p + iter.buf_pos.offset - 1;
+  const lit_utf8_byte_t *end_p = str_curr_p - 1;
 
   if (begin_p > end_p)
   {
@@ -771,7 +769,7 @@ ecma_utf8_string_to_number (const lit_utf8_byte_t *str_p, /**< utf-8 string */
 
   return num;
 #endif /* CONFIG_ECMA_NUMBER_TYPE == CONFIG_ECMA_NUMBER_FLOAT32 */
-} /* ecma_utf8_string_to_number */
+} /* ecma_cesu8_string_to_number */
 
 /**
  * ECMA-defined conversion of UInt32 to String (zero-terminated).

jerry-core/ecma/base/ecma-helpers-string.cpp

@@ -918,7 +918,7 @@ ecma_string_to_number (const ecma_string_t *str_p) /**< ecma-string */
                                                           (ssize_t) string_size);
       JERRY_ASSERT (bytes_copied > 0);
 
-      num = ecma_utf8_string_to_number (str_buffer_p, string_size);
+      num = ecma_cesu8_string_to_number (str_buffer_p, string_size);
 
       MEM_FINALIZE_LOCAL_ARRAY (str_buffer_p);
 

jerry-core/ecma/base/ecma-helpers.h

@@ -317,7 +317,7 @@ extern void
 ecma_free_external_pointer_in_property (ecma_property_t *prop_p);
 
 /* ecma-helpers-conversion.cpp */
-extern ecma_number_t ecma_utf8_string_to_number (const lit_utf8_byte_t *str_p, lit_utf8_size_t str_size);
+extern ecma_number_t ecma_cesu8_string_to_number (const lit_utf8_byte_t *str_p, lit_utf8_size_t str_size);
 extern ssize_t ecma_uint32_to_cesu8_string (uint32_t value, lit_utf8_byte_t *out_buffer_p, ssize_t buffer_size);
 extern uint32_t ecma_number_to_uint32 (ecma_number_t value);
 extern int32_t ecma_number_to_int32 (ecma_number_t value);

jerry-core/ecma/builtin-objects/ecma-builtin-date.cpp

@@ -48,27 +48,22 @@
  * @return NaN if cannot read from string, ToNumber() otherwise
  */
 static ecma_number_t
-ecma_date_parse_date_chars (lit_utf8_iterator_t *iter, /**< iterator of the utf8 string */
+ecma_date_parse_date_chars (lit_utf8_byte_t **str_p, /**< pointer to the cesu8 string */
+                            const lit_utf8_byte_t *str_end_p, /**< pointer to the end of the string */
                             uint32_t num_of_chars) /**< number of characters to read and convert */
 {
   JERRY_ASSERT (num_of_chars > 0);
-
-  lit_utf8_size_t copy_size = 0;
-  const lit_utf8_byte_t *str_start_p = iter->buf_p + iter->buf_pos.offset;
+  const lit_utf8_byte_t *str_start_p = *str_p;
 
   while (num_of_chars--)
   {
-    if (lit_utf8_iterator_is_eos (iter)
-        || !lit_char_is_decimal_digit (lit_utf8_iterator_peek_next (iter)))
+    if (*str_p >= str_end_p || !lit_char_is_decimal_digit (lit_cesu8_read_next (str_p)))
     {
       return ecma_number_make_nan ();
     }
-
-    copy_size += lit_get_unicode_char_size_by_cesu8_first_byte (*(iter->buf_p + iter->buf_pos.offset));
-    lit_utf8_iterator_incr (iter);
   }
 
-  return ecma_utf8_string_to_number (str_start_p, copy_size);
+  return ecma_cesu8_string_to_number (str_start_p, (lit_utf8_size_t) (*str_p - str_start_p));
 } /* ecma_date_parse_date_chars */
 
 /**
@@ -211,10 +206,11 @@ ecma_builtin_date_parse (ecma_value_t this_arg __attr_unused___, /**< this argum
   ssize_t sz = ecma_string_to_cesu8_string (date_str_p, date_start_p, (ssize_t) date_str_size);
   JERRY_ASSERT (sz >= 0);
 
-  lit_utf8_iterator_t iter = lit_utf8_iterator_create (date_start_p, date_str_size);
+  lit_utf8_byte_t *date_str_curr_p = date_start_p;
+  const lit_utf8_byte_t *date_str_end_p = date_start_p + date_str_size;
 
   /* 1. read year */
-  ecma_number_t year = ecma_date_parse_date_chars (&iter, 4);
+  ecma_number_t year = ecma_date_parse_date_chars (&date_str_curr_p, date_str_end_p, 4);
 
   if (!ecma_number_is_nan (year)
       && year >= 0)
@@ -224,12 +220,12 @@ ecma_builtin_date_parse (ecma_value_t this_arg __attr_unused___, /**< this argum
     ecma_number_t time = ECMA_NUMBER_ZERO;
 
     /* 2. read month if any */
-    if (!lit_utf8_iterator_is_eos (&iter)
-        && lit_utf8_iterator_peek_next (&iter) == '-')
+    if (date_str_curr_p < date_str_end_p
+        && lit_cesu8_peek_next (&date_str_curr_p) == '-')
     {
       /* eat up '-' */
-      lit_utf8_iterator_incr (&iter);
-      month = ecma_date_parse_date_chars (&iter, 2);
+      lit_cesu8_incr (&date_str_curr_p);
+      month = ecma_date_parse_date_chars (&date_str_curr_p, date_str_end_p, 2);
 
       if (month > 12 || month < 1)
       {
@@ -238,12 +234,12 @@ ecma_builtin_date_parse (ecma_value_t this_arg __attr_unused___, /**< this argum
     }
 
     /* 3. read day if any */
-    if (!lit_utf8_iterator_is_eos (&iter)
-        && lit_utf8_iterator_peek_next (&iter) == '-')
+    if (date_str_curr_p < date_str_end_p
+        && lit_cesu8_peek_next (&date_str_curr_p) == '-')
     {
       /* eat up '-' */
-      lit_utf8_iterator_incr (&iter);
-      day = ecma_date_parse_date_chars (&iter, 2);
+      lit_cesu8_incr (&date_str_curr_p);
+      day = ecma_date_parse_date_chars (&date_str_curr_p, date_str_end_p, 2);
 
       if (day < 1 || day > 31)
       {
@@ -252,24 +248,24 @@ ecma_builtin_date_parse (ecma_value_t this_arg __attr_unused___, /**< this argum
     }
 
     /* 4. read time if any */
-    if (!lit_utf8_iterator_is_eos (&iter)
-        && lit_utf8_iterator_peek_next (&iter) == 'T')
+    if (date_str_curr_p < date_str_end_p
+        && lit_cesu8_peek_next (&date_str_curr_p) == 'T')
     {
+      /* eat up 'T' */
+      lit_cesu8_incr (&date_str_curr_p);
+
       ecma_number_t hours = ECMA_NUMBER_ZERO;
       ecma_number_t minutes = ECMA_NUMBER_ZERO;
       ecma_number_t seconds = ECMA_NUMBER_ZERO;
       ecma_number_t milliseconds = ECMA_NUMBER_ZERO;
 
-      ecma_length_t num_of_visited_chars = lit_utf8_iterator_get_index (&iter);
-      ecma_length_t date_str_len = lit_utf8_string_length (iter.buf_p, iter.buf_size) - 1;
+      ecma_length_t remaining_length = lit_utf8_string_length (date_str_curr_p,
+                                                               (lit_utf8_size_t) (date_str_end_p - date_str_curr_p));
 
-      if ((date_str_len - num_of_visited_chars) >= 5)
+      if (remaining_length >= 5)
       {
-        /* eat up 'T' */
-        lit_utf8_iterator_incr (&iter);
-
         /* 4.1 read hours and minutes */
-        hours = ecma_date_parse_date_chars (&iter, 2);
+        hours = ecma_date_parse_date_chars (&date_str_curr_p, date_str_end_p, 2);
 
         if (hours < 0 || hours > 24)
         {
@@ -281,33 +277,35 @@ ecma_builtin_date_parse (ecma_value_t this_arg __attr_unused___, /**< this argum
         }
 
         /* eat up ':' */
-        lit_utf8_iterator_incr (&iter);
+        lit_cesu8_incr (&date_str_curr_p);
 
-        minutes = ecma_date_parse_date_chars (&iter, 2);
+        minutes = ecma_date_parse_date_chars (&date_str_curr_p, date_str_end_p, 2);
 
         if (minutes < 0 || minutes > 59)
         {
           minutes = ecma_number_make_nan ();
         }
 
         /* 4.2 read seconds if any */
-        if (!lit_utf8_iterator_is_eos (&iter) && lit_utf8_iterator_peek_next (&iter) == ':')
+        if (date_str_curr_p < date_str_end_p
+            && lit_cesu8_peek_next (&date_str_curr_p) == ':')
         {
           /* eat up ':' */
-          lit_utf8_iterator_incr (&iter);
-          seconds = ecma_date_parse_date_chars (&iter, 2);
+          lit_cesu8_incr (&date_str_curr_p);
+          seconds = ecma_date_parse_date_chars (&date_str_curr_p, date_str_end_p, 2);
 
           if (seconds < 0 || seconds > 59)
           {
             seconds = ecma_number_make_nan ();
           }
 
           /* 4.3 read milliseconds if any */
-          if (!lit_utf8_iterator_is_eos (&iter) && lit_utf8_iterator_peek_next (&iter) == '.')
+          if (date_str_curr_p < date_str_end_p
+              && lit_cesu8_peek_next (&date_str_curr_p) == '.')
           {
             /* eat up '.' */
-            lit_utf8_iterator_incr (&iter);
-            milliseconds = ecma_date_parse_date_chars (&iter, 3);
+            lit_cesu8_incr (&date_str_curr_p);
+            milliseconds = ecma_date_parse_date_chars (&date_str_curr_p, date_str_end_p, 3);
 
             if (milliseconds < 0)
             {
@@ -324,34 +322,35 @@ ecma_builtin_date_parse (ecma_value_t this_arg __attr_unused___, /**< this argum
       }
 
       /* 4.4 read timezone if any */
-      if (!lit_utf8_iterator_is_eos (&iter)
-          && lit_utf8_iterator_peek_next (&iter) == 'Z'
+      if (date_str_curr_p < date_str_end_p
+          && lit_cesu8_peek_next (&date_str_curr_p) == 'Z'
           && !ecma_number_is_nan (time))
       {
-        lit_utf8_iterator_incr (&iter);
+        lit_cesu8_incr (&date_str_curr_p);
         time = ecma_date_make_time (hours, minutes, seconds, milliseconds);
       }
-      else if (!lit_utf8_iterator_is_eos (&iter)
-               && (lit_utf8_iterator_peek_next (&iter) == '+'
-                   || lit_utf8_iterator_peek_next (&iter) == '-'))
+      else if (date_str_curr_p < date_str_end_p
+               && (lit_cesu8_peek_next (&date_str_curr_p) == '+'
+                   || lit_cesu8_peek_next (&date_str_curr_p) == '-'))
       {
-        ecma_length_t num_of_visited_chars = lit_utf8_iterator_get_index (&iter);
-        ecma_length_t date_str_len = lit_utf8_string_length (iter.buf_p, iter.buf_size) - 1;
+        ecma_length_t remaining_length;
+        remaining_length = lit_utf8_string_length (date_str_curr_p,
+                                                   (lit_utf8_size_t) (date_str_end_p - date_str_curr_p)) - 1;
 
-        if ((date_str_len - num_of_visited_chars) == 5)
+        if (remaining_length == 5)
         {
           bool is_negative = false;
 
-          if (lit_utf8_iterator_peek_next (&iter) == '-')
+          if (lit_cesu8_peek_next (&date_str_curr_p) == '-')
           {
             is_negative = true;
           }
 
           /* eat up '+/-' */
-          lit_utf8_iterator_incr (&iter);
+          lit_cesu8_incr (&date_str_curr_p);
 
           /* read hours and minutes */
-          hours = ecma_date_parse_date_chars (&iter, 2);
+          hours = ecma_date_parse_date_chars (&date_str_curr_p, date_str_end_p, 2);
 
           if (hours < 0 || hours > 24)
           {
@@ -363,9 +362,9 @@ ecma_builtin_date_parse (ecma_value_t this_arg __attr_unused___, /**< this argum
           }
 
           /* eat up ':' */
-          lit_utf8_iterator_incr (&iter);
+          lit_cesu8_incr (&date_str_curr_p);
 
-          minutes = ecma_date_parse_date_chars (&iter, 2);
+          minutes = ecma_date_parse_date_chars (&date_str_curr_p, date_str_end_p, 2);
 
           if (minutes < 0 || minutes > 59)
           {
@@ -384,7 +383,7 @@ ecma_builtin_date_parse (ecma_value_t this_arg __attr_unused___, /**< this argum
       }
     }
 
-    if (lit_utf8_iterator_is_eos (&iter))
+    if (date_str_curr_p >= date_str_end_p)
     {
       ecma_number_t date = ecma_date_make_day (year, month - 1, day);
       *date_num_p = ecma_date_make_date (date, time);