oj 3.13.14 → 3.13.17

data/ext/oj/parse.c

@@ -183,6 +183,36 @@ static void unicode_to_chars(ParseInfo pi, Buf buf, uint32_t code) {
     }
 }
 
+static inline const char *scan_string_noSIMD(const char *str, const char *end) {
+    for (; '"' != *str; str++) {
+        if (end <= str || '\0' == *str || '\\' == *str) {
+            break;
+        }
+    }
+    return str;
+}
+
+#if defined(OJ_USE_SSE4_2)
+#include <nmmintrin.h>
+
+static inline const char *scan_string_SIMD(const char *str, const char *end) {
+    static const char chars[16] = "\x00\\\"";
+    const __m128i terminate = _mm_loadu_si128((const __m128i *)&chars[0]);
+    const char *_end = (const char *)(end - 16);
+
+    for (; str <= _end; str += 16) {
+        const __m128i string = _mm_loadu_si128((const __m128i *)str);
+        const int r = _mm_cmpestri(terminate, 3, string, 16, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_LEAST_SIGNIFICANT);
+        if (r != 16) {
+            str = (char*)(str + r);
+            return str;
+        }
+    }
+
+    return scan_string_noSIMD(str, end);
+}
+#endif
+
 // entered at /
 static void read_escaped_str(ParseInfo pi, const char *start) {
     struct _buf buf;
@@ -192,11 +222,15 @@ static void read_escaped_str(ParseInfo pi, const char *start) {
     Val         parent = stack_peek(&pi->stack);
 
     buf_init(&buf);
-    if (0 < cnt) {
-        buf_append_string(&buf, start, cnt);
-    }
-    for (s = pi->cur; '"' != *s; s++) {
-        if (s >= pi->end) {
+    buf_append_string(&buf, start, cnt);
+
+    for (s = pi->cur; '"' != *s;) {
+#if defined(OJ_USE_SSE4_2)
+        const char *scanned = scan_string_SIMD(s, pi->end);
+#else
+        const char *scanned = scan_string_noSIMD(s, pi->end);
+#endif
+        if (scanned >= pi->end) {
             oj_set_error_at(pi,
                             oj_parse_error_class,
                             __FILE__,
@@ -204,7 +238,12 @@ static void read_escaped_str(ParseInfo pi, const char *start) {
                             "quoted string not terminated");
             buf_cleanup(&buf);
             return;
-        } else if ('\\' == *s) {
+        }
+
+        buf_append_string(&buf, s, (size_t)(scanned - s));
+        s = scanned;
+
+        if ('\\' == *s) {
             s++;
             switch (*s) {
             case 'n': buf_append(&buf, '\n'); break;
@@ -273,8 +312,7 @@ static void read_escaped_str(ParseInfo pi, const char *start) {
                 buf_cleanup(&buf);
                 return;
             }
-        } else {
-            buf_append(&buf, *s);
+            s++;
         }
     }
     if (0 == parent) {
@@ -331,22 +369,28 @@ static void read_str(ParseInfo pi) {
     const char *str    = pi->cur;
     Val         parent = stack_peek(&pi->stack);
 
-    for (; '"' != *pi->cur; pi->cur++) {
-        if (pi->end <= pi->cur) {
-            oj_set_error_at(pi,
-                            oj_parse_error_class,
-                            __FILE__,
-                            __LINE__,
-                            "quoted string not terminated");
-            return;
-        } else if ('\0' == *pi->cur) {
-            oj_set_error_at(pi, oj_parse_error_class, __FILE__, __LINE__, "NULL byte in string");
-            return;
-        } else if ('\\' == *pi->cur) {
-            read_escaped_str(pi, str);
-            return;
-        }
+#if defined(OJ_USE_SSE4_2)
+    pi->cur = scan_string_SIMD(pi->cur, pi->end);
+#else
+    pi->cur = scan_string_noSIMD(pi->cur, pi->end);
+#endif
+    if (RB_UNLIKELY(pi->end <= pi->cur)) {
+        oj_set_error_at(pi,
+                        oj_parse_error_class,
+                        __FILE__,
+                        __LINE__,
+                        "quoted string not terminated");
+        return;
+    }
+    if (RB_UNLIKELY('\0' == *pi->cur)) {
+        oj_set_error_at(pi, oj_parse_error_class, __FILE__, __LINE__, "NULL byte in string");
+        return;
     }
+    if ('\\' == *pi->cur) {
+        read_escaped_str(pi, str);
+        return;
+    }
+
     if (0 == parent) {  // simple add
         pi->add_cstr(pi, str, pi->cur - str, str);
     } else {

data/ext/oj/parser.c

@@ -533,6 +533,7 @@ static void calc_num(ojParser p) {
         // nothing to do
         break;
     }
+    p->type = OJ_NONE;
 }
 
 static void big_change(ojParser p) {
@@ -598,6 +599,8 @@ static void parse(ojParser p, const byte *json) {
     const byte *b = json;
     int         i;
 
+    p->line = 1;
+    p->col  = -1;
 #if DEBUG
     printf("*** parse - mode: %c %s\n", p->map[256], (const char *)json);
 #endif
@@ -652,6 +655,7 @@ static void parse(ojParser p, const byte *json) {
             }
             buf_append_string(&p->buf, (const char *)start, b - start);
             if ('"' == *b) {
+                p->cur = b - json;
                 p->funcs[p->stack[p->depth]].add_str(p);
                 p->map = (0 == p->depth) ? value_map : after_map;
                 break;
@@ -661,12 +665,14 @@ static void parse(ojParser p, const byte *json) {
             p->next_map = (0 == p->depth) ? value_map : after_map;
             break;
         case OPEN_OBJECT:
+            p->cur = b - json;
             p->funcs[p->stack[p->depth]].open_object(p);
             p->depth++;
             p->stack[p->depth] = OBJECT_FUN;
             p->map             = key1_map;
             break;
         case NUM_CLOSE_OBJECT:
+            p->cur = b - json;
             calc_num(p);
             // flow through
         case CLOSE_OBJECT:
@@ -677,15 +683,18 @@ static void parse(ojParser p, const byte *json) {
                 return;
             }
             p->depth--;
+            p->cur = b - json;
             p->funcs[p->stack[p->depth]].close_object(p);
             break;
         case OPEN_ARRAY:
+            p->cur = b - json;
             p->funcs[p->stack[p->depth]].open_array(p);
             p->depth++;
             p->stack[p->depth] = ARRAY_FUN;
             p->map             = value_map;
             break;
         case NUM_CLOSE_ARRAY:
+            p->cur = b - json;
             calc_num(p);
             // flow through
         case CLOSE_ARRAY:
@@ -696,9 +705,11 @@ static void parse(ojParser p, const byte *json) {
                 return;
             }
             p->depth--;
+            p->cur = b - json;
             p->funcs[p->stack[p->depth]].close_array(p);
             break;
         case NUM_COMMA:
+            p->cur = b - json;
             calc_num(p);
             if (0 < p->depth && OBJECT_FUN == p->stack[p->depth]) {
                 p->map = key_map;
@@ -860,8 +871,14 @@ static void parse(ojParser p, const byte *json) {
             b--;
             p->map = big_exp_map;
             break;
-        case NUM_SPC: calc_num(p); break;
-        case NUM_NEWLINE: calc_num(p); b++;
+        case NUM_SPC:
+            p->cur = b - json;
+            calc_num(p);
+            break;
+        case NUM_NEWLINE:
+            p->cur = b - json;
+            calc_num(p);
+            b++;
 #ifdef SPACE_JUMP
             // for (uint32_t *sj = (uint32_t*)b; 0x20202020 == *sj; sj++) { b += 4; }
             for (uint16_t *sj = (uint16_t *)b; 0x2020 == *sj; sj++) {
@@ -882,6 +899,7 @@ static void parse(ojParser p, const byte *json) {
                 buf_append_string(&p->buf, (const char *)start, b - start);
             }
             if ('"' == *b) {
+                p->cur = b - json;
                 p->funcs[p->stack[p->depth]].add_str(p);
                 p->map = p->next_map;
                 break;
@@ -890,6 +908,7 @@ static void parse(ojParser p, const byte *json) {
             break;
         case STR_SLASH: p->map = esc_map; break;
         case STR_QUOTE:
+            p->cur = b - json;
             p->funcs[p->stack[p->depth]].add_str(p);
             p->map = p->next_map;
             break;
@@ -967,6 +986,7 @@ static void parse(ojParser p, const byte *json) {
         case VAL_NULL:
             if ('u' == b[1] && 'l' == b[2] && 'l' == b[3]) {
                 b += 3;
+                p->cur = b - json;
                 p->funcs[p->stack[p->depth]].add_null(p);
                 p->map = (0 == p->depth) ? value_map : after_map;
                 break;
@@ -992,6 +1012,7 @@ static void parse(ojParser p, const byte *json) {
         case VAL_TRUE:
             if ('r' == b[1] && 'u' == b[2] && 'e' == b[3]) {
                 b += 3;
+                p->cur = b - json;
                 p->funcs[p->stack[p->depth]].add_true(p);
                 p->map = (0 == p->depth) ? value_map : after_map;
                 break;
@@ -1017,6 +1038,7 @@ static void parse(ojParser p, const byte *json) {
         case VAL_FALSE:
             if ('a' == b[1] && 'l' == b[2] && 's' == b[3] && 'e' == b[4]) {
                 b += 4;
+                p->cur = b - json;
                 p->funcs[p->stack[p->depth]].add_false(p);
                 p->map = (0 == p->depth) ? value_map : after_map;
                 break;
@@ -1050,6 +1072,7 @@ static void parse(ojParser p, const byte *json) {
                         parse_error(p, "expected null");
                         return;
                     }
+                    p->cur = b - json;
                     p->funcs[p->stack[p->depth]].add_null(p);
                     p->map = (0 == p->depth) ? value_map : after_map;
                 }
@@ -1061,6 +1084,7 @@ static void parse(ojParser p, const byte *json) {
                         parse_error(p, "expected false");
                         return;
                     }
+                    p->cur = b - json;
                     p->funcs[p->stack[p->depth]].add_false(p);
                     p->map = (0 == p->depth) ? value_map : after_map;
                 }
@@ -1072,6 +1096,7 @@ static void parse(ojParser p, const byte *json) {
                         parse_error(p, "expected true");
                         return;
                     }
+                    p->cur = b - json;
                     p->funcs[p->stack[p->depth]].add_true(p);
                     p->map = (0 == p->depth) ? value_map : after_map;
                 }
@@ -1089,6 +1114,9 @@ static void parse(ojParser p, const byte *json) {
             p->map = trail_map;
         }
     }
+    if (0 < p->depth) {
+        parse_error(p, "parse error, not closed");
+    }
     if (0 == p->depth) {
         switch (p->map[256]) {
         case '0':
@@ -1099,7 +1127,10 @@ static void parse(ojParser p, const byte *json) {
         case 'D':
         case 'g':
         case 'B':
-        case 'Y': calc_num(p); break;
+        case 'Y':
+            p->cur = b - json;
+            calc_num(p);
+            break;
         }
     }
     return;
@@ -1456,7 +1487,7 @@ static VALUE saj_parser = Qundef;
 /* Document-method: saj
  * call-seq: saj
  *
- * Returns the default saj parser. Note the default SAJ parser can not be used
+ * Returns the default SAJ parser. Note the default SAJ parser can not be used
  * concurrently in more than one thread.
  */
 static VALUE parser_saj(VALUE self) {

data/ext/oj/parser.h

@@ -80,6 +80,7 @@ typedef struct _ojParser {
 
     char     token[8];
     long     line;
+    long     cur;  // only set before call to a function
     long     col;
     int      ri;
     uint32_t ucode;

data/ext/oj/rails.c

@@ -320,7 +320,6 @@ static void dump_time(VALUE obj, int depth, Out out, bool as_ok) {
     long long sec;
     long long nsec;
 
-#ifdef HAVE_RB_TIME_TIMESPEC
     if (16 <= sizeof(struct timespec)) {
         struct timespec ts = rb_time_timespec(obj);
 
@@ -330,10 +329,6 @@ static void dump_time(VALUE obj, int depth, Out out, bool as_ok) {
         sec  = NUM2LL(rb_funcall2(obj, oj_tv_sec_id, 0, 0));
         nsec = NUM2LL(rb_funcall2(obj, oj_tv_nsec_id, 0, 0));
     }
-#else
-    sec  = NUM2LL(rb_funcall2(obj, oj_tv_sec_id, 0, 0));
-    nsec = NUM2LL(rb_funcall2(obj, oj_tv_nsec_id, 0, 0));
-#endif
     dump_sec_nano(obj, sec, nsec, out);
 }