oj 3.13.23 → 3.16.9

data/ext/oj/dump.c

@@ -15,6 +15,7 @@
 #endif
 
 #include "cache8.h"
+#include "mem.h"
 #include "odd.h"
 #include "oj.h"
 #include "trace.h"
@@ -197,7 +198,18 @@ inline static long rails_xss_friendly_size(const uint8_t *str, size_t len) {
 }
 
 inline static size_t rails_friendly_size(const uint8_t *str, size_t len) {
-    return calculate_string_size(str, len, rails_friendly_chars);
+    long    size = 0;
+    size_t  i    = len;
+    uint8_t hi   = 0;
+
+    for (; 0 < i; str++, i--) {
+        size += rails_friendly_chars[*str];
+        hi |= *str & 0x80;
+    }
+    if (0 == hi) {
+        return size - len * (size_t)'0';
+    }
+    return -(size - len * (size_t)'0');
 }
 
 const char *oj_nan_str(VALUE obj, int opt, int mode, bool plus, int *lenp) {
@@ -305,8 +317,8 @@ static const char *dump_unicode(const char *str, const char *end, Out out, const
         uint32_t c1;
 
         code -= 0x00010000;
-        c1          = ((code >> 10) & 0x000003FF) + 0x0000D800;
-        code        = (code & 0x000003FF) + 0x0000DC00;
+        c1   = ((code >> 10) & 0x000003FF) + 0x0000D800;
+        code = (code & 0x000003FF) + 0x0000DC00;
         APPEND_CHARS(out->cur, "\\u", 2);
         for (i = 3; 0 <= i; i--) {
             *out->cur++ = hex_chars[(uint8_t)(c1 >> (i * 4)) & 0x0F];
@@ -463,7 +475,7 @@ void oj_dump_time(VALUE obj, Out out, int withZone) {
 }
 
 void oj_dump_ruby_time(VALUE obj, Out out) {
-    volatile VALUE rstr = rb_funcall(obj, oj_to_s_id, 0);
+    volatile VALUE rstr = oj_safe_string_convert(obj);
 
     oj_dump_cstr(RSTRING_PTR(rstr), (int)RSTRING_LEN(rstr), 0, 0, out);
 }
@@ -609,7 +621,7 @@ void oj_write_obj_to_file(VALUE obj, const char *path, Options copts) {
 
     oj_out_init(&out);
 
-    out.omit_nil  = copts->dump_opts.omit_nil;
+    out.omit_nil = copts->dump_opts.omit_nil;
     oj_dump_obj_to_json(obj, copts, &out);
     size = out.cur - out.buf;
     if (0 == (f = fopen(path, "w"))) {
@@ -657,7 +669,7 @@ void oj_write_obj_to_stream(VALUE obj, VALUE stream, Options copts) {
 
     oj_out_init(&out);
 
-    out.omit_nil  = copts->dump_opts.omit_nil;
+    out.omit_nil = copts->dump_opts.omit_nil;
     oj_dump_obj_to_json(obj, copts, &out);
     size = out.cur - out.buf;
     if (oj_stringio_class == clas) {
@@ -697,7 +709,7 @@ void oj_dump_str(VALUE obj, int depth, Out out, bool as_ok) {
 
     if (oj_utf8_encoding_index != idx) {
         rb_encoding *enc = rb_enc_from_index(idx);
-        obj = rb_str_conv_enc(obj, enc, oj_utf8_encoding);
+        obj              = rb_str_conv_enc(obj, enc, oj_utf8_encoding);
     }
     oj_dump_cstr(RSTRING_PTR(obj), (int)RSTRING_LEN(obj), 0, 0, out);
 }
@@ -726,8 +738,11 @@ static void debug_raise(const char *orig, size_t cnt, int line) {
 
 void oj_dump_raw_json(VALUE obj, int depth, Out out) {
     if (oj_string_writer_class == rb_obj_class(obj)) {
-        StrWriter sw  = (StrWriter)DATA_PTR(obj);
-        size_t    len = sw->out.cur - sw->out.buf;
+        StrWriter sw;
+        size_t    len;
+
+        sw  = oj_str_writer_unwrap(obj);
+        len = sw->out.cur - sw->out.buf;
 
         if (0 < len) {
             len--;
@@ -736,13 +751,9 @@ void oj_dump_raw_json(VALUE obj, int depth, Out out) {
     } else {
         volatile VALUE jv;
 
-        if (RB_UNLIKELY(Yes == out->opts->trace)) {
-            oj_trace("raw_json", obj, __FILE__, __LINE__, depth + 1, TraceRubyIn);
-        }
+        TRACE(out->opts->trace, "raw_json", obj, depth + 1, TraceRubyIn);
         jv = rb_funcall(obj, oj_raw_json_id, 2, RB_INT2NUM(depth), RB_INT2NUM(out->indent));
-        if (RB_UNLIKELY(Yes == out->opts->trace)) {
-            oj_trace("raw_json", obj, __FILE__, __LINE__, depth + 1, TraceRubyOut);
-        }
+        TRACE(out->opts->trace, "raw_json", obj, depth + 1, TraceRubyOut);
         oj_dump_raw(RSTRING_PTR(jv), (size_t)RSTRING_LEN(jv), out);
     }
 }
@@ -750,8 +761,9 @@ void oj_dump_raw_json(VALUE obj, int depth, Out out) {
 void oj_dump_cstr(const char *str, size_t cnt, bool is_sym, bool escape1, Out out) {
     size_t      size;
     char       *cmap;
-    const char *orig   = str;
-    bool        has_hi = false;
+    const char *orig                  = str;
+    bool        has_hi                = false;
+    bool        do_unicode_validation = false;
 
     switch (out->opts->escape_mode) {
     case NLEsc:
@@ -764,16 +776,17 @@ void oj_dump_cstr(const char *str, size_t cnt, bool is_sym, bool escape1, Out ou
         break;
     case SlashEsc:
         has_hi = true;
-        cmap = slash_friendly_chars;
-        size = slash_friendly_size((uint8_t *)str, cnt);
+        cmap   = slash_friendly_chars;
+        size   = slash_friendly_size((uint8_t *)str, cnt);
         break;
     case XSSEsc:
         cmap = xss_friendly_chars;
         size = xss_friendly_size((uint8_t *)str, cnt);
         break;
     case JXEsc:
-        cmap = hixss_friendly_chars;
-        size = hixss_friendly_size((uint8_t *)str, cnt);
+        cmap                  = hixss_friendly_chars;
+        size                  = hixss_friendly_size((uint8_t *)str, cnt);
+        do_unicode_validation = true;
         break;
     case RailsXEsc: {
         long sz;
@@ -786,12 +799,22 @@ void oj_dump_cstr(const char *str, size_t cnt, bool is_sym, bool escape1, Out ou
         } else {
             size = (size_t)sz;
         }
+        do_unicode_validation = true;
         break;
     }
-    case RailsEsc:
+    case RailsEsc: {
+        long sz;
         cmap = rails_friendly_chars;
-        size = rails_friendly_size((uint8_t *)str, cnt);
+        sz   = rails_friendly_size((uint8_t *)str, cnt);
+        if (sz < 0) {
+            has_hi = true;
+            size   = (size_t)-sz;
+        } else {
+            size = (size_t)sz;
+        }
+        do_unicode_validation = true;
         break;
+    }
     case JSONEsc:
     default: cmap = hibit_friendly_chars; size = hibit_friendly_size((uint8_t *)str, cnt);
     }
@@ -822,7 +845,7 @@ void oj_dump_cstr(const char *str, size_t cnt, bool is_sym, bool escape1, Out ou
         for (; str < end; str++) {
             switch (cmap[(uint8_t)*str]) {
             case '1':
-                if ((JXEsc == out->opts->escape_mode || RailsXEsc == out->opts->escape_mode) && check_start <= str) {
+                if (do_unicode_validation && check_start <= str) {
                     if (0 != (0x80 & (uint8_t)*str)) {
                         if (0xC0 == (0xC0 & (uint8_t)*str)) {
                             check_start = check_unicode(str, end, orig);
@@ -846,8 +869,7 @@ void oj_dump_cstr(const char *str, size_t cnt, bool is_sym, bool escape1, Out ou
                 }
                 break;
             case '3':  // Unicode
-                if (0xe2 == (uint8_t)*str && (JXEsc == out->opts->escape_mode || RailsXEsc == out->opts->escape_mode) &&
-                    2 <= end - str) {
+                if (0xe2 == (uint8_t)*str && do_unicode_validation && 2 <= end - str) {
                     if (0x80 == (uint8_t)str[1] && (0xa8 == (uint8_t)str[2] || 0xa9 == (uint8_t)str[2])) {
                         str = dump_unicode(str, end, out, orig);
                     } else {
@@ -860,11 +882,13 @@ void oj_dump_cstr(const char *str, size_t cnt, bool is_sym, bool escape1, Out ou
                 break;
             case '6':  // control characters
                 if (*(uint8_t *)str < 0x80) {
+                    if (0 == (uint8_t)*str && out->opts->dump_opts.omit_null_byte) {
+                        break;
+                    }
                     APPEND_CHARS(out->cur, "\\u00", 4);
                     dump_hex((uint8_t)*str, out);
                 } else {
-                    if (0xe2 == (uint8_t)*str &&
-                        (JXEsc == out->opts->escape_mode || RailsXEsc == out->opts->escape_mode) && 2 <= end - str) {
+                    if (0xe2 == (uint8_t)*str && do_unicode_validation && 2 <= end - str) {
                         if (0x80 == (uint8_t)str[1] && (0xa8 == (uint8_t)str[2] || 0xa9 == (uint8_t)str[2])) {
                             str = dump_unicode(str, end, out, orig);
                         } else {
@@ -881,8 +905,7 @@ void oj_dump_cstr(const char *str, size_t cnt, bool is_sym, bool escape1, Out ou
         }
         *out->cur++ = '"';
     }
-    if ((JXEsc == out->opts->escape_mode || RailsXEsc == out->opts->escape_mode) && 0 < str - orig &&
-        0 != (0x80 & *(str - 1))) {
+    if (do_unicode_validation && 0 < str - orig && 0 != (0x80 & *(str - 1))) {
         uint8_t c = (uint8_t) * (str - 1);
         int     i;
         int     scnt = (int)(str - orig);
@@ -932,7 +955,7 @@ void oj_dump_class(VALUE obj, int depth, Out out, bool as_ok) {
 }
 
 void oj_dump_obj_to_s(VALUE obj, Out out) {
-    volatile VALUE rstr = rb_funcall(obj, oj_to_s_id, 0);
+    volatile VALUE rstr = oj_safe_string_convert(obj);
 
     oj_dump_cstr(RSTRING_PTR(rstr), (int)RSTRING_LEN(rstr), 0, 0, out);
 }
@@ -944,15 +967,15 @@ void oj_dump_raw(const char *str, size_t cnt, Out out) {
 }
 
 void oj_out_init(Out out) {
-    out->buf = out->stack_buffer;
-    out->cur = out->buf;
-    out->end = out->buf + sizeof(out->stack_buffer) - BUFFER_EXTRA;
+    out->buf       = out->stack_buffer;
+    out->cur       = out->buf;
+    out->end       = out->buf + sizeof(out->stack_buffer) - BUFFER_EXTRA;
     out->allocated = false;
 }
 
 void oj_out_free(Out out) {
     if (out->allocated) {
-        xfree(out->buf); // TBD
+        OJ_R_FREE(out->buf);  // TBD
     }
 }
 
@@ -966,9 +989,9 @@ void oj_grow_out(Out out, size_t len) {
         size += len;
     }
     if (out->allocated) {
-        REALLOC_N(buf, char, (size + BUFFER_EXTRA));
+        OJ_R_REALLOC_N(buf, char, (size + BUFFER_EXTRA));
     } else {
-        buf            = ALLOC_N(char, (size + BUFFER_EXTRA));
+        buf            = OJ_R_ALLOC_N(char, (size + BUFFER_EXTRA));
         out->allocated = true;
         memcpy(buf, out->buf, out->end - out->buf + BUFFER_EXTRA);
     }
@@ -983,26 +1006,60 @@ void oj_grow_out(Out out, size_t len) {
 void oj_dump_nil(VALUE obj, int depth, Out out, bool as_ok) {
     assure_size(out, 4);
     APPEND_CHARS(out->cur, "null", 4);
-    *out->cur   = '\0';
+    *out->cur = '\0';
 }
 
 void oj_dump_true(VALUE obj, int depth, Out out, bool as_ok) {
     assure_size(out, 4);
     APPEND_CHARS(out->cur, "true", 4);
-    *out->cur   = '\0';
+    *out->cur = '\0';
 }
 
 void oj_dump_false(VALUE obj, int depth, Out out, bool as_ok) {
     assure_size(out, 5);
     APPEND_CHARS(out->cur, "false", 5);
-    *out->cur   = '\0';
+    *out->cur = '\0';
+}
+
+static const char digits_table[] = "\
+00010203040506070809\
+10111213141516171819\
+20212223242526272829\
+30313233343536373839\
+40414243444546474849\
+50515253545556575859\
+60616263646566676869\
+70717273747576777879\
+80818283848586878889\
+90919293949596979899";
+
+char *oj_longlong_to_string(long long num, bool negative, char *buf) {
+    while (100 <= num) {
+        unsigned idx = num % 100 * 2;
+        *buf--       = digits_table[idx + 1];
+        *buf--       = digits_table[idx];
+        num /= 100;
+    }
+    if (num < 10) {
+        *buf-- = num + '0';
+    } else {
+        *buf-- = digits_table[num * 2 + 1];
+        *buf-- = digits_table[num * 2];
+    }
+
+    if (negative) {
+        *buf = '-';
+    } else {
+        buf++;
+    }
+    return buf;
 }
 
 void oj_dump_fixnum(VALUE obj, int depth, Out out, bool as_ok) {
     char      buf[32];
-    char *    b              = buf + sizeof(buf) - 1;
+    char     *b              = buf + sizeof(buf) - 1;
     long long num            = NUM2LL(obj);
-    int       neg            = 0;
+    bool      neg            = false;
     size_t    cnt            = 0;
     bool      dump_as_string = false;
 
@@ -1011,7 +1068,7 @@ void oj_dump_fixnum(VALUE obj, int depth, Out out, bool as_ok) {
         dump_as_string = true;
     }
     if (0 > num) {
-        neg = 1;
+        neg = true;
         num = -num;
     }
     *b-- = '\0';
@@ -1020,14 +1077,7 @@ void oj_dump_fixnum(VALUE obj, int depth, Out out, bool as_ok) {
         *b-- = '"';
     }
     if (0 < num) {
-        for (; 0 < num; num /= 10, b--) {
-            *b = (num % 10) + '0';
-        }
-        if (neg) {
-            *b = '-';
-        } else {
-            b++;
-        }
+        b = oj_longlong_to_string(num, neg, b);
     } else {
         *b = '0';
     }
@@ -1173,7 +1223,7 @@ void oj_dump_float(VALUE obj, int depth, Out out, bool as_ok) {
     } else if (d == (double)(long long int)d) {
         cnt = snprintf(buf, sizeof(buf), "%.1f", d);
     } else if (0 == out->opts->float_prec) {
-        volatile VALUE rstr = rb_funcall(obj, oj_to_s_id, 0);
+        volatile VALUE rstr = oj_safe_string_convert(obj);
 
         cnt = (int)RSTRING_LEN(rstr);
         if ((int)sizeof(buf) <= cnt) {
@@ -1195,7 +1245,7 @@ int oj_dump_float_printf(char *buf, size_t blen, VALUE obj, double d, const char
     // Round off issues at 16 significant digits so check for obvious ones of
     // 0001 and 9999.
     if (17 <= cnt && (0 == strcmp("0001", buf + cnt - 4) || 0 == strcmp("9999", buf + cnt - 4))) {
-        volatile VALUE rstr = rb_funcall(obj, oj_to_s_id, 0);
+        volatile VALUE rstr = oj_safe_string_convert(obj);
 
         strcpy(buf, RSTRING_PTR(rstr));
         cnt = (int)RSTRING_LEN(rstr);

data/ext/oj/dump.h

@@ -93,10 +93,7 @@ inline static void dump_ulong(unsigned long num, Out out) {
 
     *b-- = '\0';
     if (0 < num) {
-        for (; 0 < num; num /= 10, b--) {
-            *b = (num % 10) + '0';
-        }
-        b++;
+        b = oj_longlong_to_string((long long)num, false, b);
     } else {
         *b = '0';
     }