Fossil SCM

upstream cson fix: output chars >0xffff as UTF16 surrogate pairs.

stephan 2015-07-02 21:43 UTC trunk
Commit 5280dbc91997819b4b566c6ce7b3eb14e6c3c710
Parent 66c3bc159d4caa9…
2 files changed +22 -8 +35 -11
~ src/cson_amalgamation.c ~ src/cson_amalgamation.h
M src/cson_amalgamation.c
+22 -8
--- src/cson_amalgamation.c
+++ src/cson_amalgamation.c
@@ -30,11 +30,11 @@
3030
 /* Determine the integer type use to parse non-floating point numbers */
3131
 #ifdef _WIN32
3232
 typedef __int64 JSON_int_t;
3333
 #define JSON_PARSER_INTEGER_SSCANF_TOKEN "%I64d"
3434
 #define JSON_PARSER_INTEGER_SPRINTF_TOKEN "%I64d"
35
-#elif __STDC_VERSION__ >= 199901L || HAVE_LONG_LONG == 1
35
+#elif (__STDC_VERSION__ >= 199901L) || (HAVE_LONG_LONG == 1)
3636
 typedef long long JSON_int_t;
3737
 #define JSON_PARSER_INTEGER_SSCANF_TOKEN "%lld"
3838
 #define JSON_PARSER_INTEGER_SPRINTF_TOKEN "%lld"
3939
 #else
4040
 typedef long JSON_int_t;
@@ -3787,11 +3787,11 @@
37873787
 unsigned char const * end = (unsigned char const *)(str ? (str + len) : NULL);
37883788
 unsigned char const * next = NULL;
37893789
 int ch;
37903790
 unsigned char clen = 0;
37913791
 char escChar[3] = {'\\',0,0};
3792
- enum { UBLen = 8 };
3792
+ enum { UBLen = 13 };
37933793
 char ubuf[UBLen];
37943794
 int rc = 0;
37953795
 rc = f(state, "\"", 1 );
37963796
 for( ; (pos < end) && (0 == rc); pos += clen )
37973797
 {
@@ -3879,17 +3879,31 @@
38793879
 { /* UTF: transform it to \uXXXX */
38803880
 #if defined(CSON_FOSSIL_MODE)
38813881
 assume_latin1:
38823882
 #endif
38833883
 memset(ubuf,0,UBLen);
3884
- rc = sprintf(ubuf, "\\u%04x",ch);
3885
- if( rc != 6 )
3886
- {
3887
- rc = cson_rc.RangeError;
3888
- break;
3884
+ if(ch <= 0xFFFF){
3885
+ rc = sprintf(ubuf, "\\u%04x",ch);
3886
+ if( rc != 6 )
3887
+ {
3888
+ rc = cson_rc.RangeError;
3889
+ break;
3890
+ }
3891
+ rc = f( state, ubuf, 6 );
3892
+ }else{ /* encode as a UTF16 surrugate pair */
3893
+ /* http://unicodebook.readthedocs.org/en/latest/unicode_encodings.html#surrogates */
3894
+ ch -= 0x10000;
3895
+ rc = sprintf(ubuf, "\\u%04x\\u%04x",
3896
+ (0xd800 | (ch>>10)),
3897
+ (0xdc00 | (ch & 0x3ff)));
3898
+ if( rc != 12 )
3899
+ {
3900
+ rc = cson_rc.RangeError;
3901
+ break;
3902
+ }
3903
+ rc = f( state, ubuf, 12 );
38893904
 }
3890
- rc = f( state, ubuf, 6 );
38913905
 continue;
38923906
 }
38933907
 }
38943908
 if( 0 == rc )
38953909
 {
38963910
--- src/cson_amalgamation.c
+++ src/cson_amalgamation.c
@@ -30,11 +30,11 @@
30 /* Determine the integer type use to parse non-floating point numbers */
31 #ifdef _WIN32
32 typedef __int64 JSON_int_t;
33 #define JSON_PARSER_INTEGER_SSCANF_TOKEN "%I64d"
34 #define JSON_PARSER_INTEGER_SPRINTF_TOKEN "%I64d"
35 #elif __STDC_VERSION__ >= 199901L || HAVE_LONG_LONG == 1
36 typedef long long JSON_int_t;
37 #define JSON_PARSER_INTEGER_SSCANF_TOKEN "%lld"
38 #define JSON_PARSER_INTEGER_SPRINTF_TOKEN "%lld"
39 #else
40 typedef long JSON_int_t;
@@ -3787,11 +3787,11 @@
3787 unsigned char const * end = (unsigned char const *)(str ? (str + len) : NULL);
3788 unsigned char const * next = NULL;
3789 int ch;
3790 unsigned char clen = 0;
3791 char escChar[3] = {'\\',0,0};
3792 enum { UBLen = 8 };
3793 char ubuf[UBLen];
3794 int rc = 0;
3795 rc = f(state, "\"", 1 );
3796 for( ; (pos < end) && (0 == rc); pos += clen )
3797 {
@@ -3879,17 +3879,31 @@
3879 { /* UTF: transform it to \uXXXX */
3880 #if defined(CSON_FOSSIL_MODE)
3881 assume_latin1:
3882 #endif
3883 memset(ubuf,0,UBLen);
3884 rc = sprintf(ubuf, "\\u%04x",ch);
3885 if( rc != 6 )
3886 {
3887 rc = cson_rc.RangeError;
3888 break;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3889 }
3890 rc = f( state, ubuf, 6 );
3891 continue;
3892 }
3893 }
3894 if( 0 == rc )
3895 {
3896
--- src/cson_amalgamation.c
+++ src/cson_amalgamation.c
@@ -30,11 +30,11 @@
30 /* Determine the integer type use to parse non-floating point numbers */
31 #ifdef _WIN32
32 typedef __int64 JSON_int_t;
33 #define JSON_PARSER_INTEGER_SSCANF_TOKEN "%I64d"
34 #define JSON_PARSER_INTEGER_SPRINTF_TOKEN "%I64d"
35 #elif (__STDC_VERSION__ >= 199901L) || (HAVE_LONG_LONG == 1)
36 typedef long long JSON_int_t;
37 #define JSON_PARSER_INTEGER_SSCANF_TOKEN "%lld"
38 #define JSON_PARSER_INTEGER_SPRINTF_TOKEN "%lld"
39 #else
40 typedef long JSON_int_t;
@@ -3787,11 +3787,11 @@
3787 unsigned char const * end = (unsigned char const *)(str ? (str + len) : NULL);
3788 unsigned char const * next = NULL;
3789 int ch;
3790 unsigned char clen = 0;
3791 char escChar[3] = {'\\',0,0};
3792 enum { UBLen = 13 };
3793 char ubuf[UBLen];
3794 int rc = 0;
3795 rc = f(state, "\"", 1 );
3796 for( ; (pos < end) && (0 == rc); pos += clen )
3797 {
@@ -3879,17 +3879,31 @@
3879 { /* UTF: transform it to \uXXXX */
3880 #if defined(CSON_FOSSIL_MODE)
3881 assume_latin1:
3882 #endif
3883 memset(ubuf,0,UBLen);
3884 if(ch <= 0xFFFF){
3885 rc = sprintf(ubuf, "\\u%04x",ch);
3886 if( rc != 6 )
3887 {
3888 rc = cson_rc.RangeError;
3889 break;
3890 }
3891 rc = f( state, ubuf, 6 );
3892 }else{ /* encode as a UTF16 surrugate pair */
3893 /* http://unicodebook.readthedocs.org/en/latest/unicode_encodings.html#surrogates */
3894 ch -= 0x10000;
3895 rc = sprintf(ubuf, "\\u%04x\\u%04x",
3896 (0xd800 | (ch>>10)),
3897 (0xdc00 | (ch & 0x3ff)));
3898 if( rc != 12 )
3899 {
3900 rc = cson_rc.RangeError;
3901 break;
3902 }
3903 rc = f( state, ubuf, 12 );
3904 }
 
3905 continue;
3906 }
3907 }
3908 if( 0 == rc )
3909 {
3910
M src/cson_amalgamation.h
+35 -11
--- src/cson_amalgamation.h
+++ src/cson_amalgamation.h
@@ -67,23 +67,47 @@
6767
 #define CSON_INT_T_PFMT "ld"
6868
 #endif
6969
7070
 /** @typedef double_or_long_double cson_double_t
7171
72
-This is the type of double value used by the library.
73
-It is only lightly tested with long double, and when using
74
-long double the memory requirements for such values goes
75
-up.
72
+ This is the type of double value used by the library.
73
+ It is only lightly tested with long double, and when using
74
+ long double the memory requirements for such values goes
75
+ up.
76
+
77
+ Note that by default cson uses C-API defaults for numeric
78
+ precision. To use a custom precision throughout the library, one
79
+ needs to define the macros CSON_DOUBLE_T_SFMT and/or
80
+ CSON_DOUBLE_T_PFMT macros to include their desired precision, and
81
+ must build BOTH cson AND the client using these same values. For
82
+ example:
83
+
84
+ @code
85
+ #define CSON_DOUBLE_T_PFMT ".8Lf" // for Modified Julian Day values
86
+ #define HAVE_LONG_DOUBLE
87
+ @endcode
88
+
89
+ (Only CSON_DOUBLE_T_PFTM should be needed for most
90
+ purposes.)
7691
 */
77
-#if 0
78
-typedef long double cson_double_t;
79
-#define CSON_DOUBLE_T_SFMT "Lf"
80
-#define CSON_DOUBLE_T_PFMT "Lf"
92
+
93
+#if defined(HAVE_LONG_DOUBLE)
94
+ typedef long double cson_double_t;
95
+# ifndef CSON_DOUBLE_T_SFMT
96
+# define CSON_DOUBLE_T_SFMT "Lf"
97
+# endif
98
+# ifndef CSON_DOUBLE_T_PFMT
99
+# define CSON_DOUBLE_T_PFMT "Lf"
100
+# endif
81101
 #else
82
-typedef double cson_double_t;
83
-#define CSON_DOUBLE_T_SFMT "f"
84
-#define CSON_DOUBLE_T_PFMT "f"
102
+ typedef double cson_double_t;
103
+# ifndef CSON_DOUBLE_T_SFMT
104
+# define CSON_DOUBLE_T_SFMT "f"
105
+# endif
106
+# ifndef CSON_DOUBLE_T_PFMT
107
+# define CSON_DOUBLE_T_PFMT "f"
108
+# endif
85109
 #endif
86110
87111
 /** @def CSON_VOID_PTR_IS_BIG
88112
89113
 ONLY define this to a true value if you know that
90114
--- src/cson_amalgamation.h
+++ src/cson_amalgamation.h
@@ -67,23 +67,47 @@
67 #define CSON_INT_T_PFMT "ld"
68 #endif
69
70 /** @typedef double_or_long_double cson_double_t
71
72 This is the type of double value used by the library.
73 It is only lightly tested with long double, and when using
74 long double the memory requirements for such values goes
75 up.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
76 */
77 #if 0
78 typedef long double cson_double_t;
79 #define CSON_DOUBLE_T_SFMT "Lf"
80 #define CSON_DOUBLE_T_PFMT "Lf"
 
 
 
 
 
81 #else
82 typedef double cson_double_t;
83 #define CSON_DOUBLE_T_SFMT "f"
84 #define CSON_DOUBLE_T_PFMT "f"
 
 
 
 
85 #endif
86
87 /** @def CSON_VOID_PTR_IS_BIG
88
89 ONLY define this to a true value if you know that
90
--- src/cson_amalgamation.h
+++ src/cson_amalgamation.h
@@ -67,23 +67,47 @@
67 #define CSON_INT_T_PFMT "ld"
68 #endif
69
70 /** @typedef double_or_long_double cson_double_t
71
72 This is the type of double value used by the library.
73 It is only lightly tested with long double, and when using
74 long double the memory requirements for such values goes
75 up.
76
77 Note that by default cson uses C-API defaults for numeric
78 precision. To use a custom precision throughout the library, one
79 needs to define the macros CSON_DOUBLE_T_SFMT and/or
80 CSON_DOUBLE_T_PFMT macros to include their desired precision, and
81 must build BOTH cson AND the client using these same values. For
82 example:
83
84 @code
85 #define CSON_DOUBLE_T_PFMT ".8Lf" // for Modified Julian Day values
86 #define HAVE_LONG_DOUBLE
87 @endcode
88
89 (Only CSON_DOUBLE_T_PFTM should be needed for most
90 purposes.)
91 */
92
93 #if defined(HAVE_LONG_DOUBLE)
94 typedef long double cson_double_t;
95 # ifndef CSON_DOUBLE_T_SFMT
96 # define CSON_DOUBLE_T_SFMT "Lf"
97 # endif
98 # ifndef CSON_DOUBLE_T_PFMT
99 # define CSON_DOUBLE_T_PFMT "Lf"
100 # endif
101 #else
102 typedef double cson_double_t;
103 # ifndef CSON_DOUBLE_T_SFMT
104 # define CSON_DOUBLE_T_SFMT "f"
105 # endif
106 # ifndef CSON_DOUBLE_T_PFMT
107 # define CSON_DOUBLE_T_PFMT "f"
108 # endif
109 #endif
110
111 /** @def CSON_VOID_PTR_IS_BIG
112
113 ONLY define this to a true value if you know that
114

Keyboard Shortcuts

Open search /
Next entry (timeline) j
Previous entry (timeline) k
Open focused entry Enter
Show this help ?
Toggle theme Top nav button