Release Version 1.4.7

Thanks @PawelWMS

CHANGELOG.md

@@ -0,0 +1,177 @@
+1.4.7
+=====
+Fixes:
+------
+* Fix `cJSONUtils_ApplyPatches`, it was completely broken and apparently nobody noticed (or at least reported it) (075a06f40bdc4f836c7dd7cad690d253a57cfc50)
+* Fix inconsistent prototype for `cJSON_GetObjectItemCaseSensitive` (51d3df6c9f7b56b860c8fb24abe7bab255cd4fa9) thanks @PawelWMS
+
+1.4.6
+=====
+Fixes:
+------
+* Several corrections in the README
+* Making clear that `valueint` should not be written to
+* Fix overflow detection in `ensure` (2683d4d9873df87c4bdccc523903ddd78d1ad250)
+* Fix a potential null pointer dereference in cJSON_Utils (795c3acabed25c9672006b2c0f40be8845064827)
+* Replace incorrect `sizeof('\0')` with `sizeof("")` (84237ff48e69825c94261c624eb0376d0c328139)
+* Add caveats section to the README (50b3c30dfa89830f8f477ce33713500740ac3b79)
+* Make cJSON locale independent (#146) Thanks @peterh for reporting
+* Fix compiling without CMake with MSVC (#147) Thanks @dertuxmalwieder for reporting
+
+1.4.5
+=====
+Fixes:
+------
+* Fix bug in `cJSON_SetNumberHelper`, thanks @mmkeeper (#138 ef34500693e8c4a2849d41a4bd66fd19c9ec46c2)
+* Workaround for internal compiler error in GCC 5.4.0 and 6.3.1 on x86 (2f65e80a3471d053fdc3f8aed23d01dd1782a5cb [GCC bugreport](https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80097))
+
+1.4.4
+=====
+Fixes:
+--------
+* Fix a theoretical integer overflow, (not sure if it is possible on actual hardware) e58f7ec027d00b7cdcbf63e518c1b5268b29b3da
+* Fix an off by one error (cc84a446be20cc283bafdc4d94c050ba1111ac02), thanks @gatzka
+* Double check the offset of the print buffer in `ensure` (1934059554b9a0971e00f79e96900f422cfdd114)
+
+Improvements:
+-------------
+* Add a note in the header about required buffer size when using `cJSON_PrintPreallocated` (4bfb88009342fb568295a7f6dc4b7fee74fbf022)
+
+1.4.3
+=====
+Fixes:
+------
+* Fix compilation of the tests on 32 bit PowerPC and potentially other systems (4ec6e76ea2eec16f54b58e8c95b4c734e59481e4)
+* Fix compilation with old GCC compilers (4.3+ were tested) (227d3398d6b967879761ebe02c1b63dbd6ea6e0d, 466eb8e3f8a65080f2b3ca4a79ab7b72bd539dba), see also #126
+
+1.4.2
+=====
+Fixes:
+------
+* Fix minimum required cmake version (30e1e7af7c63db9b55f5a3cda977a6c032f0b132)
+* Fix detection of supported compiler flags (76e5296d0d05ceb3018a9901639e0e171b44a557)
+* Run `cJSON_test` and `cJSON_test_utils` along with unity tests (c597601cf151a757dcf800548f18034d4ddfe2cb)
+
+1.4.1
+=====
+Fix: Make `print_number` abort with a failure in out of memory situations (cf1842dc6f64c49451a022308b4415e4d468be0a)
+
+1.4.0
+=====
+Features
+--------
+* Functions to check the type of an item (#120)
+* Use dllexport on windows and fvisibility on Unix systems for public functions (#116), thanks @mjerris
+* Remove trailing zeroes from printed numbers (#123)
+* Expose the internal boolean type `cJSON_bool` in the header (2d3520e0b9d0eb870e8886e8a21c571eeddbb310)
+
+Fixes
+-----
+* Fix handling of NULL pointers in `cJSON_ArrayForEach` (b47d0e34caaef298edfb7bd09a72cfff21d231ff)
+* Make it compile with GCC 7 (fix -Wimplicit-fallthrough warning) (9d07917feb1b613544a7513d19233d4c851ad7ad)
+
+Other Improvements
+------------------
+* internally use realloc if available (#110)
+* builtin support for fuzzing with [afl](http://lcamtuf.coredump.cx/afl/) (#111)
+* unit tests for the print functions (#112)
+* Always use buffered printing (#113)
+* simplify the print functions (#114)
+* Add the compiler flags `-Wdouble-conversion`, `-Wparentheses` and `-Wcomma` (#122)
+
+1.3.2
+=====
+Fix:
+----
+- Don't build the unity library if testing is disabled ( #121 ). Thanks @ffontaine
+
+1.3.1
+=====
+Bugfix release that fixes an out of bounds read #118. This shouldn't have any security implications.
+
+1.3.0
+=====
+This release includes a lot of rework in the parser and includes the Cunity unit testing framework, as well as some fixes. I increased the minor version number because there were quite a lot of internal changes.
+
+Features:
+---------
+- New type for cJSON structs: `cJSON_Invalid` (#108)
+
+Fixes:
+------
+- runtime checks for a lot of potential integer overflows
+- fix incorrect return in cJSON_PrintBuffered (cf9d57d56cac21fc59465b8d26cf29bf6d2a87b3)
+- fix several potential issues found by [Coverity](https://scan.coverity.com/projects/cjson)
+- fix potentially undefined behavior when assigning big numbers to `valueint` (41e2837df1b1091643aff073f2313f6ff3cc10f4)
+  - Numbers exceeding `INT_MAX` or lower than `INT_MIN` will be explicitly assigned to `valueint` as `INT_MAX` and `INT_MIN` respectively (saturation on overflow).
+  - fix the `cJSON_SetNumberValue` macro (87f77274de6b3af00fb9b9a7f3b900ef382296c2), this slightly changes the behavior, see commit message
+
+Introduce unit tests
+--------------------
+
+Started writing unit tests with the [Cunity](https://github.com/ThrowTheSwitch/Unity) testing framework. Currently this covers the parser functions.
+
+Also:
+- Support for running the tests with [Valgrind](http://valgrind.org)
+- Support for compiling the tests with [AddressSanitizer](https://github.com/google/sanitizers) and [UndefinedBehaviorSanitizer](https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html).
+- `travis.yml` file for running unit tests on travis. (not enabled for the repository yet though #102)
+
+Simplifications
+---------------
+
+After having unit tests for the parser function in place, I started refactoring the parser functions (as well as others) and making them easier to read and maintain.
+- Use `strtod` from the standard library for parsing numbers (074766997246481dfc72bfa78f07898a2716473f)
+- Use goto-fail in several parser functions (#100)
+- Rewrite/restructure all of the parsing functions to be easier to understand and have less code paths doing the same as another. (#109)
+- Simplify the buffer allocation strategy to always doubling the needed amount (9f6fa94c91a87b71e4c6868dbf2ce431a48517b0)
+- Combined `cJSON_AddItemToObject` and `cJSON_AddItemToObjectCS` to one function (cf862d0fed7f9407e4b046d78d3d8050d2080d12)
+
+Other changes
+-------------
+- Prevent the usage of incompatible C and header versions via preprocessor directive (123bb1af7bfae41d805337fef4b41045ef6c7d25)
+- Let CMake automatically detect compiler flags
+- Add new compiler flags (`-Wundef`, `-Wswitch-default`, `-Wconversion`, `-fstack-protector-strong`) (#98)
+- Change internal sizes from `int` to `size_t` (ecd5678527a6bc422da694e5be9e9979878fe6a0)
+- Change internal strings from `char*` to `unsigned char*` (28b9ba4334e0f7309e867e874a31f395c0ac2474)
+- Add `const` in more places
+
+1.2.1
+=====
+Fixes:
+------
+- Fixes a potential null pointer dereference in cJSON_Utils, discovered using clang's static analyzer by @bnason-nf (#96)
+
+1.2.0
+=====
+Features:
+---------
+- Add a new type of cJSON item for raw JSON and support printing it. Thanks @loigu (#65, #90)
+
+Fixes:
+------
+- Compiler warning if const is casted away, Thanks @gatzka (#83)
+- Fix compile error with strict-overflow on PowerPC, (#85)
+- Fix typo in the README, thanks @MicroJoe (#88)
+- Add compile flag for compatibility with C++ compilers
+
+1.1.0
+=====
+- Add a function `cJSON_PrintPreallocated` to print to a preallocated buffer, thanks @ChisholmKyle (#72)
+- More compiler warnings when using Clang or GCC, thanks @gatzka (#75, #78)
+- fixed a memory leak in `cJSON_Duplicate`, thanks @alperakcan (#81)
+- fix the `ENABLE_CUSTOM_COMPILER_FLAGS` cmake option
+
+1.0.2
+=====
+Rename internal boolean type, see #71.
+
+1.0.1
+=====
+Small bugfix release.
+- Fixes a bug with the use of the cJSON structs type in cJSON_Utils, see d47339e2740360e6e0994527d5e4752007480f3a
+- improve code readability
+- initialize all variables
+
+1.0.0
+=====
+This is the first official versioned release of cJSON. It provides an API version for the shared library and improved Makefile and CMake build files.

CMakeLists.txt

@@ -7,7 +7,7 @@ project(cJSON C)
 
 set(PROJECT_VERSION_MAJOR 1)
 set(PROJECT_VERSION_MINOR 4)
-set(PROJECT_VERSION_PATCH 5)
+set(PROJECT_VERSION_PATCH 7)
 set(CJSON_VERSION_SO 1)
 set(CJSON_UTILS_VERSION_SO 1)
 set(PROJECT_VERSION "${PROJECT_VERSION_MAJOR}.${PROJECT_VERSION_MINOR}.${PROJECT_VERSION_PATCH}")

Makefile

@@ -10,7 +10,7 @@ UTILS_TEST_SRC = cJSON.c cJSON_Utils.c test_utils.c
 
 LDLIBS = -lm
 
-LIBVERSION = 1.4.5
+LIBVERSION = 1.4.7
 CJSON_SOVERSION = 1
 UTILS_SOVERSION = 1
 

README.md

@@ -9,6 +9,7 @@ Ultralightweight JSON parser in ANSI C.
   * [Building](#building)
   * [Some JSON](#some-json)
   * [Here's the structure](#heres-the-structure)
+  * [Caveats](#caveats)
   * [Enjoy cJSON!](#enjoy-cjson)
 
 ## License
@@ -136,13 +137,19 @@ What's the framerate?
 
 ```c
 cJSON *format = cJSON_GetObjectItem(root, "format");
-int framerate = cJSON_GetObjectItem(format, "frame rate")->valueint;
+cJSON *framerate_item = cJSON_GetObjectItem(format, "frame rate");
+double framerate = 0;
+if (cJSON_IsNumber(framerate_item))
+{
+  framerate = framerate_item->valuedouble;
+}
 ```
 
 Want to change the framerate?
 
 ```c
-cJSON_GetObjectItem(format, "frame rate")->valueint = 25;
+cJSON *framerate_item = cJSON_GetObjectItem(format, "frame rate");
+cJSON_SetNumberValue(framerate_item, 25);
 ```
 
 Back to disk?
@@ -201,7 +208,7 @@ typedef struct cJSON {
     int type;
 
     char *valuestring;
-    int valueint;
+    int valueint; /* writing to valueint is DEPRECATED, please use cJSON_SetNumberValue instead */
     double valuedouble;
 
     char *string;
@@ -217,8 +224,7 @@ A `child` entry will have `prev == 0`, but next potentially points on. The last
 The type expresses *Null*/*True*/*False*/*Number*/*String*/*Array*/*Object*, all of which are `#defined` in
 `cJSON.h`.
 
-A *Number* has `valueint` and `valuedouble`. If you're expecting an `int`, read `valueint`, if not read
-`valuedouble`.
+A *Number* has `valueint` and `valuedouble`. `valueint` is a relict of the past, so always use `valuedouble`.
 
 Any entry which is in the linked list which is the child of an object will have a `string`
 which is the "name" of the entry. When I said "name" in the above example, that's `string`.
@@ -235,8 +241,8 @@ void parse_and_callback(cJSON *item, const char *prefix)
 {
     while (item)
     {
-        char *newprefix = malloc(strlen(prefix) + strlen(item->name) + 2);
-        sprintf(newprefix, "%s/%s", prefix, item->name);
+        char *newprefix = malloc(strlen(prefix) + strlen(item->string) + 2);
+        sprintf(newprefix, "%s/%s", prefix, item->string);
         int dorecurse = callback(newprefix, item->type, item);
         if (item->child && dorecurse)
         {
@@ -259,22 +265,22 @@ int callback(const char *name, int type, cJSON *item)
     {
         /* populate name */
     }
-    else if (!strcmp(name, "format/type")
+    else if (!strcmp(name, "format/type"))
     {
         /* handle "rect" */ }
-    else if (!strcmp(name, "format/width")
+    else if (!strcmp(name, "format/width"))
     {
         /* 800 */
     }
-    else if (!strcmp(name, "format/height")
+    else if (!strcmp(name, "format/height"))
     {
         /* 600 */
     }
-    else if (!strcmp(name, "format/interlace")
+    else if (!strcmp(name, "format/interlace"))
     {
         /* false */
     }
-    else if (!strcmp(name, "format/frame rate")
+    else if (!strcmp(name, "format/frame rate"))
     {
         /* 24 */
     }
@@ -367,6 +373,37 @@ The `test.c` code shows how to handle a bunch of typical cases. If you uncomment
 the code, it'll load, parse and print a bunch of test files, also from [json.org](http://json.org),
 which are more complex than I'd care to try and stash into a `const char array[]`.
 
+### Caveats
+
+#### Zero Character
+
+cJSON doesn't support strings that contain the zero character `'\0'` or `\u0000`. This is impossible with the current API because strings are zero terminated.
+
+#### Character Encoding
+
+cJSON only supports UTF-8 encoded input and will always produce UTF-8 as output (If the input contained invalid UTF-8, it will most likely propagate it through to the output, thereby making the output non-valid UTF-8).
+
+#### C Standard
+
+cJSON is written in ANSI C (or C89, C90). If your compiler or C library doesn't follow this standard, correct behavior is not guaranteed.
+
+NOTE: ANSI C is not C++ therefore it shouldn't be compiled by a C++ compiler. You can compile it with a C compiler and link it with your C++ code however. Although compiling with a C++ compiler might work, correct behavior is not guaranteed.
+
+#### Floating Point Numbers
+
+cJSON does not officially support any `double` implementations other than IEE754 double precision floating point numbers. It might still work with other implementations but bugs with these will be considered invalid.
+
+The maximum length of a floating point literal that cJSON supports is currently 63 characters.
+
+#### Thread Safety
+
+In general cJSON is **not thread safe**.
+
+However it is thread safe under the following conditions:
+* `cJSON_GetErrorPtr` is never used (the `return_parse_end` parameter of `cJSON_ParseWithOpts` can be used instead)
+* `cJSON_InitHooks` is only ever called before using cJSON in any threads.
+* `setlocale` is never called before all calls to cJSON functions have returned.
+
 # Enjoy cJSON!
 
 - Dave Gamble, Aug 2009

cJSON.c

@@ -31,6 +31,7 @@
 #include <float.h>
 #include <limits.h>
 #include <ctype.h>
+#include <locale.h>
 #pragma GCC visibility pop
 
 #include "cJSON.h"
@@ -47,7 +48,7 @@ CJSON_PUBLIC(const char *) cJSON_GetErrorPtr(void)
 }
 
 /* This is a safeguard to prevent copy-pasters from using incompatible C and header files */
-#if (CJSON_VERSION_MAJOR != 1) || (CJSON_VERSION_MINOR != 4) || (CJSON_VERSION_PATCH != 5)
+#if (CJSON_VERSION_MAJOR != 1) || (CJSON_VERSION_MINOR != 4) || (CJSON_VERSION_PATCH != 7)
     #error cJSON.h and cJSON.c have different versions. Make sure that both have the same.
 #endif
 
@@ -100,7 +101,7 @@ static unsigned char* cJSON_strdup(const unsigned char* str, const internal_hook
         return NULL;
     }
 
-    len = strlen((const char*)str) + 1;
+    len = strlen((const char*)str) + sizeof("");
     if (!(copy = (unsigned char*)hooks->allocate(len)))
     {
         return NULL;
@@ -177,19 +178,63 @@ CJSON_PUBLIC(void) cJSON_Delete(cJSON *c)
     }
 }
 
+/* get the decimal point character of the current locale */
+static unsigned char get_decimal_point(void)
+{
+    struct lconv *lconv = localeconv();
+    return (unsigned char) lconv->decimal_point[0];
+}
+
 /* Parse the input text to generate a number, and populate the result into item. */
 static const unsigned char *parse_number(cJSON * const item, const unsigned char * const input)
 {
     double number = 0;
     unsigned char *after_end = NULL;
+    unsigned char number_c_string[64];
+    unsigned char decimal_point = get_decimal_point();
+    size_t i = 0;
 
     if (input == NULL)
     {
         return NULL;
     }
 
-    number = strtod((const char*)input, (char**)&after_end);
-    if (input == after_end)
+    /* copy the number into a temporary buffer and replace '.' with the decimal point
+     * of the current locale (for strtod) */
+    for (i = 0; (i < (sizeof(number_c_string) - 1)) && (input[i] != '\0'); i++)
+    {
+        switch (input[i])
+        {
+            case '0':
+            case '1':
+            case '2':
+            case '3':
+            case '4':
+            case '5':
+            case '6':
+            case '7':
+            case '8':
+            case '9':
+            case '+':
+            case '-':
+            case 'e':
+            case 'E':
+                number_c_string[i] = input[i];
+                break;
+
+            case '.':
+                number_c_string[i] = decimal_point;
+                break;
+
+            default:
+                goto loop_end;
+        }
+    }
+loop_end:
+    number_c_string[i] = '\0';
+
+    number = strtod((const char*)number_c_string, (char**)&after_end);
+    if (number_c_string == after_end)
     {
         return NULL; /* parse_error */
     }
@@ -212,7 +257,7 @@ static const unsigned char *parse_number(cJSON * const item, const unsigned char
 
     item->type = cJSON_Number;
 
-    return after_end;
+    return input + (after_end - number_c_string);
 }
 
 /* don't ask me, but the original cJSON_SetNumberValue returns an integer or double */
@@ -276,7 +321,7 @@ static unsigned char* ensure(printbuffer * const p, size_t needed, const interna
     }
 
     /* calculate new buffer size */
-    if (newsize > (INT_MAX / 2))
+    if (needed > (INT_MAX / 2))
     {
         /* overflow of int, use INT_MAX if possible */
         if (needed <= INT_MAX)
@@ -336,34 +381,24 @@ static void update_offset(printbuffer * const buffer)
 }
 
 /* Removes trailing zeroes from the end of a printed number */
-static cJSON_bool trim_trailing_zeroes(printbuffer * const buffer)
+static int trim_trailing_zeroes(const unsigned char * const number, int length, const unsigned char decimal_point)
 {
-    size_t offset = 0;
-    unsigned char *content = NULL;
-
-    if ((buffer == NULL) || (buffer->buffer == NULL) || (buffer->offset < 1))
+    if ((number == NULL) || (length <= 0))
     {
-        return false;
+        return -1;
     }
 
-    offset = buffer->offset - 1;
-    content = buffer->buffer;
-
-    while ((offset > 0) && (content[offset] == '0'))
+    while ((length > 0) && (number[length - 1] == '0'))
     {
-        offset--;
+        length--;
     }
-    if ((offset > 0) && (content[offset] == '.'))
+    if ((length > 0) && (number[length - 1] == decimal_point))
     {
-        offset--;
+        /* remove trailing decimal_point */
+        length--;
     }
 
-    offset++;
-    content[offset] = '\0';
-
-    buffer->offset = offset;
-
-    return true;
+    return length;
 }
 
 /* Render the number nicely from the given item into a string. */
@@ -372,54 +407,75 @@ static cJSON_bool print_number(const cJSON * const item, printbuffer * const out
     unsigned char *output_pointer = NULL;
     double d = item->valuedouble;
     int length = 0;
-    cJSON_bool trim_zeroes = true; /* should at the end be removed? */
+    size_t i = 0;
+    cJSON_bool trim_zeroes = true; /* should zeroes at the end be removed? */
+    unsigned char number_buffer[64]; /* temporary buffer to print the number into */
+    unsigned char decimal_point = get_decimal_point();
 
     if (output_buffer == NULL)
     {
         return false;
     }
 
-    /* This is a nice tradeoff. */
-    output_pointer = ensure(output_buffer, 64, hooks);
-    if (output_pointer == NULL)
-    {
-        return false;
-    }
-
     /* This checks for NaN and Infinity */
     if ((d * 0) != 0)
     {
-        length = sprintf((char*)output_pointer, "null");
+        length = sprintf((char*)number_buffer, "null");
     }
     else if ((fabs(floor(d) - d) <= DBL_EPSILON) && (fabs(d) < 1.0e60))
     {
         /* integer */
-        length = sprintf((char*)output_pointer, "%.0f", d);
+        length = sprintf((char*)number_buffer, "%.0f", d);
         trim_zeroes = false; /* don't remove zeroes for "big integers" */
     }
     else if ((fabs(d) < 1.0e-6) || (fabs(d) > 1.0e9))
     {
-        length = sprintf((char*)output_pointer, "%e", d);
+        length = sprintf((char*)number_buffer, "%e", d);
         trim_zeroes = false; /* don't remove zeroes in engineering notation */
     }
     else
     {
-        length = sprintf((char*)output_pointer, "%f", d);
+        length = sprintf((char*)number_buffer, "%f", d);
     }
 
-    /* sprintf failed */
-    if (length < 0)
+    /* sprintf failed or buffer overrun occured */
+    if ((length < 0) || (length > (int)(sizeof(number_buffer) - 1)))
     {
         return false;
     }
 
-    output_buffer->offset += (size_t)length;
-
     if (trim_zeroes)
     {
-        return trim_trailing_zeroes(output_buffer);
+        length = trim_trailing_zeroes(number_buffer, length, decimal_point);
+        if (length <= 0)
+        {
+            return false;
+        }
     }
 
+    /* reserve appropriate space in the output */
+    output_pointer = ensure(output_buffer, (size_t)length, hooks);
+    if (output_pointer == NULL)
+    {
+        return false;
+    }
+
+    /* copy the printed number to the output and replace locale
+     * dependent decimal point with '.' */
+    for (i = 0; i < ((size_t)length); i++)
+    {
+        if (number_buffer[i] == decimal_point)
+        {
+            output_pointer[i] = '.';
+            continue;
+        }
+
+        output_pointer[i] = number_buffer[i];
+    }
+    output_pointer[i] = '\0';
+
+    output_buffer->offset += (size_t)length;
+
     return true;
 }
 
@@ -628,7 +684,7 @@ static const unsigned char *parse_string(cJSON * const item, const unsigned char
 
         /* This is at most how much we need for the output */
         allocation_length = (size_t) (input_end - input) - skipped_bytes;
-        output = (unsigned char*)hooks->allocate(allocation_length + sizeof('\0'));
+        output = (unsigned char*)hooks->allocate(allocation_length + sizeof(""));
         if (output == NULL)
         {
             goto fail; /* allocation failure */
@@ -1101,7 +1157,7 @@ static cJSON_bool print_value(const cJSON * const item, const size_t depth, cons
                 return false;
             }
 
-            raw_length = strlen(item->valuestring) + sizeof('\0');
+            raw_length = strlen(item->valuestring) + sizeof("");
             output = ensure(output_buffer, raw_length, hooks);
             if (output == NULL)
             {

cJSON.h

@@ -31,7 +31,7 @@ extern "C"
 /* project version */
 #define CJSON_VERSION_MAJOR 1
 #define CJSON_VERSION_MINOR 4
-#define CJSON_VERSION_PATCH 5
+#define CJSON_VERSION_PATCH 7
 
 #include <stddef.h>
 
@@ -63,7 +63,7 @@ typedef struct cJSON
 
     /* The item's string, if type==cJSON_String  and type == cJSON_Raw */
     char *valuestring;
-    /* The item's number, if type==cJSON_Number */
+    /* writing to valueint is DEPRECATED, use cJSON_SetNumberValue instead */
     int valueint;
     /* The item's number, if type==cJSON_Number */
     double valuedouble;
@@ -88,9 +88,10 @@ typedef int cJSON_bool;
 /* When compiling for windows, we specify a specific calling convention to avoid issues where we are being called from a project with a different default calling convention.  For windows you have 2 define options:
 
 CJSON_HIDE_SYMBOLS - Define this in the case where you don't want to ever dllexport symbols
-CJSON_EXPORT_SYMBOLS - Define this on library build when you want to dllexport symbols
+CJSON_EXPORT_SYMBOLS - Define this on library build when you want to dllexport symbols (default)
+CJSON_IMPORT_SYMBOLS - Define this if you want to dllimport symbol
 
-For *nix builds that support visibility attribute, you can define similar behavior by 
+For *nix builds that support visibility attribute, you can define similar behavior by
 
 setting default visibility to hidden by adding
 -fvisibility=hidden (for gcc)
@@ -102,11 +103,16 @@ then using the CJSON_API_VISIBILITY flag to "export" the same symbols the way CJ
 
 */
 
+/* export symbols by default, this is necessary for copy pasting the C and header file */
+#if !defined(CJSON_HIDE_SYMBOLS) && !defined(CJSON_IMPORT_SYMBOLS) && !defined(CJSON_EXPORT_SYMBOLS)
+#define CJSON_EXPORT_SYMBOLS
+#endif
+
 #if defined(CJSON_HIDE_SYMBOLS)
 #define CJSON_PUBLIC(type)   type __stdcall
 #elif defined(CJSON_EXPORT_SYMBOLS)
 #define CJSON_PUBLIC(type)   __declspec(dllexport) type __stdcall
-#else
+#elif defined(CJSON_IMPORT_SYMBOLS)
 #define CJSON_PUBLIC(type)   __declspec(dllimport) type __stdcall
 #endif
 #else /* !WIN32 */
@@ -133,7 +139,7 @@ CJSON_PUBLIC(char *) cJSON_PrintUnformatted(const cJSON *item);
 /* Render a cJSON entity to text using a buffered strategy. prebuffer is a guess at the final size. guessing well reduces reallocation. fmt=0 gives unformatted, =1 gives formatted */
 CJSON_PUBLIC(char *) cJSON_PrintBuffered(const cJSON *item, int prebuffer, cJSON_bool fmt);
 /* Render a cJSON entity to text using a buffer already allocated in memory with given length. Returns 1 on success and 0 on failure. */
-/* NOTE: If you are printing numbers, the buffer hat to be 63 bytes bigger then the printed JSON (worst case) */
+/* NOTE: cJSON is not always 100% accurate in estimating how much memory it will use, so to be safe allocate 5 bytes more than you actually need */
 CJSON_PUBLIC(cJSON_bool) cJSON_PrintPreallocated(cJSON *item, char *buffer, const int length, const cJSON_bool format);
 /* Delete a cJSON entity and all subentities. */
 CJSON_PUBLIC(void) cJSON_Delete(cJSON *c);
@@ -143,7 +149,7 @@ CJSON_PUBLIC(int) cJSON_GetArraySize(const cJSON *array);
 /* Retrieve item number "item" from array "array". Returns NULL if unsuccessful. */
 CJSON_PUBLIC(cJSON *) cJSON_GetArrayItem(const cJSON *array, int item);
 /* Get item "string" from object. Case insensitive. */
-CJSON_PUBLIC(cJSON *) cJSON_GetObjectItem(const cJSON *object, const char *string);
+CJSON_PUBLIC(cJSON *) cJSON_GetObjectItem(const cJSON * const object, const char * const string);
 CJSON_PUBLIC(cJSON *) cJSON_GetObjectItemCaseSensitive(const cJSON *object, const char *string);
 CJSON_PUBLIC(cJSON_bool) cJSON_HasObjectItem(const cJSON *object, const char *string);
 /* For analysing failed parses. This returns a pointer to the parse error. You'll probably need to look a few chars back to make sense of it. Defined when cJSON_Parse() returns 0. 0 when cJSON_Parse() succeeds. */

cJSON_Utils.c

@@ -468,7 +468,7 @@ static int cJSONUtils_ApplyPatch(cJSON *object, cJSON *patch)
     cJSONUtils_InplaceDecodePointerString(childptr);
 
     /* add, remove, replace, move, copy, test. */
-    if (!parent)
+    if ((parent == NULL) || (childptr == NULL))
     {
         /* Couldn't find object to add to. */
         free(parentptr);
@@ -504,12 +504,7 @@ CJSON_PUBLIC(int) cJSONUtils_ApplyPatches(cJSON *object, cJSON *patches)
 {
     int err = 0;
 
-    if (patches == NULL)
-    {
-        return 1;
-    }
-
-    if (cJSON_IsArray(patches))
+    if (!cJSON_IsArray(patches))
     {
         /* malformed patches. */
         return 1;

fuzzing/afl.c

@@ -56,7 +56,7 @@ static char *read_file(const char *filename)
     }
 
     /* allocate content buffer */
-    content = (char*)malloc((size_t)length + sizeof('\0'));
+    content = (char*)malloc((size_t)length + sizeof(""));
     if (content == NULL)
     {
         goto cleanup;

test.c

@@ -53,8 +53,8 @@ static int print_preallocated(cJSON *root)
     out = cJSON_Print(root);
 
     /* create buffer to succeed */
-    /* the extra 64 bytes are in case a floating point value is printed */
-    len = strlen(out) + 64;
+    /* the extra 5 bytes are because of inaccuracies when reserving memory */
+    len = strlen(out) + 5;
     buf = (char*)malloc(len);
     if (buf == NULL)
     {

tests/common.c

@@ -70,7 +70,7 @@ CJSON_PUBLIC(char*) read_file(const char *filename)
     }
 
     /* allocate content buffer */
-    content = (char*)malloc((size_t)length + sizeof('\0'));
+    content = (char*)malloc((size_t)length + sizeof(""));
     if (content == NULL)
     {
         goto cleanup;

tests/print_number.c

@@ -89,17 +89,11 @@ static void print_number_should_print_non_number(void)
 
 static void trim_trailing_zeroes_should_trim_trailing_zeroes(void)
 {
-    printbuffer buffer;
-    unsigned char number[100];
-    buffer.length = sizeof(number);
-    buffer.buffer = number;
-
-    strcpy((char*)number, "10.00");
-    buffer.offset = sizeof("10.00") - 1;
-    TEST_ASSERT_TRUE(trim_trailing_zeroes(&buffer));
-    TEST_ASSERT_EQUAL_UINT8('\0', buffer.buffer[buffer.offset]);
-    TEST_ASSERT_EQUAL_STRING("10", number);
-    TEST_ASSERT_EQUAL_UINT(sizeof("10") - 1, buffer.offset);
+    TEST_ASSERT_EQUAL_INT(2, trim_trailing_zeroes((const unsigned char*)"10.00", (int)(sizeof("10.00") - 1), '.'));
+    TEST_ASSERT_EQUAL_INT(0, trim_trailing_zeroes((const unsigned char*)".00", (int)(sizeof(".00") - 1), '.'));
+    TEST_ASSERT_EQUAL_INT(0, trim_trailing_zeroes((const unsigned char*)"00", (int)(sizeof("00") - 1), '.'));
+    TEST_ASSERT_EQUAL_INT(-1, trim_trailing_zeroes(NULL, 10, '.'));
+    TEST_ASSERT_EQUAL_INT(-1, trim_trailing_zeroes((const unsigned char*)"", 0, '.'));
 }
 
 int main(void)