/* Copyright (C) 2003-2015 LiveCode Ltd.

This file is part of LiveCode.

LiveCode is free software; you can redistribute it and/or modify it under

the terms of the GNU General Public License v3 as published by the Free

Software Foundation.

LiveCode is distributed in the hope that it will be useful, but WITHOUT ANY

WARRANTY; without even the implied warranty of MERCHANTABILITY or

FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License

for more details.

You should have received a copy of the GNU General Public License

along with LiveCode. If not see <http://www.gnu.org/licenses/>. */

#include <foundation.h>

#include <foundation-auto.h>

#include <foundation-unicode.h>

#include "foundation-private.h"

#include "foundation-bidi.h"

#include "foundation-chunk.h"

#ifdef __LINUX__

#include <errno.h>

#include <iconv.h>

#include <langinfo.h>

#include <locale.h>

#endif

#ifdef __WINDOWS__

#include <Windows.h>

#endif

////////////////////////////////////////////////////////////////////////////////

// This method ensures there is 'count' chars of empty space starting at 'at'

// in 'string'. It returns false if allocation fails.

static bool __MCStringExpandAt(MCStringRef string, uindex_t at, uindex_t count);

// This method removes 'count' chars from string starting at 'at'.

static void __MCStringShrinkAt(MCStringRef string, uindex_t at, uindex_t count);

// This method clamps the given range to the valid limits for the string.

static void __MCStringClampRange(MCStringRef string, MCRange& x_range);

// This method forces a nativization of a string even if there is already a native char ptr.

static bool __MCStringNativize(MCStringRef string, uindex_t &r_char_count);

static bool __MCStringNativize(MCStringRef string);

// This method ensures there is a unichar ptr.

static bool __MCStringUnnativize(MCStringRef self);

// This method marks the string as changed.

static void __MCStringChanged(MCStringRef string, uindex_t simple = kMCStringFlagNoChange, uindex_t combined = kMCStringFlagNoChange, uindex_t native = kMCStringFlagNoChange);

// Creates an indirect mutable string with contents.

static bool __MCStringCreateIndirect(__MCString *contents, __MCString*& r_string);

// Returns true if the string is indirect.

static bool __MCStringIsIndirect(__MCString *self);

static bool __MCStringMakeImmutable(__MCString *self);

// Creates an immutable string from this one, changing 'self' to indirect.

static bool __MCStringMakeIndirect(__MCString *self);

// Ensures the given mutable but indirect string is direct.

static bool __MCStringResolveIndirect(__MCString *self);

// Makes direct mutable string indirect, referencing r_new_string.

static bool __MCStringCopyMutable(__MCString *self, __MCString*& r_new_string);

// Copy the given unicode chars into the target unicode buffer and return true

// if all the chars being copied in could be native.

static bool __MCStringCopyChars(unichar_t *target, const unichar_t *source, uindex_t count, bool target_can_be_native);

static bool MCStringSplitByDelimiterNative(MCStringRef self, MCStringRef p_elem_del, MCStringOptions p_options, MCProperListRef& r_list);

// Check the string and set CanBeNative, Basic and Trivial flags accordingly

static void __MCStringCheck(MCStringRef self);

////////////////////////////////////////////////////////////////////////////////

// AL-2015-02-06: [[ Bug 14504 ]] Add wrappers for string flag and length checking,

// for internal use when a string is known to be direct.

static bool __MCStringIsNative(MCStringRef self)

{

MCAssert(!__MCStringIsIndirect(self));

return (self -> flags & kMCStringFlagIsNotNative) == 0;

}

static bool __MCStringIsChecked(MCStringRef self)

{

MCAssert(!__MCStringIsIndirect(self));

return (self -> flags & kMCStringFlagIsChecked) != 0;

}

static bool __MCStringIsTrivial(MCStringRef self)

{

MCAssert(!__MCStringIsIndirect(self));

__MCStringCheck(self);

return (self -> flags & kMCStringFlagIsNotNative) == 0 || (self -> flags & kMCStringFlagIsTrivial) != 0;

}

static bool __MCStringIsBasic(MCStringRef self)

{

MCAssert(!__MCStringIsIndirect(self));

__MCStringCheck(self);

return (self -> flags & kMCStringFlagIsNotNative) == 0 || (self -> flags & kMCStringFlagIsBasic) != 0;

}

static bool __MCStringCanBeNative(MCStringRef self)

{

MCAssert(!__MCStringIsIndirect(self));

return (self -> flags & kMCStringFlagIsNotNative) == 0 || (self -> flags & kMCStringFlagCanBeNative) != 0;

}

static bool __MCStringCantBeEqualToNative(MCStringRef self, MCStringOptions p_options)

{

// If self can't be native, then we check the comparison options to see if

// it could still be native after normalization.

if (!__MCStringCanBeNative(self))

{

// At this point self must contain unicode characters which don't directly

// map to native. Thus the only way we could be equal to a native string is

// if we contain combining sequences which compose to a native char.

switch(p_options)

{

case kMCStringOptionCompareExact:

case kMCStringOptionCompareFolded:

// If no normalization is taking place, then no composition can occur

// so we can't be equal to native.

return true;

case kMCStringOptionCompareNonliteral:

case kMCStringOptionCompareCaseless:

// If the string has been checked then we have more information.

if (__MCStringIsChecked(self))

{

// If there are no combining chars, then normalization is not

// going to make a difference - there's no way this string

// can be native.

if (__MCStringIsTrivial(self))

return true;

// If the string is not simple, then even though it contains

// combining chars it can't be native.

if (!__MCStringIsBasic(self))

return true;

}

break;

default:

MCUnreachableReturn(false);

}

}

return false;

}

static bool __MCStringCopyChars(unichar_t *p_dst, const unichar_t *p_src, uindex_t p_count, bool p_dst_can_be_native)

{

// If the dst cannot be native, then there's no point checking if the src can be.

if (!p_dst_can_be_native)

{

MCMemoryCopy(p_dst, p_src, p_count * sizeof(unichar_t));

return false;

}

// Copy the unicode chars to our dst checking if we can nativize as we go.

for(uindex_t i = 0; i < p_count; i++)

{

// If we fail to convert the char to native, then we can't be native so

// finish up with a direct copy.

char_t t_nchar;

if (!MCUnicodeCharMapToNative(p_src[i], t_nchar))

{

MCMemoryCopy(p_dst + i, p_src + i, (p_count - i) * sizeof(unichar_t));

return false;

}

// We still copy across unicode char.

p_dst[i] = p_src[i];

}

// If we get here, then both src and dst can be native.

return true;

}

static uindex_t __MCStringGetLength(MCStringRef self)

{

MCAssert(!__MCStringIsIndirect(self));

return self -> char_count;

}

static bool __MCStringIsEmpty(MCStringRef string)

{

return string == nil || __MCStringGetLength(string) == 0;

}

////////////////////////////////////////////////////////////////////////////////

#include "foundation-string-native.cpp.h"

////////////////////////////////////////////////////////////////////////////////

// This method creates a 'constant' MCStringRef from the given c-string. At some

// point we'll make it work 'magically' at compile/build time. For now, uniquing

// and returning that has a similar effect (if slightly slower).

MC_DLLEXPORT_DEF

MCStringRef MCSTR(const char *p_cstring)

{

MCAssert(nil != p_cstring);

MCStringRef t_string;

/* UNCHECKED */ MCStringCreateWithNativeChars((const char_t *)p_cstring, strlen(p_cstring), t_string);

MCValueRef t_unique_string;

/* UNCHECKED */ MCValueInter(t_string, t_unique_string);

MCValueRelease(t_string);

return (MCStringRef)t_unique_string;

}

////////////////////////////////////////////////////////////////////////////////

MC_DLLEXPORT_DEF

bool MCStringCreateWithCString(const char* p_cstring, MCStringRef& r_string)

{

return MCStringCreateWithNativeChars((const char_t*)p_cstring, p_cstring == nil ? 0 : strlen(p_cstring), r_string);

}

MC_DLLEXPORT_DEF

bool MCStringCreateWithCStringAndRelease(char* p_cstring, MCStringRef& r_string)

{

MCAssert(nil != p_cstring);

if (MCStringCreateWithNativeChars((const char_t *)p_cstring, p_cstring == nil ? 0 : strlen((const char*)p_cstring), r_string))

{

delete[] p_cstring;

return true;

}

return false;

}

MC_DLLEXPORT_DEF

const char *MCStringGetCString(MCStringRef p_string)

{

if (p_string == nil)

return nil;

__MCAssertIsString(p_string);

MCStringNativize(p_string);

const char *t_cstring;

t_cstring = (const char *)MCStringGetNativeCharPtr(p_string);

MCAssert(t_cstring != nil);

return t_cstring;

}

MC_DLLEXPORT_DEF

bool MCStringIsEqualToCString(MCStringRef p_string, const char *p_cstring, MCStringOptions p_options)

{

__MCAssertIsString(p_string);

return MCStringIsEqualToNativeChars(p_string, (const char_t *)p_cstring, strlen(p_cstring), p_options);

}

MC_DLLEXPORT_DEF

bool MCStringSubstringIsEqualToCString(MCStringRef p_string, MCRange p_range, const char *p_cstring, MCStringOptions p_options)

{

__MCAssertIsString(p_string);

return MCStringSubstringIsEqualToNativeChars(p_string, p_range, (const char_t *)p_cstring, strlen(p_cstring), p_options);

}

// Create an immutable string from the given bytes, interpreting them using

// the specified encoding.

MC_DLLEXPORT_DEF

bool MCStringCreateWithBytes(const byte_t *p_bytes, uindex_t p_byte_count, MCStringEncoding p_encoding, bool p_is_external_rep, MCStringRef& r_string)

{

if (0 == p_byte_count)

{

if (nil != kMCEmptyString)

{

r_string = MCValueRetain(kMCEmptyString);

return true;

}

}

else

{

MCAssert(nil != p_bytes);

}

MCAssert(!p_is_external_rep);

switch (p_encoding)

{

case kMCStringEncodingASCII:

case kMCStringEncodingNative:

return MCStringCreateWithNativeChars(p_bytes, p_byte_count, r_string);

case kMCStringEncodingUTF16:

return MCStringCreateWithChars((unichar_t *)p_bytes, p_byte_count / 2, r_string);

// AL-2014-31-03: [[ Bug 12067 ]] Fix conversion from little endian bytes.

case kMCStringEncodingUTF16LE:

case kMCStringEncodingUTF16BE:

{

unichar_t *t_buffer;

uindex_t t_length = p_byte_count / 2;

if (!MCMemoryNewArray(t_length, t_buffer))

return false;

for (uindex_t i = 0; i < t_length; i++)

{

if (p_encoding == kMCStringEncodingUTF16BE)

t_buffer[i] = (unichar_t)MCSwapInt16BigToHost(((unichar_t *)p_bytes)[i]);

else

t_buffer[i] = (unichar_t)MCSwapInt16LittleToHost(((unichar_t *)p_bytes)[i]);

}

return MCStringCreateWithCharsAndRelease(t_buffer, t_length, r_string);

}

case kMCStringEncodingUTF8:

{

unichar_t *t_chars;

uindex_t t_char_count;

t_char_count = MCUnicodeCharsMapFromUTF8(p_bytes, p_byte_count, nil, 0);

if (!MCMemoryNewArray(t_char_count, t_chars))

return false;

MCUnicodeCharsMapFromUTF8(p_bytes, p_byte_count, t_chars, t_char_count);

if (!MCStringCreateWithCharsAndRelease(t_chars, t_char_count, r_string))

{

MCMemoryDeleteArray(t_chars);

return false;

}

return true;

}

case kMCStringEncodingUTF32:

case kMCStringEncodingUTF32LE:

case kMCStringEncodingUTF32BE:

{

// Round the byte count to a multiple of UTF-32 units

p_byte_count = ((p_byte_count + sizeof(uint32_t) - 1) & ~(sizeof(uint32_t) - 1));

// Convert the string to UTF-16 first.

MCAutoArray<unichar_t> t_buffer;

if (!t_buffer.Extend(p_byte_count / sizeof(uint32_t)))

return false;

uindex_t t_in_offset;

uindex_t t_out_offset = 0;

for (t_in_offset = 0; t_in_offset < p_byte_count; t_in_offset += sizeof(uint32_t))

{

// BMP characters are output unchanged, non-BMP requires surrogate pairs

codepoint_t t_codepoint;

t_codepoint = *(uint32_t*)&p_bytes[t_in_offset];

if (p_encoding == kMCStringEncodingUTF32BE)

t_codepoint = MCSwapInt32BigToHost(t_codepoint);

else if (p_encoding == kMCStringEncodingUTF32LE)

t_codepoint = MCSwapInt32LittleToHost(t_codepoint);

if (t_codepoint < 0x10000)

{

t_buffer[t_out_offset] = unichar_t(t_codepoint);

t_out_offset += 1;

}

else

{

// Split to surrogate pairs

// SN-2015-07-03: [[ Bug 15571 ]] Creating a surrogate pair

// makes the UTF-16 string longer.

if (!t_buffer . Extend(t_buffer . Size() + 1))

return false;

unichar_t t_lead, t_trail;

t_lead = unichar_t((t_codepoint - 0x10000) >> 10) + 0xD800;

t_trail = unichar_t((t_codepoint - 0x10000) & 0x3FF) + 0xDC00;

t_buffer[t_out_offset] = t_lead;

t_buffer[t_out_offset + 1] = t_trail;

t_out_offset += 2;

}

}

return MCStringCreateWithChars(t_buffer.Ptr(), t_out_offset, r_string);

}

#if !defined(__ISO_8859_1__)

case kMCStringEncodingISO8859_1:

break;

#endif

#ifndef __WINDOWS__

case kMCStringEncodingWindows1252:

break;

#endif

#if !defined(__MAC__) && !defined(__IOS__)

case kMCStringEncodingMacRoman:

break;

#endif

}

return false;

}

MC_DLLEXPORT_DEF

bool MCStringCreateWithBytesAndRelease(byte_t *p_bytes, uindex_t p_byte_count, MCStringEncoding p_encoding, bool p_is_external_rep, MCStringRef& r_string)

{

if (p_byte_count == 0)

{

if (kMCEmptyString != nil)

{

r_string = MCValueRetain(kMCEmptyString);

free(p_bytes);

return true;

}

}

else

{

MCAssert(nil != p_bytes);

}

MCStringRef t_string;

t_string = nil;

switch (p_encoding)

{

case kMCStringEncodingASCII:

case kMCStringEncodingNative:

break;

default:

if (!MCStringCreateWithBytes(p_bytes, p_byte_count, p_encoding, p_is_external_rep, t_string))

return false;

r_string = t_string;

free(p_bytes);

return true;

}

bool t_success;

t_success = true;

if (t_success)

t_success = __MCValueCreate(kMCValueTypeCodeString, t_string);

if (t_success)

t_success = MCMemoryReallocate(p_bytes, p_byte_count + 1, p_bytes);

if (t_success)

{

p_bytes[p_byte_count] = '\0';

t_string -> native_chars = p_bytes;

t_string -> char_count = p_byte_count;

r_string = t_string;

}

else

MCMemoryDelete(t_string);

return t_success;

}

////////////////////////////////////////////////////////////////////////////////

MC_DLLEXPORT_DEF

bool MCStringCreateWithChars(const unichar_t *p_chars, uindex_t p_char_count, MCStringRef& r_string)

{

if (p_char_count == 0)

{

if (kMCEmptyString != nil)

{

r_string = MCValueRetain(kMCEmptyString);

return true;

}

}

else

{

MCAssert(nil != p_chars);

}

bool t_success;

t_success = true;

__MCString *self;

self = nil;

if (t_success)

t_success = __MCValueCreate(kMCValueTypeCodeString, self);

bool t_not_native;

t_not_native = false;

if (t_success)

t_success = MCMemoryNewArray(p_char_count + 1, self -> native_chars);

if (t_success)

{

uindex_t i;

for(i = 0; i < p_char_count; i++)

if (!MCUnicodeCharMapToNative(p_chars[i], self -> native_chars[i]))

{

t_not_native = true;

break;

}

if (t_not_native)

{

MCMemoryDeleteArray(self -> native_chars);

t_success = MCMemoryNewArray(p_char_count + 1, self -> chars);

}

}

if (t_success)

{

if (t_not_native)

{

MCStrCharsMapFromUnicode(p_chars, p_char_count, self -> chars, self -> char_count);

self -> flags |= kMCStringFlagIsNotNative;

}

else

self -> char_count = p_char_count;

r_string = self;

}

else

{

if (self != nil)

MCMemoryDeleteArray(self -> chars);

MCMemoryDelete(self);

}

return t_success;

}

MC_DLLEXPORT_DEF

bool MCStringCreateWithCharsAndRelease(unichar_t *p_chars, uindex_t p_char_count, MCStringRef& r_string)

{

MCAssert(nil != p_chars);

if (MCStringCreateWithChars(p_chars, p_char_count, r_string))

{

free(p_chars);

return true;

}

return false;

}

MC_DLLEXPORT_DEF

bool MCStringCreateWithWString(const unichar_t *p_wstring, MCStringRef& r_string)

{

MCAssert(nil != p_wstring);

uindex_t t_length;

for(t_length = 0; p_wstring[t_length] != 0; t_length++)

;

return MCStringCreateWithChars(p_wstring, t_length, r_string);

}

MC_DLLEXPORT_DEF

bool MCStringCreateWithWStringAndRelease(unichar_t* p_wstring, MCStringRef& r_string)

{

MCAssert(nil != p_wstring);

if (MCStringCreateWithWString(p_wstring, r_string))

{

free(p_wstring);

return true;

}

return false;

}

MC_DLLEXPORT_DEF

bool MCStringCreateWithNativeChars(const char_t *p_chars, uindex_t p_char_count, MCStringRef& r_string)

{

bool t_success;

t_success = true;

if ((p_chars == nil || p_char_count == 0) && kMCEmptyString != nil)

{

r_string = MCValueRetain(kMCEmptyString);

return true;

}

MCAssert(nil != p_chars);

__MCString *self;

self = nil;

if (t_success)

t_success = __MCValueCreate(kMCValueTypeCodeString, self);

if (t_success)

t_success = MCMemoryNewArray(p_char_count + 1, self -> native_chars);

if (t_success)

MCMemoryCopy(self -> native_chars, p_chars, p_char_count);

else

{

if (self != nil)

MCMemoryDeleteArray(self -> native_chars);

MCMemoryDelete(self);

}

if (t_success)

{

self -> char_count = p_char_count;

r_string = self;

}

return t_success;

}

MC_DLLEXPORT_DEF

bool MCStringCreateWithNativeCharsAndRelease(char_t *p_chars, uindex_t p_char_count, MCStringRef& r_string)

{

return MCStringCreateWithNativeCharBufferAndRelease(p_chars, p_char_count, p_char_count, r_string);

}

MC_DLLEXPORT_DEF

bool MCStringCreateWithNativeCharBufferAndRelease(char_t* p_chars, uindex_t p_char_count, uindex_t p_buffer_length, MCStringRef& r_string)

{

MCAssert(nil != p_chars);

bool t_success;

t_success = true;

if (p_char_count == 0 && kMCEmptyString != nil)

{

r_string = MCValueRetain(kMCEmptyString);

MCMemoryDeallocate(p_chars);

return true;

}

__MCString *self;

self = nil;

if (t_success)

t_success = __MCValueCreate(kMCValueTypeCodeString, self);

uindex_t t_capacity = p_buffer_length;

if (t_success && t_capacity < p_char_count + 1)

{

t_capacity = p_char_count + 1;

t_success = MCMemoryReallocate(p_chars, t_capacity, p_chars);

}

if (t_success)

{

p_chars[p_char_count] = '\0';

self -> native_chars = p_chars;

self -> char_count = p_char_count;

self -> capacity = t_capacity;

r_string = self;

}

else

MCMemoryDelete(self);

return t_success;

}

static bool MCStringCreateMutableUnicode(uindex_t p_initial_capacity, MCStringRef& r_string)

{

bool t_success;

t_success = true;

__MCString *self;

self = nil;

if (t_success)

t_success = __MCValueCreate(kMCValueTypeCodeString, self);

if (t_success)

{

self -> flags |= kMCStringFlagIsNotNative;

t_success = __MCStringExpandAt(self, 0, p_initial_capacity);

}

if (t_success)

{

self -> flags |= kMCStringFlagIsMutable;

self->char_count = 0;

r_string = self;

}

else

{

MCValueRelease (self);

}

return t_success;

}

MC_DLLEXPORT_DEF

bool MCStringCreateMutable(uindex_t p_initial_capacity, MCStringRef& r_string)

{

bool t_success;

t_success = true;

__MCString *self;

self = nil;

if (t_success)

t_success = __MCValueCreate(kMCValueTypeCodeString, self);

if (t_success)

t_success = __MCStringExpandAt(self, 0, p_initial_capacity);

if (t_success)

{

self -> flags |= kMCStringFlagIsMutable;

self->char_count = 0;

r_string = self;

}

else

{

MCValueRelease (self);

}

return t_success;

}

////////////////////////////////////////////////////////////////////////////////

MC_DLLEXPORT_DEF

bool MCStringEncode(MCStringRef p_string, MCStringEncoding p_encoding, bool p_is_external_rep, MCDataRef& r_data)

{

__MCAssertIsString(p_string);

byte_t *t_bytes;

uindex_t t_byte_count;

if (!MCStringConvertToBytes(p_string, p_encoding, p_is_external_rep, t_bytes, t_byte_count))

return false;

if (!MCDataCreateWithBytesAndRelease(t_bytes, t_byte_count, r_data))

{

free(t_bytes);

return false;

}

return true;

}

MC_DLLEXPORT_DEF

bool MCStringEncodeAndRelease(MCStringRef p_string, MCStringEncoding p_encoding, bool p_is_external_rep, MCDataRef& r_data)

{

__MCAssertIsString(p_string);

MCDataRef t_data;

if (!MCStringEncode(p_string, p_encoding, p_is_external_rep, t_data))

return false;

MCValueRelease(p_string);

r_data = t_data;

return true;

}

MC_DLLEXPORT_DEF

bool MCStringDecode(MCDataRef p_data, MCStringEncoding p_encoding, bool p_is_external_rep, MCStringRef& r_string)

{

return MCStringCreateWithBytes(MCDataGetBytePtr(p_data), MCDataGetLength(p_data), p_encoding, p_is_external_rep, r_string);

}

MC_DLLEXPORT_DEF

bool MCStringDecodeAndRelease(MCDataRef p_data, MCStringEncoding p_encoding, bool p_is_external_rep, MCStringRef& r_string)

{

__MCAssertIsData(p_data);

MCStringRef t_string;

if (!MCStringDecode(p_data, p_encoding, p_is_external_rep, t_string))

return false;

MCValueRelease(p_data);

r_string = t_string;

return true;

}

// SN-2015-07-27: [[ Bug 15379 ]] This function is only used for internal

// purposes, when a LiveCode function parameter can be data - in which case

// no char translation must occur between the bytes in the DataRef and the

// Unicode string that the engine function takes (in MCR_exec for instance)

bool MCStringCreateUnicodeStringFromData(MCDataRef p_data, bool p_is_external_rep, MCStringRef& r_string)

{

__MCAssertIsData(p_data);

MCAssert(!p_is_external_rep);

if (MCDataIsEmpty(p_data))

{

r_string = MCValueRetain(kMCEmptyString);

return true;

}

bool t_success;

t_success = true;

__MCString *self;

self = nil;

if (t_success)

t_success = __MCValueCreate(kMCValueTypeCodeString, self);

uint32_t t_byte_count;

t_byte_count = MCDataGetLength(p_data);

const byte_t* t_bytes = MCDataGetBytePtr(p_data);

if (t_success)

t_success = MCMemoryNewArray(t_byte_count + 1, self -> chars);

if (t_success)

{

uindex_t i;

for(i = 0; i < t_byte_count; i++)

self -> chars[i] = t_bytes[i];

}

if (t_success)

{

self -> flags |= kMCStringFlagIsNotNative;

self -> char_count = t_byte_count;

r_string = self;

}

else

{

if (self != nil)

MCMemoryDeleteArray(self -> chars);

MCMemoryDelete(self);

}

return t_success;

}

////////////////////////////////////////////////////////////////////////////////

MC_DLLEXPORT_DEF

bool MCStringFormatV(MCStringRef& r_string, const char *p_format, va_list p_args)

{

MCAssert(nil != p_format);

MCStringRef t_buffer;

if (!MCStringCreateMutable(0, t_buffer))

return false;

bool t_success;

t_success = true;

const char *t_format_ptr;

t_format_ptr = p_format;

while (t_success && *t_format_ptr != '\0')

{

const char *t_format_start_ptr, *t_format_end_ptr;

t_format_start_ptr = t_format_end_ptr = t_format_ptr;

bool t_has_range;

t_has_range = false;

// We need to track integer, floating-point and 'long double' argument

// counts separately. The last group needs to be tracked individually as

// it is passed in different registers on x86_64 to other floating-point

// values.

//

// The integer types are themselves broken down by type, to ensure we

// always use the correct type in the call to va_arg. We don't need to

// track anything smaller than an int as the rules for '...' params say

// that parameters are promoted to int/double.

size_t t_int_arg_count = 0;

size_t t_long_arg_count = 0;

size_t t_llong_arg_count = 0;

size_t t_ptr_arg_count = 0;

size_t t_float_arg_count = 0;

size_t t_ldouble_arg_count = 0;

// Whether we are dealing with an MCValueRef or not

bool t_is_valueref = false;

// Loop until we encounter something that can't be handled by the C

// library (i.e a ValueRef formatting command) or reach the end of the

// formatting string.

while (!t_is_valueref && *t_format_ptr != '\0')

{

// Is this a formatting command?

if (*t_format_ptr != '%')

{

// Not a formatting command; continue

t_format_ptr++;

t_format_end_ptr = t_format_ptr;

}

else

{

// Move past the '%' and begin processing the formatting command

t_format_ptr++;

// Skip any flag characters at the beginning of the format

while (*t_format_ptr == '-'

|| *t_format_ptr == '+'

|| *t_format_ptr == ' '

|| *t_format_ptr == '#'

|| *t_format_ptr == '0')

{

t_format_ptr++;

}

// Skip any width specifier that is present

bool t_indirect_width = false;

if (*t_format_ptr == '*')

{

// Width will be specified via an argument of type int

t_int_arg_count++;

t_format_ptr++;

t_indirect_width = true;

}

else

{

// Skip any digits specifying the width

while (isdigit(*t_format_ptr))

t_format_ptr++;

}

// Skip any precision specifier that is present

bool t_indirect_precision = false;

if (*t_format_ptr == '.')

{

t_format_ptr++;

if (*t_format_ptr == '*')

{

// Precision will be specified via an argument of type int

t_int_arg_count++;

t_format_ptr++;

t_indirect_precision = true;

}

else

{

// Skip any digits specifying the precision

while (isdigit(*t_format_ptr))

t_format_ptr++;

}

}

// Process any size specifiers that are present

size_t* t_type_count = NULL;

switch (*t_format_ptr)

{

case 'h':

// Char or short, depending on the number of 'h' chars

t_format_ptr++;

if (*t_format_ptr == 'h')

t_format_ptr++;

t_type_count = &t_int_arg_count;

break;

case 'l':

// Long or long-long, depending on the number of 'l' chars

t_format_ptr++;

if (*t_format_ptr == 'l')

{

// Type is long-long

t_format_ptr++;

t_type_count = &t_llong_arg_count;

}

else

{

// Type is long

t_type_count = &t_long_arg_count;

}

break;

case 'j':

// intmax_t or uintmax_t - these are aliases for long

// long on all of our suppoerted platforms.

t_format_ptr++;

t_type_count = &t_llong_arg_count;

break;

case 'z':

case 't':

// size_t/ptrdiff_t - what's the underlying type?

t_format_ptr++;

if (sizeof(size_t) == sizeof(long long))

t_type_count = &t_llong_arg_count;

else if (sizeof(size_t) == sizeof(long))

t_type_count = &t_long_arg_count;

else /* if (sizeof(size_t) == sizeof(int)) */

t_type_count = &t_int_arg_count;

break;

case 'L':

// Long double