Computer Shopper 242/Issue 242 - April 2008 - DPCS0408DVD.ISO/Open Source/AutoHotKey/Source/AutoHotkey104705_source.exe/source/util.h

C/C++ Source or Header | 2007-06-18 | 27.7 KB | 577 lines

/*
AutoHotkey

Copyright 2003-2007 Chris Mallett ([e-mail])

This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.

This program 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.
*/

#ifndef util_h
#define util_h

#include "stdafx.h" // pre-compiled headers
#include "defines.h"
EXTERN_G;  // For ITOA() and related functions' use of g.FormatIntAsHex

#define IS_SPACE_OR_TAB(c) (c == ' ' || c == '\t')
#define IS_SPACE_OR_TAB_OR_NBSP(c) (c == ' ' || c == '\t' || c == -96) // Use a negative to support signed chars.

// v1.0.43.04: The following are macros to avoid crash bugs caused by improper casting, namely a failure to cast
// a signed char to UCHAR before promoting it to LPSTR, which crashes since CharLower/Upper would interpret
// such a high unsigned value as an address rather than a single char.
#define ltolower(ch) CharLower((LPSTR)(UCHAR)(ch))  // "L" prefix stands for "locale", like lstrcpy.
#define ltoupper(ch) CharUpper((LPSTR)(UCHAR)(ch))  // For performance, some callers don't want return value cast to char.

// NOTE: MOVING THINGS OUT OF THIS FILE AND INTO util.cpp can hurt benchmarks by 10% or more, so be careful
// when doing so (even when the change seems inconsequential, it can impact benchmarks due to quirks of code
// generation and caching).


inline char *StrToTitleCase(char *aStr)
// Inline functions like the following probably don't actually get put inline by the compiler (since it knows
// it's not worth it).  However, changing this to be non-inline has a significant impact on benchmarks even
// though the function is never called by those benchmarks, probably due to coincidences in how the generated
// code gets cached in the CPU.  So it seems best not to mess with anything without making sure it doesn't
// drop the benchmarks significantly.
{
    if (!aStr) return aStr;
    char *aStr_orig = aStr;
    for (bool convert_next_alpha_char_to_upper = true; *aStr; ++aStr)
    {
        if (IsCharAlpha(*aStr)) // Use this to better support chars from non-English languages.
        {
            if (convert_next_alpha_char_to_upper)
            {
                *aStr = (char)ltoupper(*aStr);
                convert_next_alpha_char_to_upper = false;
            }
            else
                *aStr = (char)ltolower(*aStr);
        }
        else
            if (isspace((UCHAR)*aStr))
                convert_next_alpha_char_to_upper = true;
        // Otherwise, it's a digit, punctuation mark, etc. so nothing needs to be done.
    }
    return aStr_orig;
}



inline char *StrChrAny(char *aStr, char *aCharList)
// Returns the position of the first char in aStr that is of any one of the characters listed in aCharList.
// Returns NULL if not found.
// Update: Yes, this seems identical to strpbrk().  However, since the corresponding code would
// have to be added to the EXE regardless of which was used, there doesn't seem to be much
// advantage to switching (especially since if the two differ in behavior at all, things might
// get broken).  Another reason is the name "strpbrk()" is not as easy to remember.
{
    if (aStr == NULL || aCharList == NULL) return NULL;
    if (!*aStr || !*aCharList) return NULL;
    // Don't use strchr() because that would just find the first occurrence
    // of the first search-char, which is not necessarily the first occurrence
    // of *any* search-char:
    char *look_for_this_char, char_being_analyzed;
    for (; *aStr; ++aStr)
        // If *aStr is any of the search char's, we're done:
        for (char_being_analyzed = *aStr, look_for_this_char = aCharList; *look_for_this_char; ++look_for_this_char)
            if (char_being_analyzed == *look_for_this_char)
                return aStr;  // Match found.
    return NULL; // No match.
}



inline char *omit_leading_whitespace(char *aBuf) // 10/17/2006: __forceinline didn't help significantly.
// While aBuf points to a whitespace, moves to the right and returns the first non-whitespace
// encountered.
{
    for (; IS_SPACE_OR_TAB(*aBuf); ++aBuf);
    return aBuf;
}



inline char *omit_leading_any(char *aBuf, char *aOmitList, size_t aLength)
// Returns the address of the first character in aBuf that isn't a member of aOmitList.
// But no more than aLength characters of aBuf will be considered.  If aBuf is composed
// entirely of omitted characters, the address of the char after the last char in the
// string will returned (that char will be the zero terminator unless aLength explicitly
// caused only part of aBuf to be considered).
{
    char *cp;
    for (size_t i = 0; i < aLength; ++i, ++aBuf)
    {
        // Check if the current char is a member of the omitted-char list:
        for (cp = aOmitList; *cp; ++cp)
            if (*aBuf == *cp) // Match found.
                break;
        if (!*cp) // No match found, so this character is not omitted, thus we immediately return it's position.
            return aBuf;
    }
    // Since the above didn't return, aBuf is the position of the zero terminator or (if aLength
    // indicated only a substring) the position of the char after the last char in the substring.
    return aBuf;
}



inline char *omit_trailing_whitespace(char *aBuf, char *aBuf_marker)
// aBuf_marker must be a position in aBuf (to the right of it).
// Starts at aBuf_marker and keeps moving to the left until a non-whitespace
// char is encountered.  Returns the position of that char.
{
    for (; aBuf_marker > aBuf && IS_SPACE_OR_TAB(*aBuf_marker); --aBuf_marker);
    return aBuf_marker;  // Can equal aBuf.
}



inline size_t omit_trailing_any(char *aBuf, char *aOmitList, char *aBuf_marker)
// aBuf_marker must be a position in aBuf (to the right of it).
// Starts at aBuf_marker and keeps moving to the left until a char that isn't a member
// of aOmitList is found.  The length of the remaining substring is returned.
// That length will be zero if the string consists entirely of omitted characters.
{
    char *cp;
    for (; aBuf_marker > aBuf; --aBuf_marker)
    {
        // Check if the current char is a member of the omitted-char list:
        for (cp = aOmitList; *cp; ++cp)
            if (*aBuf_marker == *cp) // Match found.
                break;
        if (!*cp) // No match found, so this character is not omitted, thus we immediately return.
            return (aBuf_marker - aBuf) + 1; // The length of the string when trailing chars are omitted.
    }
    // Since the above didn't return, aBuf_marker is now equal to aBuf.  If this final character is itself
    // a member of the omitted-list, the length returned will be zero.  Otherwise it will be 1:
    for (cp = aOmitList; *cp; ++cp)
        if (*aBuf_marker == *cp) // Match found.
            return 0;
    return 1;
}



inline size_t ltrim(char *aStr, size_t aLength = -1)
// Caller must ensure that aStr is not NULL.
// v1.0.25: Returns the length if it was discovered as a result of the operation, or aLength otherwise.
// This greatly improves the performance of PerformAssign().
// NOTE: THIS VERSION trims only tabs and spaces.  It specifically avoids
// trimming newlines because some callers want to retain those.
{
    if (!*aStr) return 0;
    char *ptr;
    // Find the first non-whitespace char (which might be the terminator):
    for (ptr = aStr; IS_SPACE_OR_TAB(*ptr); ++ptr); // Self-contained loop.
    // v1.0.25: If no trimming needed, don't do the memmove.  This seems to make a big difference
    // in the performance of critical sections of the program such as PerformAssign():
    size_t offset;
    if (offset = ptr - aStr) // Assign.
    {
        if (aLength == -1)
            aLength = strlen(ptr); // Set aLength as new/trimmed length, for use below and also as the return value.
        else // v1.0.25.05 bug-fix: Must adjust the length provided by caller to reflect what we did here.
            aLength -= offset;
        memmove(aStr, ptr, aLength + 1); // +1 to include the '\0'.  memmove() permits source & dest to overlap.
    }
    return aLength; // This will return -1 if the block above didn't execute and caller didn't specify the length.
}

inline size_t rtrim(char *aStr, size_t aLength = -1)
// Caller must ensure that aStr is not NULL.
// To improve performance, caller may specify a length (e.g. when it is already known).
// v1.0.25: Always returns the new length of the string.  This greatly improves the performance of
// PerformAssign().
// NOTE: THIS VERSION trims only tabs and spaces.  It specifically avoids trimming newlines because
// some callers want to retain those.
{
    if (!*aStr) return 0; // The below relies upon this check having been done.
    // It's done this way in case aStr just happens to be address 0x00 (probably not possible
    // on Intel & Intel-clone hardware) because otherwise --cp would decrement, causing an
    // underflow since pointers are probably considered unsigned values, which would
    // probably cause an infinite loop.  Extremely unlikely, but might as well try
    // to be thorough:
    if (aLength == -1)
        aLength = strlen(aStr); // Set aLength for use below and also as the return value.
    for (char *cp = aStr + aLength - 1; ; --cp, --aLength)
    {
        if (!IS_SPACE_OR_TAB(*cp))
        {
            cp[1] = '\0';
            return aLength;
        }
        // Otherwise, it is a space or tab...
        if (cp == aStr) // ... and we're now at the first character of the string...
        {
            if (IS_SPACE_OR_TAB(*cp)) // ... and that first character is also a space or tab...
            {
                *cp = '\0'; // ... so the entire string is made empty...
                return 0; // Fix for v1.0.39: Must return 0 not aLength in this case.
            }
            return aLength; // ... and we return in any case.
        }
        // else it's a space or tab, and there are still more characters to check.  Let the loop
        // do its decrements.
    }
}

inline size_t rtrim_with_nbsp(char *aStr, size_t aLength = -1)
// Returns the new length of the string.
// Caller must ensure that aStr is not NULL.
// To improve performance, caller may specify a length (e.g. when it is already known).
// Same as rtrim but also gets rid of those annoying nbsp (non breaking space) chars that sometimes
// wind up on the clipboard when copied from an HTML document, and thus get pasted into the text
// editor as part of the code (such as the sample code in some of the examples).
{
    if (!*aStr) return 0; // The below relies upon this check having been done.
    if (aLength == -1)
        aLength = strlen(aStr); // Set aLength for use below and also as the return value.
    for (char *cp = aStr + aLength - 1; ; --cp, --aLength)
    {
        if (!IS_SPACE_OR_TAB_OR_NBSP(*cp))
        {
            cp[1] = '\0';
            return aLength;
        }
        if (cp == aStr)
        {
            if (IS_SPACE_OR_TAB_OR_NBSP(*cp)) // ... and that first character is also a space or tab...
            {
                *cp = '\0'; // ... so the entire string is made empty...
                return 0; // Fix for v1.0.39: Must return 0 not aLength in this case.
            }
            return aLength; // ... and we return in any case.
        }
    }
}

inline size_t trim(char *aStr, size_t aLength = -1)
// Caller must ensure that aStr is not NULL.
// Returns new length of aStr.
// To improve performance, caller may specify a length (e.g. when it is already known).
// NOTE: THIS VERSION trims only tabs and spaces.  It specifically avoids
// trimming newlines because some callers want to retain those.
{
    aLength = ltrim(aStr, aLength);  // It may return -1 to indicate that it still doesn't know the length.
    return rtrim(aStr, aLength);
    // v1.0.25: rtrim() always returns the new length of the string.  This greatly improves the
    // performance of PerformAssign() and possibly other things.
}

inline size_t strip_trailing_backslash(char *aPath)
// Removes any backslash (if there is one).
// Returns length of the new string to allow some callers to avoid another strlen() call.
{
    size_t length = strlen(aPath);
    if (!length) // Below relies on this check having been done to prevent underflow.
        return length;
    char *cp = aPath + length - 1;
    if (*cp == '\\')
    {
        *cp = '\0';
        return length - 1;
    }
    // Otherwise there no slash to remove, so return the current length.
    return length;
}


// Transformation is the same in either direction because the end bytes are swapped
// and the middle byte is left as-is:
#define bgr_to_rgb(aBGR) rgb_to_bgr(aBGR)
inline COLORREF rgb_to_bgr(DWORD aRGB)
// Fancier methods seem prone to problems due to byte alignment or compiler issues.
{
    return RGB(GetBValue(aRGB), GetGValue(aRGB), GetRValue(aRGB));
}



inline bool IsHex(char *aBuf) // 10/17/2006: __forceinline worsens performance, but physically ordering it near ATOI64() [via /ORDER] boosts by 3.5%.
// Note: AHK support for hex ints reduces performance by only 10% for decimal ints, even in the tightest
// of math loops that have SetBatchLines set to -1.
{
    // For whatever reason, omit_leading_whitespace() benches consistently faster (albeit slightly) than
    // the same code put inline (confirmed again on 10/17/2006, though the difference is hardly anything):
    //for (; IS_SPACE_OR_TAB(*aBuf); ++aBuf);
    aBuf = omit_leading_whitespace(aBuf); // i.e. caller doesn't have to have ltrimmed.
    if (!*aBuf)
        return false;
    if (*aBuf == '-' || *aBuf == '+')
        ++aBuf;
    // The "0x" prefix must be followed by at least one hex digit, otherwise it's not considered hex:
    #define IS_HEX(buf) (*buf == '0' && (*(buf + 1) == 'x' || *(buf + 1) == 'X') && isxdigit(*(buf + 2)))
    return IS_HEX(aBuf);
}



// As of v1.0.30, ATOI(), ITOA() and the other related functions below are no longer macros
// because there are too many places where something like ATOI(++cp) is done, which would be a
// bug if not caught since cp would be incremented more than once if the macro referred to that
// arg more than once.  In addition, a non-comprehensive, simple benchmark shows that the
// macros don't perform any better anyway, probably in part because there are many times when
// something like ATOI(ARG1) is called, which forces the ARG1 macro to be expanded two or more
// times within ATOI (when it was a macro).  So for now, the below are declared as inline.
// However, it seems that the compiler chooses not to make them truly inline, which as it
// turns out is probably the right decision since a simple benchmark shows that even with
// __forceinline in effect for all of them (which is confirmed to actually force inline),
// the performance isn't any better.

inline __int64 ATOI64(char *buf)
// The following comment only applies if the code is a macro or actually put inline by the compiler,
// which is no longer true:
// A more complex macro is used for ATOI64(), since it is more often called from places where
// performance matters (e.g. ACT_ADD).  It adds about 500 bytes to the code size  in exchance for
// a 8% faster math loops.  But it's probably about 8% slower when used with hex integers, but
// those are so rare that the speed-up seems worth the extra code size:
//#define ATOI64(buf) _strtoi64(buf, NULL, 0) // formerly used _atoi64()
{
    return IsHex(buf) ? _strtoi64(buf, NULL, 16) : _atoi64(buf);  // _atoi64() has superior performance, so use it when possible.
}

inline unsigned __int64 ATOU64(char *buf)
{
    return _strtoui64(buf, NULL, IsHex(buf) ? 16 : 10);
}

inline int ATOI(char *buf)
{
    // Below has been updated because values with leading zeros were being intepreted as
    // octal, which is undesirable.
    // Formerly: #define ATOI(buf) strtol(buf, NULL, 0) // Use zero as last param to support both hex & dec.
    return IsHex(buf) ? strtol(buf, NULL, 16) : atoi(buf); // atoi() has superior performance, so use it when possible.
}

// v1.0.38.01: Make ATOU a macro that refers to ATOI64() to improve performance (takes advantage of _atoi64()
// being considerably faster than strtoul(), at least when the number is non-hex).  This relies on the fact
// that ATOU() and (UINT)ATOI64() produce the same result due to the way casting works.  For example:
// ATOU("-1") == (UINT)ATOI64("-1")
// ATOU("-0xFFFFFFFF") == (UINT)ATOI64("-0xFFFFFFFF")
#define ATOU(buf) (UINT)ATOI64(buf)
//inline unsigned long ATOU(char *buf)
//{
//    // As a reminder, strtoul() also handles negative numbers.  For example, ATOU("-1") is
//    // 4294967295 (0xFFFFFFFF) and ATOU("-2") is 4294967294.
//    return strtoul(buf, NULL, IsHex(buf) ? 16 : 10);
//}

inline double ATOF(char *buf)
// Unlike some Unix versions of strtod(), the VC++ version does not seem to handle hex strings
// such as "0xFF" automatically.  So this macro must check for hex because some callers rely on that.
// Also, it uses _strtoi64() vs. strtol() so that more of a double's capacity can be utilized:
{
    return IsHex(buf) ? (double)_strtoi64(buf, NULL, 16) : atof(buf);
}

inline char *ITOA(int value, char *buf)
{
    if (g.FormatIntAsHex)
    {
        char *our_buf_temp = buf;
        // Negative hex numbers need special handling, otherwise something like zero minus one would create
        // a huge 0xffffffffffffffff value, which would subsequently not be read back in correctly as
        // a negative number (but UTOA() doesn't need this since there can't be negatives in that case).
        if (value < 0)
            *our_buf_temp++ = '-';
        *our_buf_temp++ = '0';
        *our_buf_temp++ = 'x';
        _itoa(value < 0 ? -(int)value : value, our_buf_temp, 16);
        // Must not return the result of the above because it's our_buf_temp and we want buf.
        return buf;
    }
    else
        return _itoa(value, buf, 10);
}

inline char *ITOA64(__int64 value, char *buf)
{
    if (g.FormatIntAsHex)
    {
        char *our_buf_temp = buf;
        if (value < 0)
            *our_buf_temp++ = '-';
        *our_buf_temp++ = '0';
        *our_buf_temp++ = 'x';
        _i64toa(value < 0 ? -(__int64)value : value, our_buf_temp, 16);
        // Must not return the result of the above because it's our_buf_temp and we want buf.
        return buf;
    }
    else
        return _i64toa(value, buf, 10);
}

inline char *UTOA(unsigned long value, char *buf)
{
    if (g.FormatIntAsHex)
    {
        *buf = '0';
        *(buf + 1) = 'x';
        _ultoa(value, buf + 2, 16);
        // Must not return the result of the above because it's buf + 2 and we want buf.
        return buf;
    }
    else
        return _ultoa(value, buf, 10);
}

// Not currently used:
//inline char *UTOA64(unsigned __int64 value, char *buf)
//{
//    if (g.FormatIntAsHex)
//    {
//        *buf = '0';
//        *(buf + 1) = 'x';
//        return _ui64toa(value, buf + 2, 16);
//    }
//    else
//        return _ui64toa(value, buf, 10);
//}



//inline char *strcatmove(char *aDst, char *aSrc)
//// Same as strcat() but allows aSrc and aDst to overlap.
//// Unlike strcat(), it doesn't return aDst.  Instead, it returns the position
//// in aDst where aSrc was appended.
//{
//    if (!aDst || !aSrc || !*aSrc) return aDst;
//    char *aDst_end = aDst + strlen(aDst);
//    return (char *)memmove(aDst_end, aSrc, strlen(aSrc) + 1);  // Add 1 to include aSrc's terminator.
//}



// v1.0.43.03: The following macros support the new "StringCaseSense Locale" setting.  This setting performs
// 1 to 10 times slower for most things, but has the benefit of seeing characters like Σ and ─ as identical
// when insensitive.  MSDN implies that lstrcmpi() is the same as:
//     CompareString(LOCALE_USER_DEFAULT, NORM_IGNORECASE, ...)
// Note that when MSDN talks about the "word sort" vs. "string sort", it does not mean that strings like
// "co-op" and "co-op" are considered equal.  Instead, they are considered closer together than the traditional
// string sort would see them, so that they wind up together in a sorted list.
// And both of them benchmark the same, so lstrcmpi is now used here and in various other places throughout
// the program when the new locale-case-insensitive mode is in effect.
#define strcmp2(str1, str2, string_case_sense) ((string_case_sense) == SCS_INSENSITIVE ? stricmp(str1, str2) \
    : ((string_case_sense) == SCS_INSENSITIVE_LOCALE ? lstrcmpi(str1, str2) : strcmp(str1, str2)))
#define g_strcmp(str1, str2) strcmp2(str1, str2, g.StringCaseSense)
// The most common mode is listed first for performance:
#define strstr2(haystack, needle, string_case_sense) ((string_case_sense) == SCS_INSENSITIVE ? strcasestr(haystack, needle) \
    : ((string_case_sense) == SCS_INSENSITIVE_LOCALE ? lstrcasestr(haystack, needle) : strstr(haystack, needle)))
#define g_strstr(haystack, needle) strstr2(haystack, needle, g.StringCaseSense)
// For the following, caller must ensure that len1 and len2 aren't beyond the terminated length of the string
// because CompareString() might not stop at the terminator when a length is specified.  Also, CompareString()
// returns 0 on failure, but failure occurs only when parameter/flag is invalid, which should never happen in
// this case.
#define lstrcmpni(str1, len1, str2, len2) (CompareString(LOCALE_USER_DEFAULT, NORM_IGNORECASE, str1, (int)(len1), str2, (int)(len2)) - 2) // -2 for maintainability

// The following macros simplify and make consistent the calls to MultiByteToWideChar().
// MSDN implies that passing -1 for cbMultiByte is the most typical and secure usage because it ensures
// that the output is null-terminated: "the resulting wide character string has a null terminator, and the
// length returned by the function includes the terminating null character."
//
// I couldn't find any info on when MB_PRECOMPOSED is needed (if ever).  It's the default anyway,
// which implies that passing zero (which is quite common in many examples I've seen) is essentially
// the same as passing MB_PRECOMPOSED.  However, some modes such as CP_UTF8 should never use MB_PRECOMPOSED
// or the function will fail.
//
// #1: FROM ANSI TO UNICODE (UTF-16).  dest_size_in_wchars includes the terminator.
// From looking at the source to mbstowcs(), it might be faster when the "C" locale is in effect (which is
// the default in the absence of setlocale()) than MultiByteToWideChar() depending on how the latter is
// implemented. This is because mbstowcs() simply casts the characters to (wchar_t)(unsigned char) without
// any other translation at all.  Although that behavior is probably identical to MultiByteToWideChar(CP_ACP...),
// it's not completely certain -- so it seems best to stick with MultiByteToWideChar() for consistency
// (also, avoiding mbstowcs slightly reduces code size).  If there's ever a case where performance is
// important, create a simple casting loop (see mbstowcs.c for an example) that converts source to dest,
// and test if it performs significantly better than MultiByteToWideChar(CP_ACP...).
#define ToWideChar(source, dest, dest_size_in_wchars) MultiByteToWideChar(CP_ACP, 0, source, -1, dest, dest_size_in_wchars)
//
// #2: FROM UTF-8 TO UNICODE (UTF-16). dest_size_in_wchars includes the terminator.  MSDN: "For UTF-8, dwFlags must be set to either 0 or MB_ERR_INVALID_CHARS. Otherwise, the function fails with ERROR_INVALID_FLAGS."
#define UTF8ToWideChar(source, dest, dest_size_in_wchars) MultiByteToWideChar(CP_UTF8, 0, source, -1, dest, dest_size_in_wchars)
//
// #3: FROM UNICODE (UTF-16) TO UTF-8. dest_size_in_bytes includes the terminator.
#define WideCharToUTF8(source, dest, dest_size_in_bytes) WideCharToMultiByte(CP_UTF8, 0, source, -1, dest, dest_size_in_bytes, NULL, NULL)

// v1.0.44.03: Callers now use the following macro rather than the old approach.  However, this change
// is meaningful only to people who use more than one keyboard layout.  In the case of hotstrings:
// It seems that the vast majority of them would want the Hotstring monitoring to adhere to the active
// window's current keyboard layout rather than the script's.  This change is somewhat less certain to
// be desirable uncondionally for the Input command (especially invisible/non-V-option Inputs); but it
// seems best to use the same approach to avoid calling ToAsciiEx() more than once in cases where a
// script has hotstrings and also uses the Input command. Calling ToAsciiEx() twice in such a case would
// be likely to aggravate its side effects with dead keys as described at length in the hook/Input code).
#define Get_active_window_keybd_layout \
    HWND active_window;\
    HKL active_window_keybd_layout = GetKeyboardLayout((active_window = GetForegroundWindow())\
        ? GetWindowThreadProcessId(active_window, NULL) : 0); // When no foreground window, the script's own layout seems like the safest default.

#define DATE_FORMAT_LENGTH 14 // "YYYYMMDDHHMISS"
#define IS_LEAP_YEAR(year) ((year) % 4 == 0 && ((year) % 100 != 0 || (year) % 400 == 0))

int GetYDay(int aMon, int aDay, bool aIsLeapYear);
int GetISOWeekNumber(char *aBuf, int aYear, int aYDay, int aWDay);
ResultType YYYYMMDDToFileTime(char *aYYYYMMDD, FILETIME &aFileTime);
DWORD YYYYMMDDToSystemTime2(char *aYYYYMMDD, SYSTEMTIME *aSystemTime);
ResultType YYYYMMDDToSystemTime(char *aYYYYMMDD, SYSTEMTIME &aSystemTime, bool aDoValidate);
char *FileTimeToYYYYMMDD(char *aBuf, FILETIME &aTime, bool aConvertToLocalTime = false);
char *SystemTimeToYYYYMMDD(char *aBuf, SYSTEMTIME &aTime);
__int64 YYYYMMDDSecondsUntil(char *aYYYYMMDDStart, char *aYYYYMMDDEnd, bool &aFailed);
__int64 FileTimeSecondsUntil(FILETIME *pftStart, FILETIME *pftEnd);

SymbolType IsPureNumeric(char *aBuf, BOOL aAllowNegative = false // BOOL vs. bool might squeeze a little more performance out of this frequently-called function.
    , BOOL aAllowAllWhitespace = true, BOOL aAllowFloat = false, BOOL aAllowImpure = false);

void strlcpy(char *aDst, const char *aSrc, size_t aDstSize);
int snprintf(char *aBuf, int aBufSize, const char *aFormat, ...);
int snprintfcat(char *aBuf, int aBufSize, const char *aFormat, ...);
// Not currently used by anything, so commented out to possibly reduce code size:
//int strlcmp (char *aBuf1, char *aBuf2, UINT aLength1 = UINT_MAX, UINT aLength2 = UINT_MAX);
int strlicmp(char *aBuf1, char *aBuf2, UINT aLength1 = UINT_MAX, UINT aLength2 = UINT_MAX);
char *strrstr(char *aStr, char *aPattern, StringCaseSenseType aStringCaseSense, int aOccurrence = 1);
char *lstrcasestr(const char *phaystack, const char *pneedle);
char *strcasestr (const char *phaystack, const char *pneedle);
UINT StrReplace(char *aHaystack, char *aOld, char *aNew, StringCaseSenseType aStringCaseSense
    , UINT aLimit = UINT_MAX, size_t aSizeLimit = -1, char **aDest = NULL, size_t *aHaystackLength = NULL);
int PredictReplacementSize(int aLengthDelta, int aReplacementCount, int aLimit, int aHaystackLength
    , int aCurrentLength, int aEndOffsetOfCurrMatch);
char *TranslateLFtoCRLF(char *aString);
bool DoesFilePatternExist(char *aFilePattern, DWORD *aFileAttr = NULL);
#ifdef _DEBUG
    ResultType FileAppend(char *aFilespec, char *aLine, bool aAppendNewline = true);
#endif
char *ConvertFilespecToCorrectCase(char *aFullFileSpec);
char *FileAttribToStr(char *aBuf, DWORD aAttr);
unsigned __int64 GetFileSize64(HANDLE aFileHandle);
char *GetLastErrorText(char *aBuf, int aBufSize, bool aUpdateLastError = false);
void AssignColor(char *aColorName, COLORREF &aColor, HBRUSH &aBrush);
COLORREF ColorNameToBGR(char *aColorName);
HRESULT MySetWindowTheme(HWND hwnd, LPCWSTR pszSubAppName, LPCWSTR pszSubIdList);
//HRESULT MyEnableThemeDialogTexture(HWND hwnd, DWORD dwFlags);
char *ConvertEscapeSequences(char *aBuf, char aEscapeChar, bool aAllowEscapedSpace);
POINT CenterWindow(int aWidth, int aHeight);
bool FontExist(HDC aHdc, char *aTypeface);
void ScreenToWindow(POINT &aPoint, HWND aHwnd);
void WindowToScreen(int &aX, int &aY);
void GetVirtualDesktopRect(RECT &aRect);
LPVOID AllocInterProcMem(HANDLE &aHandle, DWORD aSize, HWND aHwnd);
void FreeInterProcMem(HANDLE aHandle, LPVOID aMem);

DWORD GetEnvVarReliable(char *aEnvVarName, char *aBuf);
DWORD ReadRegString(HKEY aRootKey, char *aSubkey, char *aValueName, char *aBuf, size_t aBufSize);

HBITMAP LoadPicture(char *aFilespec, int aWidth, int aHeight, int &aImageType, int aIconNumber
    , bool aUseGDIPlusIfAvailable);
HBITMAP IconToBitmap(HICON ahIcon, bool aDestroyIcon);
int CALLBACK FontEnumProc(ENUMLOGFONTEX *lpelfe, NEWTEXTMETRICEX *lpntme, DWORD FontType, LPARAM lParam);
bool IsStringInList(char *aStr, char *aList, bool aFindExactMatch);

#endif