4:36 _I|Mjq؂File not found. +lY> j- e_I|+?// © 2016 and later: Unicode, Inc. and others. // License & terms of use: http://www.unicode.org/copyright.html /* ****************************************************************************** * * Copyright (C) 1998-2015, International Business Machines * Corporation and others. All Rights Reserved. * ****************************************************************************** * * File ustdio.h * * Modification History: * * Date Name Description * 10/16/98 stephen Creation. * 11/06/98 stephen Modified per code review. * 03/12/99 stephen Modified for new C API. * 07/19/99 stephen Minor doc update. * 02/01/01 george Added sprintf & sscanf with all of its variants ****************************************************************************** */ #ifndef USTDIO_H #define USTDIO_H #include #include #include "unicode/utypes.h" #include "unicode/ucnv.h" #include "unicode/utrans.h" #include "unicode/localpointer.h" #include "unicode/unum.h" #if !UCONFIG_NO_CONVERSION /* TODO The following is a small list as to what is currently wrong/suggestions for ustdio. * Make sure that * in the scanf format specification works for all formats. * Each UFILE takes up at least 2KB. Look into adding setvbuf() for configurable buffers. * This library does buffering. The OS should do this for us already. Check on this, and remove it from this library, if this is the case. Double buffering wastes a lot of time and space. * Test stdin and stdout with the u_f* functions * Testing should be done for reading and writing multi-byte encodings, and make sure that a character that is contained across buffer boundries works even for incomplete characters. * Make sure that the last character is flushed when the file/string is closed. * snprintf should follow the C99 standard for the return value, which is return the number of characters (excluding the trailing '\0') which would have been written to the destination string regardless of available space. This is like pre-flighting. * Everything that uses %s should do what operator>> does for UnicodeString. It should convert one byte at a time, and once a character is converted then check to see if it's whitespace or in the scanset. If it's whitespace or in the scanset, put all the bytes back (do nothing for sprintf/sscanf). * If bad string data is encountered, make sure that the function fails without memory leaks and the unconvertable characters are valid substitution or are escaped characters. * u_fungetc() can't unget a character when it's at the beginning of the internal conversion buffer. For example, read the buffer size # of characters, and then ungetc to get the previous character that was at the end of the last buffer. * u_fflush() and u_fclose should return an int32_t like C99 functions. 0 is returned if the operation was successful and EOF otherwise. * u_fsettransliterator does not support U_READ side of transliteration. * The format specifier should limit the size of a format or honor it in order to prevent buffer overruns. (e.g. %256.256d). * u_fread and u_fwrite don't exist. They're needed for reading and writing data structures without any conversion. * u_file_read and u_file_write are used for writing strings. u_fgets and u_fputs or u_fread and u_fwrite should be used to do this. * The width parameter for all scanf formats, including scanset, needs better testing. This prevents buffer overflows. * Figure out what is suppose to happen when a codepage is changed midstream. Maybe a flush or a rewind are good enough. * Make sure that a UFile opened with "rw" can be used after using u_fflush with a u_frewind. * scanf(%i) should detect what type of number to use. * Add more testing of the alternate format, %# * Look at newline handling of fputs/puts * T