📄 strtod.cxx
字号:
//===========================================================================//// strtod.cxx//// ANSI String to double-precision floating point conversion////===========================================================================//####COPYRIGHTBEGIN####// // ------------------------------------------- // The contents of this file are subject to the Red Hat eCos Public License // Version 1.1 (the "License"); you may not use this file except in // compliance with the License. You may obtain a copy of the License at // http://www.redhat.com/ // // Software distributed under the License is distributed on an "AS IS" // basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See the // License for the specific language governing rights and limitations under // the License. // // The Original Code is eCos - Embedded Configurable Operating System, // released September 30, 1998. // // The Initial Developer of the Original Code is Red Hat. // Portions created by Red Hat are // Copyright (C) 1998, 1999, 2000 Red Hat, Inc. // All Rights Reserved. // ------------------------------------------- // //####COPYRIGHTEND####//===========================================================================//#####DESCRIPTIONBEGIN####//// Author(s): jlarmour// Contributors: jlarmour// Date: 1998-02-13// Purpose: // Description: // Usage: ////####DESCRIPTIONEND####////===========================================================================// CONFIGURATION#include <pkgconf/libc.h> // Configuration header// Include the C library? And strtod()?#if defined(CYGPKG_LIBC) && defined(CYGFUN_LIBC_strtod)// INCLUDES#include <cyg/infra/cyg_type.h> // Common type definitions and support#include <cyg/infra/cyg_trac.h> // Tracing support#include <cyg/infra/cyg_ass.h> // Assertion support#include <stddef.h> // NULL, wchar_t and size_t from compiler#include <stdlib.h> // Main header for stdlib functions#include <ctype.h> // isspace() and isdigit()#include "clibincl/stdlibsupp.hxx" // Support for stdlib functions#include <float.h> // DBL_MIN_10_EXP and DBL_MAX_10_EXP#include <math.h> // HUGE_VAL#include <errno.h> // errno// EXPORTED SYMBOLSexternC doublestrtod( const char *, char ** ) CYGPRI_LIBC_WEAK_ALIAS("_strtod");// CONSTANTS#define MAXE (DBL_MAX_10_EXP)#define MINE (DBL_MIN_10_EXP)// flags#define SIGN 0x01#define ESIGN 0x02#define DECP 0x04// MACROS#define Ise(c) ((c == 'e') || (c == 'E') || (c == 'd') || (c == 'D'))#define Issign(c) ((c == '-') || (c == '+'))#define Val(c) ((c - '0'))// FUNCTIONS/* * [atw] multiply 64 bit accumulator by 10 and add digit. * The KA/CA way to do this should be to use * a 64-bit integer internally and use "adjust" to * convert it to float at the end of processing. */static intten_mul(double *acc, int digit){ /* [atw] Crude, but effective (at least on a KB)... */ *acc *= 10; *acc += digit; return 0; /* no overflow */} // ten_mul()/* * compute 10**x by successive squaring. */static const doubleexp10(unsigned x){ static double powtab[] = {1.0, 10.0, 100.0, 1000.0, 10000.0}; if (x < (sizeof(powtab)/sizeof(double))) return powtab[x]; else if (x & 1) return 10.0 * exp10(x-1); else return exp10(x/2) * exp10(x/2);} // exp10()/* * return (*acc) scaled by 10**dexp. */static doubleadjust(double *acc, int dexp, int sign) /* *acc the 64 bit accumulator */ /* dexp decimal exponent */ /* sign sign flag */{ double r; if (dexp > MAXE) { errno = ERANGE; return (sign) ? -HUGE_VAL : HUGE_VAL; } else if (dexp < MINE) { errno = ERANGE; return 0.0; } r = *acc; if (sign) r = -r; if (dexp==0) return r; if (dexp < 0) return r / exp10(abs(dexp)); else return r * exp10(dexp);} // adjust()externC double_strtod( const char *nptr, char **endptr ){ const char *start=nptr; double accum = 0.0; int flags = 0; int texp = 0; int e = 0; int conv_done = 0; double retval; CYG_REPORT_FUNCNAMETYPE( "_strtod", "returning %f" ); CYG_CHECK_DATA_PTR( nptr, "nptr is an invalid pointer!" ); // endptr is allowed to be NULL, but if it isn't, we check it if (endptr != NULL) CYG_CHECK_DATA_PTR( endptr, "endptr is an invalid pointer!" ); while(isspace(*nptr)) nptr++; if(*nptr == '\0') { /* just leading spaces */ if(endptr != NULL) *endptr = (char *)start; return 0.0; } if(Issign(*nptr)) { if(*nptr == '-') flags = SIGN; if(*++nptr == '\0') { /* "+|-" : should be an error ? */ if(endptr != NULL) *endptr = (char *)start; return 0.0; } } for(; (isdigit(*nptr) || (*nptr == '.')); nptr++) { conv_done = 1; if(*nptr == '.') flags |= DECP; else { if( ten_mul(&accum, Val(*nptr)) ) texp++; if(flags & DECP) texp--; } } if(Ise(*nptr)) { conv_done = 1; if(*++nptr != '\0') /* skip e|E|d|D */ { /* ! ([nptr]xxx[.[yyy]]e) */ while(isspace(*nptr)) nptr++; /* Ansi allows spaces after e */ if(*nptr != '\0') { /* ! ([nptr]xxx[.[yyy]]e[space]) */ if(Issign(*nptr)) if(*nptr++ == '-') flags |= ESIGN; if(*nptr != '\0') { /* ! ([nptr]xxx[.[yyy]]e[nptr]) -- error?? */ for(; isdigit(*nptr); nptr++) if (e < MAXE) /* prevent from grossly overflowing */ e = e*10 + Val(*nptr); /* dont care what comes after this */ if(flags & ESIGN) texp -= e; else texp += e; } } } } if(endptr != NULL) *endptr = (char *)((conv_done) ? nptr : start); retval = adjust(&accum, (int)texp, (int)(flags & SIGN)); CYG_REPORT_RETVAL( retval ); return retval;} // _strtod()#endif // if defined(CYGPKG_LIBC) && defined(CYGFUN_LIBC_strtod)// EOF strtod.cxx
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -