Source code

Revision control

Copy as Markdown

Other Tools

/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* vim: set ts=8 sts=2 et sw=2 tw=80: */
/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
/*
* Portable safe sprintf code.
*
* Code based on mozilla/nsprpub/src/io/prprf.c rev 3.7
*
* Contributor(s):
* Kipp E.B. Hickman <kipp@netscape.com> (original author)
* Frank Yung-Fong Tang <ftang@netscape.com>
* Daniele Nicolodi <daniele@grinta.net>
*/
#include <stddef.h>
#include <stdio.h>
#include <string.h>
#include "prdtoa.h"
#include "mozilla/Logging.h"
#include "mozilla/Sprintf.h"
#include "nsCRTGlue.h"
#include "nsTextFormatter.h"
struct nsTextFormatter::SprintfStateStr {
int (*stuff)(SprintfStateStr* aState, const char16_t* aStr, uint32_t aLen);
char16_t* base;
char16_t* cur;
uint32_t maxlen;
void* stuffclosure;
};
#define _LEFT 0x1
#define _SIGNED 0x2
#define _SPACED 0x4
#define _ZEROS 0x8
#define _NEG 0x10
#define _UNSIGNED 0x20
#define ELEMENTS_OF(array_) (sizeof(array_) / sizeof(array_[0]))
/*
** Fill into the buffer using the data in src
*/
int nsTextFormatter::fill2(SprintfStateStr* aState, const char16_t* aSrc,
int aSrcLen, int aWidth, int aFlags) {
char16_t space = ' ';
int rv;
aWidth -= aSrcLen;
/* Right adjusting */
if ((aWidth > 0) && ((aFlags & _LEFT) == 0)) {
if (aFlags & _ZEROS) {
space = '0';
}
while (--aWidth >= 0) {
rv = (*aState->stuff)(aState, &space, 1);
if (rv < 0) {
return rv;
}
}
}
/* Copy out the source data */
rv = (*aState->stuff)(aState, aSrc, aSrcLen);
if (rv < 0) {
return rv;
}
/* Left adjusting */
if ((aWidth > 0) && ((aFlags & _LEFT) != 0)) {
while (--aWidth >= 0) {
rv = (*aState->stuff)(aState, &space, 1);
if (rv < 0) {
return rv;
}
}
}
return 0;
}
/*
** Fill a number. The order is: optional-sign zero-filling conversion-digits
*/
int nsTextFormatter::fill_n(nsTextFormatter::SprintfStateStr* aState,
const char16_t* aSrc, int aSrcLen, int aWidth,
int aPrec, int aFlags) {
int zerowidth = 0;
int precwidth = 0;
int signwidth = 0;
int leftspaces = 0;
int rightspaces = 0;
int cvtwidth;
int rv;
char16_t sign;
char16_t space = ' ';
char16_t zero = '0';
if ((aFlags & _UNSIGNED) == 0) {
if (aFlags & _NEG) {
sign = '-';
signwidth = 1;
} else if (aFlags & _SIGNED) {
sign = '+';
signwidth = 1;
} else if (aFlags & _SPACED) {
sign = ' ';
signwidth = 1;
}
}
cvtwidth = signwidth + aSrcLen;
if (aPrec > 0) {
if (aPrec > aSrcLen) {
/* Need zero filling */
precwidth = aPrec - aSrcLen;
cvtwidth += precwidth;
}
}
if ((aFlags & _ZEROS) && (aPrec < 0)) {
if (aWidth > cvtwidth) {
/* Zero filling */
zerowidth = aWidth - cvtwidth;
cvtwidth += zerowidth;
}
}
if (aFlags & _LEFT) {
if (aWidth > cvtwidth) {
/* Space filling on the right (i.e. left adjusting) */
rightspaces = aWidth - cvtwidth;
}
} else {
if (aWidth > cvtwidth) {
/* Space filling on the left (i.e. right adjusting) */
leftspaces = aWidth - cvtwidth;
}
}
while (--leftspaces >= 0) {
rv = (*aState->stuff)(aState, &space, 1);
if (rv < 0) {
return rv;
}
}
if (signwidth) {
rv = (*aState->stuff)(aState, &sign, 1);
if (rv < 0) {
return rv;
}
}
while (--precwidth >= 0) {
rv = (*aState->stuff)(aState, &space, 1);
if (rv < 0) {
return rv;
}
}
while (--zerowidth >= 0) {
rv = (*aState->stuff)(aState, &zero, 1);
if (rv < 0) {
return rv;
}
}
rv = (*aState->stuff)(aState, aSrc, aSrcLen);
if (rv < 0) {
return rv;
}
while (--rightspaces >= 0) {
rv = (*aState->stuff)(aState, &space, 1);
if (rv < 0) {
return rv;
}
}
return 0;
}
/*
** Convert a 64-bit integer into its printable form
*/
int nsTextFormatter::cvt_ll(SprintfStateStr* aState, uint64_t aNum, int aWidth,
int aPrec, int aRadix, int aFlags,
const char16_t* aHexStr) {
char16_t cvtbuf[100];
char16_t* cvt;
int digits;
/* according to the man page this needs to happen */
if (aPrec == 0 && aNum == 0) {
return 0;
}
/*
** Converting decimal is a little tricky. In the unsigned case we
** need to stop when we hit 10 digits. In the signed case, we can
** stop when the number is zero.
*/
cvt = &cvtbuf[0] + ELEMENTS_OF(cvtbuf);
digits = 0;
while (aNum != 0) {
uint64_t quot = aNum / aRadix;
uint64_t rem = aNum % aRadix;
*--cvt = aHexStr[rem & 0xf];
digits++;
aNum = quot;
}
if (digits == 0) {
*--cvt = '0';
digits++;
}
/*
** Now that we have the number converted without its sign, deal with
** the sign and zero padding.
*/
return fill_n(aState, cvt, digits, aWidth, aPrec, aFlags);
}
/*
** Convert a double precision floating point number into its printable
** form.
*/
int nsTextFormatter::cvt_f(SprintfStateStr* aState, double aDouble, int aWidth,
int aPrec, const char16_t aType, int aFlags) {
int mode = 2;
int decpt;
int sign;
char buf[256];
char* bufp = buf;
int bufsz = 256;
char num[256];
char* nump;
char* endnum;
int numdigits = 0;
char exp = 'e';
if (aPrec == -1) {
aPrec = 6;
} else if (aPrec > 50) {
// limit precision to avoid PR_dtoa bug 108335
// and to prevent buffers overflows
aPrec = 50;
}
switch (aType) {
case 'f':
numdigits = aPrec;
mode = 3;
break;
case 'E':
exp = 'E';
[[fallthrough]];
case 'e':
numdigits = aPrec + 1;
mode = 2;
break;
case 'G':
exp = 'E';
[[fallthrough]];
case 'g':
if (aPrec == 0) {
aPrec = 1;
}
numdigits = aPrec;
mode = 2;
break;
default:
NS_ERROR("invalid aType passed to cvt_f");
}
if (PR_dtoa(aDouble, mode, numdigits, &decpt, &sign, &endnum, num, bufsz) ==
PR_FAILURE) {
buf[0] = '\0';
return -1;
}
numdigits = endnum - num;
nump = num;
if (sign) {
*bufp++ = '-';
} else if (aFlags & _SIGNED) {
*bufp++ = '+';
}
if (decpt == 9999) {
while ((*bufp++ = *nump++)) {
}
} else {
switch (aType) {
case 'E':
case 'e':
*bufp++ = *nump++;
if (aPrec > 0) {
*bufp++ = '.';
while (*nump) {
*bufp++ = *nump++;
aPrec--;
}
while (aPrec-- > 0) {
*bufp++ = '0';
}
}
*bufp++ = exp;
::snprintf(bufp, bufsz - (bufp - buf), "%+03d", decpt - 1);
break;
case 'f':
if (decpt < 1) {
*bufp++ = '0';
if (aPrec > 0) {
*bufp++ = '.';
while (decpt++ && aPrec-- > 0) {
*bufp++ = '0';
}
while (*nump && aPrec-- > 0) {
*bufp++ = *nump++;
}
while (aPrec-- > 0) {
*bufp++ = '0';
}
}
} else {
while (*nump && decpt-- > 0) {
*bufp++ = *nump++;
}
while (decpt-- > 0) {
*bufp++ = '0';
}
if (aPrec > 0) {
*bufp++ = '.';
while (*nump && aPrec-- > 0) {
*bufp++ = *nump++;
}
while (aPrec-- > 0) {
*bufp++ = '0';
}
}
}
*bufp = '\0';
break;
case 'G':
case 'g':
if ((decpt < -3) || ((decpt - 1) >= aPrec)) {
*bufp++ = *nump++;
numdigits--;
if (numdigits > 0) {
*bufp++ = '.';
while (*nump) {
*bufp++ = *nump++;
}
}
*bufp++ = exp;
::snprintf(bufp, bufsz - (bufp - buf), "%+03d", decpt - 1);
} else {
if (decpt < 1) {
*bufp++ = '0';
if (aPrec > 0) {
*bufp++ = '.';
while (decpt++) {
*bufp++ = '0';
}
while (*nump) {
*bufp++ = *nump++;
}
}
} else {
while (*nump && decpt-- > 0) {
*bufp++ = *nump++;
numdigits--;
}
while (decpt-- > 0) {
*bufp++ = '0';
}
if (numdigits > 0) {
*bufp++ = '.';
while (*nump) {
*bufp++ = *nump++;
}
}
}
*bufp = '\0';
}
}
}
char16_t rbuf[256];
char16_t* rbufp = rbuf;
bufp = buf;
// cast to char16_t
while ((*rbufp++ = *bufp++)) {
}
*rbufp = '\0';
return fill2(aState, rbuf, NS_strlen(rbuf), aWidth, aFlags);
}
/*
** Convert a string into its printable form. |aWidth| is the output
** width. |aPrec| is the maximum number of characters of |aStr| to output,
** where -1 means until NUL.
*/
int nsTextFormatter::cvt_S(SprintfStateStr* aState, const char16_t* aStr,
int aWidth, int aPrec, int aFlags) {
int slen;
if (aPrec == 0) {
return 0;
}
/* Limit string length by precision value */
slen = aStr ? NS_strlen(aStr) : 6;
if (aPrec > 0) {
if (aPrec < slen) {
slen = aPrec;
}
}
/* and away we go */
return fill2(aState, aStr ? aStr : u"(null)", slen, aWidth, aFlags);
}
/*
** Convert a string into its printable form. |aWidth| is the output
** width. |aPrec| is the maximum number of characters of |aStr| to output,
** where -1 means until NUL.
*/
int nsTextFormatter::cvt_s(nsTextFormatter::SprintfStateStr* aState,
const char* aStr, int aWidth, int aPrec,
int aFlags) {
// Be sure to handle null the same way as %S.
if (aStr == nullptr) {
return cvt_S(aState, nullptr, aWidth, aPrec, aFlags);
}
NS_ConvertUTF8toUTF16 utf16Val(aStr);
return cvt_S(aState, utf16Val.get(), aWidth, aPrec, aFlags);
}
/*
** The workhorse sprintf code.
*/
int nsTextFormatter::dosprintf(SprintfStateStr* aState, const char16_t* aFmt,
mozilla::Span<BoxedValue> aValues) {
static const char16_t space = ' ';
static const char16_t hex[] = u"0123456789abcdef";
static const char16_t HEX[] = u"0123456789ABCDEF";
static const BoxedValue emptyString(u"");
char16_t c;
int flags, width, prec, radix;
const char16_t* hexp;
// Next argument for non-numbered arguments.
size_t nextNaturalArg = 0;
// True if we ever saw a numbered argument.
bool sawNumberedArg = false;
while ((c = *aFmt++) != 0) {
int rv;
if (c != '%') {
rv = (*aState->stuff)(aState, aFmt - 1, 1);
if (rv < 0) {
return rv;
}
continue;
}
// Save the location of the "%" in case we decide it isn't a
// format and want to just emit the text from the format string.
const char16_t* percentPointer = aFmt - 1;
/*
** Gobble up the % format string. Hopefully we have handled all
** of the strange cases!
*/
flags = 0;
c = *aFmt++;
if (c == '%') {
/* quoting a % with %% */
rv = (*aState->stuff)(aState, aFmt - 1, 1);
if (rv < 0) {
return rv;
}
continue;
}
// Check for a numbered argument.
bool sawWidth = false;
const BoxedValue* thisArg = nullptr;
if (c >= '0' && c <= '9') {
size_t argNumber = 0;
while (c && c >= '0' && c <= '9') {
argNumber = (argNumber * 10) + (c - '0');
c = *aFmt++;
}
if (c == '$') {
// Mixing numbered arguments and implicit arguments is
// disallowed.
if (nextNaturalArg > 0) {
return -1;
}
c = *aFmt++;
// Numbered arguments start at 1.
--argNumber;
if (argNumber >= aValues.Length()) {
// A correctness issue but not a safety issue.
MOZ_ASSERT(false);
thisArg = &emptyString;
} else {
thisArg = &aValues[argNumber];
}
sawNumberedArg = true;
} else {
width = argNumber;
sawWidth = true;
}
}
if (!sawWidth) {
/*
* Examine optional flags. Note that we do not implement the
* '#' flag of sprintf(). The ANSI C spec. of the '#' flag is
* somewhat ambiguous and not ideal, which is perhaps why
* the various sprintf() implementations are inconsistent
* on this feature.
*/
while ((c == '-') || (c == '+') || (c == ' ') || (c == '0')) {
if (c == '-') {
flags |= _LEFT;
}
if (c == '+') {
flags |= _SIGNED;
}
if (c == ' ') {
flags |= _SPACED;
}
if (c == '0') {
flags |= _ZEROS;
}
c = *aFmt++;
}
if (flags & _SIGNED) {
flags &= ~_SPACED;
}
if (flags & _LEFT) {
flags &= ~_ZEROS;
}
/* width */
if (c == '*') {
// Not supported with numbered arguments.
if (sawNumberedArg) {
return -1;
}
if (nextNaturalArg >= aValues.Length() ||
!aValues[nextNaturalArg].IntCompatible()) {
// A correctness issue but not a safety issue.
MOZ_ASSERT(false);
width = 0;
} else {
width = aValues[nextNaturalArg++].mValue.mInt;
}
c = *aFmt++;
} else {
width = 0;
while ((c >= '0') && (c <= '9')) {
width = (width * 10) + (c - '0');
c = *aFmt++;
}
}
}
/* precision */
prec = -1;
if (c == '.') {
c = *aFmt++;
if (c == '*') {
// Not supported with numbered arguments.
if (sawNumberedArg) {
return -1;
}
if (nextNaturalArg >= aValues.Length() ||
!aValues[nextNaturalArg].IntCompatible()) {
// A correctness issue but not a safety issue.
MOZ_ASSERT(false);
} else {
prec = aValues[nextNaturalArg++].mValue.mInt;
}
c = *aFmt++;
} else {
prec = 0;
while ((c >= '0') && (c <= '9')) {
prec = (prec * 10) + (c - '0');
c = *aFmt++;
}
}
}
// If the argument isn't known yet, find it now. This is done
// after the width and precision code, in case '*' was used.
if (thisArg == nullptr) {
// Mixing numbered arguments and implicit arguments is
// disallowed.
if (sawNumberedArg) {
return -1;
}
if (nextNaturalArg >= aValues.Length()) {
// A correctness issue but not a safety issue.
MOZ_ASSERT(false);
thisArg = &emptyString;
} else {
thisArg = &aValues[nextNaturalArg++];
}
}
/* Size. Defaults to 32 bits. */
uint64_t mask = UINT32_MAX;
if (c == 'h') {
c = *aFmt++;
mask = UINT16_MAX;
} else if (c == 'L') {
c = *aFmt++;
mask = UINT64_MAX;
} else if (c == 'l') {
c = *aFmt++;
if (c == 'l') {
c = *aFmt++;
mask = UINT64_MAX;
} else {
mask = UINT32_MAX;
}
}
/* format */
hexp = hex;
radix = 10;
// Several `MOZ_ASSERT`s below check for argument compatibility
// with the format specifier. These are only debug assertions,
// not release assertions, and exist to catch problems in C++
// callers of `nsTextFormatter`, as we do not have compile-time
// checking of format strings. In release mode, these assertions
// will be no-ops, and we will fall through to printing the
// argument based on the known type of the argument.
switch (c) {
case 'd':
case 'i': /* decimal/integer */
MOZ_ASSERT(thisArg->IntCompatible());
break;
case 'o': /* octal */
MOZ_ASSERT(thisArg->IntCompatible());
radix = 8;
flags |= _UNSIGNED;
break;
case 'u': /* unsigned decimal */
MOZ_ASSERT(thisArg->IntCompatible());
radix = 10;
flags |= _UNSIGNED;
break;
case 'x': /* unsigned hex */
MOZ_ASSERT(thisArg->IntCompatible());
radix = 16;
flags |= _UNSIGNED;
break;
case 'X': /* unsigned HEX */
MOZ_ASSERT(thisArg->IntCompatible());
radix = 16;
hexp = HEX;
flags |= _UNSIGNED;
break;
case 'e':
case 'E':
case 'f':
case 'g':
case 'G':
MOZ_ASSERT(thisArg->mKind == DOUBLE);
// Type-based printing below.
break;
case 'S':
case 's':
MOZ_ASSERT(thisArg->mKind == STRING || thisArg->mKind == STRING16);
// Type-based printing below.
break;
case 'c': {
if (!thisArg->IntCompatible()) {
MOZ_ASSERT(false);
// Type-based printing below.
break;
}
if ((flags & _LEFT) == 0) {
while (width-- > 1) {
rv = (*aState->stuff)(aState, &space, 1);
if (rv < 0) {
return rv;
}
}
}
char16_t ch = thisArg->mValue.mInt;
rv = (*aState->stuff)(aState, &ch, 1);
if (rv < 0) {
return rv;
}
if (flags & _LEFT) {
while (width-- > 1) {
rv = (*aState->stuff)(aState, &space, 1);
if (rv < 0) {
return rv;
}
}
}
}
continue;
case 'p':
if (!thisArg->PointerCompatible()) {
MOZ_ASSERT(false);
break;
}
static_assert(sizeof(uint64_t) >= sizeof(void*),
"pointers are larger than 64 bits");
rv = cvt_ll(aState, uintptr_t(thisArg->mValue.mPtr), width, prec, 16,
flags | _UNSIGNED, hexp);
if (rv < 0) {
return rv;
}
continue;
case 'n':
if (thisArg->mKind != INTPOINTER) {
return -1;
}
if (thisArg->mValue.mIntPtr != nullptr) {
*thisArg->mValue.mIntPtr = aState->cur - aState->base;
}
continue;
default:
/* Not a % token after all... skip it */
rv = (*aState->stuff)(aState, percentPointer, aFmt - percentPointer);
if (rv < 0) {
return rv;
}
continue;
}
// If we get here, we want to handle the argument according to its
// actual type; modified by the flags as appropriate.
switch (thisArg->mKind) {
case INT:
case UINT: {
int64_t val = thisArg->mValue.mInt;
if ((flags & _UNSIGNED) == 0 && val < 0) {
val = -val;
flags |= _NEG;
}
rv = cvt_ll(aState, uint64_t(val) & mask, width, prec, radix, flags,
hexp);
} break;
case INTPOINTER:
case POINTER:
// Always treat these as unsigned hex, no matter the format.
static_assert(sizeof(uint64_t) >= sizeof(void*),
"pointers are larger than 64 bits");
rv = cvt_ll(aState, uintptr_t(thisArg->mValue.mPtr), width, prec, 16,
flags | _UNSIGNED, hexp);
break;
case DOUBLE:
if (c != 'f' && c != 'E' && c != 'e' && c != 'G' && c != 'g') {
// Pick some default.
c = 'g';
}
rv = cvt_f(aState, thisArg->mValue.mDouble, width, prec, c, flags);
break;
case STRING:
rv = cvt_s(aState, thisArg->mValue.mString, width, prec, flags);
break;
case STRING16:
rv = cvt_S(aState, thisArg->mValue.mString16, width, prec, flags);
break;
default:
// Can't happen.
MOZ_ASSERT(0);
}
if (rv < 0) {
return rv;
}
}
return 0;
}
/************************************************************************/
int nsTextFormatter::StringStuff(nsTextFormatter::SprintfStateStr* aState,
const char16_t* aStr, uint32_t aLen) {
ptrdiff_t off = aState->cur - aState->base;
nsAString* str = static_cast<nsAString*>(aState->stuffclosure);
str->Append(aStr, aLen);
aState->base = str->BeginWriting();
aState->cur = aState->base + off;
return 0;
}
void nsTextFormatter::vssprintf(nsAString& aOut, const char16_t* aFmt,
mozilla::Span<BoxedValue> aValues) {
SprintfStateStr ss;
ss.stuff = StringStuff;
ss.base = 0;
ss.cur = 0;
ss.maxlen = 0;
ss.stuffclosure = &aOut;
aOut.Truncate();
dosprintf(&ss, aFmt, aValues);
}
/*
** Stuff routine that discards overflow data
*/
int nsTextFormatter::LimitStuff(SprintfStateStr* aState, const char16_t* aStr,
uint32_t aLen) {
uint32_t limit = aState->maxlen - (aState->cur - aState->base);
if (aLen > limit) {
aLen = limit;
}
while (aLen) {
--aLen;
*aState->cur++ = *aStr++;
}
return 0;
}
uint32_t nsTextFormatter::vsnprintf(char16_t* aOut, uint32_t aOutLen,
const char16_t* aFmt,
mozilla::Span<BoxedValue> aValues) {
SprintfStateStr ss;
MOZ_ASSERT((int32_t)aOutLen > 0);
if ((int32_t)aOutLen <= 0) {
return 0;
}
ss.stuff = LimitStuff;
ss.base = aOut;
ss.cur = aOut;
ss.maxlen = aOutLen;
int result = dosprintf(&ss, aFmt, aValues);
if (ss.cur == ss.base) {
return 0;
}
// Append a NUL. However, be sure not to count it in the returned
// length.
if (ss.cur - ss.base >= ptrdiff_t(ss.maxlen)) {
--ss.cur;
}
*ss.cur = '\0';
// Check the result now, so that an unterminated string can't
// possibly escape.
if (result < 0) {
return -1;
}
return ss.cur - ss.base;
}