static void Stringize (StrBuf* Source, StrBuf* Target)
/* Stringize the given string: Add double quotes at start and end and preceed
* each occurance of " and \ by a backslash.
*/
{
char C;
/* Add a starting quote */
SB_AppendChar (Target, '\"');
/* Replace any characters inside the string may not be part of a string
* unescaped.
*/
while ((C = SB_Get (Source)) != '\0') {
switch (C) {
case '\"':
case '\\':
SB_AppendChar (Target, '\\');
/* FALLTHROUGH */
default:
SB_AppendChar (Target, C);
break;
}
}
/* Add the closing quote */
SB_AppendChar (Target, '\"');
}
static int GetComma (StrBuf* B)
/* Expects and skips a comma in B. Prints an error and returns zero if no
* comma is found. Return a value <> 0 otherwise.
*/
{
SB_SkipWhite (B);
if (SB_Get (B) != ',') {
Error ("Comma expected");
return 0;
}
SB_SkipWhite (B);
return 1;
}
static void FuncSPrintF (void)
/* Handle the .SPRINTF function */
{
StrBuf Format = STATIC_STRBUF_INITIALIZER; /* User supplied format */
StrBuf R = STATIC_STRBUF_INITIALIZER; /* Result string */
StrBuf F1 = STATIC_STRBUF_INITIALIZER; /* One format spec from F */
StrBuf R1 = STATIC_STRBUF_INITIALIZER; /* One result */
char C;
int Done;
long IVal; /* Integer value */
/* Skip the .SPRINTF token */
NextTok ();
/* Left paren expected */
ConsumeLParen ();
/* First argument is a format string. Remember and skip it */
if (!LookAtStrCon ()) {
return;
}
SB_Copy (&Format, &CurTok.SVal);
NextTok ();
/* Walk over the format string, generating the function result in R */
while (1) {
/* Get the next char from the format string and check for EOS */
if (SB_Peek (&Format) == '\0') {
break;
}
/* Check for a format specifier */
if (SB_Peek (&Format) != '%') {
/* No format specifier, just copy */
SB_AppendChar (&R, SB_Get (&Format));
continue;
}
SB_Skip (&Format);
if (SB_Peek (&Format) == '%') {
/* %% */
SB_AppendChar (&R, '%');
SB_Skip (&Format);
continue;
}
if (SB_Peek (&Format) == '\0') {
InvalidFormatString ();
break;
}
/* Since a format specifier follows, we do expect anotehr argument for
* the .sprintf function.
*/
ConsumeComma ();
/* We will copy the format spec into F1 checking for the things we
* support, and later use xsprintf to do the actual formatting. This
* is easier than adding another printf implementation...
*/
SB_Clear (&F1);
SB_AppendChar (&F1, '%');
/* Check for flags */
Done = 0;
while ((C = SB_Peek (&Format)) != '\0' && !Done) {
switch (C) {
case '-': /* FALLTHROUGH */
case '+': /* FALLTHROUGH */
case ' ': /* FALLTHROUGH */
case '#': /* FALLTHROUGH */
case '0': SB_AppendChar (&F1, SB_Get (&Format)); break;
default: Done = 1; break;
}
}
/* We do only support a numerical width field */
while (IsDigit (SB_Peek (&Format))) {
SB_AppendChar (&F1, SB_Get (&Format));
}
/* Precision - only positive numerical fields supported */
if (SB_Peek (&Format) == '.') {
SB_AppendChar (&F1, SB_Get (&Format));
while (IsDigit (SB_Peek (&Format))) {
SB_AppendChar (&F1, SB_Get (&Format));
}
}
/* Length modifiers aren't supported, so read the conversion specs */
switch (SB_Peek (&Format)) {
case 'd':
case 'i':
case 'o':
case 'u':
case 'X':
case 'x':
/* Our ints are actually longs, so we use the 'l' modifier when
* calling xsprintf later. Terminate the format string.
*/
SB_AppendChar (&F1, 'l');
SB_AppendChar (&F1, SB_Get (&Format));
SB_Terminate (&F1);
/* The argument must be a constant expression */
IVal = ConstExpression ();
/* Format this argument according to the spec */
SB_Printf (&R1, SB_GetConstBuf (&F1), IVal);
/* Append the formatted argument to the result */
SB_Append (&R, &R1);
break;
case 's':
/* Add the format spec and terminate the format */
SB_AppendChar (&F1, SB_Get (&Format));
SB_Terminate (&F1);
/* The argument must be a string constant */
if (!LookAtStrCon ()) {
/* Make it one */
SB_CopyStr (&CurTok.SVal, "**undefined**");
}
/* Format this argument according to the spec */
SB_Printf (&R1, SB_GetConstBuf (&F1), SB_GetConstBuf (&CurTok.SVal));
/* Skip the string constant */
NextTok ();
/* Append the formatted argument to the result */
SB_Append (&R, &R1);
break;
case 'c':
/* Add the format spec and terminate the format */
SB_AppendChar (&F1, SB_Get (&Format));
SB_Terminate (&F1);
/* The argument must be a constant expression */
IVal = ConstExpression ();
/* Check for a valid character range */
if (IVal <= 0 || IVal > 255) {
Error ("Char argument out of range");
IVal = ' ';
}
/* Format this argument according to the spec. Be sure to pass
* an int as the char value.
*/
SB_Printf (&R1, SB_GetConstBuf (&F1), (int) IVal);
/* Append the formatted argument to the result */
SB_Append (&R, &R1);
break;
default:
Error ("Invalid format string");
SB_Skip (&Format);
break;
}
}
/* Terminate the result string */
SB_Terminate (&R);
/* We expect a closing parenthesis, but will not skip it but replace it
* by the string token just created.
*/
if (CurTok.Tok != TOK_RPAREN) {
Error ("`)' expected");
} else {
CurTok.Tok = TOK_STRCON;
SB_Copy (&CurTok.SVal, &R);
SB_Terminate (&CurTok.SVal);
}
/* Delete the string buffers */
SB_Done (&Format);
SB_Done (&R);
SB_Done (&F1);
SB_Done (&R1);
}
static void FuncIdent (void)
/* Handle the .IDENT function */
{
StrBuf Buf = STATIC_STRBUF_INITIALIZER;
token_t Id;
unsigned I;
/* Skip it */
NextTok ();
/* Left paren expected */
ConsumeLParen ();
/* The function expects a string argument */
if (!LookAtStrCon ()) {
return;
}
/* Check that the string contains a valid identifier. While doing so,
* determine if it is a cheap local, or global one.
*/
SB_Reset (&CurTok.SVal);
/* Check for a cheap local symbol */
if (SB_Peek (&CurTok.SVal) == LocalStart) {
SB_Skip (&CurTok.SVal);
Id = TOK_LOCAL_IDENT;
} else {
Id = TOK_IDENT;
}
/* Next character must be a valid identifier start */
if (!IsIdStart (SB_Get (&CurTok.SVal))) {
NoIdent ();
return;
}
for (I = SB_GetIndex (&CurTok.SVal); I < SB_GetLen (&CurTok.SVal); ++I) {
if (!IsIdChar (SB_AtUnchecked (&CurTok.SVal, I))) {
NoIdent ();
return;
}
}
if (IgnoreCase) {
UpcaseSVal ();
}
/* If anything is ok, save and skip the string. Check that the next token
* is a right paren, then replace the token by an identifier token.
*/
SB_Copy (&Buf, &CurTok.SVal);
NextTok ();
if (CurTok.Tok != TOK_RPAREN) {
Error ("`)' expected");
} else {
CurTok.Tok = Id;
SB_Copy (&CurTok.SVal, &Buf);
SB_Terminate (&CurTok.SVal);
}
/* Free buffer memory */
SB_Done (&Buf);
}
static void IFNextChar (CharSource* S)
/* Read the next character from the input file */
{
/* Check for end of line, read the next line if needed */
while (SB_GetIndex (&S->V.File.Line) >= SB_GetLen (&S->V.File.Line)) {
unsigned Len;
/* End of current line reached, read next line */
SB_Clear (&S->V.File.Line);
while (1) {
int N = fgetc (S->V.File.F);
if (N == EOF) {
/* End of file. Accept files without a newline at the end */
if (SB_NotEmpty (&S->V.File.Line)) {
break;
}
/* No more data - add an empty line to the listing. This
** is a small hack needed to keep the PC output in sync.
*/
NewListingLine (&EmptyStrBuf, S->V.File.Pos.Name, FCount);
C = EOF;
return;
/* Check for end of line */
} else if (N == '\n') {
/* End of line */
break;
/* Collect other stuff */
} else {
/* Append data to line */
SB_AppendChar (&S->V.File.Line, N);
}
}
/* If we come here, we have a new input line. To avoid problems
** with strange line terminators, remove all whitespace from the
** end of the line, then add a single newline.
*/
Len = SB_GetLen (&S->V.File.Line);
while (Len > 0 && IsSpace (SB_AtUnchecked (&S->V.File.Line, Len-1))) {
--Len;
}
SB_Drop (&S->V.File.Line, SB_GetLen (&S->V.File.Line) - Len);
SB_AppendChar (&S->V.File.Line, '\n');
/* Terminate the string buffer */
SB_Terminate (&S->V.File.Line);
/* One more line */
S->V.File.Pos.Line++;
/* Remember the new line for the listing */
NewListingLine (&S->V.File.Line, S->V.File.Pos.Name, FCount);
}
/* Set the column pointer */
S->V.File.Pos.Col = SB_GetIndex (&S->V.File.Line);
/* Return the next character from the buffer */
C = SB_Get (&S->V.File.Line);
}
static void NumericConst (void)
/* Parse a numeric constant */
{
unsigned Base; /* Temporary number base */
unsigned Prefix; /* Base according to prefix */
StrBuf S = STATIC_STRBUF_INITIALIZER;
int IsFloat;
char C;
unsigned DigitVal;
unsigned long IVal; /* Value */
/* Check for a leading hex or octal prefix and determine the possible
** integer types.
*/
if (CurC == '0') {
/* Gobble 0 and examine next char */
NextChar ();
if (toupper (CurC) == 'X') {
Base = Prefix = 16;
NextChar (); /* gobble "x" */
} else {
Base = 10; /* Assume 10 for now - see below */
Prefix = 8; /* Actual prefix says octal */
}
} else {
Base = Prefix = 10;
}
/* Because floating point numbers don't have octal prefixes (a number
** with a leading zero is decimal), we first have to read the number
** before converting it, so we can determine if it's a float or an
** integer.
*/
while (IsXDigit (CurC) && HexVal (CurC) < Base) {
SB_AppendChar (&S, CurC);
NextChar ();
}
SB_Terminate (&S);
/* The following character tells us if we have an integer or floating
** point constant. Note: Hexadecimal floating point constants aren't
** supported in C89.
*/
IsFloat = (CurC == '.' ||
(Base == 10 && toupper (CurC) == 'E') ||
(Base == 16 && toupper (CurC) == 'P' && IS_Get (&Standard) >= STD_C99));
/* If we don't have a floating point type, an octal prefix results in an
** octal base.
*/
if (!IsFloat && Prefix == 8) {
Base = 8;
}
/* Since we do now know the correct base, convert the remembered input
** into a number.
*/
SB_Reset (&S);
IVal = 0;
while ((C = SB_Get (&S)) != '\0') {
DigitVal = HexVal (C);
if (DigitVal >= Base) {
Error ("Numeric constant contains digits beyond the radix");
}
IVal = (IVal * Base) + DigitVal;
}
/* We don't need the string buffer any longer */
SB_Done (&S);
/* Distinguish between integer and floating point constants */
if (!IsFloat) {
unsigned Types;
int HaveSuffix;
/* Check for a suffix and determine the possible types */
HaveSuffix = 1;
if (toupper (CurC) == 'U') {
/* Unsigned type */
NextChar ();
if (toupper (CurC) != 'L') {
Types = IT_UINT | IT_ULONG;
} else {
NextChar ();
Types = IT_ULONG;
}
} else if (toupper (CurC) == 'L') {
/* Long type */
NextChar ();
if (toupper (CurC) != 'U') {
Types = IT_LONG | IT_ULONG;
} else {
NextChar ();
Types = IT_ULONG;
}
} else {
HaveSuffix = 0;
if (Prefix == 10) {
/* Decimal constants are of any type but uint */
Types = IT_INT | IT_LONG | IT_ULONG;
} else {
/* Octal or hex constants are of any type */
Types = IT_INT | IT_UINT | IT_LONG | IT_ULONG;
}
}
/* Check the range to determine the type */
if (IVal > 0x7FFF) {
/* Out of range for int */
Types &= ~IT_INT;
/* If the value is in the range 0x8000..0xFFFF, unsigned int is not
** allowed, and we don't have a type specifying suffix, emit a
** warning, because the constant is of type long.
*/
if (IVal <= 0xFFFF && (Types & IT_UINT) == 0 && !HaveSuffix) {
Warning ("Constant is long");
}
}
if (IVal > 0xFFFF) {
/* Out of range for unsigned int */
Types &= ~IT_UINT;
}
if (IVal > 0x7FFFFFFF) {
/* Out of range for long int */
Types &= ~IT_LONG;
}
/* Now set the type string to the smallest type in types */
if (Types & IT_INT) {
NextTok.Type = type_int;
} else if (Types & IT_UINT) {
NextTok.Type = type_uint;
} else if (Types & IT_LONG) {
NextTok.Type = type_long;
} else {
NextTok.Type = type_ulong;
}
/* Set the value and the token */
NextTok.IVal = IVal;
NextTok.Tok = TOK_ICONST;
} else {
/* Float constant */
Double FVal = FP_D_FromInt (IVal); /* Convert to double */
/* Check for a fractional part and read it */
if (CurC == '.') {
Double Scale;
/* Skip the dot */
NextChar ();
/* Read fractional digits */
Scale = FP_D_Make (1.0);
while (IsXDigit (CurC) && (DigitVal = HexVal (CurC)) < Base) {
/* Get the value of this digit */
Double FracVal = FP_D_Div (FP_D_FromInt (DigitVal * Base), Scale);
/* Add it to the float value */
FVal = FP_D_Add (FVal, FracVal);
/* Scale base */
Scale = FP_D_Mul (Scale, FP_D_FromInt (DigitVal));
/* Skip the digit */
NextChar ();
}
}
/* Check for an exponent and read it */
if ((Base == 16 && toupper (CurC) == 'F') ||
(Base == 10 && toupper (CurC) == 'E')) {
unsigned Digits;
unsigned Exp;
/* Skip the exponent notifier */
NextChar ();
/* Read an optional sign */
if (CurC == '-') {
NextChar ();
} else if (CurC == '+') {
NextChar ();
}
/* Read exponent digits. Since we support only 32 bit floats
** with a maximum exponent of +-/127, we read the exponent
** part as integer with up to 3 digits and drop the remainder.
** This avoids an overflow of Exp. The exponent is always
** decimal, even for hex float consts.
*/
Digits = 0;
Exp = 0;
while (IsDigit (CurC)) {
if (++Digits <= 3) {
Exp = Exp * 10 + HexVal (CurC);
}
NextChar ();
}
/* Check for errors: We must have exponent digits, and not more
** than three.
*/
if (Digits == 0) {
Error ("Floating constant exponent has no digits");
} else if (Digits > 3) {
Warning ("Floating constant exponent is too large");
}
/* Scale the exponent and adjust the value accordingly */
if (Exp) {
FVal = FP_D_Mul (FVal, FP_D_Make (pow (10, Exp)));
}
}
/* Check for a suffix and determine the type of the constant */
if (toupper (CurC) == 'F') {
NextChar ();
NextTok.Type = type_float;
} else {
NextTok.Type = type_double;
}
/* Set the value and the token */
NextTok.FVal = FVal;
NextTok.Tok = TOK_FCONST;
}
}
static void ParsePragma (void)
/* Parse the contents of the _Pragma statement */
{
pragma_t Pragma;
StrBuf Ident = AUTO_STRBUF_INITIALIZER;
/* Create a string buffer from the string literal */
StrBuf B = AUTO_STRBUF_INITIALIZER;
SB_Append (&B, GetLiteralStrBuf (CurTok.SVal));
/* Skip the string token */
NextToken ();
/* Get the pragma name from the string */
SB_SkipWhite (&B);
if (!SB_GetSym (&B, &Ident, "-")) {
Error ("Invalid pragma");
goto ExitPoint;
}
/* Search for the name */
Pragma = FindPragma (&Ident);
/* Do we know this pragma? */
if (Pragma == PRAGMA_ILLEGAL) {
/* According to the ANSI standard, we're not allowed to generate errors
* for unknown pragmas, but warn about them if enabled (the default).
*/
if (IS_Get (&WarnUnknownPragma)) {
Warning ("Unknown pragma `%s'", SB_GetConstBuf (&Ident));
}
goto ExitPoint;
}
/* Check for an open paren */
SB_SkipWhite (&B);
if (SB_Get (&B) != '(') {
Error ("'(' expected");
goto ExitPoint;
}
/* Skip white space before the argument */
SB_SkipWhite (&B);
/* Switch for the different pragmas */
switch (Pragma) {
case PRAGMA_ALIGN:
IntPragma (&B, &DataAlignment, 1, 4096);
break;
case PRAGMA_BSSSEG:
Warning ("#pragma bssseg is obsolete, please use #pragma bss-name instead");
/* FALLTHROUGH */
case PRAGMA_BSS_NAME:
SegNamePragma (&B, SEG_BSS);
break;
case PRAGMA_CHARMAP:
CharMapPragma (&B);
break;
case PRAGMA_CHECKSTACK:
Warning ("#pragma checkstack is obsolete, please use #pragma check-stack instead");
/* FALLTHROUGH */
case PRAGMA_CHECK_STACK:
FlagPragma (&B, &CheckStack);
break;
case PRAGMA_CODESEG:
Warning ("#pragma codeseg is obsolete, please use #pragma code-name instead");
/* FALLTHROUGH */
case PRAGMA_CODE_NAME:
SegNamePragma (&B, SEG_CODE);
break;
case PRAGMA_CODESIZE:
IntPragma (&B, &CodeSizeFactor, 10, 1000);
break;
case PRAGMA_DATASEG:
Warning ("#pragma dataseg is obsolete, please use #pragma data-name instead");
/* FALLTHROUGH */
case PRAGMA_DATA_NAME:
SegNamePragma (&B, SEG_DATA);
break;
case PRAGMA_LOCAL_STRINGS:
FlagPragma (&B, &LocalStrings);
break;
case PRAGMA_OPTIMIZE:
FlagPragma (&B, &Optimize);
break;
case PRAGMA_REGVARADDR:
FlagPragma (&B, &AllowRegVarAddr);
break;
case PRAGMA_REGVARS:
Warning ("#pragma regvars is obsolete, please use #pragma register-vars instead");
/* FALLTHROUGH */
case PRAGMA_REGISTER_VARS:
FlagPragma (&B, &EnableRegVars);
break;
case PRAGMA_RODATASEG:
Warning ("#pragma rodataseg is obsolete, please use #pragma rodata-name instead");
/* FALLTHROUGH */
case PRAGMA_RODATA_NAME:
SegNamePragma (&B, SEG_RODATA);
break;
case PRAGMA_SIGNEDCHARS:
Warning ("#pragma signedchars is obsolete, please use #pragma signed-chars instead");
/* FALLTHROUGH */
case PRAGMA_SIGNED_CHARS:
FlagPragma (&B, &SignedChars);
break;
case PRAGMA_STATICLOCALS:
Warning ("#pragma staticlocals is obsolete, please use #pragma static-locals instead");
/* FALLTHROUGH */
case PRAGMA_STATIC_LOCALS:
FlagPragma (&B, &StaticLocals);
break;
case PRAGMA_WARN:
WarnPragma (&B);
break;
case PRAGMA_WRITABLE_STRINGS:
FlagPragma (&B, &WritableStrings);
break;
case PRAGMA_ZPSYM:
StringPragma (&B, MakeZPSym);
break;
default:
Internal ("Invalid pragma");
}
/* Closing paren expected */
SB_SkipWhite (&B);
if (SB_Get (&B) != ')') {
Error ("')' expected");
goto ExitPoint;
}
SB_SkipWhite (&B);
/* Allow an optional semicolon to be compatible with the old syntax */
if (SB_Peek (&B) == ';') {
SB_Skip (&B);
SB_SkipWhite (&B);
}
/* Make sure nothing follows */
if (SB_Peek (&B) != '\0') {
Error ("Unexpected input following pragma directive");
}
ExitPoint:
/* Release the string buffers */
SB_Done (&B);
SB_Done (&Ident);
}
static int ParseChar (StrBuf* B)
/* Parse a character. Converts \n into EOL, etc. */
{
unsigned I;
unsigned Val;
int C;
/* Check for escape chars */
if ((C = SB_Get (B)) == '\\') {
switch (SB_Peek (B)) {
case '?':
C = '?';
SB_Skip (B);
break;
case 'a':
C = '\a';
SB_Skip (B);
break;
case 'b':
C = '\b';
SB_Skip (B);
break;
case 'f':
C = '\f';
SB_Skip (B);
break;
case 'r':
C = '\r';
SB_Skip (B);
break;
case 'n':
C = '\n';
SB_Skip (B);
break;
case 't':
C = '\t';
SB_Skip (B);
break;
case 'v':
C = '\v';
SB_Skip (B);
break;
case '\"':
C = '\"';
SB_Skip (B);
break;
case '\'':
C = '\'';
SB_Skip (B);
break;
case '\\':
C = '\\';
SB_Skip (B);
break;
case 'x':
case 'X':
/* Hex character constant */
SB_Skip (B);
C = HexVal (SB_Get (B)) << 4;
C |= HexVal (SB_Get (B));
break;
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
/* Octal constant */
I = 0;
Val = SB_Get (B) - '0';
while (SB_Peek (B) >= '0' && SB_Peek (B) <= '7' && ++I <= 3) {
Val = (Val << 3) | (SB_Get (B) - '0');
}
C = (int) Val;
if (Val > 256) {
Error ("Character constant out of range");
C = ' ';
}
break;
default:
Error ("Illegal character constant 0x%02X", SB_Get (B));
C = ' ';
break;
}
}
/* Return the character */
return C;
}
