6502
/
cc65
mirror of https://github.com/cc65/cc65.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
cc65/src/cc65/preproc.c

3410 lines
103 KiB
C
Raw Permalink Blame History

/*****************************************************************************/
/* */
/* preproc.c */
/* */
/* cc65 preprocessor */
/* */
/* */
/* */
/* (C) 1998-2010, Ullrich von Bassewitz */
/* Roemerstrasse 52 */
/* D-70794 Filderstadt */
/* EMail: uz@cc65.org */
/* */
/* */
/* This software is provided 'as-is', without any expressed or implied */
/* warranty. In no event will the authors be held liable for any damages */
/* arising from the use of this software. */
/* */
/* Permission is granted to anyone to use this software for any purpose, */
/* including commercial applications, and to alter it and redistribute it */
/* freely, subject to the following restrictions: */
/* */
/* 1. The origin of this software must not be misrepresented; you must not */
/* claim that you wrote the original software. If you use this software */
/* in a product, an acknowledgment in the product documentation would be */
/* appreciated but is not required. */
/* 2. Altered source versions must be plainly marked as such, and must not */
/* be misrepresented as being the original software. */
/* 3. This notice may not be removed or altered from any source */
/* distribution. */
/* */
/*****************************************************************************/
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <errno.h>
/* common */
#include "chartype.h"
#include "check.h"
#include "inline.h"
#include "print.h"
#include "xmalloc.h"
/* cc65 */
#include "codegen.h"
#include "error.h"
#include "global.h"
#include "ident.h"
#include "incpath.h"
#include "input.h"
#include "lineinfo.h"
#include "macrotab.h"
#include "ppexpr.h"
#include "preproc.h"
#include "scanner.h"
#include "scanstrbuf.h"
#include "standard.h"
/*****************************************************************************/
/* Data */
/*****************************************************************************/
/* Macro scanner mode flags */
#define MSM_NONE 0x00U /* Default */
#define MSM_MULTILINE 0x01U /* Read from multiple lines */
#define MSM_IN_DIRECTIVE 0x02U /* In PP directives scan */
#define MSM_IN_ARG_LIST 0x04U /* In macro argument scan */
#define MSM_IN_ARG_EXPANSION 0x08U /* In expansion on arguments */
#define MSM_OP_DEFINED 0x10U /* Handle the "defined" operator */
#define MSM_OP_HAS_INCLUDE 0x20U /* Handle the "__has_include" operator */
#define MSM_OP_HAS_C_ATTRIBUTE 0x40U /* Handle the "__has_c_attribute" operator */
#define MSM_TOK_HEADER 0x80U /* Support header tokens */
/* Macro expansion state flags */
#define MES_NONE 0x00U /* Nothing */
#define MES_FIRST_TOKEN 0x01U /* Mark for detecting pp-token count in the sequence */
#define MES_MULTIPLE_TOKEN 0x02U /* Multiple pp-tokens are detected in the sequence */
#define MES_BEGIN_WITH_IDENT 0x04U /* The first pp-token of the sequence is an identifier */
#define MES_HAS_REPLACEMENT 0x10U /* Macro argument has cached replacement result */
#define MES_NO_VA_COMMA 0x20U /* Variadic macro called w/o the ',' in front of variable argument */
#define MES_ERROR 0x80U /* Error has occurred in macro expansion */
/* Management data for #if */
#define IFCOND_NONE 0x00U
#define IFCOND_SKIP 0x01U
#define IFCOND_ELSE 0x02U
#define IFCOND_NEEDTERM 0x04U
/* Current PP if stack */
static PPIfStack* PPStack;
/* Struct for rescan */
typedef struct RescanInputStack RescanInputStack;
struct RescanInputStack {
Collection Lines;
Collection LastTokLens;
StrBuf* PrevTok;
};
/* Input backup for rescan */
static RescanInputStack* CurRescanStack;
/* Intermediate input buffers */
static StrBuf* PLine; /* Buffer for macro expansion */
static StrBuf* MLine; /* Buffer for macro expansion in #pragma */
static StrBuf* OLine; /* Buffer for #pragma output */
/* Newlines to be added to preprocessed text */
static unsigned PendingNewLines;
static unsigned ContinuedLines;
static int FileChanged;
/* Structure used when expanding macros */
typedef struct MacroExp MacroExp;
struct MacroExp {
Collection Args; /* Actual arguments (for function-like) */
Collection HideSets; /* Macros hidden from expansion */
StrBuf Tokens; /* Originally read sequence */
unsigned IdentCount; /* Count of identifiers in the pp-token sequence */
unsigned Flags; /* Macro argument flags */
MacroExp* Replaced; /* Macro-replaced version of this pp-token sequence */
unsigned FirstTokLen; /* Length of the first pp-token */
unsigned LastTokLen; /* Length of the last pp-token */
};
typedef struct HideRange HideRange;
struct HideRange
{
HideRange* Next;
unsigned Start;
unsigned End;
};
typedef struct HiddenMacro HiddenMacro;
struct HiddenMacro {
const Macro* M;
HideRange* HS;
};
/*****************************************************************************/
/* Forwards */
/*****************************************************************************/
static void TranslationPhase3 (StrBuf* Source, StrBuf* Target);
/* Mimic Translation Phase 3. Handle old and new style comments. Collapse
** non-newline whitespace sequences.
*/
static void PreprocessDirective (StrBuf* Source, StrBuf* Target, unsigned ModeFlags);
/* Preprocess a single line. Handle specified tokens and operators, remove
** whitespace and comments, then do macro replacement.
*/
static int ParseDirectives (unsigned ModeFlags);
/* Handle directives. Return 1 if any whitespace or newlines are parsed. */
static unsigned ReplaceMacros (StrBuf* Source, StrBuf* Target, MacroExp* E, unsigned ModeFlags);
/* Scan for and perform macro replacement. Return the count of identifiers and
** right parentheses in the replacement result.
*/
static MacroExp* InitMacroExp (MacroExp* E);
/* Initialize a MacroExp structure */
static void DoneMacroExp (MacroExp* E);
/* Cleanup after use of a MacroExp structure */
static int CheckPastePPTok (StrBuf* Target, unsigned TokLen, char Next);
/* Return 1 if the last pp-tokens from Source could be concatenated with any
** characters from Appended to form a new valid one.
*/
static void LazyCheckNextPPTok (const StrBuf* Prev, unsigned LastTokLen);
/* Memorize the previous pp-token(s) to later check for potential pp-token
** concatenation.
*/
/*****************************************************************************/
/* Low level preprocessor token handling */
/*****************************************************************************/
/* Types of preprocessor directives */
typedef enum {
PPD_ILLEGAL = -1,
PPD_DEFINE,
PPD_ELIF,
PPD_ELSE,
PPD_ENDIF,
PPD_ERROR,
PPD_IF,
PPD_IFDEF,
PPD_IFNDEF,
PPD_INCLUDE,
PPD_LINE,
PPD_PRAGMA,
PPD_UNDEF,
PPD_WARNING,
} ppdirective_t;
/* Preprocessor directive tokens mapping table */
static const struct PPDType {
const char* Tok; /* Token */
ppdirective_t Type; /* Type */
} PPDTypes[] = {
{ "define", PPD_DEFINE },
{ "elif", PPD_ELIF },
{ "else", PPD_ELSE },
{ "endif", PPD_ENDIF },
{ "error", PPD_ERROR },
{ "if", PPD_IF },
{ "ifdef", PPD_IFDEF },
{ "ifndef", PPD_IFNDEF },
{ "include", PPD_INCLUDE },
{ "line", PPD_LINE },
{ "pragma", PPD_PRAGMA },
{ "undef", PPD_UNDEF },
{ "warning", PPD_WARNING },
};
/* Number of preprocessor directive types */
#define PPDTOKEN_COUNT (sizeof(PPDTypes) / sizeof(PPDTypes[0]))
static int CmpToken (const void* Key, const void* Elem)
/* Compare function for bsearch */
{
return strcmp ((const char*) Key, ((const struct PPDType*) Elem)->Tok);
}
static ppdirective_t FindPPDirectiveType (const char* Ident)
/* Find a preprocessor directive type and return it. Return PPD_ILLEGAL if the
** identifier is not a valid preprocessor directive token.
*/
{
struct PPDType* P;
P = bsearch (Ident, PPDTypes, PPDTOKEN_COUNT, sizeof (PPDTypes[0]), CmpToken);
return P? P->Type : PPD_ILLEGAL;
}
/*****************************************************************************/
/* MacroExp helpers */
/*****************************************************************************/
static HideRange* NewHideRange (unsigned Start, unsigned Len)
/* Create a hide range */
{
HideRange* HS = xmalloc (sizeof (HideRange));
HS->Next = 0;
HS->Start = Start;
HS->End = Start + Len;
return HS;
}
static void FreeHideRange (HideRange* HS)
/* Free a hide range */
{
xfree (HS);
}
static HiddenMacro* NewHiddenMacro (const Macro* M)
/* Create a new struct for the hidden macro */
{
HiddenMacro* MHS = xmalloc (sizeof (HiddenMacro));
MHS->M = M;
MHS->HS = 0;
return MHS;
}
static void FreeHiddenMacro (HiddenMacro* MHS)
/* Free the struct and all ranges of the hidden macro */
{
HideRange* This;
HideRange* Next;
if (MHS == 0) {
return;
}
for (This = MHS->HS; This != 0; This = Next) {
Next = This->Next;
FreeHideRange (This);
}
xfree (MHS);
}
/*****************************************************************************/
/* struct MacroExp */
/*****************************************************************************/
static HiddenMacro* ME_FindHiddenMacro (const MacroExp* E, const Macro* M)
/* Find the macro hide set */
{
unsigned I;
for (I = 0; I < CollCount (&E->HideSets); ++I) {
HiddenMacro* MHS = CollAtUnchecked (&E->HideSets, I);
if (MHS->M == M) {
return MHS;
}
}
return 0;
}
static void ME_HideMacro (unsigned Idx, unsigned Count, MacroExp* E, const Macro* M)
/* Hide the macro from the Idx'th identifier */
{
if (Count > 0) {
/* Find the macro hideset */
HiddenMacro* MHS = ME_FindHiddenMacro (E, M);
HideRange** This;
/* New hidden section */
HideRange* NewHS = NewHideRange (Idx, Count);
/* New macro to hide */
if (MHS == 0) {
MHS = NewHiddenMacro (M);
CollAppend (&E->HideSets, MHS);
}
This = &MHS->HS;
if (*This == 0) {
*This = NewHS;
} else {
/* Insert */
while (1) {
if (*This == 0 || NewHS->Start <= (*This)->Start) {
/* Insert before */
NewHS->Next = *This;
*This = NewHS;
break;
} else if (NewHS->Start <= (*This)->End) {
/* Insert after */
NewHS->Next = (*This)->Next;
(*This)->Next = NewHS;
break;
}
/* Advance */
This = &(*This)->Next;
}
/* Merge */
while (*This != 0) {
HideRange* Next = (*This)->Next;
if (Next != 0 && (*This)->End >= Next->Start) {
/* Expand this to the next */
if ((*This)->End < Next->End) {
(*This)->End = Next->End;
}
/* Remove next */
(*This)->Next = Next->Next;
FreeHideRange (Next);
/* Advance */
This = &(*This)->Next;
} else {
/* No more */
break;
}
}
}
}
}
static int ME_CanExpand (unsigned Idx, const MacroExp* E, const Macro* M)
/* Return 1 if the macro can be expanded with the Idx'th identifier */
{
if (E != 0) {
/* Find the macro hideset */
HiddenMacro* MHS = ME_FindHiddenMacro (E, M);
if (MHS != 0) {
/* Check if the macro is hidden from this identifier */
HideRange* HS = MHS->HS;
while (HS != 0) {
/* If the macro name overlaps with the range where the macro is hidden,
** the macro cannot be expanded.
*/
if (Idx >= HS->Start && Idx < HS->End) {
return 0;
}
HS = HS->Next;
}
}
}
return 1;
}
static void ME_OffsetHideSets (unsigned Idx, unsigned Offs, MacroExp* E)
/* Adjust all macro hide set ranges for the macro expansion when the identifier
** at Idx is replaced with a count of Offs + 1 (if Offs > 0) of identifiers.
*/
{
if (Offs != 0) {
unsigned I;
for (I = 0; I < CollCount (&E->HideSets); ++I) {
HiddenMacro* MHS = CollAtUnchecked (&E->HideSets, I);
HideRange* This;
for (This = MHS->HS; This != 0; This = This->Next) {
if (Idx < This->Start) {
This->Start += Offs;
This->End += Offs;
} else if (Idx < This->End) {
This->End += Offs;
}
}
}
}
}
static void ME_RemoveToken (unsigned Idx, unsigned Count, MacroExp* E)
/* Remove the Idx'th identifier token from tracking and offset all hidden
** ranges accordingly.
*/
{
unsigned I;
for (I = 0; I < CollCount (&E->HideSets); ++I) {
HiddenMacro* MHS = CollAtUnchecked (&E->HideSets, I);
HideRange* This;
HideRange** Prev;
for (Prev = &MHS->HS, This = *Prev; This != 0; This = *Prev) {
if (Idx < This->Start) {
if (This->Start - Idx >= Count) {
This->Start -= Count;
This->End -= Count;
} else {
if (This->End - Idx > Count) {
This->Start = Idx;
This->End -= Count;
} else {
/* Remove */
(*Prev) = This->Next;
FreeHideRange (This);
continue;
}
}
} else if (Idx < This->End) {
if (This->End - Idx > Count) {
This->End -= Count;
} else {
This->End = Idx;
}
if (This->End == This->Start) {
/* Remove */
(*Prev) = This->Next;
FreeHideRange (This);
continue;
}
}
Prev = &This->Next;
}
}
}
static void ME_HandleSemiNestedMacro (unsigned NameIdx, unsigned LastIdx, MacroExp* E)
/* Unhide the macro name from all hidesets if it was expanded with an unhidden
** right parenthesis. This is unspecified but allowed behavior according to
** ISO/IEC 9899:2018, 6.10.3.4ff.
*/
{
unsigned I;
for (I = 0; I < CollCount (&E->HideSets); ++I) {
HiddenMacro* MHS = CollAtUnchecked (&E->HideSets, I);
HideRange* This;
HideRange** Prev;
for (Prev = &MHS->HS, This = *Prev; This != 0; This = *Prev) {
if (NameIdx < This->End) {
if (NameIdx >= This->Start && LastIdx >= This->End) {
This->End = NameIdx;
if (This->End == This->Start) {
/* Remove */
(*Prev) = This->Next;
FreeHideRange (This);
continue;
}
}
Prev = &This->Next;
} else {
break;
}
}
}
}
static void ME_AddArgHideSets (unsigned Idx, const MacroExp* A, MacroExp* Parent)
/* Propagate the macro hide sets of the substituted argument starting as the
** Idx'th identifier of the result.
*/
{
unsigned I;
/* Move the hide set generated with in the argument as it will be freed later */
for (I = 0; I < CollCount (&A->HideSets); ++I) {
HiddenMacro* MHS = CollAtUnchecked (&A->HideSets, I);
HideRange* HS;
for (HS = MHS->HS; HS != 0; HS = HS->Next) {
ME_HideMacro (Idx + HS->Start, HS->End - HS->Start, Parent, MHS->M);
}
}
}
static void ME_DoneHideSets (MacroExp* E)
/* Free all of hidden macros for the macro expansion */
{
unsigned I;
for (I = 0; I < CollCount (&E->HideSets); ++I) {
HiddenMacro* MHS = CollAtUnchecked (&E->HideSets, I);
FreeHiddenMacro (MHS);
}
DoneCollection (&E->HideSets);
}
static void ME_SetTokLens (MacroExp* E, unsigned TokLen)
/* Set token lengths and flags for macro expansion struct */
{
E->LastTokLen = TokLen;
if ((E->Flags & MES_FIRST_TOKEN) != 0) {
E->Flags &= ~MES_FIRST_TOKEN;
E->FirstTokLen = E->LastTokLen;
} else {
E->Flags |= MES_MULTIPLE_TOKEN;
}
}
static MacroExp* ME_MakeReplaced (MacroExp* A)
/* Make a replaced version of the argument */
{
/* Replace the parameter with actual argument tokens */
if ((A->Flags & MES_HAS_REPLACEMENT) == 0) {
A->Replaced = xmalloc (sizeof (MacroExp));
InitMacroExp (A->Replaced);
SB_Reset (&A->Tokens);
/* Propagate the hide sets */
ME_AddArgHideSets (0, A, A->Replaced);
/* Do macro expansion on the argument */
A->Replaced->IdentCount = ReplaceMacros (&A->Tokens,
&A->Replaced->Tokens,
A->Replaced,
MSM_IN_ARG_EXPANSION);
A->Flags |= MES_HAS_REPLACEMENT;
}
return A->Replaced != 0 ? A->Replaced : A;
}
static MacroExp* ME_GetOriginalArg (const MacroExp* E, unsigned Index)
/* Return an actual macro argument with the given index */
{
return CollAt (&E->Args, Index);
}
static MacroExp* ME_GetReplacedArg (const MacroExp* E, unsigned Index)
/* Return a replaced macro argument with the given index */
{
return ME_MakeReplaced (CollAt (&E->Args, Index));
}
static MacroExp* ME_AppendArg (MacroExp* E, MacroExp* Arg)
/* Add a copy of Arg to the list of actual macro arguments.
** NOTE: This function will clear the token sequence of Arg!
*/
{
MacroExp* A = xmalloc (sizeof (MacroExp));
/* Initialize our MacroExp structure */
InitMacroExp (A);
/* Copy info about the original strings */
A->IdentCount = Arg->IdentCount;
A->Flags = Arg->Flags;
A->FirstTokLen = Arg->FirstTokLen;
A->LastTokLen = Arg->LastTokLen;
/* Move the contents of Arg to A */
SB_Move (&A->Tokens, &Arg->Tokens);
/* Add A to the list of actual arguments */
CollAppend (&E->Args, A);
return A;
}
static void ME_ClearArgs (MacroExp* E)
/* Clear all read arguments for macro expansion */
{
unsigned I;
/* Delete the list with actual arguments */
for (I = 0; I < CollCount (&E->Args); ++I) {
MacroExp* A = CollAtUnchecked (&E->Args, I);
/* Destroy the macro expansion structure and then free memory allocated
** for it.
*/
DoneMacroExp (A);
xfree (A);
}
DoneCollection (&E->Args);
InitCollection (&E->Args);
}
static int ME_IsNextArgVariadic (const MacroExp* E, const Macro* M)
/* Return true if the next actual argument we will add is a variadic one */
{
return (M->Variadic &&
M->ParamCount == (int) CollCount (&E->Args) + 1);
}
static MacroExp* InitMacroExp (MacroExp* E)
/* Initialize a MacroExp structure */
{
InitCollection (&E->Args);
InitCollection (&E->HideSets);
SB_Init (&E->Tokens);
E->IdentCount = 0;
E->Flags = MES_FIRST_TOKEN;
E->Replaced = 0;
E->FirstTokLen = 0;
E->LastTokLen = 0;
return E;
}
static void DoneMacroExp (MacroExp* E)
/* Cleanup after use of a MacroExp structure */
{
ME_ClearArgs (E);
ME_DoneHideSets (E);
SB_Done (&E->Tokens);
if (E->Replaced != 0) {
DoneMacroExp (E->Replaced);
}
}
/*****************************************************************************/
/* Rescan input stack */
/*****************************************************************************/
static void PushRescanLine (RescanInputStack* RIS, StrBuf* L, unsigned LastTokLen)
/* Push an input line to the rescan input stack */
{
CollAppend (&RIS->Lines, L);
/* Abuse the pointer to store an unsigned */
CollAppend (&RIS->LastTokLens, (void*)(uintptr_t)LastTokLen);
}
static void PopRescanLine (void)
/* Pop and free a rescan input line if it reaches the end */
{
if (CurC == '\0' && CollCount (&CurRescanStack->Lines) > 1) {
FreeStrBuf (CollPop (&CurRescanStack->Lines));
InitLine (CollLast (&CurRescanStack->Lines));
CollPop (&CurRescanStack->LastTokLens);
}
}
static void InitRescanInputStack (RescanInputStack* RIS)
/* Init a RescanInputStack struct */
{
InitCollection (&RIS->Lines);
InitCollection (&RIS->LastTokLens);
RIS->PrevTok = 0;
}
static void DoneRescanInputStack (RescanInputStack* RIS)
/* Free a RescanInputStack struct. RIS must be non-NULL. */
{
/* Free pushed input lines */
while (CollCount (&RIS->Lines) > 1) {
FreeStrBuf (CollPop (&RIS->Lines));
}
/* Switch back to the old input stack */
InitLine (CollPop (&RIS->Lines));
/* Free any remaining pp-tokens used for concatenation check */
FreeStrBuf (RIS->PrevTok);
/* Done */
DoneCollection (&RIS->Lines);
DoneCollection (&RIS->LastTokLens);
}
/*****************************************************************************/
/* Code */
/*****************************************************************************/
static int MacName (char* Ident)
/* Get a macro symbol name into Ident. If we have an error, print a
** diagnostic message and clear the line.
*/
{
if (IsSym (Ident) == 0) {
if (CurC != '\0') {
PPError ("Macro name must be an identifier");
} else {
PPError ("Missing macro name");
}
ClearLine ();
return 0;
} else {
return 1;
}
}
static void CheckForBadIdent (const char* Ident, int Std, const Macro* M)
/* Check for and warning on problematic identifiers */
{
if (Std >= STD_C99 &&
(M == 0 || !M->Variadic) &&
strcmp (Ident, "__VA_ARGS__") == 0) {
/* __VA_ARGS__ cannot be used as a macro parameter name in post-C89
** mode.
*/
PPWarning ("__VA_ARGS__ can only appear in the expansion of a C99 variadic macro");
}
}
static void AddPreLine (StrBuf* Str)
/* Add newlines to the string buffer */
{
/* No need to prettify the non-exist output */
if (!PreprocessOnly) {
PendingNewLines = 0;
ContinuedLines = 0;
return;
}
/* We'll adjust the line number later if necessary */
PendingNewLines += ContinuedLines;
if (FileChanged || PendingNewLines > 6) {
/* Output #line directives as source info */
StrBuf Comment = AUTO_STRBUF_INITIALIZER;
if (SB_NotEmpty (Str) && SB_LookAtLast (Str) != '\n') {
SB_AppendChar (Str, '\n');
}
SB_Printf (&Comment, "#line %u \"%s\"\n",
GetCurrentLineNum () - ContinuedLines, GetCurrentFileName ());
SB_Append (Str, &Comment);
} else {
/* Output new lines */
while (PendingNewLines > 0) {
SB_AppendChar (Str, '\n');
--PendingNewLines;
}
}
FileChanged = 0;
PendingNewLines = 0;
ContinuedLines = 0;
}
static void AppendIndent (StrBuf* Str, int Count)
/* Add Count of spaces ' ' to the string buffer */
{
while (Count > 0) {
SB_AppendChar (Str, ' ');
--Count;
}
}
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 void OldStyleComment (void)
/* Remove an old style C comment from line */
{
/* Remember the current line number, so we can output better error
** messages if the comment is not terminated in the current file.
*/
unsigned StartingLine = GetCurrentLineNum ();
/* Skip the start of comment chars */
NextChar ();
NextChar ();
/* Skip the comment */
while (CurC != '*' || NextC != '/') {
if (CurC == '\0') {
if (NextLine () == 0) {
PPError ("End-of-file reached in comment starting at line %u",
StartingLine);
return;
}
++PendingNewLines;
} else {
if (CurC == '/' && NextC == '*') {
PPWarning ("'/*' found inside a comment");
}
NextChar ();
}
}
/* Skip the end of comment chars */
NextChar ();
NextChar ();
}
static void NewStyleComment (void)
/* Remove a new style C comment from line */
{
/* Diagnose if this is unsupported */
if (IS_Get (&Standard) < STD_C99 && !AllowNewComments) {
PPError ("C++ style comments are not allowed in C89");
PPNote ("(this will be reported only once per input file)");
AllowNewComments = 1;
}
/* Beware: Because line continuation chars are handled when reading
** lines, we may only skip until the end of the source line, which
** may not be the same as the end of the input line. The end of the
** source line is denoted by a lf (\n) character.
*/
do {
NextChar ();
} while (CurC != '\n' && CurC != '\0');
if (CurC == '\n') {
NextChar ();
}
}
static int SkipWhitespace (int SkipLines)
/* Skip white space and comments in the input stream. If skipLines is true,
** also skip newlines and add that count to global PendingNewLines. Return 1
** if the last skipped character was a white space other than a newline '\n',
** otherwise return -1 if there were any newline characters skipped, otherwise
** return 0 if nothing was skipped.
*/
{
int Skipped = 0;
int NewLine = 0;
/* Rescanning */
while (1) {
if (IsSpace (CurC)) {
NextChar ();
Skipped = 1;
} else if (CurC == '/' && NextC == '*') {
OldStyleComment ();
Skipped = 1;
} else if (CurC == '/' && NextC == '/') {
NewStyleComment ();
Skipped = 1;
} else if (CurC == '\0') {
/* End of line */
if (CurRescanStack != 0 &&
CollCount (&CurRescanStack->Lines) > 1 &&
Line == CollLast (&CurRescanStack->Lines)) {
unsigned LastTokLen = (unsigned)(uintptr_t)CollLast (&CurRescanStack->LastTokLens);
/* Check for potentially merged tokens */
if (Skipped == 0 && LastTokLen != 0) {
/* Get the following input */
StrBuf* Next = CollAtUnchecked (&CurRescanStack->Lines,
CollCount (&CurRescanStack->Lines) - 2);
char C = SB_Peek (Next);
/* We cannot check right now if the next pp-token may be a
** macro.
*/
if (IsIdent (C)) {
/* Memorize the previous pp-token and check it later */
LazyCheckNextPPTok (Line, LastTokLen);
} else if (C != '\0' && !IsSpace (C)) {
/* If the two adjacent pp-tokens could be put together
** to form a new one, we have to separate them with an
** additional space.
*/
Skipped = CheckPastePPTok (Line, LastTokLen, SB_Peek (Next));
}
}
/* switch back to previous input */
PopRescanLine ();
} else if (SkipLines) {
/* Read next line */
if (NextLine () != 0) {
++PendingNewLines;
NewLine = 1;
Skipped = 0;
} else {
/* End of input */
break;
}
} else {
break;
}
} else {
/* No more white space */
break;
}
}
return Skipped != 0 ? Skipped : -(NewLine != 0);
}
static void CopyHeaderNameToken (StrBuf* Target)
/* Copy a header name from the input to Target. */
{
/* Remember the quote character, copy it to the target buffer and skip it */
char Quote = CurC == '"' ? '"' : '>';
SB_AppendChar (Target, CurC);
NextChar ();
/* Copy the characters inside the string */
while (CurC != '\0' && CurC != Quote) {
/* Keep an escaped char */
if (CurC == '\\') {
SB_AppendChar (Target, CurC);
NextChar ();
}
/* Copy the character */
SB_AppendChar (Target, CurC);
NextChar ();
}
/* If we had a terminating quote, copy it */
if (CurC != '\0') {
SB_AppendChar (Target, CurC);
NextChar ();
}
}
static int IsQuotedString (void)
/* Retrun 1 if the incoming characters indicate a string literal or character
** constant, otherwise return 0.
*/
{
return IsQuote (CurC) || IsWideQuoted (CurC, NextC);
}
static void CopyQuotedString (StrBuf* Target)
/* Copy a single or double quoted string from the input to Target. */
{
/* Remember the quote character, copy it to the target buffer and skip it */
char Quote;
if (CurC == 'L') {
SB_AppendChar (Target, CurC);
NextChar ();
}
Quote = CurC;
SB_AppendChar (Target, CurC);
NextChar ();
/* Copy the characters inside the string */
while (CurC != '\0' && CurC != Quote) {
/* Keep an escaped char */
if (CurC == '\\') {
SB_AppendChar (Target, CurC);
NextChar ();
}
/* Copy the character */
SB_AppendChar (Target, CurC);
NextChar ();
}
/* If we had a terminating quote, copy it */
if (CurC != '\0') {
SB_AppendChar (Target, CurC);
NextChar ();
} else {
PPWarning ("Missing terminating %c character", Quote);
}
}
static int GetPunc (char* S)
/* Parse a punctuator token. Return 1 and store the parsed token string into S
** on success, otherwise just return 0.
*/
{
char C;
switch (CurC) {
case '[':
case ']':
case '(':
case ')':
case '{':
case '}':
case '~':
case '?':
case ':':
case ';':
case ',':
/* C */
*S++ = CurC;
NextChar ();
break;
case '=':
case '#':
/* C or CC */
C = *S++ = CurC;
NextChar ();
if (CurC == C) {
*S++ = C;
NextChar ();
}
break;
case '*':
case '/':
case '%':
case '^':
case '!':
/* C or C= */
*S++ = CurC;
NextChar ();
if (CurC == '=') {
*S++ = CurC;
NextChar ();
}
break;
case '+':
case '&':
case '|':
/* C, CC or C= */
C = *S++ = CurC;
NextChar ();
if (CurC == C || CurC == '=') {
*S++ = CurC;
NextChar ();
}
break;
case '<':
case '>':
/* C, CC, C= or CC= */
C = *S++ = CurC;
NextChar ();
if (CurC == C) {
*S++ = CurC;
if (NextC == '=') {
*S++ = NextC;
NextChar ();
}
NextChar ();
} else if (CurC == '=') {
*S++ = CurC;
NextChar ();
}
break;
case '-':
/* C, CC, C= or C> */
*S++ = CurC;
NextChar ();
switch (CurC) {
case '-':
case '=':
case '>':
*S++ = CurC;
NextChar ();
break;
default:
break;
}
break;
case '.':
/* C or CCC */
*S++ = CurC;
NextChar ();
if (CurC == '.' && NextC == '.') {
*S++ = CurC;
*S++ = NextC;
NextChar ();
NextChar ();
}
break;
default:
return 0;
}
*S = '\0';
return 1;
}
static int CheckPastePPTok (StrBuf* Source, unsigned TokLen, char Next)
/* Return 1 if the last pp-tokens from Source could be concatenated with any
** characters from Appended to form a new valid one.
*/
{
char C;
unsigned NewTokLen;
StrBuf* OldSource;
StrBuf Src = AUTO_STRBUF_INITIALIZER;
StrBuf Buf = AUTO_STRBUF_INITIALIZER;
if (TokLen == 0 || IsBlank (SB_LookAtLast (Source))) {
return 0;
}
PRECONDITION (SB_GetLen (Source) >= TokLen);
/* Special casing "..", "/ /" and "/ *" that are not pp-tokens but still
** need be separated.
*/
C = SB_LookAt (Source, SB_GetLen (Source) - TokLen);
if ((C == '.' && Next == '.') || (C == '/' && (Next == '/' || Next == '*'))) {
return 1;
}
SB_CopyBuf (&Src, SB_GetConstBuf (Source) + SB_GetLen (Source) - TokLen, TokLen);
SB_AppendChar (&Src, Next);
SB_Reset (&Src);
OldSource = InitLine (&Src);
if (IsPPNumber (CurC, NextC)) {
/* PP-number */
CopyPPNumber (&Buf);
} else if (IsQuotedString ()) {
/* Quoted string */
CopyQuotedString (&Buf);
} else {
ident Ident;
if (GetPunc (Ident)) {
/* Punctuator */
SB_CopyStr (&Buf, Ident);
} else if (IsSym (Ident)) {
/* Identifier */
SB_CopyStr (&Buf, Ident);
}
}
NewTokLen = SB_GetLen (&Buf);
SB_Done (&Buf);
SB_Done (&Src);
/* Restore old source */
InitLine (OldSource);
/* Return if concatenation succeeded */
return NewTokLen != TokLen;
}
static int TryPastePPTok (StrBuf* Target,
StrBuf* Appended,
unsigned FirstTokLen,
unsigned SecondTokLen)
/* Paste the whole appened pp-token sequence onto the end of the target
** pp-token sequence. Diagnose if it fails to form a valid pp-token with the
** two pp-tokens pasted together. Return 1 if succeeds.
*/
{
unsigned TokLen;
StrBuf* OldSource;
StrBuf Src = AUTO_STRBUF_INITIALIZER;
StrBuf Buf = AUTO_STRBUF_INITIALIZER;
if (FirstTokLen == 0 || SecondTokLen == 0) {
SB_Append (Target, Appended);
return 1;
}
/* Since we need to concatenate the token sequences, remove the
** last whitespace that was added to target, since it must come
** from the input.
*/
if (IsBlank (SB_LookAtLast (Target))) {
SB_Drop (Target, 1);
}
PRECONDITION (SB_GetLen (Target) >= FirstTokLen &&
SB_GetLen (Appended) >= SecondTokLen);
/* Special casing "..", "/ /" and "/ *" */
if (FirstTokLen == 1) {
char C = SB_LookAt (Target, SB_GetLen (Target) - FirstTokLen);
char N = SB_LookAt (Appended, 0);
/* Avoid forming a comment introducer or an ellipsis. Note that an
** ellipsis pp-token cannot be formed with macros anyway.
*/
if ((C == '.' && N == '.') || (C == '/' && (N == '/' || N == '*'))) {
SB_AppendChar (Target, ' ');
SB_Append (Target, Appended);
PPWarning ("Pasting formed \"%c%c\", an invalid preprocessing token", C, N);
return 0;
}
}
SB_CopyBuf (&Src, SB_GetConstBuf (Target) + SB_GetLen (Target) - FirstTokLen, FirstTokLen);
if (SecondTokLen == SB_GetLen (Appended) || IsSpace (SB_LookAt (Appended, SecondTokLen))) {
SB_AppendBuf (&Src, SB_GetConstBuf (Appended), SecondTokLen);
} else {
SB_AppendBuf (&Src, SB_GetConstBuf (Appended), SecondTokLen + 1);
}
SB_Reset (&Src);
OldSource = InitLine (&Src);
if (IsPPNumber (CurC, NextC)) {
/* PP-number */
CopyPPNumber (&Buf);
} else if (IsQuotedString ()) {
/* Quoted string */
CopyQuotedString (&Buf);
} else {
ident Ident;
if (GetPunc (Ident)) {
/* Punctuator */
SB_CopyStr (&Buf, Ident);
} else if (IsSym (Ident)) {
/* Identifier */
SB_CopyStr (&Buf, Ident);
} else {
/* Unknown */
}
}
TokLen = SB_GetLen (&Buf);
if (TokLen < FirstTokLen + SecondTokLen) {
/* The pasting doesn't form a valid pp-token */
while (SB_GetLen (&Buf) < FirstTokLen + SecondTokLen) {
SB_AppendChar (&Buf, CurC);
NextChar ();
}
SB_Terminate (&Buf);
PPWarning ("Pasting formed \"%s\", an invalid preprocessing token",
SB_GetConstBuf (&Buf));
/* Add a space between the tokens to avoid problems in rescanning */
if (TokLen > FirstTokLen) {
SB_AppendChar (Target, ' ');
}
/* Append all remaining tokens */
SB_Append (Target, Appended);
/* No concatenation */
TokLen = FirstTokLen;
} else {
/* Add a space after the merged token if necessary */
SB_AppendBuf (Target, SB_GetConstBuf (Appended), SecondTokLen);
if (TokLen > FirstTokLen + SecondTokLen) {
SB_AppendChar (Target, ' ');
}
/* Append all remaining tokens */
SB_AppendBuf (Target,
SB_GetConstBuf (Appended) + SecondTokLen,
SB_GetLen (Appended) - SecondTokLen);
}
SB_Done (&Buf);
SB_Done (&Src);
/* Restore old source */
InitLine (OldSource);
/* Return if concatenation succeeded */
return TokLen != FirstTokLen;
}
static void SeparatePPTok (StrBuf* Target, char Next)
/* Add a space to target if the previous pp-token could be concatenated with
** the following character.
*/
{
if (CurRescanStack->PrevTok != 0) {
unsigned Len = SB_GetLen (CurRescanStack->PrevTok) - SB_GetIndex (CurRescanStack->PrevTok);
/* Check for pp-token pasting */
if (CheckPastePPTok (CurRescanStack->PrevTok, Len, Next)) {
SB_AppendChar (Target, ' ');
}
FreeStrBuf (CurRescanStack->PrevTok);
CurRescanStack->PrevTok = 0;
}
}
static void LazyCheckNextPPTok (const StrBuf* Prev, unsigned LastTokLen)
/* Memorize the previous pp-token(s) to later check for potential pp-token
** concatenation.
*/
{
char C;
int CheckEllipsis = 0;
unsigned NewIndex = SB_GetLen (Prev) - LastTokLen;
PRECONDITION (SB_GetLen (Prev) >= LastTokLen);
/* Check for some special cases */
C = SB_AtUnchecked (Prev, NewIndex);
/* We may exclude certain punctuators for speedups. As newer C standards
** could add more punctuators such as "[[", "]]", "::" and so on, this
** check might need changes accordingly.
*/
if (C == '[' || C == ']' || C == '(' || C == ')' ||
C == '{' || C == '}' || C == '~' || C == '?' ||
C == ':' || C == ';' || C == ',') {
/* These punctuators cannot be concatenated */
return;
}
/* Special check for .. */
if (NewIndex > 0 &&
C == '.' &&
SB_AtUnchecked (Prev, NewIndex - 1) == '.') {
/* Save the preceding '.' as well */
CheckEllipsis = 1;
}
if (CurRescanStack->PrevTok != 0) {
unsigned OldIndex = SB_GetIndex (CurRescanStack->PrevTok);
unsigned OldLen = SB_GetLen (CurRescanStack->PrevTok) - OldIndex;
unsigned NewLen = SB_GetLen (Prev) - NewIndex;
if (OldLen == NewLen &&
strncmp (SB_GetConstBuf (CurRescanStack->PrevTok) + OldIndex - CheckEllipsis,
SB_GetConstBuf (Prev) + NewIndex - CheckEllipsis,
OldLen + CheckEllipsis) == 0) {
/* Same pp-token, keep using the old one */
} else {
/* Logic error */
SB_Terminate (CurRescanStack->PrevTok);
Internal ("Unchecked pp-token concatenation: \"%s\"",
SB_GetConstBuf (CurRescanStack->PrevTok) + SB_GetIndex (CurRescanStack->PrevTok));
}
} else {
/* Memorize the current line */
CurRescanStack->PrevTok = NewStrBuf ();
SB_CopyBuf (CurRescanStack->PrevTok,
SB_GetConstBuf (Prev) + NewIndex - CheckEllipsis,
LastTokLen + CheckEllipsis);
SB_Reset (CurRescanStack->PrevTok);
}
}
static int CheckExtraTokens (const char* Name)
/* Check for extra tokens at the end of the directive. Return 1 if there are
** extra tokens, otherwise 0.
*/
{
SkipWhitespace (0);
if (SB_GetIndex (Line) != SB_GetLen (Line)) {
PPWarning ("Extra tokens at end of #%s directive", Name);
return 1;
}
return 0;
}
/*****************************************************************************/
/* Macro stuff */
/*****************************************************************************/
static unsigned ReadMacroArgs (unsigned NameIdx, MacroExp* E, const Macro* M, int MultiLine)
/* Identify the arguments to a macro call as-is. Return the total count of
** identifiers and right parentheses in the read argument list.
*/
{
unsigned Idx = 0;
unsigned CountInArg = 0;
unsigned Parens = 0; /* Number of open parenthesis */
ident Ident;
MacroExp Arg;
InitMacroExp (&Arg);
/* Eat the left paren */
NextChar ();
/* Read the actual macro arguments */
while (1) {
/* Squeeze runs of blanks within an arg */
unsigned OldPendingNewLines = PendingNewLines;
int Skipped = SkipWhitespace (MultiLine);
/* Directives can only be found in an argument list that spans
** multiple lines.
*/
if (MultiLine && OldPendingNewLines < PendingNewLines && CurC == '#') {
unsigned Newlines = 0;
while (OldPendingNewLines < PendingNewLines && CurC == '#') {
Newlines += PendingNewLines - OldPendingNewLines;
PendingNewLines = OldPendingNewLines;
OldPendingNewLines = 0;
Skipped = ParseDirectives (MSM_IN_ARG_LIST) || Skipped;
}
PendingNewLines += Newlines;
}
/* Append a space as a separator */
if (Skipped && SB_NotEmpty (&Arg.Tokens)) {
SB_AppendChar (&Arg.Tokens, ' ');
Skipped = 1;
} else {
Skipped = 0;
}
/* Finish reading the current argument if we are not inside nested
** parentheses or a variadic macro argument.
*/
if (Parens == 0 &&
((CurC == ',' && !ME_IsNextArgVariadic (E, M)) || CurC == ')')) {
/* End of actual argument. Remove whitespace from the end. */
while (IsBlank (SB_LookAtLast (&Arg.Tokens))) {
SB_Drop (&Arg.Tokens, 1);
}
/* If this is not the single empty argument for a macro with an
** empty argument list, remember it.
*/
if (CurC != ')' ||
CollCount (&E->Args) > 0 ||
SB_NotEmpty (&Arg.Tokens) ||
M->ParamCount > 0) {
MacroExp* A = ME_AppendArg (E, &Arg);
unsigned I;
/* Copy the hide sets from the argument list */
for (I = 0; I < CollCount (&E->HideSets); ++I) {
HiddenMacro* MHS = CollAtUnchecked (&E->HideSets, I);
HideRange* HS;
for (HS = MHS->HS; HS != 0; HS = HS->Next) {
/* Get the correct hide range */
unsigned Start = NameIdx + 1 + Idx;
unsigned Len;
if (HS->Start < Start) {
if (HS->End > Start) {
Len = HS->End - Start;
} else {
/* Out of the range */
continue;
}
Start = 0;
} else {
Len = HS->End - HS->Start;
Start = HS->Start - Start;
}
if (Start + Len > Idx + CountInArg) {
if (Idx + CountInArg > Start) {
Len = Idx + CountInArg - Start;
} else {
/* Out of the range */
break;
}
}
ME_HideMacro (Start, Len, A, MHS->M);
}
}
/* More argument info */
A->IdentCount = CountInArg;
}
Idx += CountInArg;
/* Check for end of macro param list */
if (CurC == ')') {
/* Count right parens */
++Idx;
NextChar ();
break;
}
/* Start the next param */
NextChar ();
DoneMacroExp (&Arg);
InitMacroExp (&Arg);
CountInArg = 0;
continue;
} else if (CurC == '\0') {
/* End of input inside macro argument list */
PPError ("Unterminated argument list invoking macro '%s'", M->Name);
Parens = -1;
ClearLine ();
E->Flags |= MES_ERROR;
Idx = 0;
break;
} else {
unsigned LastLen = SB_GetLen (&Arg.Tokens);
if (IsSym (Ident)) {
/* Just copy the identifier */
SB_AppendStr (&Arg.Tokens, Ident);
/* Count identifiers */
++CountInArg;
/* Used for concatentation check */
if ((Arg.Flags & MES_FIRST_TOKEN) != 0) {
Arg.Flags |= MES_BEGIN_WITH_IDENT;
}
} else if (IsPPNumber (CurC, NextC)) {
/* Copy a pp-number */
CopyPPNumber (&Arg.Tokens);
} else if (IsQuotedString ()) {
/* Quoted string - just copy */
CopyQuotedString (&Arg.Tokens);
} else if (GetPunc (Ident)) {
/* Check nested parentheses */
if (Ident[0] == '(') {
/* Opening nested parenthesis */
++Parens;
} else if (Ident[0] == ')') {
/* Closing nested parenthesis */
--Parens;
/* Count right parens */
++CountInArg;
}
/* Just copy the punctuator */
SB_AppendStr (&Arg.Tokens, Ident);
} else {
/* Just copy the character */
SB_AppendChar (&Arg.Tokens, CurC);
NextChar ();
/* But don't count it */
++LastLen;
}
/* Used for concatentation check */
ME_SetTokLens (&Arg, SB_GetLen (&Arg.Tokens) - LastLen - Skipped);
}
}
/* Compare formal and actual argument count */
if (CollCount (&E->Args) < (unsigned) M->ParamCount) {
/* Check further only when the parentheses are paired */
if (Parens == 0) {
/* Specially casing variable argument */
if (M->Variadic &&
M->ParamCount > 0 &&
CollCount (&E->Args) + 1 == (unsigned) M->ParamCount) {
/* The variable argument is left out entirely */
E->Flags |= MES_NO_VA_COMMA;
if (IS_Get (&Standard) < STD_CC65) {
PPWarning ("ISO C does not permit leaving out the comma before the variable argument");
}
} else {
/* Too few argument */
PPError ("Macro \"%s\" passed only %u arguments, but requires %u",
M->Name, CollCount (&E->Args), (unsigned) M->ParamCount);
}
}
/* Be sure to make enough empty arguments available */
DoneMacroExp (&Arg);
InitMacroExp (&Arg);
Arg.Flags |= MES_HAS_REPLACEMENT;
while (CollCount (&E->Args) < (unsigned) M->ParamCount) {
ME_AppendArg (E, &Arg);
}
} else if (Parens == 0 && CollCount (&E->Args) > (unsigned) M->ParamCount) {
/* Too many arguments */
PPError ("Macro \"%s\" passed %u arguments, but takes just %u",
M->Name, CollCount (&E->Args), (unsigned) M->ParamCount);
}
/* Deallocate argument resources */
DoneMacroExp (&Arg);
/* Return the total count of identifiers and right parentheses in the
** argument list.
*/
return Idx;
}
static unsigned SubstMacroArgs (unsigned NameIdx, StrBuf* Target, MacroExp* E, Macro* M, unsigned* IdentCount)
/* Argument substitution according to ISO/IEC 9899:1999 (E), 6.10.3.1ff.
** Return the length of the last pp-token in the result and output the count
** of identifiers and right parentheses in the result to *IdentCount.
*/
{
unsigned Idx = NameIdx;
ident Ident;
unsigned TokLen = 0;
int ParamIdx;
StrBuf* OldSource;
int HaveSpace = 0;
StrBuf Buf = AUTO_STRBUF_INITIALIZER; /* Temporary buffer */
/* Remember the current input stack and disable it for now */
Collection* OldInputStack = UseInputStack (0);
/* Remember the current input and switch to the macro replacement */
unsigned OldIndex = SB_GetIndex (&M->Replacement);
SB_Reset (&M->Replacement);
OldSource = InitLine (&M->Replacement);
/* If the macro expansion replaces an function-like macro with an argument
** list containing a right parenthesis outside the hidesets of previously
** replaced macros, stop those hidesets from this replacement. This is not
** required by the standard but just to match up with other major C
** compilers.
*/
ME_HandleSemiNestedMacro (NameIdx, NameIdx + *IdentCount, E);
/* Substitution loop */
while (CurC != '\0') {
int NeedPaste = 0;
/* If we have an identifier, check if it's a macro */
if (IsSym (Ident)) {
/* Remember and skip any following whitespace */
HaveSpace = SkipWhitespace (0);
/* Check if it's a macro parameter */
if ((ParamIdx = FindMacroParam (M, Ident)) >= 0) {
/* If a ## operator follows, we have to insert the actual
** argument as-is, otherwise it must be macro-replaced.
*/
if (CurC == '#' && NextC == '#') {
/* Get the corresponding actual argument */
const MacroExp* A = ME_GetOriginalArg (E, ParamIdx);
/* Separate with a white space if necessary */
if (CheckPastePPTok (Target, TokLen, SB_Peek (&A->Tokens))) {
SB_AppendChar (Target, ' ');
}
/* For now we need no placemarkers */
SB_Append (Target, &A->Tokens);
/* Adjust tracking */
ME_OffsetHideSets (Idx, A->IdentCount, E);
Idx += A->IdentCount;
/* This will be used for concatenation */
TokLen = A->LastTokLen;
} else {
/* Get the corresponding macro-replaced argument */
const MacroExp* A = ME_GetReplacedArg (E, ParamIdx);
/* Separate with a white space if necessary */
if (CheckPastePPTok (Target, TokLen, SB_Peek (&A->Tokens))) {
SB_AppendChar (Target, ' ');
}
/* Append the replaced string */
SB_Append (Target, &A->Tokens);
/* Insert the range of identifiers to parent preceding this argument */
ME_OffsetHideSets (Idx, A->IdentCount, E);
/* Add hide range */
ME_AddArgHideSets (Idx, A, E);
/* Adjust tracking */
Idx += A->IdentCount;
/* May be used for later pp-token merge check */
TokLen = A->LastTokLen;
}
} else {
/* An identifier, keep it */
SB_AppendStr (Target, Ident);
/* Adjust tracking */
ME_OffsetHideSets (Idx, 1, E);
++Idx;
/* May be used for later concatenation */
TokLen = strlen (Ident);
}
/* Special casing for 'L' prefixing '#' */
if (TokLen == 1 && SB_LookAtLast (Target) == 'L' && CurC == '#') {
HaveSpace = 1;
}
/* Squeeze and add the skipped whitespace back for consistency */
if (HaveSpace && !IsBlank (SB_LookAtLast (Target))) {
SB_AppendChar (Target, ' ');
}
/* Done with this substituted argument */
continue;
} else if (CurC == '#' && NextC == '#') {
/* ## operator */
NextChar ();
NextChar ();
SkipWhitespace (0);
/* If the next token is an identifier which is a macro argument,
** replace it, otherwise just add it.
*/
if (IsSym (Ident)) {
unsigned NewCount = 1;
/* Check if it's a macro parameter */
if ((ParamIdx = FindMacroParam (M, Ident)) >= 0) {
/* Get the corresponding actual argument */
MacroExp* A = ME_GetOriginalArg (E, ParamIdx);
/* Insert the range of identifiers to parent preceding this argument */
ME_OffsetHideSets (Idx, A->IdentCount, E);
/* Add hide range */
ME_AddArgHideSets (Idx, A, E);
/* Adjust tracking */
NewCount = A->IdentCount;
/* If the preceding pp-token is not a placemarker and is
** concatenated to with an identifier, the count of tracked
** identifiers is then one less.
*/
if (TryPastePPTok (Target, &A->Tokens, TokLen, A->FirstTokLen)) {
if (TokLen > 0 && (A->Flags & MES_BEGIN_WITH_IDENT) != 0) {
--NewCount;
ME_RemoveToken (Idx, 1, E);
}
if ((A->Flags & MES_MULTIPLE_TOKEN) == 0) {
TokLen += A->FirstTokLen;
} else {
TokLen = A->LastTokLen;
}
} else {
TokLen = A->LastTokLen;
}
} else {
unsigned Len;
/* Just an ordinary identifier - add as-is */
SB_CopyStr (&Buf, Ident);
/* If the preceding pp-token is not a placemarker and is
** concatenated to with an identifier, the count of tracked
** identifiers is then one less.
*/
Len = SB_GetLen (&Buf);
if (TryPastePPTok (Target, &Buf, TokLen, Len)) {
if (TokLen > 0) {
--NewCount;
}
TokLen += Len;
} else {
TokLen = Len;
}
/* Adjust tracking */
ME_OffsetHideSets (Idx, NewCount, E);
}
/* Adjust tracking */
Idx += NewCount;
/* Keep the whitespace for consistency */
HaveSpace = SkipWhitespace (0);
if (HaveSpace && !IsBlank (SB_LookAtLast (Target))) {
SB_AppendChar (Target, ' ');
}
/* Done with this concatenated identifier */
continue;
}
if (CurC != '\0') {
/* Non-identifiers may still be pasted together */
NeedPaste = 1;
}
}
/* Use the temporary buffer */
SB_Clear (&Buf);
if (IsPPNumber (CurC, NextC)) {
CopyPPNumber (&Buf);
} else if (IsQuotedString ()) {
CopyQuotedString (&Buf);
} else {
if (CurC == '#' && M->ParamCount >= 0) {
/* A # operator within a macro expansion of a function-like
** macro. Read the following identifier and check if it's a
** macro parameter.
*/
NextChar ();
SkipWhitespace (0);
if (!IsSym (Ident) || (ParamIdx = FindMacroParam (M, Ident)) < 0) {
/* Should not happen, but still */
Internal ("'#' is not followed by a macro parameter");
} else {
/* Make a valid string from Replacement */
MacroExp* A = ME_GetOriginalArg (E, ParamIdx);
SB_Reset (&A->Tokens);
Stringize (&A->Tokens, &Buf);
}
} else if (GetPunc (Ident)) {
/* Count right parens. This is OK since they cannot be pasted
** to form different punctuators with others.
*/
if (Ident[0] == ')') {
/* Adjust tracking */
ME_OffsetHideSets (Idx, 1, E);
++Idx;
}
SB_AppendStr (&Buf, Ident);
} else if (CurC != '\0') {
SB_AppendChar (&Buf, CurC);
NextChar ();
}
}
/* Squeeze any whitespace for consistency. Especially, comments must
** be consumed before fed to the punctuator parser, or their leading
** '/' characters would be parsed wrongly as division operators.
*/
HaveSpace = SkipWhitespace (0);
if (NeedPaste) {
unsigned Len = SB_GetLen (&Buf);
/* Concatenate pp-tokens */
if (TryPastePPTok (Target, &Buf, TokLen, Len)) {
TokLen += Len;
} else {
TokLen = Len;
}
} else {
/* Just append the token */
SB_Append (Target, &Buf);
TokLen = SB_GetLen (&Buf);
}
if (HaveSpace && !IsBlank (SB_LookAtLast (Target))) {
SB_AppendChar (Target, ' ');
}
}
/* Done with the temporary buffer */
SB_Done (&Buf);
/* Remove the macro name itself together with the arguments (if any) */
ME_RemoveToken (Idx, 1 + *IdentCount, E);
/* Hide this macro for the whole result of this expansion */
ME_HideMacro (NameIdx, Idx - NameIdx, E, M);
/* Switch back the input */
UseInputStack (OldInputStack);
InitLine (OldSource);
SB_SetIndex (&M->Replacement, OldIndex);
/* Set the count of identifiers and right parentheses in the result */
*IdentCount = Idx - NameIdx;
/* Return the length of the last pp-token */
return TokLen;
}
static unsigned ExpandMacro (unsigned Idx, StrBuf* Target, MacroExp* E, Macro* M, int MultiLine)
/* Expand a macro into Target. Return the length of the last pp-token in the
** result of the expansion.
*/
{
unsigned Count = 0; /* Count of identifiers and right parentheses */
unsigned Len = 0; /* Length of the last pp-token in the result */
/* Disable previous pp-token spacing checking */
StrBuf* PrevTok = CurRescanStack->PrevTok;
CurRescanStack->PrevTok = 0;
#if DEV_CC65_DEBUG
static unsigned V = 0;
printf ("Expanding (%u) %s\n", ++V, M->Name);
#endif
/* Check if this is a function like macro */
if (M->ParamCount >= 0) {
/* Read the actual macro arguments (with the enclosing parentheses) */
Count = ReadMacroArgs (Idx, E, M, MultiLine);
}
if ((E->Flags & MES_ERROR) == 0) {
/* Replace macro parameters with arguments handling the # and ## operators */
Len = SubstMacroArgs (Idx, Target, E, M, &Count);
} else {
SB_CopyStr (Target, M->Name);
}
if (CollCount (&E->Args) > 0) {
/* Clear all arguments */
ME_ClearArgs (E);
}
#if DEV_CC65_DEBUG
printf ("Expanded (%u) %s to %d ident(s) at %u: %s\n",
V--, M->Name, Count, Idx, SB_GetConstBuf (Target));
#endif
/* Reenable previous pp-token concatenation checking */
FreeStrBuf (CurRescanStack->PrevTok);
CurRescanStack->PrevTok = PrevTok;
/* Return the length of the last pp-token in the expansion result */
return Len;
}
static unsigned ReplaceMacros (StrBuf* Source, StrBuf* Target, MacroExp* E, unsigned ModeFlags)
/* Scan for and perform macro replacement. Return the count of identifiers and
** right parentheses in the replacement result.
*/
{
unsigned Count = 0;
StrBuf* TmpTarget = NewStrBuf ();
/* Remember the current input and switch to Source */
StrBuf* OldSource = InitLine (Source);
RescanInputStack RescanStack;
RescanInputStack* OldRescanStack = CurRescanStack;
InitRescanInputStack (&RescanStack);
PushRescanLine (&RescanStack, Line, 0);
CurRescanStack = &RescanStack;
/* Loop substituting macros */
while (CurC != '\0') {
int Skipped = 0;
ident Ident;
/* If we have an identifier, check if it's a macro */
if (IsSym (Ident)) {
/* Check for bad identifier names */
if ((ModeFlags & (MSM_MULTILINE | MSM_IN_DIRECTIVE | MSM_IN_ARG_LIST)) != 0 &&
(CollCount (&CurRescanStack->Lines) == 1 || CurC == '\0')) {
CheckForBadIdent (Ident, IS_Get (&Standard), 0);
}
if ((ModeFlags & MSM_OP_DEFINED) != 0 && strcmp (Ident, "defined") == 0) {
/* Handle the "defined" operator */
int HaveParen = 0;
/* Eat the "defined" operator */
ME_RemoveToken (Count, 1, E);
SkipWhitespace (0);
if (CurC == '(') {
HaveParen = 1;
NextChar ();
SkipWhitespace (0);
}
/* Add a space to separate the result if necessary */
SeparatePPTok (Target, '0');
if (IsSym (Ident)) {
/* Eat the identifier */
ME_RemoveToken (Count, 1, E);
SB_AppendChar (Target, IsMacro (Ident) ? '1' : '0');
if (HaveParen) {
SkipWhitespace (0);
if (CurC != ')') {
PPError ("')' expected");
ClearLine ();
} else {
/* Eat the right parenthesis */
ME_RemoveToken (Count, 1, E);
NextChar ();
}
}
} else {
PPError ("Macro name must be an identifier");
ClearLine ();
SB_AppendChar (Target, '0');
}
} else {
Macro* M = FindMacro (Ident);
/* Check if it's an expandable macro */
if (M != 0 && ME_CanExpand (Count, E, M)) {
int MultiLine = (ModeFlags & MSM_MULTILINE) != 0;
unsigned LastTokLen;
/* Check if this is a function-like macro */
if (M->ParamCount >= 0) {
int HaveSpace = SkipWhitespace (MultiLine) > 0;
/* A function-like macro name without an immediately
** following argument list is not subject to expansion.
*/
if (CurC != '(') {
/* No expansion */
++Count;
/* Add a space to separate the macro name if necessary */
SeparatePPTok (Target, M->Name[0]);
SB_AppendStr (Target, M->Name);
/* Keep tracking pp-token lengths */
if ((ModeFlags & MSM_IN_ARG_EXPANSION) != 0) {
/* Used for concatentation check */
if ((E->Flags & MES_FIRST_TOKEN) != 0) {
E->Flags |= MES_BEGIN_WITH_IDENT;
}
ME_SetTokLens (E, strlen (M->Name));
}
/* Since we have already got on hold of the next
** line, we have to reuse it as the next line
** instead of reading a new line from the source.
*/
if (PendingNewLines > 0 && MultiLine) {
unsigned I = SB_GetIndex (Line);
/* There is no way a function-like macro call
** detection could span multiple lines within
** the range of another just expanded macro.
*/
CHECK (CollCount (&CurRescanStack->Lines) == 1);
/* Revert one newline */
--PendingNewLines;
/* Align indention */
while (I > 0) {
--I;
if (SB_GetBuf (Line)[I] == '\n') {
++I;
break;
}
SB_GetBuf (Line)[I] = ' ';
}
/* Set start index */
SB_SetIndex (Line, I);
/* Add newlines */
AddPreLine (Target);
/* Reuse this line as the next line */
ReuseInputLine ();
/* Quit this loop */
break;
}
/* Append back the whitespace */
if (HaveSpace) {
SB_AppendChar (Target, ' ');
}
/* Loop */
goto Loop;
}
}
/* Either an object-like or function-like macro */
MultiLine = MultiLine && M->ParamCount >= 0;
/* If we were going to support #pragma in macro argument
** list, it would be output to OLine.
*/
if (MultiLine && OLine == 0) {
OLine = TmpTarget;
LastTokLen = ExpandMacro (Count, TmpTarget, E, M, MultiLine);
OLine = 0;
} else {
LastTokLen = ExpandMacro (Count, TmpTarget, E, M, MultiLine);
}
/* Check for errors in expansion */
if ((E->Flags & MES_ERROR) != 0) {
break;
}
/* Pop the current line if it is at the end */
PopRescanLine ();
if (SB_GetLen (TmpTarget) > 0) {
/* Start rescanning from the temporary result */
SB_Reset (TmpTarget);
InitLine (TmpTarget);
PushRescanLine (CurRescanStack, TmpTarget, LastTokLen);
/* Add a space before a '#' at the beginning of the line */
if (CurC == '#' &&
NextC != '#' &&
(SB_IsEmpty (Target) || SB_LookAtLast (Target) == '\n')) {
SB_AppendChar (Target, ' ');
}
/* Switch the buffers */
TmpTarget = NewStrBuf ();
} else if (PendingNewLines > 0 && MultiLine) {
/* Cancel remaining check for pp-tokens separation
** if there is since ther have been newlines that
** can always separate them.
*/
if (CurRescanStack->PrevTok != 0) {
FreeStrBuf (CurRescanStack->PrevTok);
CurRescanStack->PrevTok = 0;
}
/* Squeeze whitespace */
SkipWhitespace (0);
/* Add indention to preprocessor output if needed */
if (CurC != '\0' && CollCount (&CurRescanStack->Lines) == 1) {
/* Add newlines */
AddPreLine (Target);
/* Align indention */
AppendIndent (Target, SB_GetIndex (Line));
}
}
/* Since we are rescanning, we needn't add the
** count of just replaced identifiers right now.
*/
continue;
}
/* An unexpandable identifier. Keep it. */
++Count;
/* Add a space to separate the macro name if necessary */
SeparatePPTok (Target, Ident[0]);
SB_AppendStr (Target, Ident);
/* Keep tracking pp-token lengths */
if ((ModeFlags & MSM_IN_ARG_EXPANSION) != 0) {
/* Used for concatentation check */
if ((E->Flags & MES_FIRST_TOKEN) != 0) {
E->Flags |= MES_BEGIN_WITH_IDENT;
}
ME_SetTokLens (E, strlen (Ident));
}
}
} else {
unsigned LastLen;
/* Add a space to separate the macro name if necessary */
SeparatePPTok (Target, CurC);
LastLen = SB_GetLen (Target);
if ((ModeFlags & MSM_TOK_HEADER) != 0 && (CurC == '<' || CurC == '\"')) {
CopyHeaderNameToken (Target);
} else if (IsPPNumber (CurC, NextC)) {
CopyPPNumber (Target);
} else if (IsQuotedString ()) {
CopyQuotedString (Target);
} else {
/* We want to squeeze whitespace until the end of the current
** input line, so we have to deal with such cases specially.
*/
if (CollCount (&CurRescanStack->Lines) > 1) {
RescanInputStack* RIS = CurRescanStack;
/* Temporarily disable input popping */
CurRescanStack = 0;
Skipped = SkipWhitespace (0);
CurRescanStack = RIS;
if (CurC == '\0') {
/* Now we are at the end of the input line */
goto Loop;
}
} else {
Skipped = SkipWhitespace (0);
}
/* Punctuators must be checked after whitespace since comments
** introducers may be misinterpreted as division operators.
*/
if (!Skipped) {
if (GetPunc (Ident)) {
if (Ident[0] == ')') {
/* Count right parens */
++Count;
}
SB_AppendStr (Target, Ident);
/* If an identifier follows immediately, it could be a macro
** expanded later that occasionally need a space to separate.
*/
if (IsIdent (CurC)) {
/* Memorize the previous pp-token and check it later */
LazyCheckNextPPTok (Target, strlen (Ident));
}
} else {
SB_AppendChar (Target, CurC);
NextChar ();
/* Don't count this character */
++LastLen;
}
}
}
/* Keep tracking pp-token lengths */
if ((ModeFlags & MSM_IN_ARG_EXPANSION) != 0) {
ME_SetTokLens (E, SB_GetLen (Target) - LastLen);
}
}
Loop:
/* Switch back to the previous input stream if we have finished
** rescanning the current one.
*/
if (CurC == '\0' && CollCount (&CurRescanStack->Lines) > 1) {
/* Check for rescan sequence end and pp-token pasting */
Skipped = SkipWhitespace (0) || Skipped;
/* Add indention to preprocessor output if needed */
if (CurC != '\0' &&
PendingNewLines > 0 &&
(ModeFlags & MSM_MULTILINE) != 0 &&
CollCount (&CurRescanStack->Lines) == 1) {
/* Add newlines */
AddPreLine (Target);
/* Align indention */
AppendIndent (Target, SB_GetIndex (Line));
Skipped = 0;
}
}
/* Append a space if there hasn't been one */
if (Skipped && !IsSpace (SB_LookAtLast (Target))) {
SB_AppendChar (Target, ' ');
}
}
/* Append the remaining result */
SB_Append (Target, TmpTarget);
/* Done with the temporary buffer */
SB_Done (TmpTarget);
/* Drop whitespace at the end */
if (IsBlank (SB_LookAtLast (Target))) {
SB_Drop (Target, 1);
}
/* Sanity check */
if ((E->Flags & MES_ERROR) == 0) {
CHECK (CollCount (&CurRescanStack->Lines) == 1);
}
/* Done with the current input stack */
DoneRescanInputStack (CurRescanStack);
CurRescanStack = OldRescanStack;
/* Switch back the input */
InitLine (OldSource);
/* Return the count of identifiers and right parentheses */
return Count;
}
/*****************************************************************************/
/* Directives */
/*****************************************************************************/
static int ParseMacroReplacement (StrBuf* Source, Macro* M)
/* Check correctness of macro definition while squeezing old and new style
** comments and other non-newline whitespace sequences. Return 1 on success
** or 0 on failure.
*/
{
/* Switch to the new input source */
StrBuf* OldSource = InitLine (Source);
int HasWhiteSpace = 0;
unsigned Len;
ident Ident;
int Std = IS_Get (&Standard);
/* Skip whitespace before the macro replacement */
SkipWhitespace (0);
/* Check for ## at start */
if (CurC == '#' && NextC == '#') {
/* Diagnose and bail out */
PPError ("'##' cannot appear at start of macro expansion");
goto Error_Handler;
}
/* Loop removing ws and comments */
while (CurC != '\0') {
if (HasWhiteSpace) {
SB_AppendChar (&M->Replacement, ' ');
} else if (IsQuotedString ()) {
CopyQuotedString (&M->Replacement);
} else if (IsSym (Ident)) {
CheckForBadIdent (Ident, Std, M);
SB_AppendStr (&M->Replacement, Ident);
} else {
if (M->ParamCount >= 0 && GetPunc (Ident)) {
Len = strlen (Ident);
/* Check for # */
if (Len == 1 && Ident[0] == '#') {
HasWhiteSpace = SkipWhitespace (0);
/* Check next pp-token */
if (!IsSym (Ident) || FindMacroParam (M, Ident) < 0) {
PPError ("'#' is not followed by a macro parameter");
goto Error_Handler;
}
/* Make the replacement */
SB_AppendChar (&M->Replacement, '#');
if (HasWhiteSpace) {
SB_AppendChar (&M->Replacement, ' ');
}
SB_AppendStr (&M->Replacement, Ident);
} else {
SB_AppendBuf (&M->Replacement, Ident, Len);
}
} else {
SB_AppendChar (&M->Replacement, CurC);
NextChar ();
}
}
HasWhiteSpace = SkipWhitespace (0);
}
/* Check for ## at end */
Len = SB_GetLen (&M->Replacement);
if (Len >= 2) {
if (SB_LookAt (&M->Replacement, Len - 1) == '#' &&
SB_LookAt (&M->Replacement, Len - 2) == '#') {
/* Diagnose and bail out */
PPError ("'##' cannot appear at end of macro expansion");
goto Error_Handler;
}
}
/* Terminate the new input line */
SB_Terminate (&M->Replacement);
/* Switch back to the old source */
InitLine (OldSource);
/* Success */
return 1;
Error_Handler:
/* Switch back to the old source */
InitLine (OldSource);
/* Failure */
return 0;
}
static void DoDefine (void)
/* Process #define directive */
{
ident Ident;
Macro* M = 0;
Macro* Existing;
int Std;
/* Read the macro name */
SkipWhitespace (0);
if (!MacName (Ident)) {
goto Error_Handler;
}
/* Remember the language standard we are in */
Std = IS_Get (&Standard);
/* Check for forbidden macro names */
if (strcmp (Ident, "defined") == 0) {
PPError ("'%s' cannot be used as a macro name", Ident);
goto Error_Handler;
}
/* Check for and warn on special identifiers */
CheckForBadIdent (Ident, Std, 0);
/* Create a new macro definition */
M = NewMacro (Ident);
/* Check if this is a function-like macro */
if (CurC == '(') {
/* Skip the left paren */
NextChar ();
/* Set the marker that this is a function-like macro */
M->ParamCount = 0;
/* Read the formal parameter list */
while (1) {
/* Skip white space and check for end of parameter list */
SkipWhitespace (0);
if (CurC == ')') {
break;
}
/* The next token must be either an identifier, or - if not in
** C89 mode - the ellipsis.
*/
if (Std >= STD_C99 && CurC == '.') {
/* Ellipsis */
NextChar ();
if (CurC != '.' || NextC != '.') {
PPError ("'...' expected");
goto Error_Handler;
}
NextChar ();
NextChar ();
/* Remember that the macro is variadic and use __VA_ARGS__ as
** the parameter name.
*/
AddMacroParam (M, "__VA_ARGS__");
M->Variadic = 1;
} else {
/* Must be macro parameter name */
if (MacName (Ident) == 0) {
goto Error_Handler;
}
/* Check for and warn on special identifiers */
CheckForBadIdent (Ident, Std, 0);
/* Add the macro parameter */
AddMacroParam (M, Ident);
}
/* If we had an ellipsis, or the next char is not a comma, we've
** reached the end of the macro parameter list.
*/
SkipWhitespace (0);
if (M->Variadic || CurC != ',') {
break;
}
NextChar ();
}
/* Check for a right paren and eat it if we find one */
if (CurC != ')') {
PPError ("')' expected for macro definition");
goto Error_Handler;
}
NextChar ();
}
/* Remove whitespace and comments from the line, store the preprocessed
** line into the macro replacement buffer.
*/
if (ParseMacroReplacement (Line, M) == 0) {
goto Error_Handler;
}
#if 0
printf ("%s: <%.*s>\n", M->Name, SB_GetLen (&M->Replacement), SB_GetConstBuf (&M->Replacement));
#endif
/* Get an existing macro definition with this name */
Existing = FindMacro (M->Name);
/* If we have an existing macro, check if the redefinition is identical.
** Print a diagnostic if not.
*/
if (Existing != 0) {
if (MacroCmp (M, Existing) != 0) {
PPError ("Macro redefinition is not identical");
}
/* Undefine the existing macro anyways */
UndefineMacro (Existing->Name);
}
/* Insert the new macro into the macro table */
InsertMacro (M);
/* Success */
return;
Error_Handler:
/* Cleanup */
ClearLine ();
if (M != 0) {
FreeMacro (M);
}
}
static int PushIf (int Skip, int Invert, int Cond)
/* Push a new if level onto the if stack */
{
/* Check for an overflow of the if stack */
if (PPStack->Index >= MAX_PP_IFS-1) {
PPError ("Too many nested #if clauses");
return 1;
}
/* Push the #if condition */
++PPStack->Index;
if (Skip) {
PPStack->Stack[PPStack->Index] = IFCOND_SKIP | IFCOND_NEEDTERM;
return 1;
} else {
PPStack->Stack[PPStack->Index] = IFCOND_NONE | IFCOND_NEEDTERM;
return (Invert ^ Cond);
}
}
static void DoError (void)
/* Print an error */
{
SkipWhitespace (0);
if (CurC == '\0') {
PPError ("#error");
} else {
StrBuf MsgLine = AUTO_STRBUF_INITIALIZER;
TranslationPhase3 (Line, &MsgLine);
PPError ("#error: %s", SB_GetConstBuf (&MsgLine) + SB_GetIndex (&MsgLine));
SB_Done (&MsgLine);
}
/* Clear the rest of line */
ClearLine ();
}
static int DoIf (int Skip)
/* Process #if directive */
{
PPExpr Expr = AUTO_PPEXPR_INITIALIZER;
if (!Skip) {
/* We're about to use a dedicated expression parser to evaluate the #if
** expression. Save the current tokens to come back here later.
*/
Token SavedCurTok = CurTok;
Token SavedNextTok = NextTok;
/* Make sure the line infos for the tokens won't get removed */
if (SavedCurTok.LI) {
UseLineInfo (SavedCurTok.LI);
}
if (SavedNextTok.LI) {
UseLineInfo (SavedNextTok.LI);
}
/* Macro-replace a single line with support for the "defined" operator */
SB_Clear (MLine);
PreprocessDirective (Line, MLine, MSM_OP_DEFINED);
/* Read from the processed line */
SB_Reset (MLine);
MLine = InitLine (MLine);
/* Add two semicolons as sentinels to the line, so the following
** expression evaluation will eat these two tokens but nothing from
** the following line.
*/
SB_AppendStr (Line, ";;");
SB_Terminate (Line);
/* Load CurTok and NextTok with tokens from the new input */
NextToken ();
NextToken ();
/* Call the expression parser */
ParsePPExprInLine (&Expr);
/* Restore input source */
MLine = InitLine (MLine);
/* Reset the old tokens */
CurTok = SavedCurTok;
NextTok = SavedNextTok;
}
/* Set the #if condition according to the expression result */
return PushIf (Skip, 1, Expr.IVal != 0);
}
static int DoIfDef (int skip, int flag)
/* Process #ifdef if flag == 1, or #ifndef if flag == 0. */
{
int Value = 0;
if (!skip) {
ident Ident;
SkipWhitespace (0);
if (MacName (Ident)) {
CheckForBadIdent (Ident, IS_Get (&Standard), 0);
Value = IsMacro (Ident);
/* Check for extra tokens */
CheckExtraTokens (flag ? "ifdef" : "ifndef");
}
}
return PushIf (skip, flag, Value);
}
static void DoInclude (void)
/* Open an include file. */
{
char RTerm;
InputType IT;
StrBuf Filename = AUTO_STRBUF_INITIALIZER;
/* Macro-replace a single line with special support for <filename> */
SB_Clear (MLine);
PreprocessDirective (Line, MLine, MSM_TOK_HEADER);
/* Read from the processed line */
SB_Reset (MLine);
MLine = InitLine (MLine);
/* Get the next char and check for a valid file name terminator. Setup
** the include directory spec (SYS/USR) by looking at the terminator.
*/
switch (CurC) {
case '\"':
RTerm = '\"';
IT = IT_USRINC;
break;
case '<':
RTerm = '>';
IT = IT_SYSINC;
break;
default:
PPError ("#include expects \"FILENAME\" or <FILENAME>");
goto Done;
}
NextChar ();
/* Get a copy of the filename */
while (CurC != '\0' && CurC != RTerm) {
SB_AppendChar (&Filename, CurC);
NextChar ();
}
SB_Terminate (&Filename);
/* Check if we got a terminator */
if (CurC == RTerm) {
/* Skip the terminator */
NextChar ();
/* Check for extra tokens following the filename */
CheckExtraTokens ("include");
/* Open the include file */
OpenIncludeFile (SB_GetConstBuf (&Filename), IT);
} else {
/* No terminator found */
PPError ("#include expects \"FILENAME\" or <FILENAME>");
}
Done:
/* Free the allocated filename data */
SB_Done (&Filename);
/* Restore input source */
MLine = InitLine (MLine);
/* Clear the remaining line so the next input will come from the new
** file (if open)
*/
ClearLine ();
}
static unsigned GetLineDirectiveNum (void)
/* Get a decimal digit-sequence from the input. Return 0 on errors. */
{
unsigned long Num = 0;
StrBuf Buf = AUTO_STRBUF_INITIALIZER;
/* The only non-decimal-numeric character allowed in the digit-sequence is
** the digit separator '\'' as of C23, but we haven't supported it yet.
*/
SkipWhitespace (0);
while (IsDigit (CurC))
{
SB_AppendChar (&Buf, CurC);
NextChar ();
}
/* Ensure the buffer is terminated with a '\0' */
SB_Terminate (&Buf);
if (SkipWhitespace (0) != 0 || CurC == '\0') {
const char* Str = SB_GetConstBuf (&Buf);
if (Str[0] == '\0') {
PPWarning ("#line directive interprets number as decimal, not octal");
} else {
Num = strtoul (Str, 0, 10);
if (Num > 2147483647) {
PPError ("#line directive requires an integer argument not greater than 2147483647");
ClearLine ();
Num = 0;
} else if (Num == 0) {
PPError ("#line directive requires a positive integer argument");
ClearLine ();
}
}
} else {
PPError ("#line directive requires a simple decimal digit sequence");
ClearLine ();
}
/* Done with the buffer */
SB_Done (&Buf);
return (unsigned)Num;
}
static void DoLine (void)
/* Process #line directive */
{
unsigned LineNum;
/* Macro-replace a single line with support for the "defined" operator */
SB_Clear (MLine);
PreprocessDirective (Line, MLine, MSM_NONE);
/* Read from the processed line */
SB_Reset (MLine);
MLine = InitLine (MLine);
/* Parse and check the specified line number */
LineNum = GetLineDirectiveNum ();
if (LineNum != 0) {
/* Parse and check the optional filename argument */
if (SB_GetIndex (Line) < SB_GetLen (Line)) {
StrBuf Filename = AUTO_STRBUF_INITIALIZER;
if (SB_GetString (Line, &Filename)) {
SB_Terminate (&Filename);
SetCurrentFileName (SB_GetConstBuf (&Filename));
} else {
PPError ("Invalid filename for #line directive");
LineNum = 0;
}
SB_Done (&Filename);
}
/* #line actually sets the line number of the next line */
if (LineNum > 0) {
SetCurrentLineNum (LineNum - 1);
/* Check for extra tokens at the end */
CheckExtraTokens ("line");
}
}
/* Restore input source */
MLine = InitLine (MLine);
}
static void DoPragma (void)
/* Handle a #pragma line by converting the #pragma preprocessor directive into
** the _Pragma() compiler operator.
*/
{
StrBuf* PragmaLine = OLine;
PRECONDITION (PragmaLine != 0);
/* Add the source info to preprocessor output if needed */
AddPreLine (PragmaLine);
/* Macro-replace a single line */
SB_Clear (MLine);
PreprocessDirective (Line, MLine, MSM_NONE);
/* Convert #pragma to _Pragma () */
SB_AppendStr (PragmaLine, "_Pragma (");
SB_Reset (MLine);
Stringize (MLine, PragmaLine);
SB_AppendChar (PragmaLine, ')');
/* End this line */
SB_SetIndex (PragmaLine, SB_GetLen (PragmaLine));
}
static void DoUndef (void)
/* Process the #undef directive */
{
ident Ident;
SkipWhitespace (0);
if (MacName (Ident)) {
CheckForBadIdent (Ident, IS_Get (&Standard), 0);
UndefineMacro (Ident);
}
/* Check for extra tokens */
CheckExtraTokens ("undef");
}
static void DoWarning (void)
/* Print a warning */
{
SkipWhitespace (0);
if (CurC == '\0') {
PPWarning ("#warning");
} else {
StrBuf MsgLine = AUTO_STRBUF_INITIALIZER;
TranslationPhase3 (Line, &MsgLine);
PPWarning ("#warning: %s", SB_GetConstBuf (&MsgLine) + SB_GetIndex (&MsgLine));
SB_Done (&MsgLine);
}
/* Clear the rest of line */
ClearLine ();
}
static int ParseDirectives (unsigned ModeFlags)
/* Handle directives. Return 1 if any whitespace or newlines are parsed. */
{
int PPSkip = 0;
ident Directive;
/* Skip white space at the beginning of the first line */
int Whitespace = SkipWhitespace (0);
/* Check for stuff to skip */
while (CurC == '\0' || CurC == '#' || PPSkip) {
/* Check for preprocessor lines lines */
if (CurC == '#') {
NextChar ();
SkipWhitespace (0);
if (CurC == '\0') {
/* Ignore the empty preprocessor directive */
continue;
}
if (!IsSym (Directive)) {
if (!PPSkip) {
PPError ("Preprocessor directive expected");
}
ClearLine ();
} else {
switch (FindPPDirectiveType (Directive)) {
case PPD_DEFINE:
if (!PPSkip) {
DoDefine ();
}
break;
case PPD_ELIF:
if (PPStack->Index >= 0) {
if ((PPStack->Stack[PPStack->Index] & IFCOND_ELSE) == 0) {
/* Handle as #else/#if combination */
if ((PPStack->Stack[PPStack->Index] & IFCOND_SKIP) == 0) {
PPSkip = !PPSkip;
}
PPStack->Stack[PPStack->Index] |= IFCOND_ELSE;
PPSkip = DoIf (PPSkip);
/* #elif doesn't need a terminator */
PPStack->Stack[PPStack->Index] &= ~IFCOND_NEEDTERM;
} else {
PPError ("Duplicate #else/#elif");
}
} else {
PPError ("Unexpected #elif");
}
break;
case PPD_ELSE:
if (PPStack->Index >= 0) {
if ((PPStack->Stack[PPStack->Index] & IFCOND_ELSE) == 0) {
if ((PPStack->Stack[PPStack->Index] & IFCOND_SKIP) == 0) {
PPSkip = !PPSkip;
}
PPStack->Stack[PPStack->Index] |= IFCOND_ELSE;
/* Check for extra tokens */
CheckExtraTokens ("else");
} else {
PPError ("Duplicate #else");
}
} else {
PPError ("Unexpected '#else'");
}
break;
case PPD_ENDIF:
if (PPStack->Index >= 0) {
/* Remove any clauses on top of stack that do not
** need a terminating #endif.
*/
while (PPStack->Index >= 0 &&
(PPStack->Stack[PPStack->Index] & IFCOND_NEEDTERM) == 0) {
--PPStack->Index;
}
/* Stack may not be empty here or something is wrong */
CHECK (PPStack->Index >= 0);
/* Remove the clause that needs a terminator */
PPSkip = (PPStack->Stack[PPStack->Index--] & IFCOND_SKIP) != 0;
/* Check for extra tokens */
CheckExtraTokens ("endif");
} else {
PPError ("Unexpected '#endif'");
}
break;
case PPD_ERROR:
if (!PPSkip) {
DoError ();
}
break;
case PPD_IF:
PPSkip = DoIf (PPSkip);
break;
case PPD_IFDEF:
PPSkip = DoIfDef (PPSkip, 1);
break;
case PPD_IFNDEF:
PPSkip = DoIfDef (PPSkip, 0);
break;
case PPD_INCLUDE:
if (!PPSkip) {
DoInclude ();
}
break;
case PPD_LINE:
if (!PPSkip) {
DoLine ();
}
break;
case PPD_PRAGMA:
if (!PPSkip) {
if ((ModeFlags & MSM_IN_ARG_LIST) == 0) {
DoPragma ();
return Whitespace;
} else {
PPError ("Embedded #pragma directive within macro arguments is unsupported");
}
}
break;
case PPD_UNDEF:
if (!PPSkip) {
DoUndef ();
}
break;
case PPD_WARNING:
/* #warning is a non standard extension */
if (IS_Get (&Standard) > STD_C99) {
if (!PPSkip) {
DoWarning ();
}
} else {
if (!PPSkip) {
PPError ("Preprocessor directive expected");
}
ClearLine ();
}
break;
default:
if (!PPSkip) {
PPError ("Preprocessor directive expected");
}
ClearLine ();
}
}
}
if (NextLine () == 0) {
break;
}
++PendingNewLines;
Whitespace = SkipWhitespace (0) || Whitespace;
}
return Whitespace;
}
void HandleSpecialMacro (Macro* M, const char* Name)
/* Handle special "magic" macros that may change */
{
if (strcmp (Name, "__COUNTER__") == 0) {
/* Replace __COUNTER__ with the current counter number */
if (IS_Get (&Standard) < STD_CC65) {
PPWarning ("__COUNTER__ is a cc65 extension");
}
SB_Printf (&M->Replacement, "%u", GetCurrentCounter ());
} else if (strcmp (Name, "__LINE__") == 0) {
/* Replace __LINE__ with the current line number */
SB_Printf (&M->Replacement, "%u", GetCurrentLineNum ());
} else if (strcmp (Name, "__FILE__") == 0) {
/* Replace __FILE__ with the current filename */
StrBuf B = AUTO_STRBUF_INITIALIZER;
SB_InitFromString (&B, GetCurrentFileName ());
SB_Clear (&M->Replacement);
Stringize (&B, &M->Replacement);
SB_Done (&B);
}
}
/*****************************************************************************/
/* Preprocessing */
/*****************************************************************************/
static void TranslationPhase3 (StrBuf* Source, StrBuf* Target)
/* Mimic Translation Phase 3. Handle old and new style comments. Collapse
** non-newline whitespace sequences.
*/
{
/* Switch to the new input source */
StrBuf* OldSource = InitLine (Source);
/* Loop removing ws and comments */
while (CurC != '\0') {
int HasWhiteSpace = 0;
while (1) {
/* Squeeze runs of blanks */
if (IsSpace (CurC)) {
NextChar ();
HasWhiteSpace = 1;
} else if (CurC == '/' && NextC == '*') {
OldStyleComment ();
HasWhiteSpace = 1;
} else if (CurC == '/' && NextC == '/') {
NewStyleComment ();
HasWhiteSpace = 1;
} else {
/* No more white space */
break;
}
}
if (HasWhiteSpace) {
SB_AppendChar (Target, ' ');
} else if (IsQuotedString ()) {
CopyQuotedString (Target);
} else {
SB_AppendChar (Target, CurC);
NextChar ();
}
}
/* Terminate the new input line */
SB_Terminate (Target);
/* Switch back to the old source */
InitLine (OldSource);
}
static void PreprocessDirective (StrBuf* Source, StrBuf* Target, unsigned ModeFlags)
/* Preprocess a single line. Handle specified tokens and operators, remove
** whitespace and comments, then do macro replacement.
*/
{
MacroExp E;
SkipWhitespace (0);
InitMacroExp (&E);
ReplaceMacros (Source, Target, &E, ModeFlags | MSM_IN_DIRECTIVE);
DoneMacroExp (&E);
}
void Preprocess (void)
/* Preprocess lines count of which is affected by directives */
{
MacroExp E;
SB_Clear (PLine);
/* Add the source info to preprocessor output if needed */
AddPreLine (PLine);
/* Parse any directives */
OLine = PLine;
ParseDirectives (MSM_MULTILINE);
OLine = 0;
/* Add the source info to preprocessor output if needed */
AddPreLine (PLine);
/* Add leading whitespace to prettify preprocessor output */
if (CurC != '\0') {
AppendIndent (PLine, SB_GetIndex (Line));
}
/* Expand macros if any */
InitMacroExp (&E);
ReplaceMacros (Line, PLine, &E, MSM_MULTILINE);
DoneMacroExp (&E);
/* Add the source info to preprocessor output if needed */
AddPreLine (PLine);
/* Read from the new line */
SB_Reset (PLine);
PLine = InitLine (PLine);
if (Verbosity > 1 && SB_NotEmpty (Line)) {
printf ("%s:%u: %.*s\n", GetCurrentFileName (), GetCurrentLineNum (),
(int) SB_GetLen (Line), SB_GetConstBuf (Line));
}
/* Free all undefined macros */
FreeUndefinedMacros ();
}
void InitPreprocess (void)
/* Init preprocessor */
{
/* Create the output buffers */
MLine = NewStrBuf ();
PLine = NewStrBuf ();
}
void DonePreprocess (void)
/* Done with preprocessor */
{
/* Done with the output buffers */
SB_Done (MLine);
SB_Done (PLine);
}
void SetPPIfStack (PPIfStack* Stack)
/* Specify which PP #if stack to use */
{
PPStack = Stack;
}
void ContinueLine (void)
/* Continue the current line ended with a '\\' */
{
++ContinuedLines;
}
void PreprocessBegin (void)
/* Initialize preprocessor with current file */
{
/* Reset #if depth */
PPStack->Index = -1;
/* Remember to update source file location in preprocess-only mode */
FileChanged = 1;
/* Enable diagnostics on new style comments in C89 mode */
AllowNewComments = 0;
}
void PreprocessEnd (void)
/* Preprocessor done with current file */
{
/* Check for missing #endif */
while (PPStack->Index >= 0) {
if ((PPStack->Stack[PPStack->Index] & IFCOND_NEEDTERM) != 0) {
PPError ("#endif expected");
}
--PPStack->Index;
}
/* Remember to update source file location in preprocess-only mode */
FileChanged = 1;
}
Reference in New Issue View Git Blame Copy Permalink