d0p1/ack
1
0
Fork 0
Code Issues Pull requests Packages Projects Releases Wiki Activity

Rip out the built-in preprocessor from cemcom.ansi, now it's no longer used.

Browse source
This commit is contained in:
David Given 2018-09-02 12:50:50 +02:00
parent 6931ffa5a4
commit cb983fac44
17 changed files with 0 additions and 2614 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
lang/cem/cemcom.ansi/BigPars
View file

@ -111,10 +111,6 @@
#define INP_READ_IN_ONE 1 /* read input file in one */
!File: nopp.h
#define NOPP 1 /* if NOT defined, use built-int preprocessor */
!File: nobitfield.h
/*#define NOBITFIELD 1 /* if NOT defined, implement bitfields */

256
lang/cem/cemcom.ansi/LLlex.c
View file

@ -27,13 +27,6 @@ int tk_nmb_at_last_syn_err = -5 /*ERR_SHADOW*/;
int idfsize = IDFSIZE;
char sp_occurred[SP_TOTAL + 1];
#ifndef NOPP
int ReplaceMacros = 1; /* replacing macros */
int AccDefined = 0; /* accept "defined(...)" */
int UnknownIdIsZero = 0; /* interpret unknown id as integer 0 */
int Unstacked = 0; /* an unstack is done */
extern int InputLevel;
#endif
int AccFileSpecifier = 0; /* return filespecifier <...> */
int EoiForNewline = 0; /* return EOI upon encountering newline */
int File_Inserted = 0; /* a file has just been inserted */
@ -48,33 +41,6 @@ extern arith full_mask[];
extern int lint_skip_comment;
#endif
#ifndef NOPP
static struct token LexStack[MAX_LL_DEPTH];
static LexSP = 0;
void skipcomment();
void skiplinecomment();
/* In PushLex() the actions are taken in order to initialise or
re-initialise the lexical scanner.
E.g. at the invocation of a sub-parser that uses LLlex(), the
state of the current parser should be saved.
*/
PushLex()
{
assert(LexSP < MAX_LL_DEPTH);
assert(ASIDE == 0); /* ASIDE = 0; */
GetToken(&ahead);
LexStack[LexSP++] = dot;
}
PopLex()
{
assert(LexSP > 0);
dot = LexStack[--LexSP];
}
#endif /* NOPP */
int LLlex()
{
/* LLlex() plays the role of Lexical Analyzer for the C parser.
@ -152,9 +118,6 @@ go_on: /* rescan, the following character has been read */
return ptok->tk_symb = EOI;
while ((ch = GetChar()), (ch == '#'
#ifndef NOPP
|| ch == '/'
#endif
|| class(ch) == STSKIP))
{
/* blanks are allowed before hashes */
@ -162,26 +125,6 @@ go_on: /* rescan, the following character has been read */
{
/* a control line follows */
domacro();
#ifndef NOPP
if (File_Inserted)
{
File_Inserted = 0;
goto firstline;
}
}
else if ((ch == '/') && !InputLevel)
{
int nch = GetChar();
if (nch == '*')
skipcomment();
else if (nch == '/')
skiplinecomment();
else
{
UnGetChar();
break;
}
#endif /* NOPP */
}
}
/* We have to loop here, because in
@ -192,9 +135,6 @@ go_on: /* rescan, the following character has been read */
case STSKIP: /* just skip the skip characters */
goto again;
case STGARB: /* garbage character */
#ifndef NOPP
garbage:
#endif
if (040 < ch && ch < 0177)
{
return ptok->tk_symb = ch;
@ -285,21 +225,6 @@ go_on: /* rescan, the following character has been read */
return ptok->tk_symb = XORAB;
break;
case '/':
#ifndef NOPP
if (!InputLevel)
{
if (nch == '*')
{
skipcomment();
goto again;
}
else if (nch == '/')
{
skiplinecomment();
goto again;
}
}
#endif
if (nch == '=')
return ptok->tk_symb = DIVAB;
break;
@ -339,17 +264,6 @@ go_on: /* rescan, the following character has been read */
register int pos = -1;
register struct idf* idef;
extern int idfsize; /* ??? */
#ifndef NOPP
int NoExpandNext = 0;
if (Unstacked)
EnableMacros(); /* unstack macro's when allowed. */
if (ch == NOEXPM)
{
NoExpandNext = 1;
ch = GetChar();
}
#endif
do
{ /* read the identifier */
if (++pos < idfsize)
@ -366,19 +280,6 @@ go_on: /* rescan, the following character has been read */
sp_occurred[idef->id_special] = 1;
idef->id_file = ptok->tk_file;
idef->id_line = ptok->tk_line;
#ifndef NOPP
if (idef->id_macro && ReplaceMacros && !NoExpandNext)
{
if (replace(idef))
goto again;
}
if (UnknownIdIsZero && idef->id_reserved != SIZEOF)
{
ptok->tk_ival = (arith)0;
ptok->tk_fund = INT;
return ptok->tk_symb = INTEGER;
}
#endif /* NOPP */
ptok->tk_symb
= (idef->id_reserved
? idef->id_reserved
@ -472,101 +373,12 @@ go_on: /* rescan, the following character has been read */
}
case STEOI: /* end of text on source file */
return ptok->tk_symb = EOI;
#ifndef NOPP
case STMSPEC:
if (!InputLevel)
goto garbage;
if (ch == TOKSEP)
goto again;
/* fallthrough shouldn't happen */
#endif
default: /* this cannot happen */
crash("bad class for char 0%o", ch);
}
/*NOTREACHED*/
}
#ifndef NOPP
void skipcomment()
{
/* The last character read has been the '*' of '/_*'. The
characters, except NL and EOI, between '/_*' and the first
occurring '*_/' are not interpreted.
NL only affects the LineNumber. EOI is not legal.
Important note: it is not possible to stop skipping comment
beyond the end-of-file of an included file.
EOI is returned by LoadChar only on encountering EOF of the
top-level file...
*/
register int c, oldc = '\0';
NoUnstack++;
c = GetChar();
#ifdef LINT
if (!lint_skip_comment)
{
lint_start_comment();
lint_comment_char(c);
}
#endif /* LINT */
do
{
while (c != '*')
{
if (class(c) == STNL)
{
++LineNumber;
}
else if (c == EOI)
{
NoUnstack--;
#ifdef LINT
if (!lint_skip_comment)
lint_end_comment();
#endif /* LINT */
return;
}
oldc = c;
c = GetChar();
#ifdef LINT
if (!lint_skip_comment)
lint_comment_char(c);
#endif /* LINT */
} /* last Character seen was '*' */
c = GetChar();
if (c != '/' && oldc == '/')
lexwarning("comment inside comment ?");
oldc = '*';
#ifdef LINT
if (!lint_skip_comment)
lint_comment_char(c);
#endif /* LINT */
} while (c != '/');
#ifdef LINT
if (!lint_skip_comment)
lint_end_comment();
#endif /* LINT */
NoUnstack--;
}
void skiplinecomment(void)
{
/* The last character read has been the '/' of '//'. We read
and discard all characters up to but not including the next
NL. */
for (;;) {
int c = GetChar();
if ((class(c) == STNL) || (c == EOI))
{
UnGetChar();
break;
}
}
}
#endif /* NOPP */
arith char_constant(nm) char* nm;
{
register arith val = 0;
@ -710,78 +522,10 @@ int GetChar()
*/
register int ch;
#ifndef NOPP
again:
#endif
LoadChar(ch);
#ifndef NOPP
/* possible trigraph sequence */
if (ch == '?')
ch = trigraph();
/* \<newline> is removed from the input stream */
if (ch == '\\')
{
LoadChar(ch);
if (ch == '\n')
{
++LineNumber;
goto again;
}
PushBack();
ch = '\\';
}
#endif
return (LexSave = ch);
}
#ifndef NOPP
int trigraph()
{
register int ch;
LoadChar(ch);
if (ch == '?')
{
LoadChar(ch);
switch (ch)
{ /* its a trigraph */
case '=':
ch = '#';
return (ch);
case '(':
ch = '[';
return (ch);
case '/':
ch = '\\';
return (ch);
case ')':
ch = ']';
return (ch);
case '\'':
ch = '^';
return (ch);
case '<':
ch = '{';
return (ch);
case '!':
ch = '|';
return (ch);
case '>':
ch = '}';
return (ch);
case '-':
ch = '~';
return (ch);
}
PushBack();
}
PushBack();
return ('?');
}
#endif
/* strflt2tok only checks the syntax of the floating-point number and
* selects the right type for the number.
*/

6
lang/cem/cemcom.ansi/LLlex.h
View file

@ -43,12 +43,6 @@ extern struct token dot, ahead, aside;
extern int token_nmb; /* number of the ahead token */
extern int tk_nmb_at_last_syn_err; /* token number at last syntax error */
#ifndef NOPP
extern int ReplaceMacros; /* "LLlex.c" */
extern int AccDefined; /* "LLlex.c" */
extern int Unstacked; /* "LLlex.c" */
extern int UnknownIdIsZero; /* "LLlex.c" */
#endif /* NOPP */
extern int EoiForNewline; /* "LLlex.c" */
extern int AccFileSpecifier; /* "LLlex.c" */
extern int File_Inserted; /* "LLlex.c" */

4
lang/cem/cemcom.ansi/SmallPars
View file

@ -111,10 +111,6 @@
/*#define INP_READ_IN_ONE 1 /* read input file in one */
!File: nopp.h
#define NOPP 1 /* if NOT defined, use built-int preprocessor */
!File: nobitfield.h
/*#define NOBITFIELD 1 /* if NOT defined, implement bitfields */

2
lang/cem/cemcom.ansi/build.lua
View file

@ -111,7 +111,6 @@ cprogram {
"./field.c",
"./fltcstoper.c",
"./idf.c",
"./init.c",
"./input.c",
"./l_comment.c",
"./l_ev_ord.c",
@ -124,7 +123,6 @@ cprogram {
"./options.c",
"./pragma.c",
"./proto.c",
"./replace.c",
"./skip.c",
"./stab.c",
"./stack.c",

860
lang/cem/cemcom.ansi/domacro.c
View file

@ -24,865 +24,6 @@ int IncludeLevel = 0;
extern char options[];
#ifndef NOPP
#include <alloc.h>
#include "class.h"
#include "macro.h"
extern char** inctable; /* list of include directories */
extern char* getwdir();
char ifstack[IFDEPTH]; /* if-stack: the content of an entry is */
/* 1 if a corresponding ELSE has been */
/* encountered. */
int nestlevel = -1;
void macro_def();
void do_define();
struct idf* GetIdentifier(skiponerr) int skiponerr; /* skip the rest of the line on error */
{
/* returns a pointer to the descriptor of the identifier that is
read from the input stream. When the input doe not contain
an identifier, the rest of the line is skipped when
skiponerr is on, and a null-pointer is returned.
The substitution of macros is disabled.
*/
int tmp = UnknownIdIsZero;
int tok;
struct token tk;
UnknownIdIsZero = ReplaceMacros = 0;
tok = GetToken(&tk);
ReplaceMacros = 1;
UnknownIdIsZero = tmp;
if (tok != IDENTIFIER)
{
if (skiponerr && tok != EOI)
SkipToNewLine();
return (struct idf*)0;
}
return tk.tk_idf;
}
/* domacro() is the control line interpreter. The '#' has already
been read by the lexical analyzer by which domacro() is called.
The token appearing directly after the '#' is obtained by calling
the basic lexical analyzing function GetToken() and is interpreted
to perform the action belonging to that token.
An error message is produced when the token is not recognized,
i.e. it is not one of "define" .. "undef" , integer or newline.
*/
domacro()
{
struct token tk; /* the token itself */
int toknum;
EoiForNewline = 1;
ReplaceMacros = 0;
toknum = GetToken(&tk);
ReplaceMacros = 1;
switch (toknum)
{ /* select control line action */
case IDENTIFIER: /* is it a macro keyword? */
switch (tk.tk_idf->id_resmac)
{
case K_DEFINE: /* "define" */
do_define();
break;
case K_ELIF: /* "elif" */
do_elif();
break;
case K_ELSE: /* "else" */
do_else();
break;
case K_ENDIF: /* "endif" */
do_endif();
break;
case K_IF: /* "if" */
do_if();
break;
case K_IFDEF: /* "ifdef" */
do_ifdef(1);
break;
case K_IFNDEF: /* "ifndef" */
do_ifdef(0);
break;
case K_INCLUDE: /* "include" */
do_include();
break;
case K_LINE: /* "line" */
/* set LineNumber and FileName according to
the arguments.
*/
if (GetToken(&tk) != INTEGER)
{
lexerror("bad #line syntax");
SkipToNewLine();
}
else
do_line((unsigned int)tk.tk_ival);
break;
case K_ERROR: /* "error" */
do_error();
break;
case K_PRAGMA: /* "pragma" */
do_pragma();
break;
case K_UNDEF: /* "undef" */
do_undef((struct idf*)0);
break;
default:
/* invalid word seen after the '#' */
lexerror("%s: unknown control", tk.tk_idf->id_text);
SkipToNewLine();
}
break;
case INTEGER: /* # <integer> [<filespecifier>]? */
do_line((unsigned int)tk.tk_ival);
break;
case EOI: /* only `#' on this line: do nothing, ignore */
break;
default: /* invalid token following '#' */
lexerror("illegal # line");
SkipToNewLine();
}
EoiForNewline = 0;
}
#ifdef LINT
int lint_skip_comment;
#endif
void skip_block(to_endif) int to_endif;
{
/* skip_block() skips the input from
1) a false #if, #ifdef, #ifndef or #elif until the
corresponding #elif (resulting in true), #else or
#endif is read.
2) a #else corresponding to a true #if, #ifdef,
#ifndef or #elif until the corresponding #endif is
seen.
*/
register int ch;
register int skiplevel = nestlevel; /* current nesting level */
struct token tk;
int toknum;
#ifdef LINT
lint_skip_comment++;
#endif
NoUnstack++;
for (;;)
{
ch = GetChar(); /* read first character after newline */
while (class(ch) == STSKIP)
ch = GetChar();
if (ch != '#')
{
if (ch == EOI)
{
NoUnstack--;
#ifdef LINT
lint_skip_comment--;
#endif
return;
}
/* A possible '/' is not pushed back */
if (ch == '/')
{
ch = GetChar();
if (ch == '/')
{
skiplinecomment();
continue;
}
else if (ch == '*')
{
skipcomment();
continue;
}
else
UnGetChar();
}
else
UnGetChar();
SkipToNewLine();
continue;
}
ReplaceMacros = 0;
toknum = GetToken(&tk);
ReplaceMacros = 1;
if (toknum != IDENTIFIER)
{
if (toknum != INTEGER)
{
lexerror("illegal # line");
}
SkipToNewLine();
continue;
}
/* an IDENTIFIER: look for #if, #ifdef and #ifndef
without interpreting them.
Interpret #else, #elif and #endif if they occur
on the same level.
*/
switch (tk.tk_idf->id_resmac)
{
default:
case K_UNKNOWN:
/* invalid word seen after the '#' */
lexwarning("%s: unknown control", tk.tk_idf->id_text);
/* fallthrough */
case K_DEFINE:
case K_ERROR:
case K_INCLUDE:
case K_LINE:
case K_PRAGMA:
case K_UNDEF:
case K_FILE:
SkipToNewLine();
break;
case K_IF:
case K_IFDEF:
case K_IFNDEF:
push_if();
SkipToNewLine();
break;
case K_ELIF:
if (ifstack[nestlevel])
lexerror("#elif after #else");
if (!to_endif && nestlevel == skiplevel)
{
nestlevel--;
push_if();
if (ifexpr())
{
NoUnstack--;
#ifdef LINT
lint_skip_comment--;
#endif
return;
}
}
else
SkipToNewLine(); /* otherwise done in ifexpr() */
break;
case K_ELSE:
if (ifstack[nestlevel])
lexerror("#else after #else");
++(ifstack[nestlevel]);
if (!to_endif && nestlevel == skiplevel)
{
if (SkipToNewLine())
{
if (!options['o'])
lexstrict("garbage following #else");
}
NoUnstack--;
#ifdef LINT
lint_skip_comment--;
#endif
return;
}
else
SkipToNewLine();
break;
case K_ENDIF:
assert(nestlevel > nestlow);
if (nestlevel == skiplevel)
{
if (SkipToNewLine())
{
if (!options['o'])
lexstrict("garbage following #endif");
}
nestlevel--;
NoUnstack--;
#ifdef LINT
lint_skip_comment--;
#endif
return;
}
else
SkipToNewLine();
nestlevel--;
break;
}
}
}
ifexpr()
{
/* ifexpr() returns whether the restricted constant
expression following #if or #elif evaluates to true. This
is done by calling the LLgen generated subparser for
constant expressions. The result of this expression will
be given in the extern long variable "ifval".
*/
extern arith ifval;
int errors = err_occurred;
ifval = (arith)0;
AccDefined = 1;
UnknownIdIsZero = 1;
PushLex(); /* NEW parser */
If_expr(); /* invoke constant expression parser */
PopLex(); /* OLD parser */
AccDefined = 0;
UnknownIdIsZero = 0;
return (errors == err_occurred) && (ifval != (arith)0);
}
do_include()
{
/* do_include() performs the inclusion of a file.
*/
char* filenm;
char* result;
int tok;
struct token tk;
AccFileSpecifier = 1;
if (((tok = GetToken(&tk)) == FILESPECIFIER) || tok == STRING)
filenm = tk.tk_bts;
else
{
lexerror("bad include syntax");
filenm = (char*)0;
}
AccFileSpecifier = 0;
if (SkipToNewLine())
{
lexerror("bad include syntax");
}
inctable[0] = WorkingDir;
if (filenm)
{
if (!InsertFile(filenm, &inctable[tok == FILESPECIFIER], &result))
{
lexerror("cannot open include file \"%s\"", filenm);
add_dependency(filenm);
free(filenm);
}
else
{
add_dependency(result);
WorkingDir = getwdir(result);
File_Inserted = 1;
FileName = result;
LineNumber = 0;
nestlow = nestlevel;
#ifdef DBSYMTAB
IncludeLevel++;
if (options['g'])
{
C_ms_stb_cst(FileName, N_BINCL, 0, (arith)0);
}
#endif /* DBSYMTAB */
if (result != filenm)
free(filenm);
}
}
}
void do_define()
{
/* do_define() interprets a #define control line.
*/
struct idf* id; /* the #defined identifier's descriptor */
int nformals = -1; /* keep track of the number of formals */
char* formals[NPARAMS]; /* pointers to the names of the formals */
char parbuf[PARBUFSIZE]; /* names of formals */
char* repl_text; /* start of the replacement text */
int length; /* length of the replacement text */
register ch;
char* get_text();
/* read the #defined macro's name */
if (!(id = GetIdentifier(1)))
{
lexerror("illegal #define line");
return;
}
/* there is a formal parameter list if the identifier is
followed immediately by a '('.
*/
ch = GetChar();
if (ch == '(')
{
if ((nformals = getparams(formals, parbuf)) == -1)
{
SkipToNewLine();
return; /* an error occurred */
}
ch = GetChar();
}
/* read the replacement text if there is any */
ch = skipspaces(ch, 0); /* find first character of the text */
assert(ch != EOI);
/* UnGetChar() is not right when replacement starts with a '/' */
ChPushBack(ch);
repl_text = get_text((nformals > 0) ? formals : 0, &length);
macro_def(id, repl_text, nformals, length, NOFLAG);
LineNumber++;
}
push_if()
{
if (nestlevel >= IFDEPTH)
fatal("too many nested #if/#ifdef/#ifndef");
else
ifstack[++nestlevel] = 0;
}
do_elif()
{
if (nestlevel <= nestlow)
{
lexerror("#elif without corresponding #if");
SkipToNewLine();
}
else
{ /* restart at this level as if a #if is detected. */
if (ifstack[nestlevel])
{
lexerror("#elif after #else");
SkipToNewLine();
}
nestlevel--;
push_if();
skip_block(1);
}
}
do_else()
{
if (SkipToNewLine())
if (!options['o'])
lexstrict("garbage following #else");
if (nestlevel <= nestlow)
lexerror("#else without corresponding #if");
else
{ /* mark this level as else-d */
if (ifstack[nestlevel])
{
lexerror("#else after #else");
}
++(ifstack[nestlevel]);
skip_block(1);
}
}
do_endif()
{
if (SkipToNewLine())
{
if (!options['o'])
lexstrict("garbage following #endif");
}
if (nestlevel <= nestlow)
{
lexerror("#endif without corresponding #if");
}
else
nestlevel--;
}
do_if()
{
push_if();
if (!ifexpr()) /* a false #if/#elif expression */
skip_block(0);
}
do_ifdef(how)
{
register struct idf* id;
/* how == 1 : ifdef; how == 0 : ifndef
*/
push_if();
if (!(id = GetIdentifier(1)))
lexerror("illegal #ifdef construction");
else if (SkipToNewLine())
if (!options['o'])
lexstrict("garbage following #%s <identifier>", how ? "ifdef" : "ifndef");
/* The next test is a shorthand for:
(how && !id->id_macro) || (!how && id->id_macro)
*/
if (how ^ (id && id->id_macro != 0))
skip_block(0);
}
/* argidf != NULL when the undef came from a -U option */
do_undef(argidf) struct idf* argidf;
{
register struct idf* id = argidf;
/* Forget a macro definition. */
if (id || (id = GetIdentifier(1)))
{
if (id->id_macro)
{ /* forget the macro */
if (id->id_macro->mc_flag & NOUNDEF)
{
lexerror("it is not allowed to undef %s", id->id_text);
}
else
{
free(id->id_macro->mc_text);
free_macro(id->id_macro);
id->id_macro = (struct macro*)0;
}
} /* else: don't complain */
if (!argidf)
{
if (SkipToNewLine())
if (!options['o'])
lexstrict("garbage following #undef");
}
}
else
lexerror("illegal #undef construction");
}
do_error()
{
int len;
char* get_text();
char* bp = get_text((char**)0, &len);
lexerror("user error: %s", bp);
free(bp);
LineNumber++;
}
int getparams(buf, parbuf) char* buf[];
char parbuf[];
{
/* getparams() reads the formal parameter list of a macro
definition.
The number of parameters is returned.
As a formal parameter list is expected when calling this
routine, -1 is returned if an error is detected, for
example:
#define one(1), where 1 is not an identifier.
Note that the '(' has already been eaten.
The names of the formal parameters are stored into parbuf.
*/
register char** pbuf = &buf[0];
register int c;
register char* ptr = &parbuf[0];
register char** pbuf2;
c = GetChar();
c = skipspaces(c, 0);
if (c == ')')
{ /* no parameters: #define name() */
*pbuf = (char*)0;
return 0;
}
for (;;)
{ /* eat the formal parameter list */
if (class(c) != STIDF && class(c) != STELL)
{
lexerror("#define: bad formal parameter");
return -1;
}
*pbuf = ptr; /* name of the formal */
*ptr++ = c;
if (ptr >= &parbuf[PARBUFSIZE])
fatal("formal parameter buffer overflow");
do
{ /* eat the identifier name */
c = GetChar();
*ptr++ = c;
if (ptr >= &parbuf[PARBUFSIZE])
fatal("formal parameter buffer overflow");
} while (in_idf(c));
*(ptr - 1) = '\0'; /* mark end of the name */
/* Check if this formal parameter is already used.
Usually, macros do not have many parameters, so ...
*/
for (pbuf2 = pbuf - 1; pbuf2 >= &buf[0]; pbuf2--)
{
if (!strcmp(*pbuf2, *pbuf))
{
lexerror("formal parameter \"%s\" already used", *pbuf);
}
}
pbuf++;
c = skipspaces(c, 0);
if (c == ')')
{ /* end of the formal parameter list */
*pbuf = (char*)0;
return pbuf - buf;
}
if (c != ',')
{
lexerror("#define: bad formal parameter list");
return -1;
}
c = GetChar();
c = skipspaces(c, 0);
}
/*NOTREACHED*/
}
void macro_def(id, text, nformals, length, flags) register struct idf* id;
char* text;
{
register struct macro* newdef = id->id_macro;
/* macro_def() puts the contents and information of a macro
definition into a structure and stores it into the symbol
table entry belonging to the name of the macro.
An error is given if there was already a definition
*/
if (newdef)
{ /* is there a redefinition? */
if (newdef->mc_flag & NOUNDEF)
{
lexerror("it is not allowed to redefine %s", id->id_text);
}
else if (!macroeq(newdef->mc_text, text))
lexerror("illegal redefine of \"%s\"", id->id_text);
free(text);
return;
}
id->id_macro = newdef = new_macro();
newdef->mc_text = text; /* replacement text */
newdef->mc_nps = nformals; /* nr of formals */
newdef->mc_length = length; /* length of repl. text */
newdef->mc_flag = flags; /* special flags */
}
int find_name(nm, index) char *nm, *index[];
{
/* find_name() returns the index of "nm" in the namelist
"index" if it can be found there. 0 is returned if it is
not there.
*/
register char** ip = &index[0];
while (*ip)
if (strcmp(nm, *ip++) == 0)
return ip - &index[0];
/* arrived here, nm is not in the name list. */
return 0;
}
#define BLANK(ch) ((ch == ' ') || (ch == '\t'))
char* get_text(formals, length) char* formals[];
int* length;
{
/* get_text() copies the replacement text of a macro
definition with zero, one or more parameters, thereby
substituting each formal parameter by a special character
(non-ascii: 0200 & (order-number in the formal parameter
list)) in order to substitute this character later by the
actual parameter. The replacement text is copied into
itself because the copied text will contain fewer or the
same amount of characters. The length of the replacement
text is returned.
Implementation:
finite automaton : we are interested in
1- white space, sequences must be mapped onto 1 single
blank.
2- identifiers, since they might be replaced by some
actual parameter.
3- strings and character constants, since replacing
variables within them is illegal, and white-space is
significant.
4- comment, same as for 1
Other tokens will not be seen as such.
*/
register int c;
struct repl repls;
register struct repl* repl = &repls;
int blank = 0;
c = GetChar();
repl->r_ptr = repl->r_text = Malloc(repl->r_size = ITEXTSIZE);
*repl->r_ptr = '\0';
while ((c != EOI) && (class(c) != STNL))
{
if (BLANK(c))
{
blank++;
c = GetChar();
continue;
}
if (c == '\'' || c == '"')
{
register int delim = c;
if (blank)
{
blank = 0;
add2repl(repl, ' ');
}
do
{
add2repl(repl, c);
if (c == '\\')
add2repl(repl, GetChar());
c = GetChar();
} while (c != delim && c != EOI && class(c) != STNL);
if (c == EOI || class(c) == STNL)
{
lexstrict("unclosed opening %c", delim);
break;
}
add2repl(repl, c);
c = GetChar();
}
else if (c == '/')
{
c = GetChar();
if (c == '*')
{
skipcomment();
blank++;
c = GetChar();
continue;
}
else if (c == '/')
{
skiplinecomment();
blank++;
c = GetChar();
continue;
}
if (blank)
{
blank = 0;
add2repl(repl, ' ');
}
add2repl(repl, '/');
}
else if (formals && (class(c) == STIDF || class(c) == STELL))
{
char id_buf[IDFSIZE + 1];
register char* idp = id_buf;
int n;
/* read identifier: it may be a formal parameter */
*idp++ = c;
do
{
c = GetChar();
if (idp <= &id_buf[IDFSIZE])
*idp++ = c;
} while (in_idf(c));
*--idp = '\0';
if (blank)
{
blank = 0;
add2repl(repl, ' ');
}
/* construct the formal parameter mark or identifier */
if (n = find_name(id_buf, formals))
add2repl(repl, FORMALP | (char)n);
else
{
idp = id_buf;
while (*idp)
add2repl(repl, *idp++);
}
}
else if (class(c) == STNUM)
{
if (blank)
{
blank = 0;
add2repl(repl, ' ');
}
add2repl(repl, c);
if (c == '.')
{
c = GetChar();
if (class(c) != STNUM)
{
continue;
}
add2repl(repl, c);
}
c = GetChar();
while (in_idf(c) || c == '.')
{
add2repl(repl, c);
if ((c = GetChar()) == 'e' || c == 'E')
{
add2repl(repl, c);
c = GetChar();
if (c == '+' || c == '-')
{
add2repl(repl, c);
c = GetChar();
}
}
}
}
else
{
if (blank)
{
blank = 0;
add2repl(repl, ' ');
}
add2repl(repl, c);
c = GetChar();
}
}
*length = repl->r_ptr - repl->r_text;
return Realloc(repl->r_text, (unsigned)(repl->r_ptr - repl->r_text + 1));
}
/* macroeq() decides whether two macro replacement texts are
identical. This version compares the texts, which occur
as strings, without taking care of the leading and trailing
blanks (spaces and tabs).
*/
macroeq(s, t) register char* s, *t;
{
/* skip leading spaces */
while (BLANK(*s))
s++;
while (BLANK(*t))
t++;
/* first non-blank encountered in both strings */
/* The actual comparison loop: */
while (*s && *s == *t)
s++, t++;
/* two cases are possible when arrived here: */
if (*s == '\0')
{ /* *s == '\0' */
while (BLANK(*t))
t++;
return *t == '\0';
}
else
{ /* *s != *t */
while (BLANK(*s))
s++;
while (BLANK(*t))
t++;
return (*s == '\0') && (*t == '\0');
}
}
#else /* NOPP */
struct idf* GetIdentifier(skiponerr) int skiponerr; /* skip the rest of the line on error */
{
/* returns a pointer to the descriptor of the identifier that is
@ -929,7 +70,6 @@ domacro()
EoiForNewline = 0;
SkipToNewLine();
}
#endif /* NOPP */
do_line(l) unsigned int l;
{

9
lang/cem/cemcom.ansi/dumpidf.c
View file

@ -106,15 +106,6 @@ dumpidf(idf, opt)
if (!idf)
return;
#ifndef NOPP
if ((opt&1) && idf->id_macro) {
if (!started++) {
newline();
print("%s:", idf->id_text);
}
print(" macro");
}
#endif /* NOPP */
if ((opt&2) && idf->id_reserved) {
if (!started++) {
newline();

13
lang/cem/cemcom.ansi/idf.str
View file

@ -8,10 +8,6 @@
#include "parameters.h"
struct id_u {
#ifndef NOPP
struct macro *idd_macro;
int idd_resmac; /* if nonzero: keyword of macroproc. */
#endif /* NOPP */
int idd_reserved; /* non-zero for reserved words */
char *idd_file; /* file containing the occurrence */
unsigned int idd_line; /* line number of the occurrence */
@ -36,14 +32,5 @@ struct id_u {
#include <idf_pkg.spec>
#ifndef NOPP
struct dependency {
struct dependency *next;
struct idf *dep_idf;
};
/* ALLOCDEF "dependency" 10 */
#endif /* NOPP */
extern int level;
extern struct idf *gen_idf();

99
lang/cem/cemcom.ansi/init.c
View file

@ -1,99 +0,0 @@
/*
* (c) copyright 1987 by the Vrije Universiteit, Amsterdam, The Netherlands.
* See the copyright notice in the ACK home directory, in the file "Copyright".
*/
/* $Id$ */
/* PREPROCESSOR: INITIALIZATION ROUTINES */
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include "parameters.h"
#ifndef NOPP
#include <system.h>
#include <alloc.h>
#include <time.h>
#include "idf.h"
#include "class.h"
#include "macro.h"
extern char *sprint();
struct mkey {
char *mk_reserved;
int mk_key;
} mkey[] = {
{"define", K_DEFINE},
{"elif", K_ELIF},
{"else", K_ELSE},
{"endif", K_ENDIF},
{"error", K_ERROR},
{"if", K_IF},
{"ifdef", K_IFDEF},
{"ifndef", K_IFNDEF},
{"include", K_INCLUDE},
{"line", K_LINE},
{"pragma", K_PRAGMA},
{"undef", K_UNDEF},
{0, K_UNKNOWN}
};
init_pp()
{
static char *months[12] = {
"Jan", "Feb", "Mar", "Apr", "May", "Jun",
"Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
};
time_t clock;
static char dbuf[30];
static char tbuf[30];
struct tm *tp;
/* Initialise the control line keywords (if, include, define, etc)
Although the lexical analyzer treats them as identifiers, the
control line handler can recognize them as keywords by the
id_resmac field of the identifier.
*/
{
register struct mkey *mk = &mkey[0];
while (mk->mk_reserved) {
register struct idf *idf = str2idf(mk->mk_reserved, 0);
if (idf->id_resmac)
fatal("maximum identifier length insufficient");
idf->id_resmac = mk->mk_key;
mk++;
}
}
/* Initialize __LINE__, __FILE__, __DATE__, __TIME__,
and __STDC__ macro definitions.
*/
clock = time(NULL);
tp = localtime(&clock);
/* __DATE__ */
sprint(dbuf, "\"%s %02d %d\"", months[tp->tm_mon],
tp->tm_mday, tp->tm_year+1900);
if (tp->tm_mday < 10) dbuf[5] = ' '; /* hack */
macro_def(str2idf("__DATE__", 0), dbuf, -1, strlen(dbuf), NOUNDEF);
/* __TIME__ */
sprint(tbuf, "\"%02d:%02d:%02d\"", tp->tm_hour, tp->tm_min, tp->tm_sec);
macro_def(str2idf("__TIME__", 0), tbuf, -1, strlen(tbuf), NOUNDEF);
/* __LINE__ */
macro_def(str2idf("__LINE__", 0), "0", -1, 1, NOUNDEF | FUNC);
/* __FILE__ */
macro_def(str2idf("__FILE__", 0), "", -1, 1, NOUNDEF | FUNC);
/* __STDC__ */
macro_def(str2idf("__STDC__", 0), "1", -1, 1, NOUNDEF);
/* defined(??) */
macro_def(str2idf("defined", 0), "", 1, 1, NOUNDEF | FUNC);
}
#endif /* NOPP */

56
lang/cem/cemcom.ansi/input.c
View file

@ -18,50 +18,10 @@ struct file_info finfo;
#include <inp_pkg.body>
#include <alloc.h>
#ifndef NOPP
#ifdef DBSYMTAB
#include <stb.h>
#include <em.h>
extern int IncludeLevel;
extern char options[];
#endif
char *
getwdir(fn)
register char *fn;
{
register char *p;
char *strrchr();
p = strrchr(fn, '/');
while (p && *(p + 1) == '\0') { /* remove trailing /'s */
*p = '\0';
p = strrchr(fn, '/');
}
if (fn[0] == '\0' || (fn[0] == '/' && p == &fn[0])) /* absolute path */
return "";
if (p) {
*p = '\0';
fn = Salloc(fn,(unsigned) (p - &fn[0] + 1));
*p = '/';
return fn;
}
return "";
}
int InputLevel;
extern int nestlevel;
#endif /* NOPP */
int NoUnstack;
AtEoIT()
{
#ifndef NOPP
InputLevel--;
unstackrepl();
#endif /* NOPP */
return 0;
}
@ -69,22 +29,6 @@ extern char *source;
AtEoIF()
{
#ifndef NOPP
if (nestlevel != nestlow) lexwarning("missing #endif");
else
#endif /* NOPP */
if (NoUnstack) lexerror("unexpected EOF");
#ifndef NOPP
nestlevel = nestlow;
#ifdef DBSYMTAB
if (options['g'] && IncludeLevel > 0) {
C_ms_stb_cst(FileName, N_EINCL, 0, (arith) 0);
}
IncludeLevel--;
#endif
/* We don't free WorkingDir and FileName here because the rest of the
* compiler may be holding pointers to them for displaying error messages.
*/
#endif /* NOPP */
return 0;
}

44
lang/cem/cemcom.ansi/macro.str
View file

@ -6,47 +6,3 @@
/* PREPROCESSOR: DEFINITION OF MACRO DESCRIPTOR */
#include "parameters.h"
#ifndef NOPP
/* The flags of the mc_flag field of the macro structure. Note that
these flags can be set simultaneously.
*/
#define NOFLAG 0 /* no special flags */
#define FUNC 0x1 /* function attached */
#define NOUNDEF 0x2 /* reserved macro */
#define NOREPLACE 0x4 /* prevent recursion */
#define FORMALP 0200 /* mask for creating macro formal parameter */
/* The macro descriptor is very simple, except the fact that the
mc_text, which points to the replacement text, contains the
non-ascii characters \201, \202, etc, indicating the position of a
formal parameter in this text.
*/
struct macro {
struct macro *next;
char * mc_text; /* the replacement text */
int mc_nps; /* number of formal parameters */
int mc_length; /* length of replacement text */
char mc_flag; /* marking this macro */
};
/* ALLOCDEF "macro" 20 */
/* `token' numbers of keywords of command-line processor
*/
#define K_UNKNOWN 0
#define K_DEFINE 1
#define K_ELIF 2
#define K_ELSE 3
#define K_ENDIF 4
#define K_ERROR 5
#define K_IF 6
#define K_IFDEF 7
#define K_IFNDEF 8
#define K_INCLUDE 9
#define K_LINE 10
#define K_PRAGMA 11
#define K_UNDEF 12
#define K_FILE 100 /* for dependency generator */
#endif /* NOPP */

196
lang/cem/cemcom.ansi/main.c
View file

@ -29,19 +29,6 @@ extern struct tokenname tkidf[];
extern char *symbol2str();
extern char options[128];
#ifndef NOPP
int inc_pos = 1; /* place where next -I goes */
int inc_total = 0;
int inc_max;
char **inctable;
extern int do_dependencies;
extern char *dep_file;
static File *dep_fd = STDOUT;
extern char *getwdir();
#endif /* NOPP */
struct sp_id special_ids[] = {
{"__setjmp", SP_SETJMP}, /* non-local goto's are registered */
{0, 0}
@ -74,10 +61,6 @@ int
union_align = AL_UNION;
#endif /* NOCROSS */
#ifndef NOPP
arith ifval; /* ifval will contain the result of the #if expression */
#endif /* NOPP */
char *prog_name;
main(argc, argv)
@ -86,17 +69,6 @@ main(argc, argv)
/* parse and interpret the command line options */
prog_name = argv[0];
#ifndef NOPP
inctable = (char **) Malloc(10 * sizeof(char *));
inctable[0] = "";
inctable[1] = 0;
inctable[2] = 0;
inc_total = 3;
inc_max = 10;
init_pp(); /* initialise the preprocessor macros */
#endif /* NOPP */
/* Note: source file "-" indicates that the source is supplied
as standard input. This is only allowed if INP_READ_IN_ONE is
not defined!
@ -122,83 +94,10 @@ main(argc, argv)
if (options['m']) Info();
#endif /* DEBUG */
#ifndef NOPP
if (do_dependencies) {
extern char *source;
list_dependencies(source);
}
#endif
sys_stop(err_occurred ? S_EXIT : S_END);
/*NOTREACHED*/
}
#ifndef NOPP
struct dependency *file_head;
extern char *strrchr();
list_dependencies(source)
char *source;
{
register struct dependency *p = file_head;
if (source) {
register char *s = strrchr(source, '.');
if (s && *(s+1)) {
s++;
*s++ = 'o';
*s = '\0';
/* the source may be in another directory than the
* object generated, so don't include the pathname
* leading to it.
*/
if (s = strrchr(source, '/')) {
source = s + 1;
}
}
else source = 0;
}
if (dep_file && !sys_open(dep_file, OP_WRITE, &dep_fd)) {
fatal("could not open %s", dep_file);
}
while (p) {
dependency(p->dep_idf->id_text, source);
p = p->next;
}
}
add_dependency(s)
char *s;
{
register struct idf *p = str2idf(s, 1);
if (! p->id_resmac) {
register struct dependency *q = new_dependency();
p->id_resmac = K_FILE;
q->dep_idf = p;
q->next = file_head;
file_head = q;
}
}
void
dependency(s, source)
char *s, *source;
{
if (options['i'] && !strncmp(s, "/usr/include/", 13)) {
return;
}
if (options['m'] && source) {
fprint(dep_fd, "%s: %s\n", source, s);
}
else fprint(dep_fd, "%s\n", s);
}
#endif /* NOPP */
char *source = 0;
#ifdef GEN_NM_LIST
@ -213,19 +112,10 @@ compile(argc, argv)
register char *destination = 0;
#endif /* LINT */
#ifdef DEBUG
#ifndef NOPP
int pp_only = options['E'] || options['P'] || options['C'];
#endif /* NOPP */
#endif
switch (argc) {
case 1:
#ifndef LINT
#ifdef DEBUG
#ifndef NOPP
if (!pp_only)
#endif /* NOPP */
#endif
fatal("%s: destination file not specified", prog_name);
#endif /* LINT */
@ -277,20 +167,7 @@ compile(argc, argv)
: 0);
#endif /* LINT */
#ifndef NOPP
WorkingDir = getwdir(source);
PushLex(); /* initialize lex machine */
#else /* NOPP */
GetToken(&ahead);
#endif /* NOPP */
#ifdef DEBUG
#ifndef NOPP
if (pp_only) /* run the preprocessor as if it is stand-alone */
preprocess();
else
#endif /* NOPP */
#endif /* DEBUG */
{
/* compile the source text */
C_program();
@ -311,9 +188,6 @@ compile(argc, argv)
dumpidftab("end of main", options['f'] ? 7 : 0);
#endif /* DEBUG */
}
#ifndef NOPP
PopLex();
#endif /* NOPP */
}
init()
@ -395,76 +269,6 @@ init_specials(si)
}
#ifdef DEBUG
#ifndef NOPP
preprocess()
{
/* preprocess() is the "stand-alone" preprocessor which
consecutively calls the lexical analyzer LLlex() to get
the tokens and prints them in a suitable way.
*/
static unsigned int lastlineno = 0;
static char *lastfilenm = "";
while (LLlex() != EOI) {
if (lastlineno != dot.tk_line) {
if (strcmp(lastfilenm, dot.tk_file) == 0) {
if (dot.tk_line - lastlineno <= 1) {
lastlineno++;
print("\n");
}
else {
lastlineno = dot.tk_line;
if (!options['P'])
print("\n#line %ld \"%s\"\n",
lastlineno,
lastfilenm
);
}
}
else {
lastfilenm = dot.tk_file;
lastlineno = dot.tk_line;
if (!options['P'])
print("\n#line %ld \"%s\"\n",
lastlineno, lastfilenm);
}
}
else
if (strcmp(lastfilenm, dot.tk_file) != 0) {
lastfilenm = dot.tk_file;
if (!options['P'])
print("\n#line %ld \"%s\"\n",
lastlineno, lastfilenm);
}
switch (DOT) {
case IDENTIFIER:
case TYPE_IDENTIFIER:
print("%s ", dot.tk_idf->id_text);
break;
case STRING:
{
char sbuf[1024]; /* a transient buffer */
print("\"%s\" ", bts2str(dot.tk_bts, dot.tk_len -
1, sbuf));
break;
}
case INTEGER:
print("%ld ", dot.tk_ival);
break;
case FLOATING:
print("%s ", dot.tk_fval);
break;
case EOI:
case EOF:
return;
default: /* very expensive... */
print("%s ", symbol2str(DOT));
}
}
}
#endif /* NOPP */
Info()
{
extern int cnt_string_cst, cnt_formal,

86
lang/cem/cemcom.ansi/options.c
View file

@ -16,16 +16,6 @@
#include "sizes.h"
#include "align.h"
#ifndef NOPP
extern char **inctable;
extern int inc_pos;
extern int inc_max;
extern int inc_total;
int do_dependencies = 0;
char *dep_file = 0;
#endif /* NOPP */
char options[128]; /* one for every char */
#ifdef LINT
char loptions[128]; /* one for every char */
@ -60,18 +50,6 @@ next_option: /* to allow combined one-char options */
goto next_option;
#ifndef LINT
#ifndef NOPP
case 'A' : /* Amake dependency generation */
do_dependencies = 1;
if (*text) {
dep_file = text;
}
break;
case 'i':
case 'm':
options[opt] = 1;
break;
#endif /* NOPP */
#endif /* LINT */
#ifdef DBSYMTAB
case 'g': /* symbol table for debugger */
@ -110,64 +88,6 @@ next_option: /* to allow combined one-char options */
goto next_option;
#endif /* LINT */
#ifndef NOPP
case 'D' : { /* -Dname : predefine name */
register char *cp = text, *name, *mactext;
unsigned maclen;
if (class(*cp) != STIDF && class(*cp) != STELL) {
error("identifier missing in -D%s", text);
break;
}
name = cp;
while (*cp && in_idf(*cp)) {
++cp;
}
if (!*cp) { /* -Dname */
maclen = 1;
mactext = Salloc("1", 2);
}
else
if (*cp == '=') { /* -Dname=text */
*cp++ = '\0'; /* end of name */
maclen = (unsigned) strlen(cp);
mactext = Salloc(cp, maclen + 1);
}
else { /* -Dname?? */
error("malformed option -D%s", text);
break;
}
macro_def(str2idf(name, 0), mactext, -1, (int)maclen, NOFLAG);
break;
}
case 'I' : /* -Ipath : insert "path" into include list */
if (*text) {
int i;
register char *new = text;
if (inc_total >= inc_max) {
inctable = (char **)
Realloc((char *)inctable,
(unsigned)((inc_max+=10)*sizeof(char *)));
}
for (i = inc_pos++; i < inc_total ; i++) {
char *tmp = inctable[i];
inctable[i] = new;
new = tmp;
}
inc_total++;
}
else inctable[inc_pos] = 0;
break;
#endif /* NOPP */
case 'M': /* maximum identifier length */
idfsize = txt2int(&text);
if (*text || idfsize <= 0)
@ -197,12 +117,6 @@ next_option: /* to allow combined one-char options */
break;
}
#ifndef NOPP
case 'U' : /* -Uname : undefine predefined */
if (*text) do_undef(str2idf(text, 0));
break;
#endif /* NOPP */
#ifndef LINT
#ifndef NOCROSS
case 'V' : /* set object sizes and alignment requirements */

12
lang/cem/cemcom.ansi/program.g
View file

@ -61,10 +61,6 @@
#include "l_lint.h"
#endif /* LINT */
#ifndef NOPP
extern arith ifval;
#endif /* NOPP */
extern error();
}
@ -75,14 +71,6 @@ control_if_expression
:
constant_expression(&exprX)
{
#ifndef NOPP
register struct expr *expr = exprX;
if (expr->ex_flags & EX_SIZEOF)
expr_error(expr,
"sizeof not allowed in preprocessor");
ifval = expr->VL_VALUE;
free_expression(expr);
#endif /* NOPP */
}
;

808
lang/cem/cemcom.ansi/replace.c
View file

@ -1,808 +0,0 @@
/*
* (c) copyright 1987 by the Vrije Universiteit, Amsterdam, The Netherlands.
* See the copyright notice in the ACK home directory, in the file "Copyright".
*/
/* $Id$ */
/* M A C R O R E P L A C E M E N T */
#include <assert.h>
#include <stdlib.h>
#include <string.h>
#include "parameters.h"
#ifndef NOPP
#include <ack_string.h>
#include <alloc.h>
#include "idf.h"
#include "input.h"
#include "macro.h"
#include "arith.h"
#include "LLlex.h"
#include "class.h"
#include "replace.h"
extern struct idf *GetIdentifier();
extern int InputLevel;
struct repl *ReplaceList; /* list of currently active macros */
void macro2buffer();
void getactuals();
void expand_defined();
int
replace(idf)
register struct idf *idf;
{
/* replace is called by the lexical analyzer to perform
macro replacement. The routine actualy functions as a
higher interface to the real thing: expand_macro().
*/
struct repl *repl;
if (!(idf->id_macro)) return 0;
if (idf->id_macro->mc_flag & NOREPLACE)
return 0;
repl = new_repl();
repl->r_ptr = repl->r_text = Malloc(repl->r_size = LAPBUF);
repl->r_args = new_args();
repl->r_idf = idf;
if (!expand_macro(repl, idf))
return 0;
InputLevel++;
InsertText(repl->r_text, (int)(repl->r_ptr - repl->r_text));
idf->id_macro->mc_flag |= NOREPLACE;
repl->r_level = InputLevel;
repl->next = ReplaceList;
ReplaceList = repl;
return 1;
}
unstackrepl()
{
Unstacked++;
}
freeargs(args)
struct args *args;
{
register int i;
/* We must don't know how many parameters were specified, so be
* prepared to free all NPARAMS parameters.
* When an expvec is !0, the rawvec will also be !0.
* When an expvec is 0, all remaining vectors will also be 0.
*/
for (i = 0; i < NPARAMS; i++) {
if (args->a_expvec[i]) {
free(args->a_expvec[i]);
free(args->a_rawvec[i]);
} else break;
}
free_args(args);
}
EnableMacros()
{
register struct repl *r = ReplaceList, *prev = 0;
assert(Unstacked > 0);
while(r) {
struct repl *nxt = r->next;
if (r->r_level > InputLevel) {
r->r_idf->id_macro->mc_flag &= ~NOREPLACE;
if (!prev) ReplaceList = nxt;
else prev->next = nxt;
free(r->r_text);
freeargs(r->r_args);
free_repl(r);
}
else prev = r;
r = nxt;
}
Unstacked = 0;
}
expand_macro(repl, idf)
register struct repl *repl;
register struct idf *idf;
{
/* expand_macro() does the actual macro replacement.
"idf" is a description of the identifier which
caused the replacement.
If the identifier represents a function-like macro
call, the number of actual parameters is checked
against the number of formal parameters. Note that
in ANSI C the parameters are expanded first;
this is done by calling getactuals().
When the possible parameters are expanded, the replace-
ment list associated with "idf" is expanded.
expand_macro() returns 1 if the replacement succeeded
and 0 if some error occurred.
A special case is "defined". This acts as a unary operator
on a single, unexpanded identifier, which may be surrounded
by parenthesis. The function expand_defined() handles this.
*/
register struct macro *mac = idf->id_macro;
struct args *args = repl->r_args;
register int ch;
if (mac->mc_nps != -1) { /* with parameter list */
if (mac->mc_flag & FUNC) {
/* the following assertion won't compile:
assert(!strcmp("defined", idf->id_text));
expand the assert macro by hand (??? dirty, temporary)
*/
#ifdef DEBUG
if (strcmp("defined", idf->id_text))
crash("in %s, %u: assertion %s failed",
__FILE__, __LINE__ - 2,
"strcmp(\"defined\", idf->id_text)");
#endif
if (!AccDefined) return 0;
expand_defined(repl);
return 1;
}
ch = GetChar();
ch = skipspaces(ch,1);
if (ch != '(') { /* no replacement if no () */
ChPushBack(ch);
return 0;
} else
getactuals(repl, idf);
}
if (mac->mc_flag & FUNC) /* this macro leads to special action */
macro_func(idf);
macro2buffer(repl, idf, args);
/* According to the ANSI definition:
#define a +
a+b; --> + + b ;
'a' must be substituded, but the result should be
three tokens: + + ID. Therefore a token separator is
inserted after the replacement.
*/
if (repl->r_text == repl->r_ptr || *(repl->r_ptr - 1) != TOKSEP) {
add2repl(repl, TOKSEP);
}
return 1;
}
void
expand_defined(repl)
register struct repl *repl;
{
register int ch = GetChar();
struct idf *id;
int parens = 0;
ch = skipspaces(ch, 0);
if (ch == '(') {
parens++;
ch = GetChar();
ch = skipspaces(ch, 0);
}
if ((class(ch) != STIDF) && (class(ch) != STELL)) {
error("identifier missing");
if (parens && ch != ')') error(") missing");
if (!parens || ch != ')') ChPushBack(ch);
add2repl(repl, '0');
return;
}
ChPushBack(ch);
id = GetIdentifier(0);
assert(id || class(ch) == STELL);
ch = GetChar();
ch = skipspaces(ch, 0);
if (parens && ch != ')') error(") missing");
if (!parens || ch != ')') ChPushBack(ch);
add2repl(repl, (id && id->id_macro) ? '1' : '0');
add2repl(repl, ' ');
}
newarg(args)
struct args *args;
{
args->a_expptr = args->a_expbuf = Malloc(args->a_expsize = ARGBUF);
args->a_rawptr = args->a_rawbuf = Malloc(args->a_rawsize = ARGBUF);
}
void
getactuals(repl, idf)
struct repl *repl;
register struct idf *idf;
{
/* Get the actual parameters from the input stream.
The hard part is done by actual(), only comma's and
other syntactic trivialities are checked here.
*/
register struct args *args = repl->r_args;
register int nps = idf->id_macro->mc_nps;
register int argcnt;
register int ch;
argcnt = 0;
newarg(args);
if ((ch = GetChar()) != ')') {
UnGetChar();
while ((ch = actual(repl)) != ')' ) {
if (ch != ',') {
lexerror("illegal macro call");
return;
}
stash(repl, '\0', 1);
args->a_expvec[argcnt] = args->a_expbuf;
args->a_rawvec[argcnt] = args->a_rawbuf;
++argcnt;
if (argcnt == STDC_NPARAMS)
lexstrict("number of parameters exceeds ANSI standard");
if (argcnt >= NPARAMS)
fatal("argument vector overflow");
newarg(args);
}
stash(repl, '\0', 1);
args->a_expvec[argcnt] = args->a_expbuf;
args->a_rawvec[argcnt] = args->a_rawbuf;
++argcnt;
}
if (argcnt < nps)
lexerror("too few macro arguments");
else if (argcnt > nps)
lexerror("too many macro arguments");
}
saveraw(repl)
struct repl *repl;
{
register struct repl *nrepl = ReplaceList;
register struct args *ap = nrepl->r_args;
register char *p;
/* stash identifier name */
for (p = nrepl->r_idf->id_text; *p != '\0'; p++)
stash(repl, *p, -1);
/* The following code deals with expanded function
like macro calls. It makes the following code
work:
#define def(a,b) x(a,b)
#define glue(a,b) a ## b
glue(abc,def(a,b))
Results in:
abcdef(a,b);
*/
if (ap->a_rawvec[0]) {
/* stash arguments */
register int i;
for (i = 0; ap->a_rawvec[i] != (char *)0; i++) {
if (i == 0) stash(repl, '(', -1);
else stash(repl, ',', -1);
for (p = ap->a_rawvec[i]; *p != '\0'; p++)
stash(repl, *p, -1);
}
stash(repl, ')', -1);
}
}
int
actual(repl)
struct repl *repl;
{
/* This routine deals with the scanning of an actual parameter.
It keeps in account the opening and closing brackets,
preprocessor numbers, strings and character constants.
*/
register int ch = 0;
register int level = 0, nostashraw = 0;
int lastch;
static int Unstacked_missed;
while (1) {
lastch = ch;
ch = GetChar();
if (nostashraw
&& nostashraw >= Unstacked_missed) {
nostashraw -= Unstacked_missed;
Unstacked_missed = 0;
}
if (Unstacked) {
nostashraw -= Unstacked;
if (nostashraw < 0) {
Unstacked_missed = -nostashraw;
nostashraw = 0;
}
EnableMacros();
}
if (class(ch) == STIDF || class(ch) == STELL) {
/* Scan a preprocessor identifier token. If the
token is a macro, it is expanded first.
*/
char buf[(IDFSIZE > NUMSIZE ? IDFSIZE : NUMSIZE) + 1];
register char *p = buf;
register struct idf *idef;
register int pos = -1;
extern int idfsize;
int NoExpandMacro;
if (ch == NOEXPM) {
NoExpandMacro= 1;
ch = GetChar();
} else NoExpandMacro = 0;
do {
if (++pos < idfsize) {
*p++ = ch;
}
ch = GetChar();
} while (in_idf(ch));
*p++ = '\0';
ch = '\0'; /* It could be an unstashed TOKSEP */
UnGetChar();
/* When the identifier has an associated macro
replacement list, it's expanded.
*/
idef = findidf(buf);
if (!idef || NoExpandMacro || !replace(idef)) {
if (NoExpandMacro
|| (idef && idef->id_macro
&& (idef->id_macro->mc_flag & NOREPLACE)))
stash(repl, NOEXPM, !nostashraw);
for (p = buf; *p != '\0'; p++)
stash(repl, *p, !nostashraw);
} else {
if (!nostashraw) saveraw(repl);
nostashraw++;
}
} else if (class(ch) == STNUM) {
/* a preprocessing number has the following
regular expression:
[0-9|"."[0-9]]{[0-9"."a-zA-Z_]|{[Ee][+-]}}*
*/
stash(repl, ch, !nostashraw);
if (ch == '.') {
ch = GetChar();
if (class(ch) != STNUM) {
ch = '\0'; /* It could be an unstashed TOKSEP */
UnGetChar();
continue;
}
else stash(repl, ch, !nostashraw);
}
ch = GetChar();
while (in_idf(ch) || ch == '.') {
stash(repl, ch, !nostashraw);
if ((ch = GetChar()) == 'e' || ch == 'E') {
stash(repl, ch, !nostashraw);
ch = GetChar();
if (ch == '+' || ch == '-') {
stash(repl, ch, !nostashraw);
ch = GetChar();
}
}
}
ch = '\0'; /* It could be an unstashed TOKSEP */
UnGetChar();
} else if (ch == '(') {
/* a comma may occur between parentheses */
level++;
stash(repl, ch, !nostashraw);
} else if (ch == ')') {
level--;
/* closing parenthesis of macro call */
if (level < 0) return ')';
stash(repl, ch, !nostashraw);
} else if (ch == ',') {
if (level <= 0) { /* comma separator for next argument */
if (level)
lexerror("unbalanced parenthesis");
if (!nostashraw)
return ','; /* ??? */
}
stash(repl, ch, !nostashraw);
} else if (ch == '\n') {
/* newlines are accepted as white spaces */
LineNumber++;
/* This piece of code needs some explanation:
consider the call of a macro defined as:
#define sum(a,b) (a+b)
in the following form:
sum(
/_* comment *_/ #include phone_number
,2);
in which case the include must be handled
interpreted as such.
*/
a_new_line: ch = GetChar();
while (class(ch) == STSKIP || ch == '/') {
if (ch == '/') {
if ((ch = GetChar()) == '*' && !InputLevel) {
skipcomment();
stash(repl, ' ', !nostashraw);
ch = GetChar();
continue;
} else {
UnGetChar();
ch = '/';
}
stash(repl, '/', !nostashraw);
break;
} else ch = GetChar();
}
if (ch == '#') {
domacro();
/* Clear File_Inserted since domacro could
* be called again, which calls GetToken().
*/
File_Inserted = 0;
goto a_new_line;
} else if (ch == EOI) {
lexerror("unterminated macro call");
return ')';
}
if (ch != '/') {
UnGetChar();
ch = ' ';
stash(repl, ' ', !nostashraw);
}
} else if (ch == '/') {
/* comments are treated as one white space token */
if ((ch = GetChar()) == '*' && !InputLevel) {
skipcomment();
stash(repl, ' ', !nostashraw);
} else {
UnGetChar();
ch = '/';
stash(repl, '/', !nostashraw);
}
} else if (ch == '\'' || ch == '"') {
/* Strings are considered as ONE token, thus no
replacement within strings.
*/
register int match = ch;
stash(repl, ch, !nostashraw);
while ((ch = GetChar()) != EOI) {
if (ch == match)
break;
if (ch == '\\') {
stash(repl, ch, !nostashraw);
ch = GetChar();
} else if (ch == '\n') {
lexerror("newline in string");
LineNumber++;
stash(repl, match, !nostashraw);
break;
}
stash(repl, ch, !nostashraw);
}
if (ch != match) {
lexerror("unterminated macro call");
return ')';
}
stash(repl, ch, !nostashraw);
} else {
if (lastch == TOKSEP && ch == TOKSEP) continue;
stash(repl, ch, !nostashraw);
}
}
}
macro_func(idef)
register struct idf *idef;
{
/* macro_func() performs the special actions needed with some
macros. These macros are __FILE__ and __LINE__ which
replacement texts must be evaluated at the time they are
used.
*/
register struct macro *mac = idef->id_macro;
static char FilNamBuf[PATHLENGTH];
switch (idef->id_text[2]) {
case 'F': /* __FILE__ */
FilNamBuf[0] = '"';
strcpy(&FilNamBuf[1], FileName);
strcat(FilNamBuf, "\"");
mac->mc_text = FilNamBuf;
mac->mc_length = strlen(FilNamBuf);
break;
case 'L': /* __LINE__ */
mac->mc_text = long2str((long)LineNumber, 10);
mac->mc_length = strlen(mac->mc_text);
break;
default:
crash("(macro_func)");
/*NOTREACHED*/
}
}
void
macro2buffer(repl, idf, args)
register struct repl *repl;
register struct idf *idf;
register struct args *args;
{
/* macro2buffer expands the replacement list and places the
result onto the replacement buffer. It deals with the #
and ## operators, and inserts the actual parameters.
The argument buffer contains the raw argument (needed
for the ## operator), and the expanded argument (for
all other parameter substitutions).
The grammar of the replacement list is:
repl_list: TOKEN repl_list
| PARAMETER repl_list
| '#' PARAMETER
| TOKEN '##' TOKEN
| PARAMETER '##' TOKEN
| TOKEN '##' PARAMETER
| PARAMETER '##' PARAMETER
;
As the grammar indicates, we could make a DFA and
use this finite state machine for the replacement
list parsing (inserting the arguments, etc.).
Currently we go through the replacement list in a
linear fashion. This is VERY expensive, something
smarter should be done (but even a DFA is O(|s|)).
*/
register char *ptr = idf->id_macro->mc_text;
int err = 0;
int func = idf->id_macro->mc_nps != -1;
char *stringify();
assert(ptr[idf->id_macro->mc_length] == '\0');
while (*ptr) {
if (*ptr == '\'' || *ptr == '"') {
register int delim = *ptr;
do {
add2repl(repl, *ptr);
if (*ptr == '\\')
add2repl(repl, *++ptr);
if (*ptr == '\0') {
lexerror("unterminated string");
return;
}
ptr++;
} while (*ptr != delim || *ptr == '\0');
add2repl(repl, *ptr++);
} else if (*ptr == '#' && (func || *(ptr+1) == '#')) {
if (*++ptr == '#') {
register int tmpindex;
/* ## - paste operator */
ptr++;
/* trim the actual replacement list */
--repl->r_ptr;
while (repl->r_ptr >= repl->r_text
&& is_wsp(*repl->r_ptr))
--repl->r_ptr;
/* ## occurred at the beginning of the replacement list.
*/
if (repl->r_ptr < repl->r_text) {
err = 1;
break;
}
if (repl->r_ptr >= repl->r_text
&& *repl->r_ptr == TOKSEP)
--repl->r_ptr;
++repl->r_ptr;
tmpindex = repl->r_ptr - repl->r_text;
/* tmpindex can be 0 */
/* skip space in macro replacement list */
while ((*ptr & FORMALP) == 0 && is_wsp(*ptr))
ptr++;
/* ## occurred at the end of the replacement list.
*/
if (*ptr & FORMALP) {
register int n = *ptr++ & 0177;
register char *p;
assert(n > 0);
p = args->a_rawvec[n-1];
if (p) { /* else macro argument missing */
while (is_wsp(*p)) p++;
if (*p == NOEXPM) p++;
while (*p)
add2repl(repl, *p++);
}
while (tmpindex > 0
&& in_idf(repl->r_text[tmpindex]))
tmpindex--;
if (tmpindex >= 0
&& repl->r_text[tmpindex] == NOEXPM)
repl->r_text[tmpindex] = TOKSEP;
} else if (*ptr == '\0') {
err = 1;
break;
} else {
if (in_idf(*ptr)) {
tmpindex--;
while (tmpindex > 0
&& in_idf(repl->r_text[tmpindex]))
tmpindex--;
if (tmpindex >= 0
&& repl->r_text[tmpindex] == NOEXPM)
repl->r_text[tmpindex] = TOKSEP;
}
}
} else { /* # operator */
ptr = stringify(repl, ptr, args);
}
} else if (*ptr & FORMALP) {
/* insert actual parameter */
register int n = *ptr++ & 0177;
register char *p, *q;
assert(n > 0);
/* This is VERY dirty, we look ahead for the
## operator. If it's found we use the raw
argument buffer instead of the expanded
one.
*/
for (p = ptr; (*p & FORMALP) == 0 && is_wsp(*p); p++)
/* EMPTY */;
if (*p == '#' && p[1] == '#')
q = args->a_rawvec[n-1];
else
q = args->a_expvec[n-1];
if (q) /* else macro argument missing */
while (*q)
add2repl(repl, *q++);
if (repl->r_text == repl->r_ptr || *(repl->r_ptr - 1) != TOKSEP)
add2repl(repl, TOKSEP);
} else {
add2repl(repl, *ptr++);
}
}
if (err)
lexerror("illegal use of the ## operator");
}
char *
stringify(repl, ptr, args)
register struct repl *repl;
register char *ptr;
register struct args *args;
{
/* If a parameter is immediately preceded by a # token
both are replaced by a single string literal that
contains the spelling of the token sequence for the
corresponding argument.
Each occurrence of white space between the argument's
tokens become a single space character in the string
literal. White spaces before the first token and after
the last token comprising the argument are deleted.
To retain the original spelling we insert backslashes
as appropriate. We only escape backslashes if they
occure within string tokens.
*/
register int space = 1; /* skip leading spaces */
register int delim = 0; /* string or character constant delim */
register int backslash = 0; /* last character was a \ */
/* skip spaces macro replacement list */
while ((*ptr & FORMALP) == 0 && is_wsp(*ptr))
ptr++;
if (*ptr & FORMALP) {
register int n = *ptr++ & 0177;
register char *p;
assert(n != 0);
p = args->a_rawvec[n-1];
add2repl(repl, '"');
while (*p) {
if (is_wsp(*p)) {
if (!space) {
space = 1;
add2repl(repl, ' ');
}
p++;
continue;
}
space = 0;
if (!delim && (*p == '"' || *p == '\''))
delim = *p;
else if (*p == delim && !backslash)
delim = 0;
backslash = *p == '\\';
if (*p == '"' || (delim && *p == '\\'))
add2repl(repl, '\\');
if (*p == TOKSEP || *p == NOEXPM) p++;
else add2repl(repl, *p++);
}
/* trim spaces in the replacement list */
for (--repl->r_ptr; is_wsp(*repl->r_ptr); repl->r_ptr--)
/* EMPTY */;
++repl->r_ptr; /* oops, one to far */
add2repl(repl, '"');
} else
error("illegal use of # operator");
return ptr;
}
/* The following routine is also called from domacro.c.
*/
add2repl(repl, ch)
register struct repl *repl;
int ch;
{
register int index = repl->r_ptr - repl->r_text;
assert(index < repl->r_size);
if (index + 2 >= repl->r_size) {
repl->r_text = Realloc(repl->r_text, (unsigned) (repl->r_size <<= 1));
repl->r_ptr = repl->r_text + index;
}
*repl->r_ptr++ = ch;
*repl->r_ptr = '\0';
}
/* If the variable stashraw is negative, we must only stash into the raw
* buffer. If the variable is zero, we must only stash into the expanded
* buffer. Otherwise, we must use both buffers.
*/
stash(repl, ch, stashraw)
struct repl *repl;
register int ch;
int stashraw;
{
/* Stash characters into the macro expansion buffer.
*/
register struct args *args = repl->r_args;
register int index = args->a_expptr - args->a_expbuf;
if (stashraw >= 0) {
assert(index < args->a_expsize);
if (index + 1 >= args->a_expsize) {
args->a_expbuf = Realloc(args->a_expbuf,
(unsigned) (args->a_expsize <<= 1));
args->a_expptr = args->a_expbuf + index;
}
*args->a_expptr++ = ch;
}
if (stashraw) {
index = args->a_rawptr - args->a_rawbuf;
assert(index < args->a_rawsize);
if (index + 1 >= args->a_rawsize) {
args->a_rawbuf = Realloc(args->a_rawbuf,
(unsigned)(args->a_rawsize <<= 1));
args->a_rawptr = args->a_rawbuf + index;
}
*args->a_rawptr++ = ch;
}
}
#endif /* NOPP */

51
lang/cem/cemcom.ansi/replace.str
View file

@ -1,51 +0,0 @@
/*
* (c) copyright 1987 by the Vrije Universiteit, Amsterdam, The Netherlands.
* See the copyright notice in the ACK home directory, in the file "Copyright".
*/
/* $Id$ */
/* DEFINITIONS FOR THE MACRO REPLACEMENT ROUTINES */
struct repl {
struct repl *next;
struct idf *r_idf; /* name of the macro */
struct args *r_args; /* replacement parameters */
int r_level; /* level of insertion */
int r_size; /* current size of replacement buffer */
char *r_ptr; /* replacement text index pointer */
char *r_text; /* replacement text */
};
/* ALLOCDEF "repl" 4 */
#define NO_REPL (struct repl *)0
/* The implementation of the ## operator is currently very clumsy.
When the the ## operator is used the arguments are taken from
the raw buffer; this buffer contains a precise copy of the
original argument. The fully expanded copy is in the arg buffer.
The two copies are here explicitely because:
#define ABC f()
#define ABCD 2
#define g(x, y) x ## y + h(x)
g(ABC, D); // gives: 2 + h(f())
In this case we need two copies: one raw copy for the pasting
operator, and an expanded one as argument for h().
*/
struct args {
char *a_expptr; /* expanded argument index pointer */
char *a_expbuf; /* expanded argument buffer pointer */
int a_expsize; /* current size of expanded buffer */
char *a_expvec[NPARAMS]; /* expanded argument vector */
char *a_rawptr; /* raw argument index pointer */
char *a_rawbuf; /* raw argument buffer pointer */
int a_rawsize; /* current size of raw buffer */
char *a_rawvec[NPARAMS]; /* raw argument vector */
};
/* ALLOCDEF "args" 2 */
#define NO_ARGS (struct args *)0

108
lang/cem/cemcom.ansi/skip.c
View file

@ -11,125 +11,17 @@
#include "class.h"
#include "input.h"
#ifndef NOPP
extern int InputLevel;
int skipspaces(ch, skipnl) register int ch;
{
/* skipspaces() skips any white space and returns the first
non-space character.
*/
register int nlseen = 0;
for (;;)
{
while (class(ch) == STSKIP)
ch = GetChar();
if (skipnl && class(ch) == STNL)
{
ch = GetChar();
LineNumber++;
nlseen++;
continue;
}
if (ch == TOKSEP && InputLevel)
{
ch = GetChar();
continue;
}
/* \\\n are handled by trigraph */
if (ch == '/')
{
ch = GetChar();
if (ch == '*' && !InputLevel)
{
skipcomment();
ch = GetChar();
}
else if (ch == '/' && !InputLevel)
{
skiplinecomment();
ch = GetChar();
}
else
{
UnGetChar();
return '/';
}
}
else if (nlseen && ch == '#')
{
domacro();
ch = GetChar();
}
else
return ch;
}
}
#endif /* NOPP */
SkipToNewLine()
{
register int ch;
register int garbage = 0;
#ifndef NOPP
register int delim = 0;
#endif
while ((ch = GetChar()) != '\n')
{
#ifndef NOPP
if (delim)
{
if (ch == '\\')
{
if (GetChar() == '\n')
break;
}
else if (ch == delim)
{
delim = 0;
}
continue;
}
else if (ch == '\'' || ch == '\"')
{
delim = ch;
garbage = 1;
}
else if (ch == '/')
{
if (!InputLevel)
{
int nch = GetChar();
if (nch == '*')
{
skipcomment();
continue;
}
else if (nch == '/')
{
skiplinecomment();
continue;
}
else
UnGetChar();
}
}
else if (ch == TOKSEP && InputLevel)
{
continue;
}
#endif
if (!is_wsp(ch))
garbage = 1;
}
#ifndef NOPP
if (delim)
strict("unclosed opening %c", delim);
#endif
++LineNumber;
return garbage;
}