mirror of
https://github.com/CloverHackyColor/CloverBootloader.git
synced 2024-12-02 13:03:28 +01:00
b1264ef1e3
Signed-off-by: Sergey Isakov <isakov-sl@bk.ru>
195 lines
5.2 KiB
C
Executable File
195 lines
5.2 KiB
C
Executable File
/* DLexer.h (formerly DLexer.cpp)
|
|
*
|
|
* This was renamed because the use of the .cpp extension caused problems
|
|
* with IDEs.
|
|
*
|
|
* SOFTWARE RIGHTS
|
|
*
|
|
* We reserve no LEGAL rights to the Purdue Compiler Construction Tool
|
|
* Set (PCCTS) -- PCCTS is in the public domain. An individual or
|
|
* company may do whatever they wish with source code distributed with
|
|
* PCCTS or the code generated by PCCTS, including the incorporation of
|
|
* PCCTS, or its output, into commerical software.
|
|
*
|
|
* We encourage users to develop software with PCCTS. However, we do ask
|
|
* that credit is given to us for developing PCCTS. By "credit",
|
|
* we mean that if you incorporate our source code into one of your
|
|
* programs (commercial product, research project, or otherwise) that you
|
|
* acknowledge this fact somewhere in the documentation, research report,
|
|
* etc... If you like PCCTS and have developed a nice tool with the
|
|
* output, please mention that you developed it using PCCTS. In
|
|
* addition, we ask that this header remain intact in our source code.
|
|
* As long as these guidelines are kept, we expect to continue enhancing
|
|
* this system and expect to make other tools available as they are
|
|
* completed.
|
|
*
|
|
* ANTLR 1.33
|
|
* Terence Parr
|
|
* Parr Research Corporation
|
|
* with Purdue University and AHPCRC, University of Minnesota
|
|
* 1989-2000
|
|
*/
|
|
|
|
#include <assert.h>
|
|
|
|
#define ZZINC {if ( track_columns ) (++_endcol);}
|
|
|
|
#define ZZGETC {ch = input->nextChar(); cl = ZZSHIFT(ch);}
|
|
|
|
#define ZZNEWSTATE (newstate = dfa[state][cl])
|
|
|
|
#ifndef ZZCOPY
|
|
#define ZZCOPY \
|
|
/* Truncate matching buffer to size (not an error) */ \
|
|
if (nextpos < lastpos){ \
|
|
*(nextpos++) = ch; \
|
|
}else{ \
|
|
bufovf = 1; \
|
|
}
|
|
#endif
|
|
|
|
void DLGLexer::
|
|
mode( int m )
|
|
{
|
|
/* points to base of dfa table */
|
|
if (m<MAX_MODE){
|
|
automaton = m;
|
|
/* have to redo class since using different compression */
|
|
cl = ZZSHIFT(ch);
|
|
}else{
|
|
sprintf((char *)ebuf,"Invalid automaton mode = %d ",m);
|
|
errstd(ebuf);
|
|
}
|
|
}
|
|
|
|
ANTLRTokenType DLGLexer::
|
|
nextTokenType(void)
|
|
{
|
|
register int state, newstate;
|
|
/* last space reserved for the null char */
|
|
register DLGChar *lastpos;
|
|
ANTLRTokenType tk;
|
|
|
|
skip:
|
|
bufovf = 0;
|
|
lastpos = &_lextext[_bufsize-1];
|
|
nextpos = _lextext;
|
|
_begcol = _endcol+1;
|
|
more:
|
|
_begexpr = nextpos;
|
|
if ( interactive ) {
|
|
/* interactive version of automaton */
|
|
/* if there is something in ch, process it */
|
|
state = newstate = dfa_base[automaton];
|
|
if (charfull){
|
|
ZZINC;
|
|
ZZCOPY;
|
|
ZZNEWSTATE;
|
|
}
|
|
while (alternatives[newstate]){
|
|
state = newstate;
|
|
ZZGETC;
|
|
ZZINC;
|
|
ZZCOPY;
|
|
ZZNEWSTATE;
|
|
}
|
|
/* figure out if last character really part of token */
|
|
if ((state != dfa_base[automaton]) && (newstate == DfaStates)){
|
|
charfull = 1;
|
|
--nextpos;
|
|
}else{
|
|
charfull = 0;
|
|
state = newstate;
|
|
}
|
|
*(nextpos) = '\0';
|
|
/* Able to transition out of start state to some non err state?*/
|
|
if ( state == dfa_base[automaton] ){
|
|
/* make sure doesn't get stuck */
|
|
advance();
|
|
}
|
|
}
|
|
else { /* non-interactive version of automaton */
|
|
if (!charfull)
|
|
advance();
|
|
else
|
|
ZZINC;
|
|
state = dfa_base[automaton];
|
|
while (ZZNEWSTATE != DfaStates) {
|
|
state = newstate;
|
|
assert(state <= sizeof(dfa)/sizeof(dfa[0]));
|
|
ZZCOPY;
|
|
ZZGETC;
|
|
ZZINC;
|
|
}
|
|
charfull = 1;
|
|
if ( state == dfa_base[automaton] ){
|
|
if (nextpos < lastpos){
|
|
*(nextpos++) = ch;
|
|
}else{
|
|
bufovf = 1;
|
|
}
|
|
*nextpos = '\0';
|
|
/* make sure doesn't get stuck */
|
|
advance();
|
|
}else{
|
|
*nextpos = '\0';
|
|
}
|
|
}
|
|
if ( track_columns ) _endcol -= charfull;
|
|
_endexpr = nextpos -1;
|
|
add_erase = 0;
|
|
#ifdef OLD
|
|
tk = (ANTLRTokenType)
|
|
(*actions[accepts[state]])(this); // must pass this manually
|
|
// actions is not a [] of pointers
|
|
// to member functions.
|
|
#endif
|
|
tk = (this->*actions[accepts[state]])();
|
|
|
|
// MR1
|
|
// MR1 11-Apr-97 Help for tracking DLG results
|
|
// MR1
|
|
|
|
#ifdef DEBUG_LEXER
|
|
|
|
/* MR1 */ if (debugLexerFlag) {
|
|
/* MR1 */ if (parser != NULL) {
|
|
/* MR1 */ /* MR23 */ printMessage(stdout, "\ntoken name=%s",parser->parserTokenName(tk));
|
|
/* MR1 */ } else {
|
|
/* MR1 */ /* MR23 */ printMessage(stdout, "\ntoken nnumber=%d",tk);
|
|
/* MR1 */ };
|
|
/* MR1 */ /* MR23 */ printMessage(stdout, " lextext=(%s) mode=%d",
|
|
/* MR1 */ (_lextext[0]=='\n' && _lextext[1]==0) ?
|
|
/* MR1 */ "newline" : _lextext,
|
|
/* MR1 */ automaton);
|
|
/* MR1 */ if (interactive && !charfull) {
|
|
/* MR1 */ /* MR23 */ printMessage(stdout, " char=empty");
|
|
/* MR1 */ } else {
|
|
/* MR1 */ if (ch=='\n') {
|
|
/* MR1 */ /* MR23 */ printMessage(stdout, " char=newline");
|
|
/* MR1 */ } else {
|
|
/* MR1 */ /* MR23 */ printMessage(stdout, " char=(%c)",ch);
|
|
/* MR1 */ };
|
|
/* MR1 */ };
|
|
/* MR1 */ /* MR23 */ printMessage(stdout, " %s\n",
|
|
/* MR1 */ (add_erase==1 ? "skip()" :
|
|
/* MR1 */ add_erase==2 ? "more()" :
|
|
/* MR1 */ ""));
|
|
/* MR1 */ };
|
|
|
|
#endif
|
|
|
|
switch (add_erase) {
|
|
case 1: goto skip;
|
|
case 2: goto more;
|
|
}
|
|
return tk;
|
|
}
|
|
|
|
void DLGLexer::
|
|
advance()
|
|
{
|
|
if ( input==NULL ) err_in();
|
|
ZZGETC; charfull = 1; ZZINC;
|
|
}
|