scan.l

C++版 词法分析、语法分析器

			}

<XLATION>{LINEFEED}		++linenum;


<SECT2PROLOG>.*{LINEFEED}/{NOT_WS}	{
			++linenum;
			ACTION_ECHO;
			MARK_END_OF_PROLOG;
			BEGIN(SECT2);
			}

<SECT2PROLOG>.*{LINEFEED}	++linenum; ACTION_ECHO;

<SECT2PROLOG><<EOF>>	MARK_END_OF_PROLOG; yyterminate();

<SECT2>^{OPTWS}{LINEFEED}	++linenum; /* allow blank lines in section 2 */

<SECT2>^"%header{".*{LINEFEED}	{
			line_directive_out( headerfile );
			BEGIN(HEADER2_BLOC);
			}
<SECT2>^({WS}|"%{")	{
			indented_code = (yytext[0] != '%');
			doing_codeblock = true;
			bracelevel = 1;

			if ( indented_code )
			    ACTION_ECHO;

			BEGIN(CODEBLOCK_2);
			}

<SECT2>"<"		BEGIN(SC); return ( '<' );
<SECT2>^"^"		return ( '^' );
<SECT2>\"		BEGIN(QUOTE); return ( '"' );
<SECT2>"{"/[0-9]		BEGIN(NUM); return ( '{' );
<SECT2>"{"[^0-9\n\r][^}\n\r]*	BEGIN(BRACEERROR);
<SECT2>"$"/[ \t\n\r]	return ( '$' );

<SECT2>{WS}"%{"		{
			bracelevel = 1;
			BEGIN(PERCENT_BRACE_ACTION);
			return ( '\n' );
			}
<SECT2>{WS}"|".*{LINEFEED}	continued_action = true; ++linenum; return ( '\n' );

<SECT2>{WS}		{
			/* this rule is separate from the one below because
			 * otherwise we get variable trailing context, so
			 * we can't build the scanner using -{f,F}
			 */
			bracelevel = 0;
			continued_action = false;
			BEGIN(ACTION);
			return ( '\n' );
			}

<SECT2>{OPTWS}/{LINEFEED}	{
			bracelevel = 0;
			continued_action = false;
			BEGIN(ACTION);
			return ( '\n' );
			}

<SECT2>^{OPTWS}{LINEFEED}	++linenum; return ( '\n' );

<SECT2>"<<EOF>>"	return ( EOF_OP );

<SECT2>^"%%".*		{
			sectnum = 3;
			BEGIN(SECT3);
			return ( EOF ); /* to stop the parser */
			}

<SECT2>"["([^\\\]\n\r]|{ESCSEQ})+"]"	{
			int cclval;

			(void) strcpy( nmstr, (char *) yytext );

			/* check to see if we've already encountered this ccl */
			if ( (cclval = ccllookup( (Char *) nmstr )) )
			    {
			    yylval = cclval;
			    ++cclreuse;
			    return ( PREVCCL );
			    }
			else
			    {
			    /* we fudge a bit.  We know that this ccl will
			     * soon be numbered as lastccl + 1 by cclinit
			     */
			    cclinstal( (Char *) nmstr, lastccl + 1 );

			    /* push back everything but the leading bracket
			     * so the ccl can be rescanned
			     */
			    PUT_BACK_STRING((Char *) nmstr, 1);

			    BEGIN(FIRSTCCL);
			    return ( '[' );
			    }
			}

<SECT2>"{"{NAME}"}"	{
			register Char *nmdefptr;
			Char *ndlookup();

			(void) strcpy( nmstr, (char *) yytext );
			nmstr[yyleng - 1] = '\0';  /* chop trailing brace */

			/* lookup from "nmstr + 1" to chop leading brace */
			if ( ! (nmdefptr = ndlookup( nmstr + 1 )) )
			    synerr( "undefined {name}" );

			else
			    { /* push back name surrounded by ()'s */
			    unput(')');
			    PUT_BACK_STRING(nmdefptr, 0);
			    unput('(');
			    }
			}

<SECT2>[/|*+?.()]	return ( yytext[0] );
<SECT2>.		RETURNCHAR;
<SECT2>{LINEFEED}		++linenum; return ( '\n' );


<SC>","			return ( ',' );
<SC>">"			BEGIN(SECT2); return ( '>' );
<SC>">"/"^"		BEGIN(CARETISBOL); return ( '>' );
<SC>{SCNAME}		RETURNNAME;
<SC>.			synerr( "bad start condition name" );

<CARETISBOL>"^"		BEGIN(SECT2); return ( '^' );


<QUOTE>[^"\n\r]		RETURNCHAR;
<QUOTE>\"		BEGIN(SECT2); return ( '"' );

<QUOTE>{LINEFEED}		{
			synerr( "missing quote" );
			BEGIN(SECT2);
			++linenum;
			return ( '"' );
			}


<FIRSTCCL>"^"/[^-\n\r]	BEGIN(CCL); return ( '^' );
<FIRSTCCL>"^"/-		return ( '^' );
<FIRSTCCL>-		BEGIN(CCL); yylval = '-'; return ( CHAR );
<FIRSTCCL>.		BEGIN(CCL); RETURNCHAR;

<CCL>-/[^\]\n\r]		return ( '-' );
<CCL>[^\]\n\r]		RETURNCHAR;
<CCL>"]"		BEGIN(SECT2); return ( ']' );


<NUM>[0-9]+		{
			yylval = myctoi( yytext );
			return ( NUMBER );
			}

<NUM>","			return ( ',' );
<NUM>"}"			BEGIN(SECT2); return ( '}' );

<NUM>.			{
			synerr( "bad character inside {}'s" );
			BEGIN(SECT2);
			return ( '}' );
			}

<NUM>{LINEFEED}			{
			synerr( "missing }" );
			BEGIN(SECT2);
			++linenum;
			return ( '}' );
			}


<BRACEERROR>"}"		synerr( "bad name in {}'s" ); BEGIN(SECT2);
<BRACEERROR>{LINEFEED}		synerr( "missing }" ); ++linenum; BEGIN(SECT2);


<PERCENT_BRACE_ACTION,CODEBLOCK_2>{OPTWS}"%}".*		bracelevel = 0;
<PERCENT_BRACE_ACTION,CODEBLOCK_2,ACTION>"reject"	{
			ACTION_ECHO;
			CHECK_REJECT(yytext);
			}
<PERCENT_BRACE_ACTION,CODEBLOCK_2,ACTION>"yymore"	{
			ACTION_ECHO;
			CHECK_YYMORE(yytext);
			}
<PERCENT_BRACE_ACTION,CODEBLOCK_2>{NAME}|{NOT_NAME}|.	ACTION_ECHO;
<PERCENT_BRACE_ACTION,CODEBLOCK_2>{LINEFEED}			{
			++linenum;
			ACTION_ECHO;
			if ( bracelevel == 0 ||
			     (doing_codeblock && indented_code) )
			    {
			    if ( ! doing_codeblock )
				fputs( "\tYY_BREAK\n", temp_action_file );
			    
			    doing_codeblock = false;
			    BEGIN(SECT2);
			    }
			}


	/* Reject and YYmore() are checked for above, in PERCENT_BRACE_ACTION */
<ACTION>"{"		ACTION_ECHO; ++bracelevel;
<ACTION>"}"		ACTION_ECHO; --bracelevel;
<ACTION>[^a-z_{}"'/\n\r]+	ACTION_ECHO;
<ACTION>{NAME}		ACTION_ECHO;
<ACTION>"//".*{LINEFEED}	++linenum;ACTION_ECHO; 
<ACTION>"/*"		ACTION_ECHO; BEGIN(ACTION_COMMENT);
<ACTION>"'"([^'\\\n\r]|\\.)*"'"	ACTION_ECHO; /* character constant */
<ACTION>\"		ACTION_ECHO; BEGIN(ACTION_STRING);
<ACTION>{LINEFEED}		{
			++linenum;
			ACTION_ECHO;
			if ( bracelevel == 0 )
			    {
			    fputs( "\tYY_BREAK\n", temp_action_file );
			    BEGIN(SECT2);
			    }
			}
<ACTION>.		ACTION_ECHO;

<ACTION_COMMENT>"*/"	ACTION_ECHO; BEGIN(ACTION);
<ACTION_COMMENT>[^*\n\r]+	ACTION_ECHO;
<ACTION_COMMENT>"*"	ACTION_ECHO;
<ACTION_COMMENT>{LINEFEED}	++linenum; ACTION_ECHO;
<ACTION_COMMENT>.	ACTION_ECHO;

<ACTION_STRING>[^"\\\n\r]+	ACTION_ECHO;
<ACTION_STRING>\\.	ACTION_ECHO;
<ACTION_STRING>{LINEFEED}	++linenum; ACTION_ECHO;
<ACTION_STRING>\"	ACTION_ECHO; BEGIN(ACTION);
<ACTION_STRING>.	ACTION_ECHO;

<ACTION,ACTION_COMMENT,ACTION_STRING><<EOF>>	{
			synerr( "EOF encountered inside an action" );
			yyterminate();
			}


<SECT2,QUOTE,CCL>{ESCSEQ}	{
			yylval = myesc( yytext );
			return ( CHAR );
			}

<FIRSTCCL>{ESCSEQ}	{
			yylval = myesc( yytext );
			BEGIN(CCL);
			return ( CHAR );
			}


<SECT3>.*({LINEFEED}?)		ECHO;
%%


int yywrap()

    {
    if ( --num_input_files > 0 )
	{
	set_input_file( *++input_files );
	return ( 0 );
	}

    else
	return ( 1 );
    }


/* set_input_file - open the given file (if NULL, stdin) for scanning */

void set_input_file( file )
char *file;

    {
    if ( file )
	{
	infilename = file;
	yyin = fopen( infilename, "r" );

	if ( yyin == NULL )
	    lerrsf( "can't open %s", file );
	}

    else
	{
	yyin = stdin;
	infilename = "<stdin>";
	}
    }