xref: /aoo42x/main/l10ntools/source/xrmlex.l (revision cdf0e10c)
1*cdf0e10cSrcweir %{
2*cdf0e10cSrcweir /*
3*cdf0e10cSrcweir  * lexer for parsing xml-property source files (*.xml)
4*cdf0e10cSrcweir  *
5*cdf0e10cSrcweir  */
6*cdf0e10cSrcweir 
7*cdf0e10cSrcweir 
8*cdf0e10cSrcweir /* enlarge token buffer to tokenize whole strings */
9*cdf0e10cSrcweir #undef YYLMAX
10*cdf0e10cSrcweir #define YYLMAX 64000
11*cdf0e10cSrcweir 
12*cdf0e10cSrcweir /* to enable debug output define LEXDEBUG */
13*cdf0e10cSrcweir #define LEXDEBUG		1
14*cdf0e10cSrcweir #ifdef LEXDEBUG
15*cdf0e10cSrcweir #define OUTPUT	fprintf
16*cdf0e10cSrcweir #else
17*cdf0e10cSrcweir #define OUTPUT(Par1,Par2);
18*cdf0e10cSrcweir #endif
19*cdf0e10cSrcweir 
20*cdf0e10cSrcweir /* table of possible token ids */
21*cdf0e10cSrcweir #include "tokens.h"
22*cdf0e10cSrcweir #include <stdlib.h>
23*cdf0e10cSrcweir #include <stdio.h>
24*cdf0e10cSrcweir 
25*cdf0e10cSrcweir #if defined __GNUC__
26*cdf0e10cSrcweir #pragma GCC system_header
27*cdf0e10cSrcweir #elif defined __SINPRO_CC
28*cdf0e10cSrcweir #pragma disable_warn
29*cdf0e10cSrcweir #elif defined _MSC_VER
30*cdf0e10cSrcweir #pragma warning(push, 1)
31*cdf0e10cSrcweir #endif
32*cdf0e10cSrcweir 
33*cdf0e10cSrcweir /* external functions (C++ code, declared as extren "C" */
34*cdf0e10cSrcweir extern int WorkOnTokenSet( int, char* );
35*cdf0e10cSrcweir extern int Argument( char * );
36*cdf0e10cSrcweir extern int InitXrmExport( char * , char * );
37*cdf0e10cSrcweir extern int EndXrmExport();
38*cdf0e10cSrcweir extern int GetError();
39*cdf0e10cSrcweir extern int SetError();
40*cdf0e10cSrcweir extern char *GetOutputFile( int argc, char* argv[]);
41*cdf0e10cSrcweir extern FILE *GetXrmFile();
42*cdf0e10cSrcweir extern int isQuiet();
43*cdf0e10cSrcweir extern void removeTempFile();
44*cdf0e10cSrcweir extern char* getFilename();
45*cdf0e10cSrcweir 
46*cdf0e10cSrcweir /* forwards */
47*cdf0e10cSrcweir void YYWarning();
48*cdf0e10cSrcweir 
49*cdf0e10cSrcweir int bText=0;
50*cdf0e10cSrcweir %}
51*cdf0e10cSrcweir 
52*cdf0e10cSrcweir %p 24000
53*cdf0e10cSrcweir %e 1200
54*cdf0e10cSrcweir %n 500
55*cdf0e10cSrcweir 
56*cdf0e10cSrcweir %%
57*cdf0e10cSrcweir 
58*cdf0e10cSrcweir "<p "[^\>]*xml:lang[^\>]*\> {
59*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_START , yytext );
60*cdf0e10cSrcweir }
61*cdf0e10cSrcweir 
62*cdf0e10cSrcweir "</p>" {
63*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_END, yytext );
64*cdf0e10cSrcweir }
65*cdf0e10cSrcweir 
66*cdf0e10cSrcweir "<h1 "[^\>]*xml:lang[^\>]*\> {
67*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_START , yytext );
68*cdf0e10cSrcweir }
69*cdf0e10cSrcweir 
70*cdf0e10cSrcweir "</h1>" {
71*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_END, yytext );
72*cdf0e10cSrcweir }
73*cdf0e10cSrcweir "<h2 "[^\>]*xml:lang[^\>]*\> {
74*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_START , yytext );
75*cdf0e10cSrcweir }
76*cdf0e10cSrcweir 
77*cdf0e10cSrcweir "</h2>" {
78*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_END, yytext );
79*cdf0e10cSrcweir }
80*cdf0e10cSrcweir "<h3 "[^\>]*xml:lang[^\>]*\> {
81*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_START , yytext );
82*cdf0e10cSrcweir }
83*cdf0e10cSrcweir 
84*cdf0e10cSrcweir "</h3>" {
85*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_END, yytext );
86*cdf0e10cSrcweir }
87*cdf0e10cSrcweir "<h4 "[^\>]*xml:lang[^\>]*\> {
88*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_START , yytext );
89*cdf0e10cSrcweir }
90*cdf0e10cSrcweir 
91*cdf0e10cSrcweir "</h4>" {
92*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_END, yytext );
93*cdf0e10cSrcweir }
94*cdf0e10cSrcweir "<h5 "[^\>]*xml:lang[^\>]*\> {
95*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_START , yytext );
96*cdf0e10cSrcweir }
97*cdf0e10cSrcweir 
98*cdf0e10cSrcweir "</h5>" {
99*cdf0e10cSrcweir 	WorkOnTokenSet( XRM_TEXT_END, yytext );
100*cdf0e10cSrcweir }
101*cdf0e10cSrcweir 
102*cdf0e10cSrcweir 
103*cdf0e10cSrcweir 
104*cdf0e10cSrcweir 
105*cdf0e10cSrcweir 
106*cdf0e10cSrcweir 
107*cdf0e10cSrcweir "<!--"	{
108*cdf0e10cSrcweir 	char c1 = 0, c2 = 0, c3 = input();
109*cdf0e10cSrcweir 	char pChar[2];
110*cdf0e10cSrcweir 	pChar[1] = 0x00;
111*cdf0e10cSrcweir 	pChar[0] = c3;
112*cdf0e10cSrcweir 
113*cdf0e10cSrcweir 	WorkOnTokenSet( COMMEND, yytext );
114*cdf0e10cSrcweir 	WorkOnTokenSet( COMMEND, pChar );
115*cdf0e10cSrcweir 
116*cdf0e10cSrcweir 	for(;;) {
117*cdf0e10cSrcweir 		if ( c3 == EOF )
118*cdf0e10cSrcweir 			break;
119*cdf0e10cSrcweir 		if ( c1 == '-' && c2 == '-' && c3 == '>' )
120*cdf0e10cSrcweir 			break;
121*cdf0e10cSrcweir 		c1 = c2;
122*cdf0e10cSrcweir 		c2 = c3;
123*cdf0e10cSrcweir 		c3 = input();
124*cdf0e10cSrcweir 		pChar[0] = c3;
125*cdf0e10cSrcweir 		WorkOnTokenSet( COMMEND, pChar );
126*cdf0e10cSrcweir 	}
127*cdf0e10cSrcweir }
128*cdf0e10cSrcweir 
129*cdf0e10cSrcweir .|\n {
130*cdf0e10cSrcweir 	if ( bText == 1 )
131*cdf0e10cSrcweir 		WorkOnTokenSet( XML_TEXTCHAR, yytext );
132*cdf0e10cSrcweir 	else
133*cdf0e10cSrcweir 		WorkOnTokenSet( UNKNOWNCHAR, yytext );
134*cdf0e10cSrcweir }
135*cdf0e10cSrcweir 
136*cdf0e10cSrcweir 
137*cdf0e10cSrcweir %%
138*cdf0e10cSrcweir 
139*cdf0e10cSrcweir /*****************************************************************************/
140*cdf0e10cSrcweir int	yywrap(void)
141*cdf0e10cSrcweir /*****************************************************************************/
142*cdf0e10cSrcweir {
143*cdf0e10cSrcweir 	return 1;
144*cdf0e10cSrcweir }
145*cdf0e10cSrcweir 
146*cdf0e10cSrcweir /*****************************************************************************/
147*cdf0e10cSrcweir void YYWarning( char *s )
148*cdf0e10cSrcweir /*****************************************************************************/
149*cdf0e10cSrcweir {
150*cdf0e10cSrcweir 	/* write warning to stderr */
151*cdf0e10cSrcweir 	fprintf( stderr,
152*cdf0e10cSrcweir 		"Warning: \"%s\" in line %d: \"%s\"\n", s, yylineno, yytext  );
153*cdf0e10cSrcweir }
154*cdf0e10cSrcweir 
155*cdf0e10cSrcweir /*****************************************************************************/
156*cdf0e10cSrcweir #ifdef GCC
157*cdf0e10cSrcweir void yyerror ( char *s, ... )
158*cdf0e10cSrcweir #else
159*cdf0e10cSrcweir void yyerror ( char *s )
160*cdf0e10cSrcweir #endif
161*cdf0e10cSrcweir /*****************************************************************************/
162*cdf0e10cSrcweir {
163*cdf0e10cSrcweir 	/* write error to stderr */
164*cdf0e10cSrcweir 	fprintf( stderr,
165*cdf0e10cSrcweir 		"Error: \"%s\" in line %d: \"%s\"\n", s, yylineno, yytext  );
166*cdf0e10cSrcweir 	SetError();
167*cdf0e10cSrcweir }
168*cdf0e10cSrcweir 
169*cdf0e10cSrcweir /*****************************************************************************/
170*cdf0e10cSrcweir int
171*cdf0e10cSrcweir #ifdef WNT
172*cdf0e10cSrcweir _cdecl
173*cdf0e10cSrcweir #endif
174*cdf0e10cSrcweir main( int argc, char* argv[])
175*cdf0e10cSrcweir /*****************************************************************************/
176*cdf0e10cSrcweir {
177*cdf0e10cSrcweir 	/* error level */
178*cdf0e10cSrcweir 	int nRetValue = 0;
179*cdf0e10cSrcweir 	char *pOutput;
180*cdf0e10cSrcweir 	FILE *pFile;
181*cdf0e10cSrcweir 
182*cdf0e10cSrcweir 	pOutput = GetOutputFile( argc, argv );
183*cdf0e10cSrcweir 
184*cdf0e10cSrcweir 	if ( !pOutput ) {
185*cdf0e10cSrcweir 		fprintf( stdout, "Syntax: XRMEX[-p Prj][-r PrjRoot]-i FileIn [-o FileOut][-m DataBase][-e][-b][-u][-NOUTF8][-L l1,l2,...]\n" );
186*cdf0e10cSrcweir 		fprintf( stdout, " Prj:      Project\n" );
187*cdf0e10cSrcweir 		fprintf( stdout, " PrjRoot:  Path to project root (..\\.. etc.)\n" );
188*cdf0e10cSrcweir 		fprintf( stdout, " FileIn:   Source files (*.src)\n" );
189*cdf0e10cSrcweir 		fprintf( stdout, " FileOut:  Destination file (*.*)\n" );
190*cdf0e10cSrcweir 		fprintf( stdout, " DataBase: Mergedata (*.sdf)\n" );
191*cdf0e10cSrcweir 		fprintf( stdout, " -e: Disable writing errorlog\n" );
192*cdf0e10cSrcweir 		fprintf( stdout, " -b: Break when Token \"HelpText\" found in source\n" );
193*cdf0e10cSrcweir 		fprintf( stdout, " -u: [english] and [german] are allowed, Id is Taken from DataBase \n" );
194*cdf0e10cSrcweir 		fprintf( stdout, " -NOUTF8: disable UTF8 as language independent encoding\n" );
195*cdf0e10cSrcweir 		fprintf( stdout, " -L: Restrict the handled languages. l1,l2,... are elements of (de,en-US,es...)\n" );
196*cdf0e10cSrcweir 		fprintf( stdout, "     A fallback language can be defined like this: l1=f1.\n" );
197*cdf0e10cSrcweir 		fprintf( stdout, "     f1, f2,... are also elements of (de,en-US,es...)\n" );
198*cdf0e10cSrcweir 		fprintf( stdout, "     Example: -L en-US,es=de\n" );
199*cdf0e10cSrcweir 		fprintf( stdout, "              Restriction to es and en-US, de will be fallback for 99\n" );
200*cdf0e10cSrcweir //		fprintf( stdout, " -ISO99: IsoCode is the full qualified ISO language code for language 99" );
201*cdf0e10cSrcweir 		return 1;
202*cdf0e10cSrcweir 	}
203*cdf0e10cSrcweir 	pFile = GetXrmFile();
204*cdf0e10cSrcweir 	InitXrmExport( pOutput , getFilename() );
205*cdf0e10cSrcweir 
206*cdf0e10cSrcweir     if ( !pFile )
207*cdf0e10cSrcweir 		return 1;
208*cdf0e10cSrcweir 
209*cdf0e10cSrcweir    	yyin = pFile;
210*cdf0e10cSrcweir 
211*cdf0e10cSrcweir 	/* create global instance of class XmlExport */
212*cdf0e10cSrcweir 	//InitXrmExport( pOutput );
213*cdf0e10cSrcweir 
214*cdf0e10cSrcweir 	/* start parser */
215*cdf0e10cSrcweir    	yylex();
216*cdf0e10cSrcweir 
217*cdf0e10cSrcweir 	/* get error info. and end export */
218*cdf0e10cSrcweir 	nRetValue = GetError();
219*cdf0e10cSrcweir 	EndXrmExport();
220*cdf0e10cSrcweir 
221*cdf0e10cSrcweir     removeTempFile();
222*cdf0e10cSrcweir 	/* return error level */
223*cdf0e10cSrcweir 	return nRetValue;
224*cdf0e10cSrcweir }
225