xref: /AOO41X/main/i18npool/source/indexentry/genindex_data.cxx (revision cdf0e10c4e3984b49a9502b011690b615761d4a3)
1*cdf0e10cSrcweir /*************************************************************************
2*cdf0e10cSrcweir  *
3*cdf0e10cSrcweir  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4*cdf0e10cSrcweir  *
5*cdf0e10cSrcweir  * Copyright 2000, 2010 Oracle and/or its affiliates.
6*cdf0e10cSrcweir  *
7*cdf0e10cSrcweir  * OpenOffice.org - a multi-platform office productivity suite
8*cdf0e10cSrcweir  *
9*cdf0e10cSrcweir  * This file is part of OpenOffice.org.
10*cdf0e10cSrcweir  *
11*cdf0e10cSrcweir  * OpenOffice.org is free software: you can redistribute it and/or modify
12*cdf0e10cSrcweir  * it under the terms of the GNU Lesser General Public License version 3
13*cdf0e10cSrcweir  * only, as published by the Free Software Foundation.
14*cdf0e10cSrcweir  *
15*cdf0e10cSrcweir  * OpenOffice.org is distributed in the hope that it will be useful,
16*cdf0e10cSrcweir  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17*cdf0e10cSrcweir  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18*cdf0e10cSrcweir  * GNU Lesser General Public License version 3 for more details
19*cdf0e10cSrcweir  * (a copy is included in the LICENSE file that accompanied this code).
20*cdf0e10cSrcweir  *
21*cdf0e10cSrcweir  * You should have received a copy of the GNU Lesser General Public License
22*cdf0e10cSrcweir  * version 3 along with OpenOffice.org.  If not, see
23*cdf0e10cSrcweir  * <http://www.openoffice.org/license.html>
24*cdf0e10cSrcweir  * for a copy of the LGPLv3 License.
25*cdf0e10cSrcweir  *
26*cdf0e10cSrcweir  ************************************************************************/
27*cdf0e10cSrcweir 
28*cdf0e10cSrcweir // MARKER(update_precomp.py): autogen include statement, do not remove
29*cdf0e10cSrcweir #include "precompiled_i18npool.hxx"
30*cdf0e10cSrcweir 
31*cdf0e10cSrcweir #include <stdio.h>
32*cdf0e10cSrcweir #include <string.h>
33*cdf0e10cSrcweir #include <stdlib.h>
34*cdf0e10cSrcweir #include <sal/main.h>
35*cdf0e10cSrcweir #include <sal/types.h>
36*cdf0e10cSrcweir #include <rtl/ustring.hxx>
37*cdf0e10cSrcweir 
38*cdf0e10cSrcweir #define MAX_ADDRESS 0x30000
39*cdf0e10cSrcweir #define MAX_INDEX MAX_ADDRESS/0x100
40*cdf0e10cSrcweir 
41*cdf0e10cSrcweir using namespace ::rtl;
42*cdf0e10cSrcweir 
43*cdf0e10cSrcweir /* Main Procedure */
44*cdf0e10cSrcweir 
45*cdf0e10cSrcweir SAL_IMPLEMENT_MAIN_WITH_ARGS(argc, argv)
46*cdf0e10cSrcweir {
47*cdf0e10cSrcweir 	FILE *fp;
48*cdf0e10cSrcweir 
49*cdf0e10cSrcweir 	if (argc < 4) exit(-1);
50*cdf0e10cSrcweir 
51*cdf0e10cSrcweir 	fp = fopen(argv[1], "rb");	// open the source file for read;
52*cdf0e10cSrcweir 	if (fp == NULL) {
53*cdf0e10cSrcweir 	    printf("Open the rule source file failed.");
54*cdf0e10cSrcweir         return 1;
55*cdf0e10cSrcweir     }
56*cdf0e10cSrcweir 
57*cdf0e10cSrcweir 
58*cdf0e10cSrcweir     sal_Int32 i, j, k;
59*cdf0e10cSrcweir     sal_Int32 address[MAX_ADDRESS];
60*cdf0e10cSrcweir     for (i=0; i<MAX_ADDRESS; i++) address[i]=-1;
61*cdf0e10cSrcweir     OUString sep=OUString(sal_Unicode('|'));
62*cdf0e10cSrcweir     OUString result=sep;
63*cdf0e10cSrcweir     sal_Int32 max=0;
64*cdf0e10cSrcweir 
65*cdf0e10cSrcweir 	sal_Char str[1024];
66*cdf0e10cSrcweir 	while (fgets(str, 1024, fp)) {
67*cdf0e10cSrcweir 	    // don't convert last new line character to Ostr.
68*cdf0e10cSrcweir         sal_Int32 len = strlen(str) - 1;
69*cdf0e10cSrcweir         // skip comment line
70*cdf0e10cSrcweir         if (len == 0 || str[0] == '#')
71*cdf0e10cSrcweir             continue;
72*cdf0e10cSrcweir 
73*cdf0e10cSrcweir 	    // input file is in UTF-8 encoding
74*cdf0e10cSrcweir 	    OUString Ostr = OUString((const sal_Char *)str, len, RTL_TEXTENCODING_UTF8);
75*cdf0e10cSrcweir         len = Ostr.getLength();
76*cdf0e10cSrcweir         if (len == 0)
77*cdf0e10cSrcweir             continue; // skip empty line.
78*cdf0e10cSrcweir 
79*cdf0e10cSrcweir         sal_Int32 nPos=0;
80*cdf0e10cSrcweir         sal_uInt32 nChar = Ostr.iterateCodePoints(&nPos, 2);
81*cdf0e10cSrcweir         if (nChar > MAX_ADDRESS) {
82*cdf0e10cSrcweir             printf("Code point 0x%lx exceeds MAX_ADDRESS 0x%x, Please increase MAX_ADDRESS", static_cast<long unsigned int>(nChar), MAX_ADDRESS);
83*cdf0e10cSrcweir             exit(1);
84*cdf0e10cSrcweir         }
85*cdf0e10cSrcweir         OUString key=Ostr.copy(nPos)+sep;
86*cdf0e10cSrcweir         sal_Int32 idx = result.indexOf(key);
87*cdf0e10cSrcweir         if (key.getLength() > max) max=key.getLength();
88*cdf0e10cSrcweir         if (idx >= 0) {
89*cdf0e10cSrcweir             address[nChar]=idx;
90*cdf0e10cSrcweir         } else {
91*cdf0e10cSrcweir             address[nChar]=result.getLength();
92*cdf0e10cSrcweir             result+=key;
93*cdf0e10cSrcweir         }
94*cdf0e10cSrcweir 	}
95*cdf0e10cSrcweir 	fclose(fp);
96*cdf0e10cSrcweir 
97*cdf0e10cSrcweir 	fp = fopen(argv[2], "wb");
98*cdf0e10cSrcweir 	if (fp == NULL) {
99*cdf0e10cSrcweir 	    printf("Can't create the C source file.");
100*cdf0e10cSrcweir         return 1;
101*cdf0e10cSrcweir 	}
102*cdf0e10cSrcweir 
103*cdf0e10cSrcweir 	fprintf(fp, "/*\n");
104*cdf0e10cSrcweir 	fprintf(fp, " * Copyright(c) 1999 - 2006, Sun Microsystems, Inc.\n");
105*cdf0e10cSrcweir 	fprintf(fp, " * All Rights Reserved.\n");
106*cdf0e10cSrcweir 	fprintf(fp, " */\n\n");
107*cdf0e10cSrcweir 	fprintf(fp, "/* !!!The file is generated automatically. DONOT edit the file manually!!! */\n\n");
108*cdf0e10cSrcweir     fprintf(fp, "#include <sal/types.h>\n");
109*cdf0e10cSrcweir     fprintf(fp, "\nextern \"C\" {\n");
110*cdf0e10cSrcweir 
111*cdf0e10cSrcweir     sal_Int32 index[MAX_INDEX];
112*cdf0e10cSrcweir     sal_Int32 max_index=0;
113*cdf0e10cSrcweir     for (i=k=0; i<MAX_INDEX; i++) {
114*cdf0e10cSrcweir         index[i] = 0xFFFF;
115*cdf0e10cSrcweir         for (j=0; j<0x100; j++) {
116*cdf0e10cSrcweir             if (address[i*0x100+j] >=0) {
117*cdf0e10cSrcweir                 max_index=i;
118*cdf0e10cSrcweir                 index[i]=0x100*k++;
119*cdf0e10cSrcweir                 break;
120*cdf0e10cSrcweir             }
121*cdf0e10cSrcweir         }
122*cdf0e10cSrcweir     }
123*cdf0e10cSrcweir 
124*cdf0e10cSrcweir 	fprintf(fp, "\nstatic const sal_uInt16 idx1[] = {");
125*cdf0e10cSrcweir     for (i = k = 0; i <= max_index;  i++) {
126*cdf0e10cSrcweir         if (k++ % 16 == 0) fprintf(fp, "\n\t");
127*cdf0e10cSrcweir         fprintf(
128*cdf0e10cSrcweir             fp, "0x%04lx, ", sal::static_int_cast< unsigned long >(index[i]));
129*cdf0e10cSrcweir     }
130*cdf0e10cSrcweir 	fprintf(fp, "\n};\n\n");
131*cdf0e10cSrcweir 
132*cdf0e10cSrcweir     sal_Int32 len=result.getLength();
133*cdf0e10cSrcweir     const sal_Unicode *ustr=result.getStr();
134*cdf0e10cSrcweir 	fprintf(fp, "\nstatic const sal_uInt16 idx2[] = {");
135*cdf0e10cSrcweir     for (i = k = 0; i <= max_index; i++) {
136*cdf0e10cSrcweir         if (index[i] != 0xFFFF) {
137*cdf0e10cSrcweir             for (j = 0; j<0x100; j++) {
138*cdf0e10cSrcweir                 if (k++ % 16 == 0) fprintf(fp, "\n\t");
139*cdf0e10cSrcweir                 sal_Int32 ad=address[i*0x100+j];
140*cdf0e10cSrcweir                 fprintf(
141*cdf0e10cSrcweir                     fp, "0x%04lx, ",
142*cdf0e10cSrcweir                     sal::static_int_cast< unsigned long >(
143*cdf0e10cSrcweir                         ad == -1 ? 0 : max == 2 ? ustr[ad] : ad));
144*cdf0e10cSrcweir             }
145*cdf0e10cSrcweir             fprintf(fp, "\n\t");
146*cdf0e10cSrcweir         }
147*cdf0e10cSrcweir     }
148*cdf0e10cSrcweir 	fprintf(fp, "\n};\n\n");
149*cdf0e10cSrcweir 
150*cdf0e10cSrcweir     if (max == 2) {
151*cdf0e10cSrcweir         fprintf(fp, "\nstatic const sal_uInt16 *idx3 = NULL;\n\n");
152*cdf0e10cSrcweir     } else {
153*cdf0e10cSrcweir         fprintf(fp, "\nstatic const sal_uInt16 idx3[] = {");
154*cdf0e10cSrcweir         for (i = k = 0; i < len;  i++) {
155*cdf0e10cSrcweir             if (k++ % 16 == 0) fprintf(fp, "\n\t");
156*cdf0e10cSrcweir             fprintf(fp, "0x%04x, ", (sep.toChar() == ustr[i]) ? 0 : ustr[i]);
157*cdf0e10cSrcweir         }
158*cdf0e10cSrcweir         fprintf(fp, "\n};\n\n");
159*cdf0e10cSrcweir     }
160*cdf0e10cSrcweir 
161*cdf0e10cSrcweir     fprintf(fp, "const sal_uInt16** get_%s(sal_Int16 &max_index)\n{\n\tstatic const sal_uInt16 *idx[]={idx1, idx2, idx3};\n\tmax_index=0x%x;\n\treturn idx;\n}\n\n", argv[3], static_cast<unsigned int>(max_index));
162*cdf0e10cSrcweir     fprintf (fp, "}\n");
163*cdf0e10cSrcweir 
164*cdf0e10cSrcweir 	fclose(fp);
165*cdf0e10cSrcweir     return 0;
166*cdf0e10cSrcweir }	// End of main
167