xref: /aoo42x/main/l10ntools/inc/wtratree.hxx (revision cdf0e10c)
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 
29*cdf0e10cSrcweir #ifndef TX3_WTRATREE_HXX
30*cdf0e10cSrcweir #define TX3_WTRATREE_HXX
31*cdf0e10cSrcweir 
32*cdf0e10cSrcweir // USED
33*cdf0e10cSrcweir 	// Base Classes
34*cdf0e10cSrcweir 	// Components
35*cdf0e10cSrcweir 	// Parameters
36*cdf0e10cSrcweir #include <tools/string.hxx>
37*cdf0e10cSrcweir 
38*cdf0e10cSrcweir const INT16		C_NR_OF_WTT_RESULTS = 5;
39*cdf0e10cSrcweir const INT16    	C_NR_OF_POSSIBLE_CHARS = 256;
40*cdf0e10cSrcweir 
41*cdf0e10cSrcweir 
42*cdf0e10cSrcweir typedef unsigned char u_char;
43*cdf0e10cSrcweir typedef const char * constr;
44*cdf0e10cSrcweir 
45*cdf0e10cSrcweir 
46*cdf0e10cSrcweir class WTT_Node;
47*cdf0e10cSrcweir 
48*cdf0e10cSrcweir 
49*cdf0e10cSrcweir /**	@task
50*cdf0e10cSrcweir 	This class implements the functionality, that class WordTransformer
51*cdf0e10cSrcweir 	offers.
52*cdf0e10cSrcweir 	WordTransformer is dependant of this class, but NOT the other way!
53*cdf0e10cSrcweir **/
54*cdf0e10cSrcweir class WordTransTree
55*cdf0e10cSrcweir {
56*cdf0e10cSrcweir   public:
57*cdf0e10cSrcweir 	enum E_Result
58*cdf0e10cSrcweir 	{
59*cdf0e10cSrcweir 		OK = 0,
60*cdf0e10cSrcweir 		HOTKEY_LOST,
61*cdf0e10cSrcweir 		OUTPUT_OVERFLOW
62*cdf0e10cSrcweir 	};
63*cdf0e10cSrcweir 
64*cdf0e10cSrcweir 
65*cdf0e10cSrcweir 	//	LIFECYCLE
66*cdf0e10cSrcweir 						WordTransTree(
67*cdf0e10cSrcweir 							CharSet				i_nWorkingCharSet = RTL_TEXTENCODING_MS_1252);
68*cdf0e10cSrcweir 	void				SetCharSet(
69*cdf0e10cSrcweir 							CharSet				i_nWorkingCharSet);
70*cdf0e10cSrcweir 						~WordTransTree();
71*cdf0e10cSrcweir 
72*cdf0e10cSrcweir 	void				AddWordPair(
73*cdf0e10cSrcweir 							const ByteString &		i_sOldString,
74*cdf0e10cSrcweir 							const ByteString &		i_sReplaceString );
75*cdf0e10cSrcweir 
76*cdf0e10cSrcweir 	// OPERATIONS
77*cdf0e10cSrcweir 	void				InitTransformation(
78*cdf0e10cSrcweir 							const char *		i_sInput,               /// [!=0], a range of i_nInputLength must be valid memory for read.
79*cdf0e10cSrcweir 							UINT32				i_nInputLength,
80*cdf0e10cSrcweir 							UINT32				i_nOutputMaxLength = STRING_MAXLEN - 12 );
81*cdf0e10cSrcweir 	E_Result			TransformNextToken();
82*cdf0e10cSrcweir 
83*cdf0e10cSrcweir 	// INQUIRY
84*cdf0e10cSrcweir 	sal_Bool				TextEndReached() const;
85*cdf0e10cSrcweir 	const char *		Output() const;
86*cdf0e10cSrcweir 
87*cdf0e10cSrcweir 		// These 3 functions are valid between two calls of
88*cdf0e10cSrcweir 		//   TransformNextToken():
89*cdf0e10cSrcweir 	E_Result			CurResult() const;
90*cdf0e10cSrcweir 	ByteString 			CurReplacedString() const;
91*cdf0e10cSrcweir 	ByteString	 		CurReplacingString() const;
92*cdf0e10cSrcweir 	char 				CurHotkey() const;
93*cdf0e10cSrcweir 
94*cdf0e10cSrcweir   private:
95*cdf0e10cSrcweir 	// SERVICE FUNCTONS
96*cdf0e10cSrcweir 	UINT8	  			CalculateBranch(
97*cdf0e10cSrcweir 							u_char 				i_cInputChar ) const;
98*cdf0e10cSrcweir 
99*cdf0e10cSrcweir 	void				Handle_Hotkey();
100*cdf0e10cSrcweir 	void				Handle_TokenToKeep();
101*cdf0e10cSrcweir 	void				Handle_TokenToTransform();
102*cdf0e10cSrcweir 
103*cdf0e10cSrcweir 	// DATA
104*cdf0e10cSrcweir 		// Fixed data
105*cdf0e10cSrcweir 	const u_char *		sInput;
106*cdf0e10cSrcweir 	UINT32              nInputLength;
107*cdf0e10cSrcweir 	const u_char *		pInputEnd;
108*cdf0e10cSrcweir 
109*cdf0e10cSrcweir 	u_char *			sOutput;				// DYN
110*cdf0e10cSrcweir 	UINT32				nOutputMaxLength;
111*cdf0e10cSrcweir 
112*cdf0e10cSrcweir 	WTT_Node *			dpParsingTreeTop;		// DYN
113*cdf0e10cSrcweir 	WTT_Node *			pUnknownAlpha;
114*cdf0e10cSrcweir 	u_char		 		cChar2Branch[C_NR_OF_POSSIBLE_CHARS];
115*cdf0e10cSrcweir 	u_char		 		c_AE, c_OE, c_UE, c_ae, c_oe, c_ue;
116*cdf0e10cSrcweir 
117*cdf0e10cSrcweir 		// Working data
118*cdf0e10cSrcweir 	const u_char *		pInputCurTokenStart;
119*cdf0e10cSrcweir 	const u_char *		pInputPosition;
120*cdf0e10cSrcweir 	u_char *			pOutputPosition;
121*cdf0e10cSrcweir 	WTT_Node *			pCurParseNode;
122*cdf0e10cSrcweir 
123*cdf0e10cSrcweir 		// Data which are valid only after a completed call to TransformNextToken()
124*cdf0e10cSrcweir 	E_Result			eCurResult;
125*cdf0e10cSrcweir 	u_char				cCurHotkey;            	// Letter wich is used as hotkey
126*cdf0e10cSrcweir 	u_char				cCurHotkeySign;       	// Letter which is used to assign hotkey ('~'or '&') .
127*cdf0e10cSrcweir };
128*cdf0e10cSrcweir 
129*cdf0e10cSrcweir 
130*cdf0e10cSrcweir 
131*cdf0e10cSrcweir 
132*cdf0e10cSrcweir 
133*cdf0e10cSrcweir 
134*cdf0e10cSrcweir 
135*cdf0e10cSrcweir inline sal_Bool
136*cdf0e10cSrcweir WordTransTree::TextEndReached() const
137*cdf0e10cSrcweir 	{ return pInputPosition == pInputEnd; }
138*cdf0e10cSrcweir inline const char *
139*cdf0e10cSrcweir WordTransTree::Output() const
140*cdf0e10cSrcweir 	{ return TextEndReached() ? (constr) sOutput : ""; }
141*cdf0e10cSrcweir inline WordTransTree::E_Result
142*cdf0e10cSrcweir WordTransTree::CurResult() const
143*cdf0e10cSrcweir 	{ return eCurResult; }
144*cdf0e10cSrcweir inline ByteString
145*cdf0e10cSrcweir WordTransTree::CurReplacedString() const
146*cdf0e10cSrcweir 	{ return ByteString((constr) pInputCurTokenStart,pInputPosition-pInputCurTokenStart); }
147*cdf0e10cSrcweir inline char
148*cdf0e10cSrcweir WordTransTree::CurHotkey() const
149*cdf0e10cSrcweir 	{ return cCurHotkey; }
150*cdf0e10cSrcweir inline UINT8
151*cdf0e10cSrcweir WordTransTree::CalculateBranch(u_char i_cInputChar) const
152*cdf0e10cSrcweir 	{ return cChar2Branch[i_cInputChar]; }
153*cdf0e10cSrcweir 
154*cdf0e10cSrcweir 
155*cdf0e10cSrcweir 
156*cdf0e10cSrcweir #endif
157*cdf0e10cSrcweir 
158*cdf0e10cSrcweir 
159*cdf0e10cSrcweir 
160