xref: /aosp_15_r20/external/icu/libicu/cts_headers/csmatch.h (revision 0e209d3975ff4a8c132096b14b0e9364a753506e)
1*0e209d39SAndroid Build Coastguard Worker // © 2016 and later: Unicode, Inc. and others.
2*0e209d39SAndroid Build Coastguard Worker // License & terms of use: http://www.unicode.org/copyright.html
3*0e209d39SAndroid Build Coastguard Worker /*
4*0e209d39SAndroid Build Coastguard Worker  **********************************************************************
5*0e209d39SAndroid Build Coastguard Worker  *   Copyright (C) 2005-2012, International Business Machines
6*0e209d39SAndroid Build Coastguard Worker  *   Corporation and others.  All Rights Reserved.
7*0e209d39SAndroid Build Coastguard Worker  **********************************************************************
8*0e209d39SAndroid Build Coastguard Worker  */
9*0e209d39SAndroid Build Coastguard Worker 
10*0e209d39SAndroid Build Coastguard Worker #ifndef __CSMATCH_H
11*0e209d39SAndroid Build Coastguard Worker #define __CSMATCH_H
12*0e209d39SAndroid Build Coastguard Worker 
13*0e209d39SAndroid Build Coastguard Worker #include "unicode/uobject.h"
14*0e209d39SAndroid Build Coastguard Worker 
15*0e209d39SAndroid Build Coastguard Worker #if !UCONFIG_NO_CONVERSION
16*0e209d39SAndroid Build Coastguard Worker 
17*0e209d39SAndroid Build Coastguard Worker U_NAMESPACE_BEGIN
18*0e209d39SAndroid Build Coastguard Worker 
19*0e209d39SAndroid Build Coastguard Worker class InputText;
20*0e209d39SAndroid Build Coastguard Worker class CharsetRecognizer;
21*0e209d39SAndroid Build Coastguard Worker 
22*0e209d39SAndroid Build Coastguard Worker /*
23*0e209d39SAndroid Build Coastguard Worker  * CharsetMatch represents the results produced by one Charset Recognizer for one input text
24*0e209d39SAndroid Build Coastguard Worker  *              Any confidence > 0 indicates a possible match, meaning that the input bytes
25*0e209d39SAndroid Build Coastguard Worker  *              are at least legal.
26*0e209d39SAndroid Build Coastguard Worker  *
27*0e209d39SAndroid Build Coastguard Worker  *              The full results of a detect are represented by an array of these
28*0e209d39SAndroid Build Coastguard Worker  *              CharsetMatch objects, each representing a possible matching charset.
29*0e209d39SAndroid Build Coastguard Worker  *
30*0e209d39SAndroid Build Coastguard Worker  *              Note that a single charset recognizer may detect multiple closely related
31*0e209d39SAndroid Build Coastguard Worker  *              charsets, and set different names depending on the exact input bytes seen.
32*0e209d39SAndroid Build Coastguard Worker  */
33*0e209d39SAndroid Build Coastguard Worker class CharsetMatch : public UMemory
34*0e209d39SAndroid Build Coastguard Worker {
35*0e209d39SAndroid Build Coastguard Worker  private:
36*0e209d39SAndroid Build Coastguard Worker     InputText               *textIn;
37*0e209d39SAndroid Build Coastguard Worker     int32_t                  confidence;
38*0e209d39SAndroid Build Coastguard Worker     const char              *fCharsetName;
39*0e209d39SAndroid Build Coastguard Worker     const char              *fLang;
40*0e209d39SAndroid Build Coastguard Worker 
41*0e209d39SAndroid Build Coastguard Worker  public:
42*0e209d39SAndroid Build Coastguard Worker     CharsetMatch();
43*0e209d39SAndroid Build Coastguard Worker 
44*0e209d39SAndroid Build Coastguard Worker     /**
45*0e209d39SAndroid Build Coastguard Worker       * fully set the state of this CharsetMatch.
46*0e209d39SAndroid Build Coastguard Worker       * Called by the CharsetRecognizers to record match results.
47*0e209d39SAndroid Build Coastguard Worker       * Default (nullptr) parameters for names will be filled by calling the
48*0e209d39SAndroid Build Coastguard Worker       *   corresponding getters on the recognizer.
49*0e209d39SAndroid Build Coastguard Worker       */
50*0e209d39SAndroid Build Coastguard Worker     void set(InputText               *input,
51*0e209d39SAndroid Build Coastguard Worker              const CharsetRecognizer *cr,
52*0e209d39SAndroid Build Coastguard Worker              int32_t                  conf,
53*0e209d39SAndroid Build Coastguard Worker              const char              *csName=nullptr,
54*0e209d39SAndroid Build Coastguard Worker              const char              *lang=nullptr);
55*0e209d39SAndroid Build Coastguard Worker 
56*0e209d39SAndroid Build Coastguard Worker     /**
57*0e209d39SAndroid Build Coastguard Worker       * Return the name of the charset for this Match
58*0e209d39SAndroid Build Coastguard Worker       */
59*0e209d39SAndroid Build Coastguard Worker     const char *getName() const;
60*0e209d39SAndroid Build Coastguard Worker 
61*0e209d39SAndroid Build Coastguard Worker     const char *getLanguage()const;
62*0e209d39SAndroid Build Coastguard Worker 
63*0e209d39SAndroid Build Coastguard Worker     int32_t getConfidence()const;
64*0e209d39SAndroid Build Coastguard Worker 
65*0e209d39SAndroid Build Coastguard Worker     int32_t getUChars(char16_t *buf, int32_t cap, UErrorCode *status) const;
66*0e209d39SAndroid Build Coastguard Worker };
67*0e209d39SAndroid Build Coastguard Worker 
68*0e209d39SAndroid Build Coastguard Worker U_NAMESPACE_END
69*0e209d39SAndroid Build Coastguard Worker 
70*0e209d39SAndroid Build Coastguard Worker #endif
71*0e209d39SAndroid Build Coastguard Worker #endif /* __CSMATCH_H */
72