ICU 77.1  77.1
All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Modules Pages
bytestriebuilder.h
Go to the documentation of this file.
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 *******************************************************************************
5 * Copyright (C) 2010-2016, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 *******************************************************************************
8 * file name: bytestriebuilder.h
9 * encoding: UTF-8
10 * tab size: 8 (not used)
11 * indentation:4
12 *
13 * created on: 2010sep25
14 * created by: Markus W. Scherer
15 */
16 
22 #ifndef __BYTESTRIEBUILDER_H__
23 #define __BYTESTRIEBUILDER_H__
24 
25 #include "unicode/utypes.h"
26 
27 #if U_SHOW_CPLUSPLUS_API
28 
29 #include "unicode/bytestrie.h"
30 #include "unicode/stringpiece.h"
32 
33 class BytesTrieTest;
34 
35 U_NAMESPACE_BEGIN
36 
37 class BytesTrieElement;
38 class CharString;
45 class U_COMMON_API BytesTrieBuilder : public StringTrieBuilder {
46 public:
53 
58  virtual ~BytesTrieBuilder();
59 
74  BytesTrieBuilder &add(StringPiece s, int32_t value, UErrorCode &errorCode);
75 
94  BytesTrie *build(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
95 
120 
128 
129 private:
130  friend class ::BytesTrieTest;
131 
132  BytesTrieBuilder(const BytesTrieBuilder &other) = delete; // no copy constructor
133  BytesTrieBuilder &operator=(const BytesTrieBuilder &other) = delete; // no assignment operator
134 
135  void buildBytes(UStringTrieBuildOption buildOption, UErrorCode &errorCode);
136 
137  virtual int32_t getElementStringLength(int32_t i) const override;
138  virtual char16_t getElementUnit(int32_t i, int32_t byteIndex) const override;
139  virtual int32_t getElementValue(int32_t i) const override;
140 
141  virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t byteIndex) const override;
142 
143  virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t byteIndex) const override;
144  virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t byteIndex, int32_t count) const override;
145  virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t byteIndex, char16_t byte) const override;
146 
147  virtual UBool matchNodesCanHaveValues() const override { return false; }
148 
149  virtual int32_t getMaxBranchLinearSubNodeLength() const override { return BytesTrie::kMaxBranchLinearSubNodeLength; }
150  virtual int32_t getMinLinearMatch() const override { return BytesTrie::kMinLinearMatch; }
151  virtual int32_t getMaxLinearMatchLength() const override { return BytesTrie::kMaxLinearMatchLength; }
152 
156  class BTLinearMatchNode : public LinearMatchNode {
157  public:
158  BTLinearMatchNode(const char *units, int32_t len, Node *nextNode);
159  virtual bool operator==(const Node &other) const override;
160  virtual void write(StringTrieBuilder &builder) override;
161  private:
162  const char *s;
163  };
164 
165  virtual Node *createLinearMatchNode(int32_t i, int32_t byteIndex, int32_t length,
166  Node *nextNode) const override;
167 
168  UBool ensureCapacity(int32_t length);
169  virtual int32_t write(int32_t byte) override;
170  int32_t write(const char *b, int32_t length);
171  virtual int32_t writeElementUnits(int32_t i, int32_t byteIndex, int32_t length) override;
172  virtual int32_t writeValueAndFinal(int32_t i, UBool isFinal) override;
173  virtual int32_t writeValueAndType(UBool hasValue, int32_t value, int32_t node) override;
174  virtual int32_t writeDeltaTo(int32_t jumpTarget) override;
175  static int32_t internalEncodeDelta(int32_t i, char intBytes[]);
176 
177  CharString *strings; // Pointer not object so we need not #include internal charstr.h.
178  BytesTrieElement *elements;
179  int32_t elementsCapacity;
180  int32_t elementsLength;
181 
182  // Byte serialization of the trie.
183  // Grows from the back: bytesLength measures from the end of the buffer!
184  char *bytes;
185  int32_t bytesCapacity;
186  int32_t bytesLength;
187 };
188 
189 U_NAMESPACE_END
190 
191 #endif /* U_SHOW_CPLUSPLUS_API */
192 
193 #endif // __BYTESTRIEBUILDER_H__
C++ API: Trie for mapping byte sequences to integer values.
Builder class for BytesTrie.
StringPiece buildStringPiece(UStringTrieBuildOption buildOption, UErrorCode &errorCode)
Builds a BytesTrie for the add()ed data and byte-serializes it.
BytesTrieBuilder(UErrorCode &errorCode)
Constructs an empty builder.
virtual ~BytesTrieBuilder()
Destructor.
BytesTrieBuilder & clear()
Removes all (byte sequence, value) pairs.
BytesTrieBuilder & add(StringPiece s, int32_t value, UErrorCode &errorCode)
Adds a (byte sequence, value) pair.
BytesTrie * build(UStringTrieBuildOption buildOption, UErrorCode &errorCode)
Builds a BytesTrie for the add()ed data.
Light-weight, non-const reader class for a BytesTrie.
Definition: bytestrie.h:55
A string-like object that points to a sized piece of memory.
Definition: stringpiece.h:61
U_EXPORT UBool operator==(const StringPiece &x, const StringPiece &y)
Global operator == for StringPiece.
C++ API: StringPiece: Read-only byte string wrapper class.
C++ API: Builder API for trie builders.
UStringTrieBuildOption
Build options for BytesTrieBuilder and CharsTrieBuilder.
int8_t UBool
The ICU boolean type, a signed-byte integer.
Definition: umachine.h:247
Basic definitions for ICU, for both C and C++ APIs.
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition: utypes.h:430
#define U_COMMON_API
Set to export library symbols from inside the common library, and to import them from outside.
Definition: utypes.h:315