15 #ifndef __STRINGTRIEBUILDER_H__
16 #define __STRINGTRIEBUILDER_H__
28 typedef struct UHashtable UHashtable;
63 #ifndef U_HIDE_INTERNAL_API
65 static UBool hashNode(
const void *node);
67 static UBool equalNodes(
const void *left,
const void *right);
78 #ifndef U_HIDE_INTERNAL_API
80 void createCompactBuilder(int32_t sizeGuess,
UErrorCode &errorCode);
82 void deleteCompactBuilder();
88 int32_t writeNode(int32_t start, int32_t limit, int32_t unitIndex);
90 int32_t writeBranchSubNode(int32_t start, int32_t limit, int32_t unitIndex, int32_t length);
95 #ifndef U_HIDE_INTERNAL_API
97 Node *makeNode(int32_t start, int32_t limit, int32_t unitIndex,
UErrorCode &errorCode);
99 Node *makeBranchSubNode(int32_t start, int32_t limit, int32_t unitIndex,
104 virtual int32_t getElementStringLength(int32_t i)
const = 0;
106 virtual UChar getElementUnit(int32_t i, int32_t unitIndex)
const = 0;
108 virtual int32_t getElementValue(int32_t i)
const = 0;
113 virtual int32_t getLimitOfLinearMatch(int32_t first, int32_t last, int32_t unitIndex)
const = 0;
117 virtual int32_t countElementUnits(int32_t start, int32_t limit, int32_t unitIndex)
const = 0;
119 virtual int32_t skipElementsBySomeUnits(int32_t i, int32_t unitIndex, int32_t count)
const = 0;
121 virtual int32_t indexOfElementWithNextUnit(int32_t i, int32_t unitIndex,
UChar unit)
const = 0;
124 virtual UBool matchNodesCanHaveValues()
const = 0;
127 virtual int32_t getMaxBranchLinearSubNodeLength()
const = 0;
129 virtual int32_t getMinLinearMatch()
const = 0;
131 virtual int32_t getMaxLinearMatchLength()
const = 0;
133 #ifndef U_HIDE_INTERNAL_API
136 static const int32_t kMaxBranchLinearSubNodeLength=5;
141 static const int32_t kMaxSplitBranchLevels=14;
189 Node(int32_t initialHash) : hash(initialHash), offset(0) {}
190 inline int32_t hashCode()
const {
return hash; }
192 static inline int32_t hashCode(
const Node *node) {
return node==
NULL ? 0 : node->hashCode(); }
223 virtual int32_t markRightEdgesFirst(int32_t edgeNumber);
227 inline void writeUnlessInsideRightEdge(int32_t firstRight, int32_t lastRight,
234 if(offset<0 && (offset<lastRight || firstRight<offset)) {
238 inline int32_t getOffset()
const {
return offset; }
267 void setValue(int32_t v) {
283 :
ValueNode(0x222222*37+hashCode(nextNode)), next(nextNode) { setValue(v); }
285 virtual int32_t markRightEdgesFirst(int32_t edgeNumber);
297 :
ValueNode((0x333333*37+len)*37+hashCode(nextNode)),
298 length(len), next(nextNode) {}
300 virtual int32_t markRightEdgesFirst(int32_t edgeNumber);
313 int32_t firstEdgeNumber;
323 virtual int32_t markRightEdgesFirst(int32_t edgeNumber);
326 void add(int32_t c, int32_t value) {
327 units[length]=(
UChar)c;
329 values[length]=value;
331 hash=(hash*37+c)*37+value;
334 void add(int32_t c,
Node *node) {
335 units[length]=(
UChar)c;
339 hash=(hash*37+c)*37+hashCode(node);
342 Node *equal[kMaxBranchLinearSubNodeLength];
344 int32_t values[kMaxBranchLinearSubNodeLength];
345 UChar units[kMaxBranchLinearSubNodeLength];
355 hashCode(lessThanNode))*37+hashCode(greaterOrEqualNode)),
356 unit(middleUnit), lessThan(lessThanNode), greaterOrEqual(greaterOrEqualNode) {}
358 virtual int32_t markRightEdgesFirst(int32_t edgeNumber);
363 Node *greaterOrEqual;
371 :
ValueNode((0x666666*37+len)*37+hashCode(subNode)),
372 length(len), next(subNode) {}
374 virtual int32_t markRightEdgesFirst(int32_t edgeNumber);
383 virtual Node *createLinearMatchNode(int32_t i, int32_t unitIndex, int32_t length,
384 Node *nextNode)
const = 0;
387 virtual int32_t write(int32_t unit) = 0;
389 virtual int32_t writeElementUnits(int32_t i, int32_t unitIndex, int32_t length) = 0;
391 virtual int32_t writeValueAndFinal(int32_t i,
UBool isFinal) = 0;
393 virtual int32_t writeValueAndType(
UBool hasValue, int32_t value, int32_t node) = 0;
395 virtual int32_t writeDeltaTo(int32_t jumpTarget) = 0;
400 #endif // __STRINGTRIEBUILDER_H__
Builds a trie more slowly, attempting to generate a shorter but equivalent serialization.
Base class for string trie builder classes.
U_EXPORT UBool operator==(const StringPiece &x, const StringPiece &y)
Global operator == for StringPiece.
#define U_NAMESPACE_BEGIN
This is used to begin a declaration of a public ICU C++ API.
UBool operator!=(const StringPiece &x, const StringPiece &y)
Global operator != for StringPiece.
#define NULL
Define NULL if necessary, to 0 for C++ and to ((void *)0) for C.
#define TRUE
The TRUE value of a UBool.
C++ API: Common ICU base class UObject.
uint16_t UChar
Define UChar to be UCHAR_TYPE, if that is #defined (for example, to char16_t), or wchar_t if that is ...
UStringTrieBuildOption
Build options for BytesTrieBuilder and CharsTrieBuilder.
#define U_NAMESPACE_END
This is used to end a declaration of a public ICU C++ API.
UErrorCode
Error code to replace exception handling, so that the code is compatible with all C++ compilers...
Basic definitions for ICU, for both C and C++ APIs.
#define FALSE
The FALSE value of a UBool.
#define U_COMMON_API
Set to export library symbols from inside the common library, and to import them from outside...
UObject is the common ICU "boilerplate" class.
int8_t UBool
The ICU boolean type.