rbbisetb.cpp | Explore in Territory

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
//
//  rbbisetb.cpp
//
/*
***************************************************************************
*   Copyright (C) 2002-2008 International Business Machines Corporation   *
*   and others. All rights reserved.                                      *
***************************************************************************
*/
//
//  RBBISetBuilder   Handles processing of Unicode Sets from RBBI rules
//                   (part of the rule building process.)
//
//      Starting with the rules parse tree from the scanner,
//
//                   -  Enumerate the set of UnicodeSets that are referenced
//                      by the RBBI rules.
//                   -  compute a set of non-overlapping character ranges
//                      with all characters within a range belonging to the same
//                      set of input unicode sets.
//                   -  Derive a set of non-overlapping UnicodeSet (like things)
//                      that will correspond to columns in the state table for
//                      the RBBI execution engine.  All characters within one
//                      of these sets belong to the same set of the original
//                      UnicodeSets from the user's rules.
//                   -  construct the trie table that maps input characters
//                      to the index of the matching non-overlapping set of set from
//                      the previous step.
//

#include "unicode/utypes.h"

#if !UCONFIG_NO_BREAK_ITERATION

#include "unicode/uniset.h"
#include "uvector.h"
#include "uassert.h"
#include "cmemory.h"
#include "cstring.h"

#include "rbbisetb.h"
#include "rbbinode.h"

U_NAMESPACE_BEGIN

const int32_t kMaxCharCategoriesFor8BitsTrie = …;
//------------------------------------------------------------------------
//
//   Constructor
//
//------------------------------------------------------------------------
RBBISetBuilder::RBBISetBuilder(RBBIRuleBuilder *rb)
{ … }


//------------------------------------------------------------------------
//
//   Destructor
//
//------------------------------------------------------------------------
RBBISetBuilder::~RBBISetBuilder()
{ … }




//------------------------------------------------------------------------
//
//   build          Build the list of non-overlapping character ranges
//                  from the Unicode Sets.
//
//------------------------------------------------------------------------
void RBBISetBuilder::buildRanges() { … }


//
// Build the Trie table for mapping UChar32 values to the corresponding
// range group number.
//
void RBBISetBuilder::buildTrie() { … }


void RBBISetBuilder::mergeCategories(IntPair categories) { … }


//-----------------------------------------------------------------------------------
//
//  getTrieSize()    Return the size that will be required to serialize the Trie.
//
//-----------------------------------------------------------------------------------
int32_t RBBISetBuilder::getTrieSize()  { … }


//-----------------------------------------------------------------------------------
//
//  serializeTrie()   Put the serialized trie at the specified address.
//                    Trust the caller to have given us enough memory.
//                    getTrieSize() MUST be called first.
//
//-----------------------------------------------------------------------------------
void RBBISetBuilder::serializeTrie(uint8_t *where) { … }

//------------------------------------------------------------------------
//
//  addValToSets     Add a runtime-mapped input value to each uset from a
//                   list of uset nodes. (val corresponds to a state table column.)
//                   For each of the original Unicode sets - which correspond
//                   directly to uset nodes - a logically equivalent expression
//                   is constructed in terms of the remapped runtime input
//                   symbol set.  This function adds one runtime input symbol to
//                   a list of sets.
//
//                   The "logically equivalent expression" is the tree for an
//                   or-ing together of all of the symbols that go into the set.
//
//------------------------------------------------------------------------
void  RBBISetBuilder::addValToSets(UVector *sets, uint32_t val) { … }

void  RBBISetBuilder::addValToSet(RBBINode *usetNode, uint32_t val) { … }


//------------------------------------------------------------------------
//
//   getNumCharCategories
//
//------------------------------------------------------------------------
int32_t  RBBISetBuilder::getNumCharCategories() const { … }


//------------------------------------------------------------------------
//
//   getDictCategoriesStart
//
//------------------------------------------------------------------------
int32_t  RBBISetBuilder::getDictCategoriesStart() const { … }


//------------------------------------------------------------------------
//
//   sawBOF
//
//------------------------------------------------------------------------
UBool  RBBISetBuilder::sawBOF() const { … }


//------------------------------------------------------------------------
//
//   getFirstChar      Given a runtime RBBI character category, find
//                     the first UChar32 that is in the set of chars 
//                     in the category.
//------------------------------------------------------------------------
UChar32  RBBISetBuilder::getFirstChar(int32_t category) const { … }


//------------------------------------------------------------------------
//
//   printRanges        A debugging function.
//                      dump out all of the range definitions.
//
//------------------------------------------------------------------------
#ifdef RBBI_DEBUG
void RBBISetBuilder::printRanges() {
    RangeDescriptor       *rlRange;
    int                    i;

    RBBIDebugPrintf("\n\n Nonoverlapping Ranges ...\n");
    for (rlRange = fRangeList; rlRange!=nullptr; rlRange=rlRange->fNext) {
        RBBIDebugPrintf("%4x-%4x  ", rlRange->fStartChar, rlRange->fEndChar);

        for (i=0; i<rlRange->fIncludesSets->size(); i++) {
            RBBINode       *usetNode    = (RBBINode *)rlRange->fIncludesSets->elementAt(i);
            UnicodeString   setName {u"anon"};
            RBBINode       *setRef = usetNode->fParent;
            if (setRef != nullptr) {
                RBBINode *varRef = setRef->fParent;
                if (varRef != nullptr  &&  varRef->fType == RBBINode::varRef) {
                    setName = varRef->fText;
                }
            }
            RBBI_DEBUG_printUnicodeString(setName); RBBIDebugPrintf("  ");
        }
        RBBIDebugPrintf("\n");
    }
}
#endif


//------------------------------------------------------------------------
//
//   printRangeGroups     A debugging function.
//                        dump out all of the range groups.
//
//------------------------------------------------------------------------
#ifdef RBBI_DEBUG
void RBBISetBuilder::printRangeGroups() {
    int                    i;

    RBBIDebugPrintf("\nRanges grouped by Unicode Set Membership...\n");
    for (RangeDescriptor *rlRange = fRangeList; rlRange!=nullptr; rlRange=rlRange->fNext) {
        if (rlRange->fFirstInGroup) {
            int groupNum = rlRange->fNum;
            RBBIDebugPrintf("%2i  ", groupNum);

            if (groupNum >= fDictCategoriesStart) { RBBIDebugPrintf(" <DICT> ");}

            for (i=0; i<rlRange->fIncludesSets->size(); i++) {
                RBBINode       *usetNode    = (RBBINode *)rlRange->fIncludesSets->elementAt(i);
                UnicodeString   setName = UNICODE_STRING("anon", 4);
                RBBINode       *setRef = usetNode->fParent;
                if (setRef != nullptr) {
                    RBBINode *varRef = setRef->fParent;
                    if (varRef != nullptr  &&  varRef->fType == RBBINode::varRef) {
                        setName = varRef->fText;
                    }
                }
                RBBI_DEBUG_printUnicodeString(setName); RBBIDebugPrintf(" ");
            }

            i = 0;
            for (RangeDescriptor *tRange = rlRange; tRange != nullptr; tRange = tRange->fNext) {
                if (tRange->fNum == rlRange->fNum) {
                    if (i++ % 5 == 0) {
                        RBBIDebugPrintf("\n    ");
                    }
                    RBBIDebugPrintf("  %05x-%05x", tRange->fStartChar, tRange->fEndChar);
                }
            }
            RBBIDebugPrintf("\n");
        }
    }
    RBBIDebugPrintf("\n");
}
#endif


//------------------------------------------------------------------------
//
//   printSets          A debugging function.
//                      dump out all of the set definitions.
//
//------------------------------------------------------------------------
#ifdef RBBI_DEBUG
void RBBISetBuilder::printSets() {
    int                   i;

    RBBIDebugPrintf("\n\nUnicode Sets List\n------------------\n");
    for (i=0; ; i++) {
        RBBINode        *usetNode;
        RBBINode        *setRef;
        RBBINode        *varRef;
        UnicodeString    setName;

        usetNode = (RBBINode *)fRB->fUSetNodes->elementAt(i);
        if (usetNode == nullptr) {
            break;
        }

        RBBIDebugPrintf("%3d    ", i);
        setName = UNICODE_STRING("anonymous", 9);
        setRef = usetNode->fParent;
        if (setRef != nullptr) {
            varRef = setRef->fParent;
            if (varRef != nullptr  &&  varRef->fType == RBBINode::varRef) {
                setName = varRef->fText;
            }
        }
        RBBI_DEBUG_printUnicodeString(setName);
        RBBIDebugPrintf("   ");
        RBBI_DEBUG_printUnicodeString(usetNode->fText);
        RBBIDebugPrintf("\n");
        if (usetNode->fLeftChild != nullptr) {
            RBBINode::printTree(usetNode->fLeftChild, true);
        }
    }
    RBBIDebugPrintf("\n");
}
#endif



//-------------------------------------------------------------------------------------
//
//  RangeDescriptor copy constructor
//
//-------------------------------------------------------------------------------------

RangeDescriptor::RangeDescriptor(const RangeDescriptor &other, UErrorCode &status) : … { … }


//-------------------------------------------------------------------------------------
//
//  RangeDesriptor default constructor
//
//-------------------------------------------------------------------------------------
RangeDescriptor::RangeDescriptor(UErrorCode &status) { … }


//-------------------------------------------------------------------------------------
//
//  RangeDesriptor Destructor
//
//-------------------------------------------------------------------------------------
RangeDescriptor::~RangeDescriptor() { … }

//-------------------------------------------------------------------------------------
//
//  RangeDesriptor::split()
//
//-------------------------------------------------------------------------------------
void RangeDescriptor::split(UChar32 where, UErrorCode &status) { … }


//-------------------------------------------------------------------------------------
//
//   RangeDescriptor::isDictionaryRange
//
//            Test whether this range includes characters from
//            the original Unicode Set named "dictionary".
//
//            This function looks through the Unicode Sets that
//            the range includes, checking for one named "dictionary"
//
//            TODO:  a faster way would be to find the set node for
//                   "dictionary" just once, rather than looking it
//                   up by name every time.
//
//-------------------------------------------------------------------------------------
bool RangeDescriptor::isDictionaryRange() { … }

U_NAMESPACE_END

#endif /* #if !UCONFIG_NO_BREAK_ITERATION */
chromium/third_party/icu/source/common/rbbisetb.cpp
