chromium/third_party/icu/source/common/rbbistbl.cpp

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
//
//  file:  rbbistbl.cpp    Implementation of the ICU RBBISymbolTable class
//
/*
***************************************************************************
*   Copyright (C) 2002-2014 International Business Machines Corporation
*   and others. All rights reserved.
***************************************************************************
*/

#include "unicode/utypes.h"

#if !UCONFIG_NO_BREAK_ITERATION

#include "unicode/unistr.h"
#include "unicode/uniset.h"
#include "unicode/uchar.h"
#include "unicode/parsepos.h"

#include "cstr.h"
#include "rbbinode.h"
#include "rbbirb.h"
#include "umutex.h"


//
//  RBBISymbolTableEntry_deleter    Used by the UHashTable to delete the contents
//                                  when the hash table is deleted.
//
U_CDECL_BEGIN
static void U_CALLCONV RBBISymbolTableEntry_deleter(void *p) { … }
U_CDECL_END



U_NAMESPACE_BEGIN

RBBISymbolTable::RBBISymbolTable(RBBIRuleScanner *rs, const UnicodeString &rules, UErrorCode &status)
    : … { … }



RBBISymbolTable::~RBBISymbolTable()
{ … }


//
//  RBBISymbolTable::lookup       This function from the abstract symbol table interface
//                                looks up a variable name and returns a UnicodeString
//                                containing the substitution text.
//
//                                The variable name does NOT include the leading $.
//
const UnicodeString  *RBBISymbolTable::lookup(const UnicodeString& s) const
{ … }



//
//  RBBISymbolTable::lookupMatcher   This function from the abstract symbol table
//                                   interface maps a single stand-in character to a
//                                   pointer to a Unicode Set.   The Unicode Set code uses this
//                                   mechanism to get all references to the same $variable
//                                   name to refer to a single common Unicode Set instance.
//
//    This implementation cheats a little, and does not maintain a map of stand-in chars
//    to sets.  Instead, it takes advantage of the fact that  the UnicodeSet
//    constructor will always call this function right after calling lookup(),
//    and we just need to remember what set to return between these two calls.
const UnicodeFunctor *RBBISymbolTable::lookupMatcher(UChar32 ch) const
{ … }

//
// RBBISymbolTable::parseReference   This function from the abstract symbol table interface
//                                   looks for a $variable name in the source text.
//                                   It does not look it up, only scans for it.
//                                   It is used by the UnicodeSet parser.
//
//                                   This implementation is lifted pretty much verbatim
//                                   from the rules based transliterator implementation.
//                                   I didn't see an obvious way of sharing it.
//
UnicodeString   RBBISymbolTable::parseReference(const UnicodeString& text,
                                                ParsePosition& pos, int32_t limit) const
{ … }



//
// RBBISymbolTable::lookupNode      Given a key (a variable name), return the
//                                  corresponding RBBI Node.  If there is no entry
//                                  in the table for this name, return nullptr.
//
RBBINode       *RBBISymbolTable::lookupNode(const UnicodeString &key) const{ … }


//
//    RBBISymbolTable::addEntry     Add a new entry to the symbol table.
//                                  Indicate an error if the name already exists -
//                                    this will only occur in the case of duplicate
//                                    variable assignments.
//
void            RBBISymbolTable::addEntry  (const UnicodeString &key, RBBINode *val, UErrorCode &err) { … }


RBBISymbolTableEntry::RBBISymbolTableEntry() : … { … }

RBBISymbolTableEntry::~RBBISymbolTableEntry() { … }


//
//  RBBISymbolTable::print    Debugging function, dump out the symbol table contents.
//
#ifdef RBBI_DEBUG
void RBBISymbolTable::rbbiSymtablePrint() const {
    RBBIDebugPrintf("Variable Definitions Symbol Table\n"
           "Name                  Node         serial  String Val\n"
           "-------------------------------------------------------------------\n");

    int32_t pos = UHASH_FIRST;
    const UHashElement  *e   = nullptr;
    for (;;) {
        e = uhash_nextElement(fHashTable,  &pos);
        if (e == nullptr ) {
            break;
        }
        RBBISymbolTableEntry  *s   = (RBBISymbolTableEntry *)e->value.pointer;

        RBBIDebugPrintf("%-19s   %8p %7d ", CStr(s->key)(), (void *)s->val, s->val->fSerialNum);
        RBBIDebugPrintf(" %s\n", CStr(s->val->fLeftChild->fText)());
    }

    RBBIDebugPrintf("\nParsed Variable Definitions\n");
    pos = -1;
    for (;;) {
        e = uhash_nextElement(fHashTable,  &pos);
        if (e == nullptr ) {
            break;
        }
        RBBISymbolTableEntry  *s   = (RBBISymbolTableEntry *)e->value.pointer;
        RBBIDebugPrintf("%s\n", CStr(s->key)());
        RBBINode::printTree(s->val, true);
        RBBINode::printTree(s->val->fLeftChild, false);
        RBBIDebugPrintf("\n");
    }
}
#endif





U_NAMESPACE_END

#endif /* #if !UCONFIG_NO_BREAK_ITERATION */