扩展的find算法 -- find_nearest(查询最接近的数据)

通常使用需要快速查找定位的数据时，会通过 set/map 等保存数据，然后使用find等算法进行查找。

但有时想查找的数据不能完全匹配，而是查找一个最接近的值（比如，在两个不同的数据结构中，想匹配出ID相同，且发生时间最接近的数据）时，STL没有提供可用的函数，一般的做法就是将数据全部遍历计算一遍，然后选取最接近的值。其算法复杂度为 O(n)。

通过扩展，写出通用的模版算法 find_nearest 可以完成这个功能，适用于 set/map。

 

PS：原本想找现成的，可惜没有找到，只有自己写一个。感觉这个人的目的和我的比较接近：http://topic.csdn.net/u/20071112/10/226d912b-05b5-4e2f-8f8e-0a93ec76eeca.html

 

废话不多说了，直接上代码和UT。

 

代码：

  //之前使用了VS中特有的 static _Kfn 函数，但在linux上不通用，因此自己写一个 template <class K> const K& ftl_Kfn(const K& Val) { return (Val); } template <class K, class V> const K& ftl_Kfn(const std::pair<const K,V> & Val) { return (Val.first); } //在set或map中根据key查找最接近的元素，通过传入的仿函数(F func)来判断哪个更接近想要的对象 //如果找到满足条件且最接近的，返回true,且 retIter 为最接近元素的迭代器；如果找不到，返回false // F 必须是如下结构的函数指针或仿函数 // int operator()(const key_type* pPre, const key_type* pWant, const key_type* pNext) // 注意： // 1.返回值的意义 : 小于0 表示 pPre 更接近， 等于 0 表示两个都不接近， 大于0 表示后一个接近 // 2.pPre 和 pNext 都可能为NULL，但 pWant 不可能为NULL template<typename C, typename F> bool find_nearest(C& container, const typename C::key_type& want, F func, typename C::iterator& retIter) { bool bResult = false; retIter = container.end(); //查看是否有想要的数据，或者该插入的位置(相等的或后一个) typename C::iterator nextIter = container.lower_bound(want); typename C::iterator preIter = nextIter; int retCmp = 0; const typename C::key_type* pPreValue = NULL; const typename C::key_type* pNextValue = NULL; //找到中间 if (nextIter != container.end()) { preIter--; if(preIter != container.end()) { //pPreValue = &container._Kfn(*preIter); pPreValue = &ftl_Kfn(*preIter); } pNextValue = &ftl_Kfn(*nextIter); //直接找到等价的 if(!container.key_comp()(want, ftl_Kfn(*nextIter))) { bResult = true; retIter = nextIter; return bResult; } } else { //在最后，说明没有数据，或者想要的是最大的值 if (!container.empty()) { nextIter = (--container.end()); pNextValue = &ftl_Kfn(*nextIter); } } retCmp = func(pPreValue,&want,pNextValue); if(retCmp < 0) { bResult = true; retIter = preIter; } else if(retCmp > 0) { bResult = true; retIter = nextIter; } //else -- 为0,表明没有找到，retIter 和 bResult 之前已经赋好值了 return bResult; }

 

UT(CppUnit)：

//比较哪个整数更接近pWant struct CompareInt { CompareInt(int maxDiff = std::numeric_limits<int>::max()) { m_maxDiff = maxDiff; } int operator()(const int* pPre, const int* pWant, const int* pNext) { int dif1 = m_maxDiff; int dif2 = m_maxDiff; if (pPre || pNext) { if (pPre) { dif1 = abs(*pWant - *pPre); } if(pNext) { dif2 = abs(*pNext-*pWant); } if( dif1 <= dif2 && dif1 <= m_maxDiff) { return -1; } else if(dif2 <= m_maxDiff) { return 1; } } return 0; } private: int m_maxDiff; }; //set中查找 void CFTLFunctionalTester::test_find_nearest_set_int() { typedef std::set<int> IntSetContainer; bool bResult = false; IntSetContainer container; //setContainer // 0, 3, 6, 9, 12, 15 for (int i = 0; i <= 15; i += 3) { container.insert(i); } CPPUNIT_ASSERT(container.size() == 6); int want; IntSetContainer::iterator nearestIter = container.end(); want = -1; nearestIter = container.end(); bResult = find_nearest(container, want, CompareInt(), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(*nearestIter == 0); want = 4; nearestIter = container.end(); bResult = find_nearest(container, want, CompareInt(), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(*nearestIter == 3); want = 5; nearestIter = container.end(); bResult = find_nearest(container, want, CompareInt(), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(*nearestIter == 6); want = 6; nearestIter = container.end(); bResult = find_nearest(container, want, CompareInt(), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(*nearestIter == 6); want = 20; nearestIter = container.end(); bResult = find_nearest(container, want, CompareInt(), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(*nearestIter == 15); } //map中查找int void CFTLFunctionalTester::test_find_nearest_map_int() { typedef std::map<int, int> IntMapContainer; bool bResult = false; IntMapContainer container; //setContainer // [0,0], [3,30], [6,60], [9,90], [12,120], [15,150] for (int i = 0; i <= 15; i += 3) { container[i] = i * 10; } CPPUNIT_ASSERT(container.size() == 6); int want; IntMapContainer::iterator nearestIter = container.end(); want = -1; nearestIter = container.end(); bResult = find_nearest(container, want, CompareInt(), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(nearestIter->first == 0); CPPUNIT_ASSERT(nearestIter->second == 0); want = 4; nearestIter = container.end(); bResult = find_nearest(container, want, CompareInt(), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(nearestIter->first == 3); CPPUNIT_ASSERT(nearestIter->second == 30); want = 5; nearestIter = container.end(); bResult = find_nearest(container, want, CompareInt(), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(nearestIter->first == 6); CPPUNIT_ASSERT(nearestIter->second == 60); want = 6; nearestIter = container.end(); bResult = find_nearest(container, want, CompareInt(), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(nearestIter->first == 6); CPPUNIT_ASSERT(nearestIter->second == 60); want = 20; nearestIter = container.end(); bResult = find_nearest(container, want, CompareInt(), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(nearestIter->first == 15); CPPUNIT_ASSERT(nearestIter->second == 150); } //呼叫信息的KEY，要找到callID相同，时间在5秒以内最接近的 struct CallInfo { int callID; //呼叫ID int tickets; //呼叫时间 CallInfo() { callID = -1; tickets = -1; } CallInfo(int callID, int tickets) { this->callID = callID; this->tickets = tickets; } bool operator < (const CallInfo & other) const { COMPARE_MEM_LESS(callID, other); COMPARE_MEM_LESS(tickets, other); return false; } }; struct CompareCallInfo { CompareCallInfo(int maxDiff) { m_maxDiff = maxDiff; } int operator()(const CallInfo* pPre, const CallInfo* pWant, const CallInfo* pNext) { int dif1 = 0x7FFFFFFF; int dif2 = 0x7FFFFFFF; if (pPre && pPre->callID == pWant->callID) { dif1 = abs(pWant->tickets - pPre->tickets); } if(pNext && pNext->callID == pWant->callID) { dif2 = abs(pNext->tickets - pWant->tickets); } if( dif1 <= dif2 && dif1 <= m_maxDiff) { return -1; } else if(dif2 <= m_maxDiff) { return 1; } return 0; } int m_maxDiff; }; //在set中查找最接近的结构体 void CFTLFunctionalTester::test_find_nearest_set_struct() { typedef std::set<CallInfo> CallInfoSetContainer; CallInfoSetContainer container; //[0,0],[0,3] //[1,10],[1,13] //[2,20],[2,23] //[3,30],[3,33] //[4,40],[4,43] //[5,50],[5,53] for (int i = 0; i <= 5; i++) { container.insert(CallInfo(i, i * 10)); container.insert(CallInfo(i, i * 10 + 3)); } CPPUNIT_ASSERT(container.size() == 12); bool bResult = false; CallInfo want; CallInfoSetContainer::iterator nearestIter = container.end(); want.callID = 1; want.tickets = 11; nearestIter = container.end(); bResult = find_nearest(container, want, CompareCallInfo(5), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(nearestIter->callID == 1); CPPUNIT_ASSERT(nearestIter->tickets == 10); want.callID = 1; want.tickets = 14; nearestIter = container.end(); bResult = find_nearest(container, want, CompareCallInfo(5), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(nearestIter->callID == 1); CPPUNIT_ASSERT(nearestIter->tickets == 13); want.callID = 3; want.tickets = 30; nearestIter = container.end(); bResult = find_nearest(container, want, CompareCallInfo(5), nearestIter); CPPUNIT_ASSERT(bResult); CPPUNIT_ASSERT(nearestIter->callID == 3); CPPUNIT_ASSERT(nearestIter->tickets == 30); //因为超过范围(大于5)，无法找到 want.callID = 3; want.tickets = 39; nearestIter = container.end(); bResult = find_nearest(container, want, CompareCallInfo(5), nearestIter); CPPUNIT_ASSERT(!bResult); CPPUNIT_ASSERT(container.end() == nearestIter ); //因为callID不同，所以无法找到 want.callID = 6; want.tickets = 10; nearestIter = container.end(); bResult = find_nearest(container, want, CompareCallInfo(5), nearestIter); CPPUNIT_ASSERT(!bResult); CPPUNIT_ASSERT(container.end() == nearestIter ); }