package heap

import (
	C "gitee.com/ljfirst/algo-go-sdk/common/constant"
	"reflect"
)

/**
 * @author ljfirst
 * @version V1.0
 * @date 2023/6/27 01:12
 * @author-Email ljfirst@mail.ustc.edu.cn
 * @blogURL https://blog.csdn.net/ljfirst
 * @description 词频统计
 * 实现统计一篇英文文章内每个单词的出现频率，并返回出现频率最高的前3个单词及其出现次数
 * */
type WordFrequency struct {
}

func (m *WordFrequency) Statistics(words []string) []string {
	if len(words) == 0 {
		return nil
	}
	statMap := make(map[string]*wordsType)
	ans := make([]string, 0)
	topFrequency := 3

	pq := NewPriorityQueue([]C.Options{C.WithQueueType(true),
		C.WithLimitSize(topFrequency),
		C.WithGT(m.gt()),
		C.WithChange(m.change()),
		C.WithCompare(m.comp())}...)

	for _, word := range words {
		w, exist := statMap[word]
		if !exist {
			w = &wordsType{Word: word, Frequency: 1}
		} else {
			w.Frequency += 1
		}
		statMap[word] = w
		if w.QueueExist {
			pq.Update(w)
		} else if pq.Size() < pq.limitSize {
			w.QueueExist = true
			pq.Offer(w)
		} else if pq.Comp(w, pq.Peak()) {
			v, ok := pq.Peak().(*wordsType)
			if ok {
				v.QueueExist = false
			}
			w.QueueExist = true
			pq.Offer(w)
		}
	}

	for pq.Size() != 0 {
		w, exist := pq.Poll().(*wordsType)
		if !exist {
			continue
		}
		ans = append(ans, w.Word)
	}
	return ans
}

func (m *WordFrequency) gt() C.GT {
	return func(arrayV interface{}, index1 int, index2 int) bool {
		value := reflect.ValueOf(arrayV)
		if value.Kind() != reflect.Array && value.Kind() != reflect.Slice {
			return false
		}
		if value.Len() <= index1 || value.Len() <= index2 {
			return false
		}
		array, ok := arrayV.([]interface{})
		if !ok {
			return false
		}
		w1, ok := array[index1].(*wordsType)
		if !ok {
			return false
		}
		w2, ok := array[index2].(*wordsType)
		if !ok {
			return false
		}
		return w1.Frequency > w2.Frequency
	}
}

func (m *WordFrequency) change() C.Change {
	return func(arrayV interface{}, index1 int, index2 int) {
		value := reflect.ValueOf(arrayV)
		if value.Kind() != reflect.Array && value.Kind() != reflect.Slice {
			return
		}
		if value.Len() <= index1 || value.Len() <= index2 {
			return
		}
		array, ok := arrayV.([]interface{})
		if !ok {
			return
		}
		array[index1], array[index2] = array[index2], array[index1]
	}
}

func (m *WordFrequency) comp() C.Compare {
	return func(value1 interface{}, value2 interface{}) bool {
		w1, ok := value1.(*wordsType)
		if !ok {
			return false
		}
		w2, ok := value2.(*wordsType)
		if !ok {
			return false
		}
		return w1.Frequency > w2.Frequency
	}
}

type wordsType struct {
	Word       string
	Frequency  int
	QueueExist bool // 用于标识元素是否存在于优先队列中，以及获取index值，便于操作
}

func (m *WordFrequency) GetAttribute() *C.Attribute {
	return &C.Attribute{
		Tags: []string{C.Heap},
		Desc: &C.Desc{
			Name:   "WordFrequency",
			NameCn: "词频统计",
		},
	}
}