redhat/analyse.go

package redhat

import (
	"bufio"
	"fmt"
	"os"
	"sort"
	"strings"
)

// HandleLine the custom func type for processing a line and push data into the map
type HandleLine func(contentMap map[string]int, line string)

// AnalyseData analyse data with sort, uniq and sort by count
func (fa *FileAnalyser) AnalyseData() error {
	contentMap, err := fa.parseContent(fa.file, processLineWithWhiteSpace)
	if err != nil {
		return err
	}

	fa.transformFromMapToDataRows(contentMap)
	fa.sortDataRows()

	return nil
}

func (fa *FileAnalyser) parseContent(file *os.File, handler HandleLine) (map[string]int, error) {
	contentMap := make(map[string]int)

	scanner := bufio.NewScanner(file)

	defer file.Close()

	for scanner.Scan() {
		handler(contentMap, scanner.Text())
	}

	if err := scanner.Err(); err != nil {
		return nil, fmt.Errorf("fail to read and process content: %s", err.Error())
	}
	return contentMap, nil
}

func (fa *FileAnalyser) sortDataRows() {
	sort.SliceStable(fa.dataRows, func(i, j int) bool {
		if firstItemHasMoreCount(fa, i, j) {
			return true
		}
		return compareRowsWhenAreSameCount(fa, i, j)
	})
}

func (fa *FileAnalyser) transformFromMapToDataRows(contentMap map[string]int) {
	for key, value := range contentMap {
		fa.dataRows = append(fa.dataRows, dataRow{
			word:  key,
			count: value,
		})
	}
}

func processLineWithWhiteSpace(contentMap map[string]int, line string) {
	words := strings.Fields(line)
	for _, word := range words {
		count, ok := contentMap[word]
		if !ok {
			contentMap[word] = 1
			continue
		}
		contentMap[word] = count + 1
	}
}

func firstItemHasMoreCount(fa *FileAnalyser, i int, j int) bool {
	return fa.dataRows[i].count > fa.dataRows[j].count
}

func compareRowsWhenAreSameCount(fa *FileAnalyser, i int, j int) bool {
	return (fa.dataRows[i].count == fa.dataRows[j].count) && fa.dataRows[i].word > fa.dataRows[j].word
}