wrapped/pkg/stats/analyzer.go

package stats

import (
	"sort"
	"time"

	"github.com/atridad/wrapped-cli/pkg/gitea"
	"github.com/atridad/wrapped-cli/pkg/models"
)

type Analyzer struct {
	repos   []gitea.GiteaRepo
	commits []models.Commit
}

func NewAnalyzer() *Analyzer {
	return &Analyzer{
		repos:   []gitea.GiteaRepo{},
		commits: []models.Commit{},
	}
}

func (a *Analyzer) AddRepos(repos []gitea.GiteaRepo) {
	a.repos = repos
}

func (a *Analyzer) AddCommits(commits []models.Commit) {
	a.commits = append(a.commits, commits...)
}

func (a *Analyzer) inferPrimaryEmail(username string) string {
	emailCount := make(map[string]int)

	for _, c := range a.commits {
		if c.AuthorEmail != "" {
			emailCount[c.AuthorEmail]++
		}
	}

	var bestEmail string
	max := 0

	for email, count := range emailCount {
		if count > max {
			max = count
			bestEmail = email
		}
	}

	return bestEmail
}

func (a *Analyzer) Generate(username string) *models.UserStats {
	primaryEmail := a.inferPrimaryEmail(username)

	var filteredCommits []models.Commit
	for _, commit := range a.commits {
		if primaryEmail != "" && commit.AuthorEmail == primaryEmail {
			filteredCommits = append(filteredCommits, commit)
		}
	}

	stats := &models.UserStats{
		Username:          username,
		TotalCommits:      len(filteredCommits),
		TotalRepositories: len(a.repos),
	}

	stats.Languages = a.generateLanguageStats()
	stats.CommitsByWeekday = a.generateWeekdayStats(filteredCommits)
	stats.CommitsByMonth = a.generateMonthStats(filteredCommits)
	stats.AverageCommitsPerDay = a.calculateAverageCommitsPerDay(filteredCommits)

	if len(stats.CommitsByMonth) > 0 {
		stats.MostActiveMonth = stats.CommitsByMonth[0].Date.Format("January")
	}
	if len(stats.CommitsByWeekday) > 0 {
		stats.MostActiveDay = stats.CommitsByWeekday[0].Day
	}

	stats.TopRepositories = a.getTopRepositories(filteredCommits, 5)

	return stats
}

func (a *Analyzer) generateLanguageStats() []models.LanguageStats {
	langCount := make(map[string]int)

	for _, repo := range a.repos {
		if repo.Language != "" {
			langCount[repo.Language]++
		}
	}

	var stats []models.LanguageStats
	for lang, count := range langCount {
		stats = append(stats, models.LanguageStats{
			Language: lang,
			Count:    count,
		})
	}

	sort.Slice(stats, func(i, j int) bool {
		return stats[i].Count > stats[j].Count
	})

	total := len(a.repos)
	for i := range stats {
		stats[i].Percent = float64(stats[i].Count) / float64(total) * 100
	}

	return stats
}

func (a *Analyzer) generateWeekdayStats(commits []models.Commit) []models.DateStats {
	dayCount := make(map[time.Weekday]int)
	dayNames := [7]string{"Sunday", "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday"}

	for _, commit := range commits {
		day := commit.Timestamp.Weekday()
		dayCount[day]++
	}

	var stats []models.DateStats
	for i := 0; i < 7; i++ {
		day := time.Weekday(i)
		stats = append(stats, models.DateStats{
			Day:   dayNames[i],
			Count: dayCount[day],
		})
	}

	sort.Slice(stats, func(i, j int) bool {
		return stats[i].Count > stats[j].Count
	})

	return stats
}

func (a *Analyzer) generateMonthStats(commits []models.Commit) []models.DateStats {
	monthCount := make(map[string]int)
	monthKeys := []string{}

	for _, commit := range commits {
		monthStr := commit.Timestamp.Format("2006-01")
		monthCount[monthStr]++
	}

	for k := range monthCount {
		monthKeys = append(monthKeys, k)
	}

	var stats []models.DateStats
	for _, monthStr := range monthKeys {
		t, _ := time.Parse("2006-01", monthStr)
		stats = append(stats, models.DateStats{
			Date:  t,
			Count: monthCount[monthStr],
		})
	}

	sort.Slice(stats, func(i, j int) bool {
		return stats[i].Count > stats[j].Count
	})

	return stats
}

func (a *Analyzer) calculateAverageCommitsPerDay(commits []models.Commit) float64 {
	if len(commits) == 0 {
		return 0
	}

	if len(commits) < 2 {
		return float64(len(commits))
	}

	sort.Slice(commits, func(i, j int) bool {
		return commits[i].Timestamp.Before(commits[j].Timestamp)
	})

	firstDay := commits[0].Timestamp
	lastDay := commits[len(commits)-1].Timestamp

	daysDiff := lastDay.Sub(firstDay).Hours() / 24
	if daysDiff == 0 {
		daysDiff = 1
	}

	return float64(len(commits)) / daysDiff
}

func (a *Analyzer) getTopRepositories(commits []models.Commit, n int) []models.Repository {
	repoCommitCount := make(map[string]int)
	repoMap := make(map[string]models.Repository)

	for _, commit := range commits {
		repoCommitCount[commit.RepoName]++
	}

	for _, repo := range a.repos {
		repoMap[repo.Name] = models.Repository{
			ID:       repo.ID,
			Name:     repo.Name,
			FullName: repo.FullName,
			HTMLURL:  repo.HTMLURL,
			Language: repo.Language,
			Topics:   repo.Topics,
		}
	}

	type repoWithCount struct {
		repo  models.Repository
		count int
	}

	var repos []repoWithCount
	for name, count := range repoCommitCount {
		if repo, exists := repoMap[name]; exists {
			repos = append(repos, repoWithCount{repo, count})
		}
	}

	sort.Slice(repos, func(i, j int) bool {
		return repos[i].count > repos[j].count
	})

	if len(repos) > n {
		repos = repos[:n]
	}

	result := make([]models.Repository, len(repos))
	for i, rc := range repos {
		result[i] = rc.repo
	}

	return result
}