puppet-catalog-tools/internal/impact/impact.go

package impact

import (
	"slices"
	"sort"
)

// Reason explains why one changed file put one host in the result.
type Reason struct {
	File   string `json:"file"`
	Rule   string `json:"rule"`
	Detail string `json:"detail"`
}

// Host is an affected node and every reason it is affected.
type Host struct {
	Certname string   `json:"certname"`
	Reasons  []Reason `json:"reasons"`
}

// Summary is the headline count.
type Summary struct {
	Hosts             int  `json:"hosts"`
	Files             int  `json:"files"`
	AllHostsTriggered bool `json:"all_hosts_triggered"`
}

// Result is the whole analysis, and the --json document.
type Result struct {
	Base          string   `json:"base"`
	Branch        string   `json:"branch"`
	ChangedFiles  []string `json:"changed_files"`
	Hosts         []Host   `json:"hosts"`
	UnmappedFiles []string `json:"unmapped_files"`
	Summary       Summary  `json:"summary"`

	// IgnoredFiles are changed files outside the modulepath — CI, docs,
	// specs — which cannot reach a catalog.
	IgnoredFiles []string `json:"-"`
	// OrphanClasses are classes a changed file defines that nothing declares
	// and no hieradata file pulls in.
	OrphanClasses []string `json:"-"`
	// EmptyFiles are changed files that mapped cleanly but reached no host,
	// which usually means the role they belong to has no members.
	EmptyFiles []string `json:"-"`
	// Truncated is set when --limit cut the host list.
	Truncated bool `json:"-"`
	// TotalHosts is the estate size the selectors were evaluated against.
	TotalHosts int `json:"-"`
}

// Certnames returns the affected hostnames, sorted.
func (r *Result) Certnames() []string {
	out := make([]string, 0, len(r.Hosts))
	for _, h := range r.Hosts {
		out = append(out, h.Certname)
	}
	return out
}

// Analyzer evaluates changed files against one estate fact table.
type Analyzer struct {
	Index *Index
	Nodes []Node
}

// Analyze maps changed files onto hosts. Files are processed in the order
// given; everything in the result is sorted, so output does not move between
// runs.
func (a *Analyzer) Analyze(base, branch string, files []string, limit int) *Result {
	resolver := newClassResolver(a.Index, a.Nodes)

	reasons := map[string][]Reason{}
	orphans := map[string]bool{}
	// hitFiles records which changed files put at least one host in the result.
	hitFiles := map[string]bool{}
	result := &Result{
		Base:         base,
		Branch:       branch,
		ChangedFiles: uniqueSorted(files),
		TotalHosts:   len(a.Nodes),
	}

	for _, file := range result.ChangedFiles {
		m := classifyPath(file, a.Index)

		switch {
		case m.ignored:
			result.IgnoredFiles = append(result.IgnoredFiles, file)
			continue
		case m.unknown:
			result.UnmappedFiles = append(result.UnmappedFiles, file)
			m.targets = []target{allTarget(RuleUnmapped,
				fmtDetail("%s matches no mapping rule; widened to the whole estate", file))}
		}

		for _, t := range m.targets {
			for _, certname := range t.sel.Select(a.Nodes) {
				reasons[certname] = append(reasons[certname],
					Reason{File: file, Rule: t.rule, Detail: t.detail})
				hitFiles[file] = true
			}
		}

		for _, ref := range m.classes {
			res := resolver.resolve(ref.name)
			if res.orphan {
				orphans[ref.name] = true
				continue
			}
			for certname, hits := range res.hosts {
				for _, h := range hits {
					reasons[certname] = append(reasons[certname],
						Reason{File: file, Rule: h.rule, Detail: ref.why + "; " + h.detail})
					hitFiles[file] = true
				}
			}
		}
	}

	// A file that mapped cleanly but reached nobody usually belongs to a role
	// with no members. That is a real answer, not a failure, but it is worth
	// showing rather than leaving as silence.
	for _, file := range result.ChangedFiles {
		if !hitFiles[file] && !slices.Contains(result.IgnoredFiles, file) {
			result.EmptyFiles = append(result.EmptyFiles, file)
		}
	}

	for name := range orphans {
		result.OrphanClasses = append(result.OrphanClasses, name)
	}
	sort.Strings(result.OrphanClasses)

	certnames := make([]string, 0, len(reasons))
	for certname := range reasons {
		certnames = append(certnames, certname)
	}
	sort.Strings(certnames)

	result.Summary = Summary{
		Hosts:             len(certnames),
		Files:             len(result.ChangedFiles),
		AllHostsTriggered: len(certnames) > 0 && len(certnames) == len(a.Nodes),
	}

	if limit > 0 && len(certnames) > limit {
		certnames = certnames[:limit]
		result.Truncated = true
	}

	for _, certname := range certnames {
		result.Hosts = append(result.Hosts, Host{Certname: certname, Reasons: dedupeReasons(reasons[certname])})
	}
	return result
}

func dedupeReasons(in []Reason) []Reason {
	sort.Slice(in, func(i, j int) bool {
		if in[i].File != in[j].File {
			return in[i].File < in[j].File
		}
		if in[i].Rule != in[j].Rule {
			return in[i].Rule < in[j].Rule
		}
		return in[i].Detail < in[j].Detail
	})
	out := in[:0]
	var prev Reason
	for idx, r := range in {
		if idx > 0 && r == prev {
			continue
		}
		prev = r
		out = append(out, r)
	}
	return out
}