169 lines
4.9 KiB
Go
169 lines
4.9 KiB
Go
package entityreconcile
|
|
|
|
import (
|
|
"sort"
|
|
"strings"
|
|
)
|
|
|
|
// ProposalResponse is the private structured response exchanged with the
|
|
// reconciliation prompt. It identifies candidates only by opaque keys.
|
|
type ProposalResponse struct {
|
|
DuplicateGroups []DuplicateGroup `json:"duplicate_groups"`
|
|
}
|
|
|
|
// DuplicateGroup proposes candidate keys that might denote one entity.
|
|
type DuplicateGroup struct {
|
|
Members []string `json:"members"`
|
|
Canonical string `json:"canonical"`
|
|
}
|
|
|
|
// Issue identifies one unsafe proposal category without prescribing a warning
|
|
// message or retry policy to a consuming normalizer.
|
|
type Issue struct {
|
|
GroupIndex int
|
|
Category string
|
|
}
|
|
|
|
// SafeGroup identifies one validated, non-overlapping duplicate group.
|
|
type SafeGroup struct {
|
|
members []string
|
|
canonical string
|
|
}
|
|
|
|
// Members returns an owned copy of the group's candidate keys.
|
|
func (g SafeGroup) Members() []string { return append([]string(nil), g.members...) }
|
|
|
|
// Canonical returns the selected canonical candidate key.
|
|
func (g SafeGroup) Canonical() string { return g.canonical }
|
|
|
|
// Assessment contains only safe groups. Its accessors return owned copies so
|
|
// callers cannot mutate retained assessment data.
|
|
type Assessment struct {
|
|
safeGroups []SafeGroup
|
|
discardedGroups int
|
|
issues []Issue
|
|
}
|
|
|
|
// SafeGroups returns validated non-overlapping groups in proposal order.
|
|
func (a Assessment) SafeGroups() []SafeGroup {
|
|
groups := make([]SafeGroup, len(a.safeGroups))
|
|
for index, group := range a.safeGroups {
|
|
groups[index] = SafeGroup{members: append([]string(nil), group.members...), canonical: group.canonical}
|
|
}
|
|
return groups
|
|
}
|
|
|
|
// DiscardedGroups returns the number of rejected proposal groups.
|
|
func (a Assessment) DiscardedGroups() int { return a.discardedGroups }
|
|
|
|
// Issues returns the deterministic rejection categories in proposal order.
|
|
func (a Assessment) Issues() []Issue { return append([]Issue(nil), a.issues...) }
|
|
|
|
// Assess validates a proposal against the opaque keys created by BuildContext.
|
|
func (m Materials) Assess(response ProposalResponse) Assessment {
|
|
all := make(map[string]struct{}, len(m.candidateKeys))
|
|
for _, key := range m.candidateKeys {
|
|
all[key] = struct{}{}
|
|
}
|
|
groups := make([]assessedGroup, len(response.DuplicateGroups))
|
|
issues := make([]Issue, 0)
|
|
for groupIndex, proposal := range response.DuplicateGroups {
|
|
groups[groupIndex] = assessGroup(proposal, all, m.eligible)
|
|
for _, category := range groups[groupIndex].issues {
|
|
issues = append(issues, Issue{GroupIndex: groupIndex, Category: category})
|
|
}
|
|
}
|
|
|
|
owners := make(map[string][]int)
|
|
for groupIndex, group := range groups {
|
|
if !group.locallyValid {
|
|
continue
|
|
}
|
|
for _, key := range group.members {
|
|
owners[key] = append(owners[key], groupIndex)
|
|
}
|
|
}
|
|
for groupIndex := range groups {
|
|
if !groups[groupIndex].locallyValid {
|
|
continue
|
|
}
|
|
for _, key := range groups[groupIndex].members {
|
|
if len(owners[key]) > 1 {
|
|
groups[groupIndex].conflicting = true
|
|
issues = append(issues, Issue{GroupIndex: groupIndex, Category: "overlapping_member"})
|
|
break
|
|
}
|
|
}
|
|
}
|
|
|
|
assessment := Assessment{issues: issues}
|
|
for _, group := range groups {
|
|
if !group.locallyValid || group.conflicting {
|
|
assessment.discardedGroups++
|
|
continue
|
|
}
|
|
assessment.safeGroups = append(assessment.safeGroups, SafeGroup{members: append([]string(nil), group.members...), canonical: group.canonical})
|
|
}
|
|
return assessment
|
|
}
|
|
|
|
type assessedGroup struct {
|
|
members []string
|
|
canonical string
|
|
issues []string
|
|
locallyValid bool
|
|
conflicting bool
|
|
}
|
|
|
|
func assessGroup(proposal DuplicateGroup, all, eligible map[string]struct{}) assessedGroup {
|
|
issues := make([]string, 0)
|
|
members := make([]string, 0, len(proposal.Members))
|
|
seen := make(map[string]struct{}, len(proposal.Members))
|
|
for _, key := range proposal.Members {
|
|
if category := keyCategory(key, all, eligible); category != "" {
|
|
issues = append(issues, "member_"+category)
|
|
continue
|
|
}
|
|
if _, exists := seen[key]; exists {
|
|
issues = append(issues, "repeated_member")
|
|
continue
|
|
}
|
|
seen[key] = struct{}{}
|
|
members = append(members, key)
|
|
}
|
|
canonicalCategory := keyCategory(proposal.Canonical, all, eligible)
|
|
if canonicalCategory != "" {
|
|
issues = append(issues, "canonical_"+canonicalCategory)
|
|
}
|
|
if len(members) < 2 {
|
|
issues = append(issues, "fewer_than_two_members")
|
|
}
|
|
if canonicalCategory == "" && !contains(members, proposal.Canonical) {
|
|
issues = append(issues, "canonical_not_member")
|
|
}
|
|
sort.Strings(members)
|
|
return assessedGroup{members: members, canonical: proposal.Canonical, issues: issues, locallyValid: len(issues) == 0}
|
|
}
|
|
|
|
func keyCategory(key string, all, eligible map[string]struct{}) string {
|
|
if strings.TrimSpace(key) == "" {
|
|
return "blank"
|
|
}
|
|
if _, ok := all[key]; !ok {
|
|
return "unknown"
|
|
}
|
|
if _, ok := eligible[key]; !ok {
|
|
return "ineligible"
|
|
}
|
|
return ""
|
|
}
|
|
|
|
func contains(values []string, want string) bool {
|
|
for _, value := range values {
|
|
if value == want {
|
|
return true
|
|
}
|
|
}
|
|
return false
|
|
}
|