Files
amcs/internal/tools/duplicate_audit.go
warkanum 3198600031
CI / build-and-test (push) Failing after 1m24s
CI / build-and-test (pull_request) Failing after 59s
feat(tools): add duplicate audit report
2026-07-14 16:28:06 +02:00

306 lines
10 KiB
Go

package tools
import (
"context"
"regexp"
"sort"
"strings"
"github.com/google/uuid"
"github.com/modelcontextprotocol/go-sdk/mcp"
"git.warky.dev/wdevs/amcs/internal/config"
"git.warky.dev/wdevs/amcs/internal/session"
"git.warky.dev/wdevs/amcs/internal/store"
thoughttypes "git.warky.dev/wdevs/amcs/internal/types"
)
const defaultDuplicateAuditLimit = 50
var duplicateWhitespace = regexp.MustCompile(`\s+`)
type DuplicateAuditTool struct {
store *store.DB
search config.SearchConfig
sessions *session.ActiveProjects
}
type DuplicateAuditInput struct {
Project string `json:"project,omitempty" jsonschema:"optional project name or id to scope thought duplicate scanning"`
Limit int `json:"limit,omitempty" jsonschema:"maximum candidate groups to return"`
ThoughtLimit int `json:"thought_limit,omitempty" jsonschema:"maximum thoughts to scan for duplicate content; defaults to search limit"`
IncludeArchived bool `json:"include_archived,omitempty" jsonschema:"include archived thoughts in the scan"`
}
type DuplicateAuditOutput struct {
Report DuplicateAuditReport `json:"report"`
}
type DuplicateAuditReport struct {
Summary DuplicateAuditSummary `json:"summary"`
ProjectCandidates []DuplicateProjectCandidate `json:"project_candidates"`
ThoughtCandidates []DuplicateThoughtCandidate `json:"thought_candidates"`
MetadataCandidates []MetadataCleanupCandidate `json:"metadata_candidates"`
RecommendedNextStep string `json:"recommended_next_step"`
}
type DuplicateAuditSummary struct {
DryRun bool `json:"dry_run"`
ScannedProjects int `json:"scanned_projects"`
ScannedThoughts int `json:"scanned_thoughts"`
ProjectCandidateGroups int `json:"project_candidate_groups"`
ThoughtCandidateGroups int `json:"thought_candidate_groups"`
MetadataCandidateGroups int `json:"metadata_candidate_groups"`
Truncated bool `json:"truncated"`
}
type DuplicateProjectCandidate struct {
MatchType string `json:"match_type"`
NormalizedValue string `json:"normalized_value"`
Confidence float64 `json:"confidence"`
RecommendedCanonicalID uuid.UUID `json:"recommended_canonical_id"`
Projects []thoughttypes.ProjectSummary `json:"projects"`
}
type DuplicateThoughtCandidate struct {
MatchType string `json:"match_type"`
NormalizedValue string `json:"normalized_value"`
Confidence float64 `json:"confidence"`
RecommendedCanonicalID uuid.UUID `json:"recommended_canonical_id"`
Thoughts []thoughttypes.Thought `json:"thoughts"`
}
type MetadataCleanupCandidate struct {
Field string `json:"field"`
NormalizedValue string `json:"normalized_value"`
Values []string `json:"values"`
Occurrences int `json:"occurrences"`
RecommendedValue string `json:"recommended_value"`
RecommendedAction string `json:"recommended_action"`
}
func NewDuplicateAuditTool(db *store.DB, search config.SearchConfig, sessions *session.ActiveProjects) *DuplicateAuditTool {
return &DuplicateAuditTool{store: db, search: search, sessions: sessions}
}
func (t *DuplicateAuditTool) Handle(ctx context.Context, req *mcp.CallToolRequest, in DuplicateAuditInput) (*mcp.CallToolResult, DuplicateAuditOutput, error) {
project, err := resolveProject(ctx, t.store, t.sessions, req, in.Project, false)
if err != nil {
return nil, DuplicateAuditOutput{}, err
}
var projectID *int64
if project != nil {
projectID = &project.NumericID
}
projects, err := t.store.ListProjects(ctx)
if err != nil {
return nil, DuplicateAuditOutput{}, err
}
thoughtLimit := normalizeLimit(in.ThoughtLimit, t.search)
if in.ThoughtLimit <= 0 {
thoughtLimit = normalizeLimit(0, t.search)
}
thoughts, err := t.store.ListThoughts(ctx, thoughttypes.ListFilter{Limit: thoughtLimit, ProjectID: projectID, IncludeArchived: in.IncludeArchived})
if err != nil {
return nil, DuplicateAuditOutput{}, err
}
if project != nil {
_ = t.store.TouchProject(ctx, project.NumericID)
}
return nil, DuplicateAuditOutput{Report: buildDuplicateAuditReport(projects, thoughts, in)}, nil
}
func buildDuplicateAuditReport(projects []thoughttypes.ProjectSummary, thoughts []thoughttypes.Thought, in DuplicateAuditInput) DuplicateAuditReport {
limit := in.Limit
if limit <= 0 {
limit = defaultDuplicateAuditLimit
}
projectCandidates, projectTruncated := duplicateProjectCandidates(projects, limit)
thoughtCandidates, thoughtTruncated := duplicateThoughtCandidates(thoughts, limit)
metadataCandidates, metadataTruncated := metadataCleanupCandidates(thoughts, limit)
return DuplicateAuditReport{
Summary: DuplicateAuditSummary{
DryRun: true,
ScannedProjects: len(projects),
ScannedThoughts: len(thoughts),
ProjectCandidateGroups: len(projectCandidates),
ThoughtCandidateGroups: len(thoughtCandidates),
MetadataCandidateGroups: len(metadataCandidates),
Truncated: projectTruncated || thoughtTruncated || metadataTruncated,
},
ProjectCandidates: projectCandidates,
ThoughtCandidates: thoughtCandidates,
MetadataCandidates: metadataCandidates,
RecommendedNextStep: "Review candidates manually, then use explicit archive/update/merge tools; this audit performs no writes.",
}
}
func duplicateProjectCandidates(projects []thoughttypes.ProjectSummary, limit int) ([]DuplicateProjectCandidate, bool) {
groups := map[string][]thoughttypes.ProjectSummary{}
for _, project := range projects {
key := normalizeDuplicateText(project.Name)
if key != "" {
groups[key] = append(groups[key], project)
}
}
keys := sortedDuplicateKeys(groups)
out := make([]DuplicateProjectCandidate, 0, min(len(keys), limit))
for _, key := range keys {
items := groups[key]
sort.SliceStable(items, func(i, j int) bool {
if items[i].ThoughtCount != items[j].ThoughtCount {
return items[i].ThoughtCount > items[j].ThoughtCount
}
return items[i].CreatedAt.Before(items[j].CreatedAt)
})
matchType := "normalized_name"
if allProjectNamesExact(items) {
matchType = "exact_name"
}
out = append(out, DuplicateProjectCandidate{MatchType: matchType, NormalizedValue: key, Confidence: 1.0, RecommendedCanonicalID: items[0].ID, Projects: items})
if len(out) == limit {
return out, len(keys) > limit
}
}
return out, false
}
func duplicateThoughtCandidates(thoughts []thoughttypes.Thought, limit int) ([]DuplicateThoughtCandidate, bool) {
groups := map[string][]thoughttypes.Thought{}
for _, thought := range thoughts {
key := normalizeDuplicateText(thought.Content)
if key != "" {
groups[key] = append(groups[key], thought)
}
}
keys := sortedDuplicateKeys(groups)
out := make([]DuplicateThoughtCandidate, 0, min(len(keys), limit))
for _, key := range keys {
items := groups[key]
sort.SliceStable(items, func(i, j int) bool { return items[i].CreatedAt.Before(items[j].CreatedAt) })
matchType := "normalized_content"
if allThoughtContentExact(items) {
matchType = "exact_content"
}
out = append(out, DuplicateThoughtCandidate{MatchType: matchType, NormalizedValue: key, Confidence: 1.0, RecommendedCanonicalID: items[0].GUID, Thoughts: items})
if len(out) == limit {
return out, len(keys) > limit
}
}
return out, false
}
func metadataCleanupCandidates(thoughts []thoughttypes.Thought, limit int) ([]MetadataCleanupCandidate, bool) {
groups := map[string]map[string]int{}
add := func(field, value string) {
trimmed := strings.TrimSpace(value)
key := normalizeDuplicateText(trimmed)
if key == "" || trimmed == "" {
return
}
bucketKey := field + "\x00" + key
if groups[bucketKey] == nil {
groups[bucketKey] = map[string]int{}
}
groups[bucketKey][trimmed]++
}
for _, thought := range thoughts {
add("type", thought.Metadata.Type)
add("source", thought.Metadata.Source)
for _, topic := range thought.Metadata.Topics {
add("topics", topic)
}
for _, person := range thought.Metadata.People {
add("people", person)
}
}
keys := make([]string, 0, len(groups))
for key, values := range groups {
if len(values) > 1 {
keys = append(keys, key)
}
}
sort.Strings(keys)
out := make([]MetadataCleanupCandidate, 0, min(len(keys), limit))
for _, key := range keys {
parts := strings.SplitN(key, "\x00", 2)
values, occurrences, recommended := valuesByCount(groups[key])
out = append(out, MetadataCleanupCandidate{
Field: parts[0],
NormalizedValue: parts[1],
Values: values,
Occurrences: occurrences,
RecommendedValue: recommended,
RecommendedAction: "preview bulk metadata remap before applying; no changes made by audit",
})
if len(out) == limit {
return out, len(keys) > limit
}
}
return out, false
}
func sortedDuplicateKeys[T any](groups map[string][]T) []string {
keys := make([]string, 0, len(groups))
for key, items := range groups {
if len(items) > 1 {
keys = append(keys, key)
}
}
sort.Strings(keys)
return keys
}
func valuesByCount(counts map[string]int) ([]string, int, string) {
values := make([]string, 0, len(counts))
occurrences := 0
for value, count := range counts {
values = append(values, value)
occurrences += count
}
sort.Slice(values, func(i, j int) bool {
if counts[values[i]] != counts[values[j]] {
return counts[values[i]] > counts[values[j]]
}
return values[i] < values[j]
})
return values, occurrences, values[0]
}
func normalizeDuplicateText(value string) string {
return duplicateWhitespace.ReplaceAllString(strings.ToLower(strings.TrimSpace(value)), " ")
}
func allProjectNamesExact(projects []thoughttypes.ProjectSummary) bool {
if len(projects) == 0 {
return false
}
first := strings.TrimSpace(projects[0].Name)
for _, project := range projects[1:] {
if strings.TrimSpace(project.Name) != first {
return false
}
}
return true
}
func allThoughtContentExact(thoughts []thoughttypes.Thought) bool {
if len(thoughts) == 0 {
return false
}
first := strings.TrimSpace(thoughts[0].Content)
for _, thought := range thoughts[1:] {
if strings.TrimSpace(thought.Content) != first {
return false
}
}
return true
}