Files
tm_back/pkg/ai_summarizer/storage.go
T
2026-05-16 11:01:13 +03:30

672 lines
22 KiB
Go

package ai_summarizer
import (
"context"
"encoding/json"
"errors"
"fmt"
"io"
"net/http"
"path/filepath"
"strings"
"time"
"github.com/minio/minio-go/v7"
"github.com/minio/minio-go/v7/pkg/credentials"
"tm/pkg/logger"
)
// procedurePrefix is prepended to every contract folder id in the MinIO key.
// The AI team's storage layout in bucket opplens-documents is:
//
// PROC_<contractFolderID>/<noticePublicationID>/tender.json
// PROC_<contractFolderID>/<noticePublicationID>/documents/{files}
const procedurePrefix = "PROC_"
// StorageClient wraps a MinIO client configured to connect to the AI service's
// MinIO bucket and retrieve tender.json files.
type StorageClient struct {
client *minio.Client
config *Config
logger logger.Logger
}
// TenderSummaryItem represents a single tender's summary retrieved from storage.
type TenderSummaryItem struct {
ContractFolderID string `json:"contract_folder_id"`
NoticeID string `json:"notice_id"`
OverallSummary string `json:"overall_summary"`
Done bool `json:"done"`
}
// noticeKeyVariants returns MinIO notice folder names to try. The pipeline may
// store either the raw TED id (e.g. "00175735-2026") or a zero-stripped form
// ("175735-2026"); we try the raw value first, then the normalized variant.
func noticeKeyVariants(noticePublicationID string) []string {
raw := strings.TrimSpace(noticePublicationID)
if raw == "" {
return nil
}
normalized := normalizeNoticeKey(raw)
if normalized == raw {
return []string{raw}
}
return []string{raw, normalized}
}
// noticeObjectPrefix returns the MinIO key prefix for one notice folder name.
func noticeObjectPrefix(contractFolderID, noticeKey string) string {
return fmt.Sprintf(
"%s%s/%s/",
procedurePrefix,
normalizeContractFolderID(contractFolderID),
strings.TrimSpace(noticeKey),
)
}
// noticeObjectPrefixes returns every prefix variant for a notice publication id.
func noticeObjectPrefixes(contractFolderID, noticePublicationID string) []string {
folderID := normalizeContractFolderID(contractFolderID)
keys := noticeKeyVariants(noticePublicationID)
prefixes := make([]string, 0, len(keys))
seen := make(map[string]struct{}, len(keys))
for _, key := range keys {
prefix := noticeObjectPrefix(folderID, key)
if _, dup := seen[prefix]; dup {
continue
}
seen[prefix] = struct{}{}
prefixes = append(prefixes, prefix)
}
return prefixes
}
// tenderJSONKeyVariants returns object keys to try for tender.json.
func tenderJSONKeyVariants(contractFolderID, noticePublicationID string) []string {
prefixes := noticeObjectPrefixes(contractFolderID, noticePublicationID)
keys := make([]string, 0, len(prefixes))
for _, prefix := range prefixes {
keys = append(keys, prefix+"tender.json")
}
return keys
}
// normalizeContractFolderID strips an accidental PROC_ prefix from stored ids.
func normalizeContractFolderID(contractFolderID string) string {
id := strings.TrimSpace(contractFolderID)
if len(id) >= 5 && strings.EqualFold(id[:5], "proc_") {
return strings.TrimSpace(id[5:])
}
return id
}
// normalizeNoticeKey trims the leading zeros that we add when persisting the
// notice publication id (e.g. "00322263-2026") so it matches the AI team's
// folder convention ("322263-2026"). Empty input is preserved.
func normalizeNoticeKey(noticePublicationID string) string {
id := strings.TrimSpace(noticePublicationID)
if id == "" {
return ""
}
// Strip leading zeros from the numeric portion only; keep the year suffix
// (and any non-numeric tail) intact.
out := strings.TrimLeft(id, "0")
if out == "" || strings.HasPrefix(out, "-") {
// All-zero numeric portion: keep a single "0" so the key is never empty
// and never starts with "-".
return "0" + out
}
return out
}
// NewStorageClient creates a new MinIO storage client for the AI service.
// It establishes a connection to the external MinIO instance and verifies
// that the configured bucket exists.
func NewStorageClient(config *Config, log logger.Logger) (*StorageClient, error) {
if config == nil {
config = DefaultConfig()
}
if err := config.ValidateStorage(); err != nil {
return nil, fmt.Errorf("ai_summarizer: invalid storage config: %w", err)
}
log.Info("Connecting to AI summarizer MinIO", map[string]interface{}{
"endpoint": config.MinioEndpoint,
"use_ssl": config.MinioUseSSL,
"region": config.MinioRegion,
"bucket": config.MinioBucket,
})
minioClient, err := minio.New(config.MinioEndpoint, &minio.Options{
Creds: credentials.NewStaticV4(config.MinioAccessKey, config.MinioSecretKey, ""),
Secure: config.MinioUseSSL,
Region: config.MinioRegion,
})
if err != nil {
log.Error("Failed to create AI summarizer MinIO client", map[string]interface{}{
"endpoint": config.MinioEndpoint,
"error": err.Error(),
})
return nil, fmt.Errorf("ai_summarizer: failed to create MinIO client: %w", err)
}
storage := &StorageClient{
client: minioClient,
config: config,
logger: log,
}
timeout := config.MinioTimeout
if timeout <= 0 {
timeout = 30 * time.Second
}
verifyCtx, cancel := context.WithTimeout(context.Background(), timeout)
defer cancel()
if err := storage.verifyConnection(verifyCtx); err != nil {
log.Error("AI summarizer MinIO connection check failed", map[string]interface{}{
"endpoint": config.MinioEndpoint,
"bucket": config.MinioBucket,
"error": err.Error(),
})
return nil, fmt.Errorf("ai_summarizer: MinIO connection check failed: %w", err)
}
log.Info("AI summarizer MinIO connection verified", map[string]interface{}{
"endpoint": config.MinioEndpoint,
"bucket": config.MinioBucket,
})
return storage, nil
}
// verifyConnection checks that MinIO is reachable and the configured bucket exists.
func (s *StorageClient) verifyConnection(ctx context.Context) error {
exists, err := s.client.BucketExists(ctx, s.config.MinioBucket)
if err != nil {
return fmt.Errorf("bucket exists check for %q: %w", s.config.MinioBucket, err)
}
if !exists {
return fmt.Errorf("%w: %q", ErrBucketNotFound, s.config.MinioBucket)
}
buckets, err := s.client.ListBuckets(ctx)
if err != nil {
s.logger.Warn("MinIO bucket exists but ListBuckets failed (credentials may be bucket-scoped)", map[string]interface{}{
"endpoint": s.config.MinioEndpoint,
"bucket": s.config.MinioBucket,
"error": err.Error(),
})
return nil
}
names := make([]string, 0, len(buckets))
for _, b := range buckets {
names = append(names, b.Name)
}
s.logger.Debug("MinIO ListBuckets succeeded", map[string]interface{}{
"endpoint": s.config.MinioEndpoint,
"bucket_count": len(names),
"target_bucket": s.config.MinioBucket,
})
return nil
}
// Ping checks that MinIO is still reachable and the configured bucket exists.
func (s *StorageClient) Ping(ctx context.Context) error {
return s.verifyConnection(ctx)
}
// resolveNoticeStoragePrefix locates the MinIO folder for a notice. It first
// checks PROC_<contractFolderID>/<notice>/ using known id variants, then falls
// back to a bucket search by notice publication id when the stored contract
// folder id does not match the AI pipeline folder (common in local testing).
func (s *StorageClient) resolveNoticeStoragePrefix(ctx context.Context, contractFolderID, noticePublicationID string) (string, error) {
for _, prefix := range noticeObjectPrefixes(contractFolderID, noticePublicationID) {
_, err := s.client.StatObject(ctx, s.config.MinioBucket, prefix+"tender.json", minio.StatObjectOptions{})
if err == nil {
return prefix, nil
}
if !isMinioNotFound(err) {
return "", fmt.Errorf("ai_summarizer: stat tender.json under %s: %w", prefix, err)
}
}
for _, noticeKey := range noticeKeyVariants(noticePublicationID) {
marker := "/" + noticeKey + "/"
var resolved string
objectCh := s.client.ListObjects(ctx, s.config.MinioBucket, minio.ListObjectsOptions{
Prefix: procedurePrefix,
Recursive: true,
})
for object := range objectCh {
if object.Err != nil {
return "", fmt.Errorf("ai_summarizer: search notice folder: %w", object.Err)
}
if idx := strings.Index(object.Key, marker); idx >= 0 {
resolved = object.Key[:idx+len(marker)]
break
}
}
if resolved != "" {
s.logger.Info("Resolved MinIO notice folder by notice publication id", map[string]interface{}{
"notice_id": noticePublicationID,
"requested_contract_folder_id": contractFolderID,
"resolved_prefix": resolved,
})
return resolved, nil
}
}
s.logger.Debug("No MinIO folder found for notice", map[string]interface{}{
"contract_folder_id": contractFolderID,
"notice_id": noticePublicationID,
})
return "", ErrObjectNotFound
}
// GetSummaryFromStorage fetches the
// PROC_<contractFolderID>/<noticePublicationID>/tender.json file from the AI
// service's MinIO bucket, parses it, and returns the overall_summary string.
//
// It returns:
// - (summary, nil) on success
// - ("", ErrObjectNotFound) if the tender.json does not exist (404)
// - ("", ErrSummaryNotReady) if the file exists but summarize_status.done is false
// - ("", ErrNoOverallSummary) if done is true but overall_summary is empty/null
// - ("", error) for any other failure
func (s *StorageClient) GetSummaryFromStorage(ctx context.Context, contractFolderID, noticePublicationID string) (string, error) {
if strings.TrimSpace(contractFolderID) == "" {
return "", fmt.Errorf("ai_summarizer: contractFolderID is required")
}
if strings.TrimSpace(noticePublicationID) == "" {
return "", fmt.Errorf("ai_summarizer: noticePublicationID is required")
}
s.logger.Debug("Fetching tender.json from MinIO", map[string]interface{}{
"bucket": s.config.MinioBucket,
"contract_folder_id": contractFolderID,
"notice_id": noticePublicationID,
})
prefix, err := s.resolveNoticeStoragePrefix(ctx, contractFolderID, noticePublicationID)
if err != nil {
return "", err
}
return s.getSummaryFromObjectKey(ctx, prefix+"tender.json", contractFolderID, noticePublicationID)
}
// getSummaryFromObjectKey reads tender.json at objectKey and returns overall_summary when ready.
// contractFolderID and noticePublicationID are used for logging only (may be empty when unknown).
func (s *StorageClient) getSummaryFromObjectKey(ctx context.Context, objectKey, contractFolderID, noticePublicationID string) (string, error) {
object, err := s.client.GetObject(ctx, s.config.MinioBucket, objectKey, minio.GetObjectOptions{})
if err != nil {
if isMinioNotFound(err) {
s.logger.Debug("MinIO object not found", map[string]interface{}{
"bucket": s.config.MinioBucket,
"object_key": objectKey,
})
return "", ErrObjectNotFound
}
var minioErr minio.ErrorResponse
if errors.As(err, &minioErr) && minioErr.StatusCode == http.StatusForbidden {
return "", fmt.Errorf("ai_summarizer: access denied to bucket %s: %w", s.config.MinioBucket, err)
}
return "", fmt.Errorf("ai_summarizer: failed to get object: %w", err)
}
defer object.Close()
data, err := io.ReadAll(object)
if err != nil {
if isMinioNotFound(err) {
s.logger.Debug("MinIO object not found on read", map[string]interface{}{
"bucket": s.config.MinioBucket,
"object_key": objectKey,
})
return "", ErrObjectNotFound
}
return "", fmt.Errorf("ai_summarizer: failed to read tender.json: %w", err)
}
var tenderJSON TenderJSON
if err := json.Unmarshal(data, &tenderJSON); err != nil {
return "", fmt.Errorf("ai_summarizer: failed to parse tender.json: %w", err)
}
done, overall := tenderJSON.resolved()
if !done {
return "", ErrSummaryNotReady
}
if overall == nil || strings.TrimSpace(*overall) == "" {
return "", ErrNoOverallSummary
}
summaryText := strings.TrimSpace(*overall)
s.logger.Info("Retrieved overall_summary from storage", map[string]interface{}{
"contract_folder_id": contractFolderID,
"notice_id": noticePublicationID,
"object_key": objectKey,
"summary_length": len(summaryText),
})
return summaryText, nil
}
// IsSummaryReady is a convenience method that checks whether the tender.json
// exists and the summarize_status.done flag is true, without returning the
// actual summary text.
func (s *StorageClient) IsSummaryReady(ctx context.Context, contractFolderID, noticePublicationID string) (bool, error) {
if strings.TrimSpace(contractFolderID) == "" || strings.TrimSpace(noticePublicationID) == "" {
return false, fmt.Errorf("ai_summarizer: contractFolderID and noticePublicationID are required")
}
prefix, err := s.resolveNoticeStoragePrefix(ctx, contractFolderID, noticePublicationID)
if errors.Is(err, ErrObjectNotFound) {
return false, nil
}
if err != nil {
return false, err
}
objectPath := prefix + "tender.json"
object, err := s.client.GetObject(ctx, s.config.MinioBucket, objectPath, minio.GetObjectOptions{})
if err != nil {
if isMinioNotFound(err) {
return false, nil
}
return false, fmt.Errorf("ai_summarizer: failed to check summary readiness: %w", err)
}
data, readErr := io.ReadAll(object)
closeErr := object.Close()
if readErr != nil {
if isMinioNotFound(readErr) {
return false, nil
}
return false, fmt.Errorf("ai_summarizer: failed to read tender.json: %w", readErr)
}
if closeErr != nil {
s.logger.Warn("Failed to close tender.json object after readiness check", map[string]interface{}{
"object_key": objectPath,
"error": closeErr.Error(),
})
}
var tenderJSON TenderJSON
if err := json.Unmarshal(data, &tenderJSON); err != nil {
return false, fmt.Errorf("ai_summarizer: failed to parse tender.json: %w", err)
}
done, _ := tenderJSON.resolved()
return done, nil
}
// ListDocuments lists document files stored under
// PROC_<contractFolderID>/<noticePublicationID>/. Non-document artefacts
// (tender.json, AI cache files, etc.) are filtered out.
func (s *StorageClient) ListDocuments(ctx context.Context, contractFolderID, noticePublicationID string) ([]StoredDocument, error) {
if strings.TrimSpace(contractFolderID) == "" {
return nil, fmt.Errorf("ai_summarizer: contractFolderID is required")
}
if strings.TrimSpace(noticePublicationID) == "" {
return nil, fmt.Errorf("ai_summarizer: noticePublicationID is required")
}
prefix, err := s.resolveNoticeStoragePrefix(ctx, contractFolderID, noticePublicationID)
if errors.Is(err, ErrObjectNotFound) {
s.logger.Info("Listed tender documents from storage", map[string]interface{}{
"contract_folder_id": contractFolderID,
"notice_id": noticePublicationID,
"documents_count": 0,
})
return nil, nil
}
if err != nil {
return nil, err
}
prefixes := []string{prefix}
s.logger.Debug("Listing tender documents from MinIO", map[string]interface{}{
"bucket": s.config.MinioBucket,
"prefixes": prefixes,
})
documents := make([]StoredDocument, 0)
seenObjects := make(map[string]struct{})
for _, prefix := range prefixes {
objectCh := s.client.ListObjects(ctx, s.config.MinioBucket, minio.ListObjectsOptions{
Prefix: prefix,
Recursive: true,
})
for object := range objectCh {
if object.Err != nil {
return nil, fmt.Errorf("ai_summarizer: failed to list documents under %s: %w", prefix, object.Err)
}
if !isTenderDocumentObject(prefix, object.Key) {
continue
}
if _, dup := seenObjects[object.Key]; dup {
continue
}
seenObjects[object.Key] = struct{}{}
documentName := filepath.Base(object.Key)
documents = append(documents, StoredDocument{
ObjectName: object.Key,
DocumentName: documentName,
Size: object.Size,
LastModified: object.LastModified.Unix(),
DocumentType: documentType(documentName),
})
}
}
s.logger.Info("Listed tender documents from storage", map[string]interface{}{
"contract_folder_id": contractFolderID,
"notice_id": noticePublicationID,
"prefixes": prefixes,
"documents_count": len(documents),
})
return documents, nil
}
// isTenderDocumentObject reports whether an object key under the given prefix
// represents a downloadable tender document (PDF/DOCX/...) rather than an
// internal artefact such as tender.json or a cache file.
func isTenderDocumentObject(prefix, objectName string) bool {
if objectName == "" || strings.HasSuffix(objectName, "/") {
return false
}
relativePath := strings.TrimPrefix(objectName, prefix)
if relativePath == "" || relativePath == objectName {
// objectName did not start with prefix; ignore.
return false
}
filename := strings.ToLower(filepath.Base(relativePath))
switch filename {
case "tender.json", "ai_cache.db":
return false
}
switch strings.ToLower(filepath.Ext(filename)) {
case ".json", ".db", ".sqlite", ".sqlite3":
return false
default:
return true
}
}
// DownloadDocument opens a document object from the configured MinIO bucket.
func (s *StorageClient) DownloadDocument(ctx context.Context, objectName string) (io.ReadCloser, error) {
objectName = strings.TrimSpace(objectName)
if objectName == "" {
return nil, fmt.Errorf("ai_summarizer: objectName is required")
}
object, err := s.client.GetObject(ctx, s.config.MinioBucket, objectName, minio.GetObjectOptions{})
if err != nil {
var minioErr minio.ErrorResponse
if errors.As(err, &minioErr) && minioErr.StatusCode == http.StatusNotFound {
return nil, ErrObjectNotFound
}
return nil, fmt.Errorf("ai_summarizer: failed to get document object: %w", err)
}
return object, nil
}
// GetAllSummaries walks the MinIO bucket for tender.json objects. It first lists
// under the PROC_ prefix; if none match, it falls back to a full-bucket recursive
// scan (slower on large buckets) so alternate key layouts still work.
func (s *StorageClient) GetAllSummaries(ctx context.Context) ([]TenderSummaryItem, error) {
results := make([]TenderSummaryItem, 0)
seen := make(map[string]struct{})
appendFromPrefix := func(prefix string) {
s.logger.Debug("Listing tender.json files from MinIO", map[string]interface{}{
"bucket": s.config.MinioBucket,
"prefix": prefix,
})
objectCh := s.client.ListObjects(ctx, s.config.MinioBucket, minio.ListObjectsOptions{
Prefix: prefix,
Recursive: true,
})
for object := range objectCh {
if object.Err != nil {
s.logger.Error("Error listing MinIO objects", map[string]interface{}{
"prefix": prefix,
"error": object.Err.Error(),
})
continue
}
key := object.Key
if !strings.HasSuffix(key, "/tender.json") {
continue
}
if _, dup := seen[key]; dup {
continue
}
contractFolderID, noticePublicationID, ok := parseTenderJSONKey(key)
if !ok {
s.logger.Debug("Skipping unparseable tender.json key", map[string]interface{}{
"key": key,
})
continue
}
seen[key] = struct{}{}
summary, err := s.GetSummaryFromStorage(ctx, contractFolderID, noticePublicationID)
if errors.Is(err, ErrObjectNotFound) {
summary, err = s.getSummaryFromObjectKey(ctx, key, contractFolderID, noticePublicationID)
}
if err != nil {
switch {
case errors.Is(err, ErrSummaryNotReady):
results = append(results, TenderSummaryItem{
ContractFolderID: contractFolderID,
NoticeID: noticePublicationID,
Done: false,
})
case errors.Is(err, ErrNoOverallSummary):
// File exists and pipeline reports done, but no summary text yet (or empty).
results = append(results, TenderSummaryItem{
ContractFolderID: contractFolderID,
NoticeID: noticePublicationID,
OverallSummary: "",
Done: true,
})
default:
s.logger.Debug("Skipping tender summary", map[string]interface{}{
"object_key": key,
"contract_folder_id": contractFolderID,
"notice_id": noticePublicationID,
"error": err.Error(),
})
}
continue
}
results = append(results, TenderSummaryItem{
ContractFolderID: contractFolderID,
NoticeID: noticePublicationID,
OverallSummary: summary,
Done: true,
})
}
}
appendFromPrefix(procedurePrefix)
if len(seen) == 0 {
s.logger.Info("No tender.json keys under PROC_; listing entire bucket for */tender.json (may be slower on large buckets)", map[string]interface{}{
"bucket": s.config.MinioBucket,
})
appendFromPrefix("")
}
s.logger.Info("Retrieved all summaries from storage", map[string]interface{}{
"total_count": len(results),
"completed_count": countDone(results),
})
return results, nil
}
// parseTenderJSONKey splits an object key ending in /tender.json into contract folder
// id and notice publication id. Accepts PROC_/proc_ prefix or plain <folder>/<notice>/tender.json.
func parseTenderJSONKey(key string) (contractFolderID, noticePublicationID string, ok bool) {
key = strings.TrimPrefix(strings.TrimSpace(key), "/")
if !strings.HasSuffix(key, "/tender.json") {
return "", "", false
}
trimmed := strings.TrimSuffix(key, "/tender.json")
if trimmed == "" {
return "", "", false
}
last := strings.LastIndex(trimmed, "/")
if last < 0 {
return "", "", false
}
noticePublicationID = strings.TrimSpace(trimmed[last+1:])
parent := strings.TrimSpace(trimmed[:last])
if noticePublicationID == "" || parent == "" {
return "", "", false
}
contractFolderID = stripLeadingProcPrefix(parent)
if contractFolderID == "" {
return "", "", false
}
return contractFolderID, noticePublicationID, true
}
func stripLeadingProcPrefix(parent string) string {
if len(parent) >= 5 && strings.EqualFold(parent[:5], "proc_") {
return strings.TrimSpace(parent[5:])
}
return parent
}
// countDone counts how many items have Done=true
func countDone(items []TenderSummaryItem) int {
count := 0
for _, item := range items {
if item.Done {
count++
}
}
return count
}
func documentType(filename string) string {
extension := strings.TrimPrefix(strings.ToLower(filepath.Ext(filename)), ".")
if extension == "" {
return "unknown"
}
return extension
}