diff options
| author | 2026-08-01 19:58:57 +0700 | |
|---|---|---|
| committer | 2026-08-01 19:58:57 +0700 | |
| commit | 684d0b6a57d2eb79730ade63baccdc6e59bebc29 (patch) | |
| tree | 956c89c81965b8c974c8497e3719d0b16d332512 /internal/storage/s3.go | |
| parent | 129e48683f6fc0615b606a737f2afcd50538bb2e (diff) | |
| download | articlator-main.tar.gz articlator-main.zip | |
feat: repurpose metadata parser as article transcoder for saya.uz andmain
telegram
Diffstat (limited to 'internal/storage/s3.go')
| -rw-r--r-- | internal/storage/s3.go | 377 |
1 files changed, 271 insertions, 106 deletions
diff --git a/internal/storage/s3.go b/internal/storage/s3.go index 5e9bcd2..5f3acc7 100644 --- a/internal/storage/s3.go +++ b/internal/storage/s3.go @@ -3,31 +3,40 @@ package storage import ( "bytes" "context" + "encoding/json" + "errors" "fmt" "io" + "log/slog" + "maps" + "net/url" + "slices" "strconv" "strings" + "sync" "time" "github.com/aws/aws-sdk-go-v2/aws" awsconfig "github.com/aws/aws-sdk-go-v2/config" "github.com/aws/aws-sdk-go-v2/credentials" "github.com/aws/aws-sdk-go-v2/service/s3" + s3types "github.com/aws/aws-sdk-go-v2/service/s3/types" "github.com/SayaAndy/saya-today-article-metadata-add/config" "github.com/SayaAndy/saya-today-article-metadata-add/internal/frontmatter" ) +const s3IndexConcurrency = 32 + var _ StorageClient = &S3StorageClient{} type S3StorageClient struct { - prefix string - bucket string - client *s3.Client - draftModeCfg *config.DraftModeConfig + prefix string + bucket string + client *s3.Client } -func NewS3StorageClient(cfg *config.StorageConfig, draftModeCfg *config.DraftModeConfig) (StorageClient, error) { +func NewS3StorageClient(cfg *config.StorageConfig) (StorageClient, error) { if cfg.Type != "s3" { return nil, fmt.Errorf("invalid storage type for S3StorageClient") } @@ -59,166 +68,322 @@ func NewS3StorageClient(cfg *config.StorageConfig, draftModeCfg *config.DraftMod client := s3.NewFromConfig(awsCfg, s3Opts...) - draftModeCfgCopy := *draftModeCfg - return &S3StorageClient{ - client: client, - bucket: s3cfg.BucketName, - prefix: s3cfg.Prefix, - draftModeCfg: &draftModeCfgCopy, + client: client, + bucket: s3cfg.BucketName, + prefix: s3cfg.Prefix, }, nil } -func (sc *S3StorageClient) Scan() ([]string, error) { - var filePaths []string +func (sc *S3StorageClient) GetMetadata(key string) (map[string]string, error) { + head, err := sc.client.HeadObject(context.Background(), &s3.HeadObjectInput{ + Bucket: aws.String(sc.bucket), + Key: aws.String(sc.prefix + key), + }) + if err != nil { + var nsk *s3types.NoSuchKey + var nf *s3types.NotFound + if errors.As(err, &nsk) || errors.As(err, &nf) { + return map[string]string{}, nil + } + return nil, fmt.Errorf("head S3 object %q: %w", sc.prefix+key, err) + } + return head.Metadata, nil +} + +func (sc *S3StorageClient) Put(key string, content []byte, metadata *frontmatter.Metadata) error { + geolocationParts := strings.Split(metadata.Geolocation, " ") + if (len(geolocationParts) == 1 && geolocationParts[0] != "") || len(geolocationParts) >= 4 { + return fmt.Errorf("invalid geolocation format, expecting '{x} {y} [areaError]' or an empty string") + } + if len(geolocationParts) >= 2 { + if _, err := strconv.ParseFloat(geolocationParts[0], 64); err != nil { + return fmt.Errorf("invalid geolocation parameter, expected float for X: %w", err) + } + if _, err := strconv.ParseFloat(geolocationParts[1], 64); err != nil { + return fmt.Errorf("invalid geolocation parameter, expected float for Y: %w", err) + } + } + if len(geolocationParts) == 3 { + if _, err := strconv.ParseFloat(geolocationParts[2], 64); err != nil { + return fmt.Errorf("invalid geolocation parameter, expected float for area error: %w", err) + } + } + + s3Metadata := map[string]string{ + "title": url.QueryEscape(metadata.Title), + "short-description": url.QueryEscape(metadata.ShortDescription), + "action-date": metadata.ActionDate, + "published-time": metadata.PublishedTime.Format(time.RFC3339), + "thumbnail": url.QueryEscape(metadata.Thumbnail), + "tags": strings.Join(metadata.Tags, ","), + "geolocation": metadata.Geolocation, + } + + targetKey := sc.prefix + key + _, err := sc.client.PutObject(context.Background(), &s3.PutObjectInput{ + Bucket: aws.String(sc.bucket), + Key: aws.String(targetKey), + Body: bytes.NewReader(content), + ContentType: aws.String("text/markdown; charset=utf-8"), + Metadata: s3Metadata, + }) + if err != nil { + return fmt.Errorf("put S3 object %q: %w", targetKey, err) + } + + return nil +} + +func (sc *S3StorageClient) BuildIndex() error { + type candidate struct { + key string + codename string + lastModified time.Time + } + var candidates []candidate paginator := s3.NewListObjectsV2Paginator(sc.client, &s3.ListObjectsV2Input{ Bucket: aws.String(sc.bucket), Prefix: aws.String(sc.prefix), }) - for paginator.HasMorePages() { page, err := paginator.NextPage(context.Background()) if err != nil { - return nil, fmt.Errorf("list S3 objects: %w", err) + return fmt.Errorf("list S3 objects: %w", err) } - for _, obj := range page.Contents { - name := aws.ToString(obj.Key) - if !strings.HasSuffix(name, ".md") { + key := aws.ToString(obj.Key) + if key == IndexFileName { continue } - - if sc.draftModeCfg.Enabled && !strings.HasSuffix(name, sc.draftModeCfg.DraftSuffix) { + if !strings.HasSuffix(key, ".md") { continue } + codename := key[strings.LastIndex(key, "/")+1 : strings.LastIndex(key, ".")] + candidates = append(candidates, candidate{key, codename, aws.ToTime(obj.LastModified)}) + } + } - filePaths = append(filePaths, strings.TrimPrefix(name, sc.prefix)) + medleys, err := sc.scanMedleys() + if err != nil { + slog.Warn("skipped reading medleys due to an error", slog.String("error", err.Error())) + } + pageToMedleyMap := make(map[string]MedleyPageEntry) + for _, medley := range medleys { + for i, page := range medley.Content { + pageToMedleyMap[page] = MedleyPageEntry{medley.Codename, i} } } - return filePaths, nil -} + type result struct { + catKey string + codename string + entry IndexEntry + } + results := make([]*result, len(candidates)) + sem := make(chan struct{}, s3IndexConcurrency) + var wg sync.WaitGroup + var firstErr error + var errMu sync.Mutex -func (sc *S3StorageClient) GetReader(path string) (io.ReadCloser, int64, error) { - key := sc.prefix + path + for i, cand := range candidates { + sem <- struct{}{} + wg.Add(1) + go func() { + defer wg.Done() + defer func() { <-sem }() - out, err := sc.client.GetObject(context.Background(), &s3.GetObjectInput{ - Bucket: aws.String(sc.bucket), - Key: aws.String(key), - }) - if err != nil { - return nil, 0, fmt.Errorf("get S3 object %q: %w", key, err) - } + head, err := sc.client.HeadObject(context.Background(), &s3.HeadObjectInput{ + Bucket: aws.String(sc.bucket), + Key: aws.String(cand.key), + }) + if err != nil { + errMu.Lock() + if firstErr == nil { + firstErr = fmt.Errorf("head S3 object %s: %w", cand.key, err) + } + errMu.Unlock() + return + } - return out.Body, aws.ToInt64(out.ContentLength), nil -} + if head.ContentType == nil || !strings.Contains(*head.ContentType, "text/markdown") { + return + } + meta := head.Metadata + if meta["title"] == "" { + return + } -func (sc *S3StorageClient) WriteMetadata(path string, metadata *frontmatter.Metadata) error { - draftKey := sc.prefix + path + publishedTime, err := time.Parse(time.RFC3339, meta["published-time"]) + if err != nil { + slog.Warn("skip entry with bad published-time", slog.String("key", cand.key), slog.String("error", err.Error())) + return + } - getOut, err := sc.client.GetObject(context.Background(), &s3.GetObjectInput{ - Bucket: aws.String(sc.bucket), - Key: aws.String(draftKey), - }) - if err != nil { - return fmt.Errorf("get draft object %q: %w", draftKey, err) - } - content, err := io.ReadAll(getOut.Body) - getOut.Body.Close() - if err != nil { - return fmt.Errorf("read draft object %q: %w", draftKey, err) - } + catKey := cand.key[:strings.Index(cand.key, "/")] - headOut, err := sc.client.HeadObject(context.Background(), &s3.HeadObjectInput{ - Bucket: aws.String(sc.bucket), - Key: aws.String(draftKey), - }) - if err != nil { - return fmt.Errorf("head draft object %q: %w", draftKey, err) + tags := strings.Split(meta["tags"], ",") + slices.Sort(tags) + + title, _ := url.QueryUnescape(meta["title"]) + shortDescription, _ := url.QueryUnescape(meta["short-description"]) + thumbnail, _ := url.QueryUnescape(meta["thumbnail"]) + + medleyName, medleyPart := "", 0 + if medley, ok := pageToMedleyMap[cand.codename]; ok { + medleyName, medleyPart = medley.Codename, medley.Position + } + + results[i] = &result{ + catKey: catKey, + codename: cand.codename, + entry: IndexEntry{ + Link: cand.key, + ModifiedTime: cand.lastModified, + Title: title, + ShortDescription: shortDescription, + ActionDate: meta["action-date"], + PublishedTime: publishedTime, + Thumbnail: thumbnail, + Tags: tags, + Geolocation: meta["geolocation"], + Medley: medleyName, + MedleyPart: medleyPart, + }, + } + }() } - draftETag := aws.ToString(headOut.ETag) + wg.Wait() - geolocationParts := strings.Split(metadata.Geolocation, " ") - if (len(geolocationParts) == 1 && geolocationParts[0] != "") || len(geolocationParts) >= 4 { - return fmt.Errorf("invalid geolocation format, expecting '{x} {y} [areaError]' or an empty string") + if firstErr != nil { + return firstErr } - if len(geolocationParts) >= 2 { - if _, err := strconv.ParseFloat(geolocationParts[0], 64); err != nil { - return fmt.Errorf("invalid geolocation parameter, expected float for X: %w", err) - } - if _, err := strconv.ParseFloat(geolocationParts[1], 64); err != nil { - return fmt.Errorf("invalid geolocation parameter, expected float for Y: %w", err) + + now := time.Now().UTC() + fresh := make(map[string]*IndexV2Category) + for _, r := range results { + if r == nil { + continue } - } - if len(geolocationParts) == 3 { - if _, err := strconv.ParseFloat(geolocationParts[2], 64); err != nil { - return fmt.Errorf("invalid geolocation parameter, expected float for area error: %w", err) + if _, ok := fresh[r.catKey]; !ok { + fresh[r.catKey] = &IndexV2Category{ + Pages: make(map[string]IndexEntry), + } } + fresh[r.catKey].Pages[r.codename] = r.entry + fresh[r.catKey].GeneratedAt = now } - medley := "" - if metadata.Medley != "" { - medley = fmt.Sprintf("%s %d", metadata.Medley, metadata.MedleyPart) + merged, err := sc.loadAndMergeIndex(fresh) + if err != nil { + return err } - s3Metadata := map[string]string{ - "title": metadata.Title, - "short-description": metadata.ShortDescription, - "action-date": metadata.ActionDate, - "published-time": metadata.PublishedTime.Format(time.RFC3339), - "thumbnail": metadata.Thumbnail, - "tags": strings.Join(metadata.Tags, ","), - "geolocation": metadata.Geolocation, - "medley": medley, - "metadata-last-update-etag": draftETag, + idx := Index{ + SchemaVersion: IndexSchemaVersion, + GeneratedAt: now, + Categories: merged, } - targetKey := draftKey - if sc.draftModeCfg.Enabled { - prodPath := strings.TrimSuffix(path, sc.draftModeCfg.DraftSuffix) + sc.draftModeCfg.ProdSuffix - targetKey = sc.prefix + prodPath + body, err := json.Marshal(idx) + if err != nil { + return fmt.Errorf("marshal index: %w", err) } _, err = sc.client.PutObject(context.Background(), &s3.PutObjectInput{ Bucket: aws.String(sc.bucket), - Key: aws.String(targetKey), - Body: bytes.NewReader(content), - ContentType: aws.String("text/markdown; charset=utf-8"), - Metadata: s3Metadata, + Key: aws.String(IndexFileName), + Body: bytes.NewReader(body), + ContentType: aws.String("application/json; charset=utf-8"), }) if err != nil { - return fmt.Errorf("put S3 object %q: %w", targetKey, err) + return fmt.Errorf("put index %q: %w", IndexFileName, err) } + totalEntries := 0 + for _, c := range merged { + totalEntries += len(c.Pages) + } + slog.Info("wrote index", + slog.String("key", IndexFileName), + slog.Int("categories", len(merged)), + slog.Int("entries", totalEntries)) return nil } -func (sc *S3StorageClient) CompareDraftAndProd(path string) bool { - draftKey := sc.prefix + path - prodPath := strings.TrimSuffix(path, sc.draftModeCfg.DraftSuffix) + sc.draftModeCfg.ProdSuffix - prodKey := sc.prefix + prodPath - - draftHead, err := sc.client.HeadObject(context.Background(), &s3.HeadObjectInput{ +func (sc *S3StorageClient) scanMedleys() ([]MedleyEntry, error) { + out, err := sc.client.GetObject(context.Background(), &s3.GetObjectInput{ Bucket: aws.String(sc.bucket), - Key: aws.String(draftKey), + Key: aws.String(MedleysIndexFileName), }) if err != nil { - return false + return nil, fmt.Errorf("get %s: %w", MedleysIndexFileName, err) + } + defer out.Body.Close() + + raw, err := io.ReadAll(out.Body) + if err != nil { + return nil, fmt.Errorf("read %s: %w", MedleysIndexFileName, err) } - prodHead, err := sc.client.HeadObject(context.Background(), &s3.HeadObjectInput{ + var entries []MedleyEntry + if err := json.Unmarshal(raw, &entries); err != nil { + return nil, fmt.Errorf("unmarshal %s: %w", MedleysIndexFileName, err) + } + + return entries, nil +} + +func (sc *S3StorageClient) loadAndMergeIndex(fresh map[string]*IndexV2Category) (map[string]*IndexV2Category, error) { + merged := make(map[string]*IndexV2Category) + + out, err := sc.client.GetObject(context.Background(), &s3.GetObjectInput{ Bucket: aws.String(sc.bucket), - Key: aws.String(prodKey), + Key: aws.String(IndexFileName), }) - if err != nil { - return true + if err == nil { + defer out.Body.Close() + raw, readErr := io.ReadAll(out.Body) + if readErr != nil { + return nil, fmt.Errorf("read existing index: %w", readErr) + } + + var legacyIdx Index + if err := json.Unmarshal(raw, &legacyIdx); err != nil { + return nil, fmt.Errorf("unmarshal existing index: %w", err) + } + + switch legacyIdx.SchemaVersion { + case 1: + for k, v := range *legacyIdx.Categories.(*map[string]*IndexV1Category) { + pages := make(map[string]IndexEntry, len(v.Pages)) + for _, page := range v.Pages { + pages[page.Link[strings.LastIndex(page.Link, "/")+1:strings.LastIndex(page.Link, ".")]] = page + } + merged[k] = &IndexV2Category{ + GeneratedAt: v.GeneratedAt, + Pages: pages, + } + } + case 2: + maps.Copy(merged, *legacyIdx.Categories.(*map[string]*IndexV2Category)) + default: + slog.Warn("unknown index schema, discarding", slog.Int("schema_version", legacyIdx.SchemaVersion)) + } + } else { + var nsk *s3types.NoSuchKey + if !errors.As(err, &nsk) { + return nil, fmt.Errorf("get existing index: %w", err) + } } - lastUpdateETag, ok := prodHead.Metadata["metadata-last-update-etag"] - if !ok { - return true + for k := range merged { + if strings.HasPrefix(k, sc.prefix) { + delete(merged, k) + } } + maps.Copy(merged, fresh) - return aws.ToString(draftHead.ETag) != lastUpdateETag + return merged, nil } |