Diffstat (limited to 'internal/blog')
| -rw-r--r-- | internal/blog/b2.go | 31 | ||||
| -rw-r--r-- | internal/blog/client.go | 1 | ||||
| -rw-r--r-- | internal/blog/index.go | 103 | ||||
| -rw-r--r-- | internal/blog/s3.go | 194 |
4 files changed, 64 insertions, 265 deletions
diff --git a/internal/blog/b2.go b/internal/blog/b2.go index f2ac890..0caa0ba 100644 --- a/internal/blog/b2.go +++ b/internal/blog/b2.go @@ -2,7 +2,6 @@ package blog import ( "context" - "encoding/json" "fmt" "slices" "strings" @@ -38,20 +37,6 @@ func NewB2Client(cfg *config.StorageConfig) (Client, error) { return &B2Client{b2cl: b2cl, bucket: bucket, prefix: b2cfg.Prefix}, nil } -func (c *B2Client) GetMedleys() ([]MedleyEntry, error) { - idxRaw, err := c.readAll(MedleysIndexFileName) - if err != nil { - return nil, fmt.Errorf("read %s: %w", MedleysIndexFileName, err) - } - - var idx []MedleyEntry - if err := json.Unmarshal(idxRaw, &idx); err != nil { - return nil, fmt.Errorf("unmarshal %s: %w", MedleysIndexFileName, err) - } - - return idx, nil -} - func (c *B2Client) Scan(prefix string) ([]*Page, error) { filePaths := []*Page{} @@ -85,15 +70,17 @@ func (c *B2Client) Scan(prefix string) ([]*Page, error) { return nil, fmt.Errorf("failed to parse published time metadata field: %w", err) } - link := obj.Name() - fileName := link[strings.LastIndex(link, "/")+1 : strings.LastIndex(link, ".")] + linkParts := strings.Split(obj.Name(), "/") + nameParts := strings.Split(linkParts[len(linkParts)-1], ".") + fileName := strings.Join(nameParts[:len(linkParts)-1], ".") + tags := strings.Split(attrs.Info["tags"], ",") slices.Sort(tags) - lang, _ := strings.CutPrefix(link[0:strings.Index(link, "/")], c.prefix) + lang, _ := strings.CutPrefix(linkParts[0], c.prefix) filePaths = append(filePaths, &Page{ - Link: link, + Link: obj.Name(), FileName: fileName, Lang: lang, ModifiedTime: attrs.LastModified, @@ -117,11 +104,7 @@ func (c *B2Client) Scan(prefix string) ([]*Page, error) { } func (c *B2Client) ReadAll(path string) ([]byte, error) { - return c.readAll(c.prefix + path) -} - -func (c *B2Client) readAll(path string) ([]byte, error) { - obj := c.bucket.Object(path) + obj := c.bucket.Object(c.prefix + path) if obj == nil { return nil, fmt.Errorf("failed to reference object in B2 bucket") } diff --git a/internal/blog/client.go b/internal/blog/client.go index 53eb0fd..56bbee4 100644 --- a/internal/blog/client.go +++ b/internal/blog/client.go @@ -17,7 +17,6 @@ type Page struct { type Client interface { Scan(prefix string) ([]*Page, error) - GetMedleys() ([]MedleyEntry, error) ReadAll(path string) ([]byte, error) ReadFrontmatter(path string) (metadata *frontmatter.Metadata, markdown []byte, err error) } diff --git a/internal/blog/index.go b/internal/blog/index.go deleted file mode 100644 index a09b59e..0000000 --- a/internal/blog/index.go +++ /dev/null @@ -1,103 +0,0 @@ -package blog - -import ( - "encoding/json" - "fmt" - "time" - - "github.com/SayaAndy/saya-today-web/internal/frontmatter" -) - -const IndexFileName = "index.json" -const MedleysIndexFileName = "medleys.json" - -const IndexSchemaVersion = 2 - -type IndexEntry struct { - Link string `json:"link"` - ModifiedTime time.Time `json:"modifiedTime"` - Title string `json:"title"` - ShortDescription string `json:"shortDescription"` - ActionDate string `json:"actionDate"` - PublishedTime time.Time `json:"publishedTime"` - Thumbnail string `json:"thumbnail"` - Tags []string `json:"tags"` - Geolocation string `json:"geolocation"` - Medley string `json:"medley,omitempty"` - MedleyPart int `json:"medleyPart,omitempty"` -} - -func (e IndexEntry) Metadata() *frontmatter.Metadata { - return &frontmatter.Metadata{ - Title: e.Title, - ShortDescription: e.ShortDescription, - ActionDate: e.ActionDate, - PublishedTime: e.PublishedTime, - Thumbnail: e.Thumbnail, - Tags: e.Tags, - Geolocation: e.Geolocation, - Medley: e.Medley, - MedleyPart: e.MedleyPart, - } -} - -type IndexV2Category struct { - GeneratedAt time.Time `json:"generatedAt"` - Pages map[string]IndexEntry `json:"pages"` -} - -type IndexV1Category struct { - GeneratedAt time.Time `json:"generatedAt"` - Pages []IndexEntry `json:"pages"` -} - -type Index struct { - SchemaVersion int `json:"schemaVersion"` - GeneratedAt time.Time `json:"generatedAt"` - Categories any `json:"categories"` -} - -func (idx *Index) UnmarshalJSON(data []byte) error { - var tmp struct { - SchemaVersion int `json:"schemaVersion"` - GeneratedAt time.Time `json:"generatedAt"` - Categories json.RawMessage `json:"categories"` - } - - if err := json.Unmarshal(data, &tmp); err != nil { - return err - } - - idx.SchemaVersion = tmp.SchemaVersion - idx.GeneratedAt = tmp.GeneratedAt - - switch tmp.SchemaVersion { - case 1: - var categories map[string]*IndexV1Category - if err := json.Unmarshal(tmp.Categories, &categories); err != nil { - return fmt.Errorf("unmarshal map[string]*IndexV1Category: %w", err) - } - idx.Categories = &categories - case 2: - var categories map[string]*IndexV2Category - if err := json.Unmarshal(tmp.Categories, &categories); err != nil { - return fmt.Errorf("unmarshal map[string]*IndexV2Category: %w", err) - } - idx.Categories = &categories - default: - return fmt.Errorf("unsupported index version: %d", tmp.SchemaVersion) - } - - return nil -} - -type MedleyEntry struct { - Codename string `json:"codename"` - Localnames map[string]string `json:"localnames"` - Content []string `json:"content"` -} - -type MedleyPageEntry struct { - Codename string `json:"codename"` - Position int `json:"position"` -} diff --git a/internal/blog/s3.go b/internal/blog/s3.go index 3c1a932..86a6463 100644 --- a/internal/blog/s3.go +++ b/internal/blog/s3.go @@ -1,16 +1,15 @@ package blog import ( - "bytes" "context" - "encoding/json" "fmt" "io" + "slices" "strings" + "time" "github.com/SayaAndy/saya-today-web/config" "github.com/SayaAndy/saya-today-web/internal/frontmatter" - "github.com/SayaAndy/saya-today-web/l10n" "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" @@ -51,7 +50,6 @@ func NewS3Client(cfg *config.StorageConfig) (Client, error) { } s3Opts = append(s3Opts, func(o *s3.Options) { o.UsePathStyle = s3cfg.UsePathStyle - o.DisableLogOutputChecksumValidationSkipped = true }) s3cl := s3.NewFromConfig(awsCfg, s3Opts...) @@ -59,120 +57,76 @@ func NewS3Client(cfg *config.StorageConfig) (Client, error) { return &S3Client{s3cfg.Prefix, s3cfg.BucketName, s3cl}, nil } -func (c *S3Client) GetMedleys() ([]MedleyEntry, error) { - idxRaw, err := c.readAll(MedleysIndexFileName) - if err != nil { - return nil, fmt.Errorf("read %s: %w", MedleysIndexFileName, err) - } - - var idx []MedleyEntry - if err := json.Unmarshal(idxRaw, &idx); err != nil { - return nil, fmt.Errorf("unmarshal %s: %w", MedleysIndexFileName, err) - } - - return idx, nil -} - func (c *S3Client) Scan(prefix string) ([]*Page, error) { - out, err := c.s3cl.GetObject(context.Background(), &s3.GetObjectInput{ - Bucket: aws.String(c.bucketName), - Key: aws.String(IndexFileName), - }) - if err != nil { - return nil, fmt.Errorf("get %s: %w", IndexFileName, err) - } - defer out.Body.Close() + pages := []*Page{} - raw, err := io.ReadAll(out.Body) - if err != nil { - return nil, fmt.Errorf("read %s: %w", IndexFileName, err) + fullPrefix := c.prefix + prefix + input := &s3.ListObjectsV2Input{ + Bucket: aws.String(c.bucketName), + Prefix: aws.String(fullPrefix), } - var idx Index - if err := json.Unmarshal(raw, &idx); err != nil { - return nil, fmt.Errorf("unmarshal %s: %w", IndexFileName, err) - } + paginator := s3.NewListObjectsV2Paginator(c.s3cl, input) - wantLang := "" - if i := strings.Index(prefix, "/"); i > 0 { - wantLang = prefix[:i] - } + for paginator.HasMorePages() { + output, err := paginator.NextPage(context.Background()) + if err != nil { + return nil, fmt.Errorf("list S3 objects: %w", err) + } - fullPrefix := c.prefix + prefix - pages := make([]*Page, 0) + for _, obj := range output.Contents { + key := aws.ToString(obj.Key) - switch idx.SchemaVersion { - case 1: - for catKey, cat := range *idx.Categories.(*map[string]*IndexV1Category) { - lang, ok := strings.CutPrefix(catKey, c.prefix) - if !ok { - continue - } - if wantLang != "" && wantLang != lang { + if !strings.HasSuffix(key, ".md") { continue } - for _, e := range cat.Pages { - if !strings.HasPrefix(e.Link, fullPrefix) { - continue - } - fileName := e.Link[strings.LastIndex(e.Link, "/")+1 : strings.LastIndex(e.Link, ".")] - pages = append(pages, &Page{ - Link: e.Link, - FileName: fileName, - Lang: lang, - ModifiedTime: e.ModifiedTime, - Metadata: &frontmatter.Metadata{ - Title: e.Title, - ShortDescription: e.ShortDescription, - ActionDate: e.ActionDate, - PublishedTime: e.PublishedTime, - Thumbnail: e.Thumbnail, - Tags: e.Tags, - Geolocation: e.Geolocation, - Medley: e.Medley, - MedleyPart: e.MedleyPart, - }, - }) + + head, err := c.s3cl.HeadObject(context.Background(), &s3.HeadObjectInput{ + Bucket: aws.String(c.bucketName), + Key: aws.String(key), + }) + if err != nil { + return nil, fmt.Errorf("head S3 object %s: %w", key, err) } - } - case 2: - for catKey, cat := range *idx.Categories.(*map[string]*IndexV2Category) { - lang, ok := strings.CutPrefix(catKey, c.prefix) - if !ok { + + if head.ContentType == nil || !strings.Contains(*head.ContentType, "text/markdown") { continue } - if wantLang != "" && wantLang != lang { + + meta := head.Metadata + if meta["title"] == "" { continue } - for codename, e := range cat.Pages { - if !strings.HasPrefix(e.Link, fullPrefix) { - continue - } - pages = append(pages, &Page{ - Link: e.Link, - FileName: codename, - Lang: lang, - ModifiedTime: e.ModifiedTime, - Metadata: &frontmatter.Metadata{ - Title: e.Title, - ShortDescription: e.ShortDescription, - ActionDate: e.ActionDate, - PublishedTime: e.PublishedTime, - Thumbnail: e.Thumbnail, - Tags: e.Tags, - Geolocation: e.Geolocation, - Medley: e.Medley, - MedleyPart: e.MedleyPart, - }, - }) + + publishedTime, err := time.Parse(time.RFC3339, meta["published-time"]) + if err != nil { + return nil, fmt.Errorf("failed to parse published time metadata field: %w", err) } - } - } - medleys, _ := c.GetMedleys() - for _, medley := range medleys { - for locale, localname := range medley.Localnames { - l10n.T.SetPath(localname, true, locale, "Medleys", medley.Codename) + linkParts := strings.Split(key, "/") + nameParts := strings.Split(linkParts[len(linkParts)-1], ".") + fileName := strings.Join(nameParts[:len(nameParts)-1], ".") + + tags := strings.Split(meta["tags"], ",") + slices.Sort(tags) + + lang, _ := strings.CutPrefix(linkParts[0], c.prefix) + + pages = append(pages, &Page{ + Link: key, + FileName: fileName, + Lang: lang, + ModifiedTime: aws.ToTime(obj.LastModified), + Metadata: &frontmatter.Metadata{ + Title: meta["title"], + ShortDescription: meta["short-description"], + ActionDate: meta["action-date"], + PublishedTime: publishedTime, + Thumbnail: meta["thumbnail"], + Tags: tags, + Geolocation: meta["geolocation"], + }, + }) } } @@ -180,13 +134,9 @@ func (c *S3Client) Scan(prefix string) ([]*Page, error) { } func (c *S3Client) ReadAll(path string) ([]byte, error) { - return c.readAll(c.prefix + path) -} - -func (c *S3Client) readAll(path string) ([]byte, error) { output, err := c.s3cl.GetObject(context.Background(), &s3.GetObjectInput{ Bucket: aws.String(c.bucketName), - Key: aws.String(path), + Key: aws.String(c.prefix + path), }) if err != nil { return nil, fmt.Errorf("get S3 object: %w", err) @@ -202,40 +152,10 @@ func (c *S3Client) readAll(path string) ([]byte, error) { } func (c *S3Client) ReadFrontmatter(path string) (metadata *frontmatter.Metadata, markdown []byte, err error) { - idxRaw, err := c.readAll(IndexFileName) - if err != nil { - return nil, nil, fmt.Errorf("read %s: %w", IndexFileName, err) - } - - var idx Index - if err := json.Unmarshal(idxRaw, &idx); err != nil { - return nil, nil, fmt.Errorf("unmarshal %s: %w", IndexFileName, err) - } - contentBytes, err := c.ReadAll(path) if err != nil { return nil, nil, fmt.Errorf("failed to read file for frontmatter parsing: %w", err) } - switch idx.SchemaVersion { - case 1: - return frontmatter.ParseFrontmatter(contentBytes) - case 2: - fullPath := c.prefix + path - page := (*idx.Categories.(*map[string]*IndexV2Category))[fullPath[:strings.LastIndex(fullPath, "/")]].Pages[fullPath[strings.LastIndex(fullPath, "/")+1:strings.LastIndex(fullPath, ".")]] - metadata = page.Metadata() - - if !bytes.HasPrefix(contentBytes, []byte("---\n")) { - return metadata, contentBytes, nil - } - - end := bytes.Index(contentBytes[4:], []byte("\n---\n")) - if end == -1 { - return metadata, contentBytes, nil - } - - return metadata, contentBytes[end+9:], nil - } - return frontmatter.ParseFrontmatter(contentBytes) } |