diff options
| author | 2025-07-15 17:36:47 +0700 | |
|---|---|---|
| committer | 2025-07-15 17:36:47 +0700 | |
| commit | 2d2138383af6d28c5b947c5a6c4989afa8f7ac56 (patch) | |
| tree | 6dfb9bf48bb640b29623cd0563c5c732afc196b0 | |
| parent | 373ec07b40178cb3b704aa83fc4fd53abfe279bc (diff) | |
| download | thumbnail-generator-2d2138383af6d28c5b947c5a6c4989afa8f7ac56.tar.gz thumbnail-generator-2d2138383af6d28c5b947c5a6c4989afa8f7ac56.zip | |
feat: first working prototype
| -rw-r--r-- | config/config.json | 2 | ||||
| -rw-r--r-- | internal/client/input/b2.go | 44 | ||||
| -rw-r--r-- | internal/client/input/input_client_interface.go | 15 | ||||
| -rw-r--r-- | internal/client/output/b2.go | 49 | ||||
| -rw-r--r-- | internal/client/output/output_client_interface.go | 20 | ||||
| -rw-r--r-- | internal/processor/processor_interface.go | 2 | ||||
| -rw-r--r-- | internal/processor/webp.go | 22 | ||||
| -rw-r--r-- | main.go | 79 |
8 files changed, 181 insertions, 52 deletions
diff --git a/config/config.json b/config/config.json index 298a03f..01b7727 100644 --- a/config/config.json +++ b/config/config.json @@ -21,7 +21,7 @@ "Config": { "Quality": 80, "Size": { - "MaxWidth": 1280, + "MaxWidth": 800, "MaxHeight": 0 } } diff --git a/internal/client/input/b2.go b/internal/client/input/b2.go index 67fb14e..54a6b09 100644 --- a/internal/client/input/b2.go +++ b/internal/client/input/b2.go @@ -5,6 +5,7 @@ import ( "fmt" "io" "slices" + "strconv" "strings" "github.com/Backblaze/blazer/b2" @@ -72,7 +73,7 @@ func (c *B2InputClient) Scan() ([]string, error) { } } - filePaths = append(filePaths, name) + filePaths = append(filePaths, strings.TrimPrefix(name, c.prefix)) } if err := iter.Err(); err != nil { @@ -82,8 +83,8 @@ func (c *B2InputClient) Scan() ([]string, error) { return filePaths, nil } -func (c *B2InputClient) ReadMetadata(path string) (map[string]string, error) { - obj := c.bucket.Object(path) +func (c *B2InputClient) ReadMetadata(path string) (*MetadataStruct, error) { + obj := c.bucket.Object(c.prefix + path) if obj == nil { return nil, fmt.Errorf("object not found in B2 bucket") } @@ -93,37 +94,38 @@ func (c *B2InputClient) ReadMetadata(path string) (map[string]string, error) { return nil, fmt.Errorf("get attributes for object: %w", err) } - metadata := attrs.Info - metadata["Name"] = attrs.Name - metadata["Size"] = fmt.Sprintf("%d", attrs.Size) - metadata["ContentType"] = attrs.ContentType - metadata["LastModified"] = attrs.LastModified.Format("2006-01-02T15:04:05Z") - metadata["SHA1"] = attrs.SHA1 - metadata["UploadTimestamp"] = attrs.UploadTimestamp.Format("2006-01-02T15:04:05Z") + metadata := MetadataStruct{ + Name: attrs.Name, + StorageType: "b2", + Hash: attrs.SHA1, + ContentType: attrs.ContentType, + FirstCreated: attrs.UploadTimestamp, + LastModified: attrs.LastModified, + Misc: attrs.Info, + } + metadata.Misc["Size"] = strconv.FormatInt(attrs.Size, 10) switch attrs.Status { case b2.Uploaded: - metadata["Status"] = "Uploaded" + metadata.Misc["b2-status"] = "Uploaded" case b2.Folder: - metadata["Status"] = "Folder" + metadata.Misc["b2-status"] = "Folder" case b2.Hider: - metadata["Status"] = "Hider" + metadata.Misc["b2-status"] = "Hider" case b2.Started: - metadata["Status"] = "Started" + metadata.Misc["b2-status"] = "Started" default: - metadata["Status"] = "Unknown" + metadata.Misc["b2-status"] = "Unknown" } - return metadata, nil + return &metadata, nil } -func (c *B2InputClient) GetReader(path string) (io.Reader, error) { - obj := c.bucket.Object(path) +func (c *B2InputClient) GetReader(path string) (io.ReadCloser, error) { + obj := c.bucket.Object(c.prefix + path) if obj == nil { return nil, fmt.Errorf("failed to reference object in B2 bucket") } - reader := obj.NewReader(context.Background()) - - return reader, nil + return obj.NewReader(context.Background()), nil } diff --git a/internal/client/input/input_client_interface.go b/internal/client/input/input_client_interface.go index 4bbbae5..c896ca4 100644 --- a/internal/client/input/input_client_interface.go +++ b/internal/client/input/input_client_interface.go @@ -2,10 +2,21 @@ package input import ( "io" + "time" ) type InputClient interface { Scan() ([]string, error) - ReadMetadata(string) (map[string]string, error) - GetReader(string) (io.Reader, error) + ReadMetadata(string) (*MetadataStruct, error) + GetReader(string) (io.ReadCloser, error) +} + +type MetadataStruct struct { + Name string + StorageType string + Hash string + ContentType string + FirstCreated time.Time + LastModified time.Time + Misc map[string]string } diff --git a/internal/client/output/b2.go b/internal/client/output/b2.go index 3c8b53c..6979e5b 100644 --- a/internal/client/output/b2.go +++ b/internal/client/output/b2.go @@ -4,9 +4,11 @@ import ( "context" "fmt" "io" + "strconv" "github.com/Backblaze/blazer/b2" "github.com/SayaAndy/saya-today-thumbnail-generator/config" + "github.com/SayaAndy/saya-today-thumbnail-generator/internal/client/input" ) var _ OutputClient = (*B2OutputClient)(nil) @@ -36,16 +38,19 @@ func NewB2OutputClient(cfg *config.OutputConfig) (*B2OutputClient, error) { return &B2OutputClient{b2cl: b2cl, bucket: bucket, prefix: b2cfg.Prefix}, nil } -func (c *B2OutputClient) GetWriter(path string) (io.Writer, error) { +func (c *B2OutputClient) GetWriter(path string, inputMetadata *input.MetadataStruct) (io.WriteCloser, error) { obj := c.bucket.Object(c.prefix + path) if obj == nil { return nil, fmt.Errorf("failed to reference object in B2 bucket") } - return obj.NewWriter(context.Background()), nil + attrs := &b2.Attrs{Info: make(map[string]string)} + attrs.Info["sha1-original"] = inputMetadata.Hash + + return obj.NewWriter(context.Background(), b2.WithAttrsOption(attrs)), nil } -func (c *B2OutputClient) ReadMetadata(path string) (map[string]string, error) { +func (c *B2OutputClient) ReadMetadata(path string) (*MetadataStruct, error) { obj := c.bucket.Object(c.prefix + path) if obj == nil { return nil, fmt.Errorf("failed to reference object in B2 bucket") @@ -56,40 +61,44 @@ func (c *B2OutputClient) ReadMetadata(path string) (map[string]string, error) { return nil, fmt.Errorf("get attributes for object: %w", err) } - metadata := attrs.Info - metadata["Name"] = attrs.Name - metadata["Size"] = fmt.Sprintf("%d", attrs.Size) - metadata["ContentType"] = attrs.ContentType - metadata["LastModified"] = attrs.LastModified.Format("2006-01-02T15:04:05Z") - metadata["SHA1"] = attrs.SHA1 - metadata["UploadTimestamp"] = attrs.UploadTimestamp.Format("2006-01-02T15:04:05Z") + metadata := MetadataStruct{ + Name: attrs.Name, + StorageType: "b2", + Hash: attrs.SHA1, + HashOriginal: attrs.Info["sha1-original"], + ContentType: attrs.ContentType, + FirstCreated: attrs.UploadTimestamp, + LastModified: attrs.LastModified, + Misc: attrs.Info, + } + metadata.Misc["Size"] = strconv.FormatInt(attrs.Size, 10) switch attrs.Status { case b2.Uploaded: - metadata["Status"] = "Uploaded" + metadata.Misc["b2-status"] = "Uploaded" case b2.Folder: - metadata["Status"] = "Folder" + metadata.Misc["b2-status"] = "Folder" case b2.Hider: - metadata["Status"] = "Hider" + metadata.Misc["b2-status"] = "Hider" case b2.Started: - metadata["Status"] = "Started" + metadata.Misc["b2-status"] = "Started" default: - metadata["Status"] = "Unknown" + metadata.Misc["b2-status"] = "Unknown" } - return metadata, nil + return &metadata, nil } -func (c *B2OutputClient) IsMissing(path string) (bool, error) { +func (c *B2OutputClient) IsMissing(path string) bool { obj := c.bucket.Object(c.prefix + path) if obj == nil { - return false, fmt.Errorf("failed to reference object in B2 bucket") + return true } attrs, err := obj.Attrs(context.Background()) if err != nil { - return true, fmt.Errorf("get attributes for object: %w", err) + return true } - return attrs.Status == b2.Hider, nil + return attrs.Status == b2.Hider } diff --git a/internal/client/output/output_client_interface.go b/internal/client/output/output_client_interface.go index 5271e18..4ec4728 100644 --- a/internal/client/output/output_client_interface.go +++ b/internal/client/output/output_client_interface.go @@ -2,10 +2,24 @@ package output import ( "io" + "time" + + "github.com/SayaAndy/saya-today-thumbnail-generator/internal/client/input" ) type OutputClient interface { - GetWriter(path string) (io.Writer, error) - ReadMetadata(string) (map[string]string, error) - IsMissing(path string) (bool, error) + GetWriter(path string, inputMetadata *input.MetadataStruct) (io.WriteCloser, error) + ReadMetadata(string) (*MetadataStruct, error) + IsMissing(path string) bool +} + +type MetadataStruct struct { + Name string + StorageType string + Hash string + HashOriginal string + ContentType string + FirstCreated time.Time + LastModified time.Time + Misc map[string]string } diff --git a/internal/processor/processor_interface.go b/internal/processor/processor_interface.go index 8a8f803..7cf20f2 100644 --- a/internal/processor/processor_interface.go +++ b/internal/processor/processor_interface.go @@ -4,5 +4,5 @@ import "io" type Processor interface { DeductOutputPath(inputPath string) string - Process(ext string, reader io.Reader, writer io.Writer) error + Process(ext string, reader io.ReadCloser, writer io.WriteCloser) error } diff --git a/internal/processor/webp.go b/internal/processor/webp.go index 984aab1..b5e089d 100644 --- a/internal/processor/webp.go +++ b/internal/processor/webp.go @@ -6,6 +6,7 @@ import ( "image/jpeg" "image/png" "io" + "log/slog" "strings" "golang.org/x/image/draw" @@ -41,10 +42,13 @@ func (p *WebpProcessor) DeductOutputPath(inputPath string) string { return strings.Join(pathParts, ".") } -func (p *WebpProcessor) Process(contentType string, reader io.Reader, writer io.Writer) error { +func (p *WebpProcessor) Process(contentType string, reader io.ReadCloser, writer io.WriteCloser) error { var src image.Image var err error + defer reader.Close() + defer writer.Close() + switch contentType { case "image/jpeg": src, err = jpeg.Decode(reader) @@ -56,6 +60,8 @@ func (p *WebpProcessor) Process(contentType string, reader io.Reader, writer io. if err != nil { return fmt.Errorf("decode png: %w", err) } + default: + return fmt.Errorf("unsupported content type: %s", contentType) } opts, err := encoder.NewLossyEncoderOptions(encoder.PresetDefault, float32(p.quality)) @@ -63,8 +69,16 @@ func (p *WebpProcessor) Process(contentType string, reader io.Reader, writer io. return fmt.Errorf("create webp encoder options: %w", err) } - xCoef := p.maxWidth / src.Bounds().Max.X - yCoef := p.maxHeight / src.Bounds().Max.Y + xCoef := float64(p.maxWidth) / float64(src.Bounds().Max.X) + if p.maxWidth == 0 { + xCoef = 1 + } + yCoef := float64(p.maxHeight) / float64(src.Bounds().Max.Y) + if p.maxHeight == 0 { + yCoef = 1 + } + slog.Debug("calculated coefficients", slog.Float64("x_coef", xCoef), slog.Float64("y_coef", yCoef)) + if xCoef > 1 && yCoef > 1 { return webp.Encode(writer, src, opts) } @@ -74,7 +88,7 @@ func (p *WebpProcessor) Process(contentType string, reader io.Reader, writer io. minCoef = yCoef } - dst := image.NewRGBA(image.Rect(0, 0, src.Bounds().Max.X*xCoef, src.Bounds().Max.Y*xCoef)) + dst := image.NewRGBA(image.Rect(0, 0, int(float64(src.Bounds().Max.X)*minCoef), int(float64(src.Bounds().Max.Y)*minCoef))) draw.CatmullRom.Scale(dst, dst.Rect, src, src.Bounds(), draw.Over, nil) return webp.Encode(writer, dst, opts) @@ -2,8 +2,12 @@ package main import ( "flag" + "log/slog" "github.com/SayaAndy/saya-today-thumbnail-generator/config" + "github.com/SayaAndy/saya-today-thumbnail-generator/internal/client/input" + "github.com/SayaAndy/saya-today-thumbnail-generator/internal/client/output" + "github.com/SayaAndy/saya-today-thumbnail-generator/internal/processor" ) var ( @@ -13,8 +17,83 @@ var ( func main() { flag.Parse() + slog.Info("starting thumbnail generator...") + slog.SetLogLoggerLevel(slog.LevelDebug) + cfg := &config.Config{} if err := config.LoadConfig(*configPath, cfg); err != nil { panic(err) } + + inputClient, err := input.NewB2InputClient(&cfg.Input) + if err != nil { + panic(err) + } + + outputClient, err := output.NewB2OutputClient(&cfg.Output) + if err != nil { + panic(err) + } + + converter, err := processor.NewWebpProcessor(&cfg.Processor) + if err != nil { + panic(err) + } + + generalLogger := slog.With( + slog.String("input_storage", cfg.Input.Storage.Type), + slog.String("processor_type", cfg.Processor.Type), + slog.String("output_storage", cfg.Output.Storage.Type), + ) + generalLogger.Info("initialized clients and processor") + + files, err := inputClient.Scan() + if err != nil { + panic(err) + } + + for _, file := range files { + outputName := converter.DeductOutputPath(file) + fileLogger := generalLogger.With(slog.String("input_path", file), slog.String("output_path", outputName)) + + isMissing := outputClient.IsMissing(outputName) + + inputMetadata, err := inputClient.ReadMetadata(file) + if err != nil { + fileLogger.Error("fail to read metadata of (supposedly existing) input file", slog.String("error", err.Error())) + continue + } + + if !isMissing { + outputMetadata, err := outputClient.ReadMetadata(outputName) + if err != nil { + fileLogger.Error("fail to read metadata of (supposedly existing) output file", slog.String("error", err.Error())) + continue + } + if inputMetadata.Hash == outputMetadata.HashOriginal { + fileLogger.Info("skip already processed file (based on equal hash)", slog.String("input_hash", inputMetadata.Hash)) + continue + } + } + + fileLogger.Info("start to process file") + reader, err := inputClient.GetReader(file) + if err != nil { + fileLogger.Error("fail to get reader for input file", slog.String("error", err.Error())) + continue + } + + writer, err := outputClient.GetWriter(outputName, inputMetadata) + if err != nil { + fileLogger.Error("fail to get writer for output file", slog.String("error", err.Error())) + continue + } + + if err := converter.Process(inputMetadata.ContentType, reader, writer); err != nil { + fileLogger.Error("fail to convert file", slog.String("error", err.Error())) + continue + } + + fileLogger.Info("successfully processed file", slog.String("input_hash", inputMetadata.Hash)) + } } |