ocrworker

README

Only for personal use!!!

Documentation

Index

• Constants
• func NaturalLanguageUnderstanding(text string, license License) (*nlu.AnalysisResults, error)
• func Vision(ocrRequest OcrRequest) ([]*pb.EntityAnnotation, error)
• func VisionObj(ocrRequest OcrRequest) ([]*pb.LocalizedObjectAnnotation, error)
• type ConvertPdf
• type FlagFunction
  • func NoOpFlagFunction() FlagFunction
• type IdentityPreprocessor
• type License
• type MockEngine
  • func (m MockEngine) ProcessRequest(ocrRequest OcrRequest) (OcrResult, error)
• type OcrEngine
  • func NewOcrEngine(engineType OcrEngineType) OcrEngine
• type OcrEngineType
  • func (e OcrEngineType) String() string
  • func (e *OcrEngineType) UnmarshalJSON(b []byte) (err error)
• type OcrHttpHandler
  • func NewOcrHttpHandler(r RabbitConfig) *OcrHttpHandler
  • func (s *OcrHttpHandler) ServeHTTP(w http.ResponseWriter, req *http.Request)
• type OcrHttpMultipartHandler
  • func NewOcrHttpMultipartHandler(r RabbitConfig) *OcrHttpMultipartHandler
  • func (s *OcrHttpMultipartHandler) ServeHTTP(w http.ResponseWriter, req *http.Request)
• type OcrRequest
  • func (o OcrRequest) String() string
• type OcrResult
  • func HandleOcrRequest(ocrRequest OcrRequest, rabbitConfig RabbitConfig) (OcrResult, error)
• type OcrRpcClient
  • func NewOcrRpcClient(rc RabbitConfig) (*OcrRpcClient, error)
  • func (c *OcrRpcClient) DecodeImage(ocrRequest OcrRequest) (OcrResult, error)
• type OcrRpcWorker
  • func NewOcrRpcWorker(rc RabbitConfig) (*OcrRpcWorker, error)
  • func (w OcrRpcWorker) Run() error
  • func (w *OcrRpcWorker) Shutdown() error
• type Preprocessor
• type PreprocessorRpcWorker
  • func NewPreprocessorRpcWorker(rc RabbitConfig, preprocessor string) (*PreprocessorRpcWorker, error)
  • func (w PreprocessorRpcWorker) Run() error
  • func (w *PreprocessorRpcWorker) Shutdown() error
• type QREngine
  • func (m QREngine) ProcessRequest(ocrRequest OcrRequest) (OcrResult, error)
• type RabbitConfig
  • func DefaultConfigFlagsOverride(flagFunction FlagFunction) RabbitConfig
  • func DefaultTestConfig() RabbitConfig
• type StrokeWidthTransformer
• type TesseractEngine
  • func (t TesseractEngine) ProcessRequest(ocrRequest OcrRequest) (OcrResult, error)
• type TesseractEngineArgs
  • func NewTesseractEngineArgs(ocrRequest OcrRequest) (*TesseractEngineArgs, error)
  • func (t TesseractEngineArgs) Export() []string
• type VisionEngine
  • func (m VisionEngine) ProcessRequest(ocrRequest OcrRequest) (OcrResult, error)
• type Yandex

Constants

const (
	ENGINE_TESSERACT = OcrEngineType(iota)
	ENGINE_VISION    = OcrEngineType(iota)
	ENGINE_QR        = OcrEngineType(iota)
	ENGINE_GO_TESSERACT
	ENGINE_MOCK
)

const MOCK_ENGINE_RESPONSE = "mock engine decoder response"

const PreprocessorConvertPdf = "convert-pdf"

const PreprocessorIdentity = "identity"

const PreprocessorStrokeWidthTransform = "stroke-width-transform"

const (
	RPC_RESPONSE_TIMEOUT = time.Second * 120
)

Functions

func NaturalLanguageUnderstanding

func NaturalLanguageUnderstanding(text string, license License) (*nlu.AnalysisResults, error)

func Vision

func Vision(ocrRequest OcrRequest) ([]*pb.EntityAnnotation, error)

func VisionObj

func VisionObj(ocrRequest OcrRequest) ([]*pb.LocalizedObjectAnnotation, error)

Types

type ConvertPdf

type ConvertPdf struct {
}

type FlagFunction

type FlagFunction func()

func NoOpFlagFunction

func NoOpFlagFunction() FlagFunction

type IdentityPreprocessor

type IdentityPreprocessor struct {
}

type License

type License struct {
	User string
	Pass string
}

type MockEngine

type MockEngine struct {
}

func (MockEngine) ProcessRequest

func (m MockEngine) ProcessRequest(ocrRequest OcrRequest) (OcrResult, error)

type OcrEngine

type OcrEngine interface {
	ProcessRequest(ocrRequest OcrRequest) (OcrResult, error)
}

func NewOcrEngine

func NewOcrEngine(engineType OcrEngineType) OcrEngine

type OcrEngineType

type OcrEngineType int

func (OcrEngineType) String

func (e OcrEngineType) String() string

func (*OcrEngineType) UnmarshalJSON

func (e *OcrEngineType) UnmarshalJSON(b []byte) (err error)

type OcrHttpHandler

type OcrHttpHandler struct {
	RabbitConfig RabbitConfig
}

func NewOcrHttpHandler

func NewOcrHttpHandler(r RabbitConfig) *OcrHttpHandler

func (*OcrHttpHandler) ServeHTTP

func (s *OcrHttpHandler) ServeHTTP(w http.ResponseWriter, req *http.Request)

type OcrHttpMultipartHandler

type OcrHttpMultipartHandler struct {
	RabbitConfig RabbitConfig
}

func NewOcrHttpMultipartHandler

func NewOcrHttpMultipartHandler(r RabbitConfig) *OcrHttpMultipartHandler

func (*OcrHttpMultipartHandler) ServeHTTP

func (s *OcrHttpMultipartHandler) ServeHTTP(w http.ResponseWriter, req *http.Request)

type OcrRequest

type OcrRequest struct {
	ImgUrl            string                 `json:"img_url"`
	ImgBase64         string                 `json:"img_base64"`
	EngineType        OcrEngineType          `json:"engine"`
	ImgBytes          []byte                 `json:"img_bytes"`
	PreprocessorChain []string               `json:"preprocessors"`
	PreprocessorArgs  map[string]interface{} `json:"preprocessor-args"`
	EngineArgs        map[string]interface{} `json:"engine_args"`

	// decode ocr in http handler rather than putting in queue
	InplaceDecode bool `json:"inplace_decode"`
}

func (OcrRequest) String

func (o OcrRequest) String() string

type OcrResult

type OcrResult struct {
	Text string
}

func HandleOcrRequest

func HandleOcrRequest(ocrRequest OcrRequest, rabbitConfig RabbitConfig) (OcrResult, error)

type OcrRpcClient

type OcrRpcClient struct {
	// contains filtered or unexported fields
}

func NewOcrRpcClient

func NewOcrRpcClient(rc RabbitConfig) (*OcrRpcClient, error)

func (*OcrRpcClient) DecodeImage

func (c *OcrRpcClient) DecodeImage(ocrRequest OcrRequest) (OcrResult, error)

type OcrRpcWorker

type OcrRpcWorker struct {
	Done chan error
	// contains filtered or unexported fields
}

func NewOcrRpcWorker

func NewOcrRpcWorker(rc RabbitConfig) (*OcrRpcWorker, error)

func (OcrRpcWorker) Run

func (w OcrRpcWorker) Run() error

func (*OcrRpcWorker) Shutdown

func (w *OcrRpcWorker) Shutdown() error

type Preprocessor

type Preprocessor interface {
	// contains filtered or unexported methods
}

type PreprocessorRpcWorker

type PreprocessorRpcWorker struct {
	Done chan error
	// contains filtered or unexported fields
}

func NewPreprocessorRpcWorker

func NewPreprocessorRpcWorker(rc RabbitConfig, preprocessor string) (*PreprocessorRpcWorker, error)

func (PreprocessorRpcWorker) Run

func (w PreprocessorRpcWorker) Run() error

func (*PreprocessorRpcWorker) Shutdown

func (w *PreprocessorRpcWorker) Shutdown() error

type QREngine

type QREngine struct {
}

func (QREngine) ProcessRequest

func (m QREngine) ProcessRequest(ocrRequest OcrRequest) (OcrResult, error)

type RabbitConfig

type RabbitConfig struct {
	AmqpURI      string
	Exchange     string
	ExchangeType string
	RoutingKey   string
	Reliable     bool
}

func DefaultConfigFlagsOverride

func DefaultConfigFlagsOverride(flagFunction FlagFunction) RabbitConfig

func DefaultTestConfig

func DefaultTestConfig() RabbitConfig

type StrokeWidthTransformer

type StrokeWidthTransformer struct {
}

type TesseractEngine

type TesseractEngine struct {
}

This variant of the TesseractEngine calls tesseract via exec

func (TesseractEngine) ProcessRequest

func (t TesseractEngine) ProcessRequest(ocrRequest OcrRequest) (OcrResult, error)

type TesseractEngineArgs

type TesseractEngineArgs struct {
	// contains filtered or unexported fields
}

func NewTesseractEngineArgs

func NewTesseractEngineArgs(ocrRequest OcrRequest) (*TesseractEngineArgs, error)

func (TesseractEngineArgs) Export

func (t TesseractEngineArgs) Export() []string

return a slice that can be passed to tesseract binary as command line args, eg, ["-c", "tessedit_char_whitelist=0123456789", "-c", "foo=bar"]

type VisionEngine

type VisionEngine struct {
	Annotations    []*pb.EntityAnnotation
	ObjAnnotations []*pb.LocalizedObjectAnnotation
	Tags           *nlu.AnalysisResults
	Translation    string
}

func (VisionEngine) ProcessRequest

func (m VisionEngine) ProcessRequest(ocrRequest OcrRequest) (OcrResult, error)

type Yandex

type Yandex struct {
	Key string
}