parser

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

type BMPParserResult struct {
	FullPath string `json:"path"`
	Text     string `json:"text"`
	Err      error  `json:"error"`
}

type BMPParserStreamResult struct {
	FullPath        string             `json:"path"`
	Text            string             `json:"text"`
	CurrentStage    ParseProgressStage `json:"stage"`
	CurrentProgress uint8              `json:"progress"`
	Err             error              `json:"error"`
}

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

type CompositeParserResult struct {
	FullPath string `json:"path"`
	Err      error  `json:"error"`
	MimeType string `json:"mimeType"`
	Inner    Result `json:"inner"`
}

type CompositeParserStreamResult struct {
	FullPath        string             `json:"path"`
	Text            string             `json:"text"`
	MimeType        string             `json:"mimeType"`
	Inner           StreamResult       `json:"inner"`
	CurrentStage    ParseProgressStage `json:"stage"`
	CurrentProgress uint8              `json:"progress"`
	Err             error              `json:"error"`
}

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

type EMLParserResult struct {
	FullPath    string              `json:"path"`
	Headers     map[string][]string `json:"headers"`
	Text        string              `json:"text"`
	Err         error               `json:"error"`
	Attachments []Result            `json:"attachments"`
}

type EMLParserStreamResult struct {
	FullPath          string             `json:"path"`
	Text              string             `json:"text"`
	CurrentStage      ParseProgressStage `json:"stage"`
	CurrentPartHeader mail.PartHeader    `json:"subResultHeader"`
	CurrentPart       StreamResult       `json:"subResult"`
	Err               error              `json:"error"`
}

type ErrMimeTypeNotSupported struct {
	MimeType *mimetype.MIME
}

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

type GIFParserResult struct {
	FullPath string `json:"path"`
	Text     string `json:"text"`
	Err      error  `json:"error"`
}

type GIFParserStreamResult struct {
	FullPath        string             `json:"path"`
	Text            string             `json:"text"`
	CurrentStage    ParseProgressStage `json:"stage"`
	CurrentProgress uint8              `json:"progress"`
	Err             error              `json:"error"`
}

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

type JPEGParserResult struct {
	FullPath string `json:"path"`
	Text     string `json:"text"`
	Err      error  `json:"error"`
}

type JPEGParserStreamResult struct {
	FullPath        string             `json:"path"`
	Text            string             `json:"text"`
	CurrentStage    ParseProgressStage `json:"stage"`
	CurrentProgress uint8              `json:"progress"`
	Err             error              `json:"error"`
}

type PDFParser struct {
}

type PDFParserResult struct {
	Metadata string                `json:"metadata"`
	Pages    []PDFParserResultPage `json:"pages"`
	Err      error                 `json:"error"`
}

type PDFParserResultPage struct {
	Text   string   `json:"text"`
	Images []Result `json:"images"`
}

type PDFParserStreamResult struct {
	FullPath        string                      `json:"path"`
	Text            string                      `json:"text"`
	CurrentStage    ParseProgressStage          `json:"stage"`
	CurrentProgress uint8                       `json:"progress"`
	Metadata        string                      `json:"metadata"`
	Pages           []PDFParserStreamResultPage `json:"pages"`
	Subfile         StreamResult                `json:"subfile"`
	Err             error                       `json:"error"`
}

type PDFParserStreamResultPage struct {
	Text   string         `json:"text"`
	Images []StreamResult `json:"images"`
}

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

type PNGParserResult struct {
	FullPath string `json:"path"`
	Text     string `json:"text"`
	Err      error  `json:"error"`
}

type PNGParserStreamResult struct {
	FullPath        string             `json:"path"`
	Text            string             `json:"text"`
	CurrentStage    ParseProgressStage `json:"stage"`
	CurrentProgress uint8              `json:"progress"`
	Err             error              `json:"error"`
}

type ParseProgressStage string

type Parser interface {
	// Returns list of supported mime types by this parser
	SupportedMimeTypes() []string
	// Parse file. Thread safe. Use path to track subfiles or use file name as hint for mime type detection.
	Parse(ctx context.Context, file io.Reader, path string) Result
	// Parse file. Thread safe. Use path to track subfiles or use file name as hint for mime type detection. Return chanel that streams results.
	ParseStream(ctx context.Context, file io.Reader, path string) chan StreamResult
}

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

type RAWBGRAParserResult struct {
	FullPath string `json:"path"`
	Text     string `json:"text"`
	Err      error  `json:"error"`
}

type RAWBGRAParserStreamResult struct {
	FullPath        string             `json:"path"`
	Text            string             `json:"text"`
	CurrentStage    ParseProgressStage `json:"stage"`
	CurrentProgress uint8              `json:"progress"`
	Err             error              `json:"error"`
}

type Result interface {
	// Get full path to the file
	Path() string
	// Convert entire result to LLM readable string
	String() string
	// Not empty if there where error
	Error() error
	// Parsed subfiles. For example files inside archives
	Subfiles() []Result
}

type StreamResult interface {
	// Get full path to the file
	Path() string
	// Current file processing progress
	Stage() ParseProgressStage
	// Progress in percents from 0 to 100
	Progress() uint8
	// Underlying result
	SubResult() StreamResult
	// Convert entire result to LLM readable string
	String() string
	// Not empty if there where error
	Error() error
}

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

type TARParserResult struct {
	FullPath        string   `json:"path"`
	SubfilesResults []Result `json:"subfiles"`
	Err             error    `json:"error"`
}

type TARParserStreamResult struct {
	FullPath       string             `json:"path"`
	CurrentStage   ParseProgressStage `json:"stage"`
	CurrentSubfile StreamResult       `json:"subResult"`
	Err            error              `json:"error"`
}

type TIFFParserStreamResult struct {
	FullPath        string             `json:"path"`
	Text            string             `json:"text"`
	CurrentStage    ParseProgressStage `json:"stage"`
	CurrentProgress uint8              `json:"progress"`
	Err             error              `json:"error"`
}

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

type TiffParserResult struct {
	FullPath string `json:"path"`
	Text     string `json:"text"`
	Err      error  `json:"error"`
}

type WEBPParserStreamResult struct {
	FullPath        string             `json:"path"`
	Text            string             `json:"text"`
	CurrentStage    ParseProgressStage `json:"stage"`
	CurrentProgress uint8              `json:"progress"`
	Err             error              `json:"error"`
}

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

type WebPParserResult struct {
	FullPath string `json:"path"`
	Text     string `json:"text"`
	Err      error  `json:"error"`
}