parser

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

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

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

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

type ErrMimeTypeNotSupported struct {
	MimeType *mimetype.MIME
}

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

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

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

type JPEGParserResult struct {
	Text string `json:"text"`
	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 PNGParser struct {
	// contains filtered or unexported fields
}

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

type Parser interface {
	// Returns list of supported mime types by this parser
	SupportedMimeTypes() []string
	// Parse file. Thread safe
	Parse(ctx context.Context, file io.Reader) Result
}

type Result interface {
	// Convert entire result to LLM readable string
	String() string
	// Not empty if there where error
	Error() error
	// Parsed subcomponents. For example images in the PDF or files inside archives
	Componets() []Result
}

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

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

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

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