parser

type Block interface {
	// Identification
	ID() string
	BlockType() BlockType
	EntityTypes() []EntityType
	Page() int

	// Content
	Text() string
	TextType() string
	Confidence() float64

	// Table specific
	RowIndex() int
	ColumnIndex() int
	RowSpan() int
	ColumnSpan() int

	// Selection elements
	SelectionStatus() string

	// Relationships
	Children() []Block
	Parents() []Block
	Relationships() []Relationship

	// Geometry
	BoundingBox() BoundingBox
	Polygon() []Point
}

type BlockProcessor interface {
	ProcessBlock(ctx context.Context, block *textract.Block) (Block, error)
}

type BlockType string

type BoundingBox struct {
	Width  float64
	Height float64
	Left   float64
	Top    float64
}

type Cell interface {
	// Content
	Text() string
	Confidence() float64
	EntityTypes() []EntityType
	IsColumnHeader() bool

	// Position
	RowIndex() int
	ColumnIndex() int
	RowSpan() int
	ColumnSpan() int

	// Navigation
	Table() Table

	// Geometry
	BoundingBox() BoundingBox
	Polygon() []Point
}

type Document interface {
	// Core access
	Pages() []Page
	Raw() *TextractResponse

	// Navigation
	GetPageByIndex(idx int) (Page, error)

	// Search and filtering
	FindKeyValuePairs(key string) []KeyValue
	FilterBlocks(opts FilterOptions) []Block

	// Metadata
	PageCount() int
	DocumentMetadata() DocumentMetadata
}

type DocumentMetadata struct {
	Pages int
}

type DocumentOption func(*DocumentOptions)

type DocumentOptions struct {
	ConfidenceThreshold float64
	EnableMergedCells   bool
	CustomProcessors    map[BlockType]BlockProcessor
}

type EntityType string

type FilterOptions struct {
	MinConfidence float64
	BlockTypes    []BlockType
	EntityTypes   []EntityType
}

type Form interface {
	// Content
	Fields() []KeyValue
	SelectionElements() []SelectionElement

	// Search
	GetFieldByKey(key string) KeyValue
	SearchFieldsByKey(key string) []KeyValue

	// Navigation
	Page() Page
}

type Geometry struct {
	BoundingBox BoundingBox
	Polygon     []Point
}

type JSONBlock struct {
	ID              string             `json:"Id"`
	BlockType       string             `json:"BlockType"`
	EntityTypes     []string           `json:"EntityTypes,omitempty"`
	Text            string             `json:"Text,omitempty"`
	Confidence      float64            `json:"Confidence"`
	Geometry        *JSONGeometry      `json:"Geometry"`
	Relationships   []JSONRelationship `json:"Relationships,omitempty"`
	RowIndex        int                `json:"RowIndex,omitempty"`
	ColumnIndex     int                `json:"ColumnIndex,omitempty"`
	RowSpan         int                `json:"RowSpan,omitempty"`
	ColumnSpan      int                `json:"ColumnSpan,omitempty"`
	SelectionStatus string             `json:"SelectionStatus,omitempty"`
	Page            int                `json:"Page,omitempty"`
}

type JSONBoundingBox struct {
	Width  float64 `json:"Width"`
	Height float64 `json:"Height"`
	Left   float64 `json:"Left"`
	Top    float64 `json:"Top"`
}

type JSONDocumentMetadata struct {
	Pages int `json:"Pages"`
}

type JSONGeometry struct {
	BoundingBox *JSONBoundingBox `json:"BoundingBox"`
	Polygon     []JSONPoint      `json:"Polygon"`
}

type JSONPoint struct {
	X float64 `json:"X"`
	Y float64 `json:"Y"`
}

type JSONRelationship struct {
	Type string   `json:"Type"`
	IDs  []string `json:"Ids"`
}

type JSONResponse struct {
	DocumentMetadata *JSONDocumentMetadata `json:"DocumentMetadata"`
	Blocks           []*JSONBlock          `json:"Blocks"`
}

type KeyValue interface {
	// Content
	Key() Block
	Value() Block

	// Helper methods
	KeyText() string
	ValueText() string
	Confidence() float64

	// Navigation
	Form() Form
}

type Line interface {
	// Content
	Text() string
	Words() []string
	Confidence() float64
	EntityTypes() []EntityType

	// Navigation
	Page() Page

	// Geometry
	BoundingBox() BoundingBox
	Polygon() []Point
}

type MergedCell interface {
	Cell
	MergedRowSpan() int
	MergedColumnSpan() int
	ContainedCells() []Cell
	EntityTypes() []EntityType
}

type Page interface {
	// Content access
	Lines() []Line
	Tables() []Table
	Forms() []Form
	Words() []string

	// Navigation
	Document() Document
	Number() int

	// Geometry
	BoundingBox() BoundingBox
	Polygon() []Point
	EntityTypes() []EntityType
}

type Parser interface {
	ParseDocument(ctx context.Context, response *TextractResponse, opts ...DocumentOption) (Document, error)
}

type Point struct {
	X float64
	Y float64
}

type Query interface {
	// Text returns the query text
	Text() string
	EntityTypes() []EntityType

	// Alias returns the query alias if one was specified
	Alias() string

	// Results returns all answers found for this query
	Results() []QueryResult

	// Page returns the parent page containing this query
	Page() Page
}

type QueryResult interface {
	// Text returns the result text
	Text() string
	EntityTypes() []EntityType

	// Confidence returns the confidence score for this result
	Confidence() float64

	// Query returns the parent query
	Query() Query

	// Block returns the underlying block
	Block() Block
}

type Relationship struct {
	Type string
	IDs  []string
}

type SelectionElement interface {
	// Status
	IsSelected() bool
	SelectionStatus() SelectionStatus
	Confidence() float64
	EntityTypes() []EntityType

	// Navigation
	Block() Block
	Form() Form

	// Geometry
	BoundingBox() BoundingBox
	Polygon() []Point
}

type SelectionElementType string

type SelectionStatus string

type Table interface {
	// Structure
	Rows() []TableRow
	Cells() [][]Cell
	MergedCells() []MergedCell
	GetHeaders() []Cell

	// Metadata
	RowCount() int
	ColumnCount() int

	// Navigation
	Page() Page
	GetCellByPosition(row, col int) (Cell, error)

	// Geometry
	BoundingBox() BoundingBox
	Polygon() []Point
	EntityTypes() []EntityType
}

type TableRow interface {
	Cells() []Cell
	RowIndex() int
	Table() Table
}

type TextType string

type TextractBlock struct {
	BlockType     *string
	Confidence    *float64
	Text          *string
	RowIndex      *int64
	ColumnIndex   *int64
	RowSpan       *int64
	ColumnSpan    *int64
	Geometry      *TextractGeometry
	ID            *string
	Relationships []*TextractRelationship
	EntityTypes   []*string
	Page          *int64
}

type TextractBoundingBox struct {
	Width  *float64
	Height *float64
	Left   *float64
	Top    *float64
}

type TextractDocument struct {
	DocumentMetadata *TextractDocumentMetadata
	Blocks           []*TextractBlock
}

type TextractDocumentMetadata struct {
	Pages *int64
}

type TextractGeometry struct {
	BoundingBox *TextractBoundingBox
	Polygon     []*TextractPoint
}

type TextractPoint struct {
	X *float64
	Y *float64
}

type TextractRelationship struct {
	Type *string
	IDs  []*string
}

type TextractResponse struct {
	Blocks   []*textract.Block
	Metadata *textract.DocumentMetadata
}