source

type CSVDialect struct {
	Delimiter string
	Quote     string
	Escape    string
	NewLine   string
	Comment   string
	SkipRows  int
	HasHeader bool
	Encoding  Encoding

	// HeaderNames are the column names exactly as they appear in the file,
	// before any de-duplication. DuckDB silently renames a repeated header to
	// keep its schema valid, which would hide a real defect, so Veritix reads
	// the header line itself as well.
	HeaderNames []string

	// SniffedTypes is DuckDB's guess at each column's type. Veritix loads all
	// columns as text regardless; this is retained so the profiler can report
	// where a naive import would have guessed wrong, silently coerced a value,
	// or turned a bad value into a null.
	SniffedTypes []SniffedColumn

	// Notes are observations about the file itself, promoted to findings later.
	Notes []Note
}

type Dataset struct {
	// Root is the common directory the dataset was discovered under.
	Root string
	// Files are the inputs Veritix understands.
	Files []File
	// Tables are the logical tables those files contain.
	Tables []TableRef
	// Skipped records files that were passed over, and why. These are worth
	// surfacing: a spreadsheet in an unreadable format is itself a finding a
	// user needs to know about, not something to silently ignore.
	Skipped []SkippedFile
}

type Encoding string

type File struct {
	// Path is the absolute path on disk.
	Path string
	// Rel is the path relative to the dataset root, used for display.
	Rel string
	// Kind is how the file will be read.
	Kind Kind
	// Size is the file size in bytes.
	Size int64
}

type Kind string

type Note struct {
	Code    string
	Message string
}

type Sheet struct {
	Name    string
	Index   int
	Visible bool

	// HeaderRow is the 1-based row the column names were found on. Anything
	// above it was a title block or a blank spacer.
	HeaderRow int
	// Columns are the header names as written.
	Columns []string
	// DataRows is the number of rows below the header.
	DataRows int
	// HiddenRows is how many data rows are hidden from a reader of the file.
	HiddenRows int
	// MergedRanges are merged cell ranges, which break the one-value-per-cell
	// assumption every downstream tool makes.
	MergedRanges []string
	// FormulaErrors maps an error code such as "#REF!" to how many cells hold it.
	FormulaErrors map[string]int
	// RaggedRows is how many rows have a different populated width than the header.
	RaggedRows int
	// BlankSeparators counts fully-blank rows inside the data, which usually
	// mean more than one table has been stacked onto a single sheet.
	BlankSeparators int

	Notes []Note
}

type SkippedFile struct {
	Path   string
	Rel    string
	Reason string
}

type SniffedColumn struct {
	Name string
	Type string
}

type TableRef struct {
	// Name is a unique, SQL-friendly identifier within the dataset.
	Name string
	// Display is the human-readable origin, e.g. "sales.xlsx#Q1".
	Display string
	// File is the file this table is read from.
	File File
	// Sheet is the worksheet name, empty for CSV.
	Sheet string
}

type Workbook struct {
	Sheets []Sheet
	Notes  []Note
}