Interface Tuning

Tuning knobs for the OCR-path geometry pass (CLI: --float-*/--para-gap/ --heading-scale/--col-*/--line-*). See DEFAULT_TUNING for defaults.

interface Tuning {
    colMaxWidth: number;
    colSplit: number;
    floatMargin: number;
    floatMaxWidth: number;
    floatMaxWords: number;
    headingScale: number;
    lineHeightRatio: number;
    lineYTol: number;
    paraGap: number;
}

Properties

colMaxWidth: number

boxes wider than this are excluded from gutter detection

colSplit: number

fractional x of the two-column midpoint

floatMargin: number

how far a fragment's left edge must sit from its column edge to be decoration

floatMaxWidth: number

fraction of page width above which a fragment is too wide to be decoration

floatMaxWords: number

fragments with more words than this are body text, not decoration

headingScale: number

lines taller than this times body height are headings

lineHeightRatio: number

same-line boxes with a larger height ratio don't merge

lineYTol: number

vertical distance within which two boxes are the same line

paraGap: number

a vertical gap this many times the line pitch is a paragraph break