Abstract

A method is disclosed for deriving a diagram from a human-authored structured text document, refusing by default and binding every drawn element to its source. Each well-formed document is classified into exactly one of four outcomes: rendered; absent, when the section a diagram would come from does not exist; out of scope, when the document type has no certified transform, refused by name; or unstructured, when the document passes that type gate and the section exists but extraction yields nothing, a fail-open condition reported, not drawn as an empty canvas. Only a rendered outcome produces drawing output. Each drawn node carries its step's unparaphrased source text as provenance and a content-derived identifier. No edges are ever drawn, and a lane label says so. A post-render assertion reports undrawn declared steps, nodes without provenance, disallowed style classes, overlap, out-of-bounds nodes, unknown citations when known documents are supplied, a fail-open result and drawing on a refusal. An uncertified type may still be drawn if a separately produced structured form classifies every step as an action and has enough steps to form a sequence; the renderer first checks that form's step count and identifiers against its own extraction and aborts with a typed error on any disagreement. Step kinds are three-valued: action, section or unknown. A kind comes from an authored marker closing a line of the step's text when one exists, otherwise from closed-class grammatical evidence in the step title, with one positive rule and unknown as the default.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS