Class DecisionTreeBlockProcessor
java.lang.Object
org.asciidoctor.extension.BaseProcessor
org.asciidoctor.extension.BlockProcessor
network.ike.docs.koncept.DecisionTreeBlockProcessor
- All Implemented Interfaces:
org.asciidoctor.extension.Processor
@Name("decision-tree")
@Contexts(":listing")
@ContentModel(":raw")
public class DecisionTreeBlockProcessor
extends org.asciidoctor.extension.BlockProcessor
The
[decision-tree] block in the documentation build (dev-decision-tree-blocks):
the static projection of the block the assistant surface renders live. It reads the block
store-free (DecisionTreeSource), rolls it back at a health price
(DecisionTreeRollback), sweeps every stated value one way (DecisionTreeSweep),
and puts in the block's place what Komet draws: the tree as decision analysts draw it, left to
right, a square where someone chooses, a circle where the world decides, a triangle where the
consequences are counted, the rolled-back cost at every glyph, the chosen alternative heavy
(DecisionTreeFigures); beneath it the tornado of the sweep, one bar per value; then the
worked table, where each orange callout's number is computed from the inputs, generated rather
than maintained.
[decision-tree, health-price=100000, symbol="$", title="Should this patient have a scan in year 2?"] ---- decision: k:ScanNow[Scan now?] @id=… … ----The figures are drawn from the same geometry the live diagram lays out, with label widths estimated rather than measured: as self-contained SVG, inline on the html5 family, so the text stays text and a wide tree scrolls, and as a content-addressed file on Prawn and DocBook, where an image block is what those converters place; on the Typst backend as native Typst drawing code (
DecisionFigureTypst), handed to the converter in a passthrough block with the role
typst, so the print pipeline draws the figure with the document's own text and never
through an embedded image. A block with no root decision renders as a literal block, so the text
is visible, never dropped; a block with gaps renders with them named in a note.
A block that names a model, [decision-tree, model=<id>], is a rendering of the
[graphlet-model] block before it (dev-decision-point-knowledge, settled 2026-11-06):
it holds no structure and no number, only how the model is viewed, the standpoint it is rolled
back from and the cohort it is counted out for. It draws the model's tree, ledger and sentences
and the worked table, and no tornado: that is the [tornado, model=<id>] block's; and no
table of assumptions: the model block draws its own parameters. A block with a body of its own
still reads as before, tornado beneath.
-
Field Summary
Fields inherited from class org.asciidoctor.extension.BlockProcessor
name -
Constructor Summary
Constructors -
Method Summary
Methods inherited from class org.asciidoctor.extension.BlockProcessor
getName, setNameMethods inherited from class org.asciidoctor.extension.BaseProcessor
createBlock, createBlock, createBlock, createBlock, createBlock, createBlock, createBlock, createDocument, createList, createList, createList, createListItem, createListItem, createPhraseNode, createPhraseNode, createPhraseNode, createPhraseNode, createPhraseNode, createPhraseNode, createSection, createSection, createSection, createSection, createSection, createTable, createTable, createTableCell, createTableCell, createTableCell, createTableCell, createTableColumn, createTableColumn, createTableRow, getConfig, log, newCursor, newCursor, newCursor, newReader, newReader, newReader, newReader, parseContent, setConfig, unwrap, updateConfig
-
Constructor Details
-
DecisionTreeBlockProcessor
public DecisionTreeBlockProcessor()Creates a new block processor instance.
-
-
Method Details
-
process
-