procgen.collapse

Lifts a section of a graph's code into a node of its own. A graph is the code of its init.luau (proc.define(function(g) ... end)), and a section is a run of whole statements of that function, named by the lines it spans (from, to): typically a few node calls that together do one job.

• The node's inputs are the locals declared before the section that its code reads. A graph input it reads is declared in the node as the graph declares it (local seed = g:input("seed", 3)); any other value becomes an input typed by what the section wires it into.
• The node's outputs are the locals the section declares that the code after it reads.
• A local function the section calls goes into the node with it: moved when nothing left in the graph calls it, copied when something does.
• The section is replaced by one call of the node, so the graph builds from the same calls and cooks to the same result: local strewn = g["~/fx/Strew"]({ seed = seed, spots = spots }). A node giving several outputs is called once and its outputs are read off the table it returns.

The node lands beside the graph (or in into) as <name>.procNode, with the description as its g:describe, .metadata and README, and is found by procgen.searchNodes and described by procgen.describeNode at once. The graph's new source is written only once it builds.

A section holding the graph's own inputs or outputs (g:input, g:output, return { ... }), or assigning to a value declared before it, is refused with the line and why.

procgen.flatten is the inverse: it writes a node's body back out in place of its call. procgen.reference puts an existing node in place of a section.

Part of procgen.toolbox, called as tools.use("procgen", "collapse", ...) (zero procgen collapse in the engine shell). Returns { node, call, inputs, outputs, copied, moved, lines, source, nodeSource, diagnostics }.