Skip to content
Documentation
How-to guides
Export diagrams

Export the graph as a diagram or a table

codegraph export writes the folded dependency graph in four formats. Two are diagrams you render with an external tool; two are data you feed to something else.

Before you start: a model.jsonl. Graphviz for DOT, a PlantUML jar or the plantuml command for PlantUML.

Choose the format and the level 

codegraph export model.jsonl --format dot      --level module --out graph.dot
codegraph export model.jsonl --format plantuml --level module --out modules.puml
codegraph export model.jsonl --format csv      --level type    --out types.csv
codegraph export model.jsonl --format json     --level type    --out types.json

--format is required; --level is module (the default) or type. Without --out the artifact goes to stdout, which is safe to redirect — every warning, fold diagnostic and summary goes to stderr.

--internal-only and --declared-only work here exactly as in analyze; see Get a facts-only answer.

Render DOT 

codegraph export model.jsonl --format dot --level type --internal-only --out types.dot
dot -Tsvg types.dot -o types.svg

The file starts with the header it was produced under, so a diagram checked into a repository still says what it is:

// codegraph — rendering of a FOLDED ANALYSIS GRAPH. Not a model.jsonl.
// level: type
// view: internalOnly
// nodes: 19, edges: 39
// base edges folded: 107, dropped: 0, unfoldable entities: 3
// encoding: solid edge = declared fact; dashed edge = contains an inference;
//           edge label and penwidth = aggregated base-edge count;
//           dashed node = stub (external, not corpus-declared).

penwidth is the aggregated count bucketed (1 / 2–4 / 5–16 / 17+) so two runs render identically on any platform. Node shape is deliberately not a channel — entity kind varies by language, so it rides in the tooltip.

Render PlantUML 

codegraph export model.jsonl --format plantuml --level module --out modules.puml
plantuml -tsvg modules.puml

The element follows the fold level: package at module level, class at type level — a module-level node carries TModule, not TType, and drawing it as a class would assert a type the model never declared. The diagram carries its own legend:

package "com.megacorp.ledger" as java_com_megacorp_ledger <<stub>> #line.dashed {
}
java_com_acme_order --> java_java_util : 15
java_com_acme_order ..> java_com_megacorp_ledger : 8
legend
  how to read this diagram
  ==
  a box | one module of the model; nothing below module level is drawn
  A --> B : n | n base edges, all provenance declared
  A ..> B : n | at least one base edge is an inference (derived / dynamic-candidate / generated)
  <<stub>> | external entity, not corpus-declared
end legend

So: solid is a declared fact, dotted contains an inference, <<stub>> is an entity your corpus does not declare.

Export data

CSV is RFC 4180 with a header row, and carries the level and view per row:

from,to,count,kinds,provenances,selfLoop,level,view
java:com.acme.order,java:com.acme.order.legacy,2,invocation;reference,declared,false,module,all
java:com.acme.order,java:com.megacorp.ledger,8,access;import;invocation;reference,declared;derived,false,module,all

JSON says what it is and does not pretend to be interchange:

{
  "kind": "codegraph.foldedGraph/1",
  "generatedBy": "@codegraph/analyzer",
  "level": "type",
  "view": { "name": "internalOnly", "filters": ["internalOnly"] },
  "nodes": [ … ],
None of these is a model. model.jsonl is the interchange contract; an export is a rendering of one folded view of it, and it cannot be read back.

If a diagram looks smaller than you expected, read the dropped and unfoldable entities counts in the header and the stderr warning. Entities with no container at the chosen level are reported, never silently omitted.

Related

Last updated on
Facts-only viewComplexity city