Visualizing a graph with Mermaid

graph_to_mermaid() (in aat/core/mermaid.py) renders an AATGraph as a Mermaid flowchart: an action is a rectangle, an agent is rounded, a target is a stadium shape, and every node with a related_node becomes a labelled edge pointing at it. By default every node is also colored by which action it clusters with, so the separate clauses in a multi-action passage are visually distinguishable.

from aat.core import graph_to_mermaid

diagram, warnings = graph_to_mermaid(graph)
print(diagram)
for w in warnings:
    print(f"Warning: {w}")

orientation controls the diagram’s layout direction – Mermaid’s own flowchart direction codes: "BT" (bottom-to-top, the default), "TB"/"TD" (top-down, synonyms), "LR", or "RL". Matched case-insensitively ("lr" works the same as "LR"); anything else raises ValueError naming the valid options, rather than silently producing invalid Mermaid syntax:

diagram, warnings = graph_to_mermaid(graph, orientation="LR")

Pass color_by_action=False for a plain, uncolored diagram. save_mermaid(graph, path, ...) takes the same orientation/color_by_action arguments and writes the diagram straight to a file (e.g. analysis.mmd).

warnings lists any node whose related_node doesn’t resolve to another node actually present in graph – normally a sign the graph failed validate() upstream (see “Analyzing multiple citable passages” above), worth checking there first – plus, if the graph has more distinct actions than the color palette has slots (currently 8), one warning that colors repeat.