Socrates/apps/web/lib/llm/analyze/proseText.ts

// Convert a ProseMirror JSON document into a plain-text string for LLM input.
// Headings get markdown-style "##"; chip nodes render as their label so the
// analyzer sees the same surface forms a human would.

import "server-only";

interface PMNode {
  type: string;
  text?: string;
  attrs?: Record<string, unknown>;
  content?: PMNode[];
}

export function proseDocToPlainText(doc: unknown): string {
  if (!doc || typeof doc !== "object") return "";
  const root = doc as PMNode;
  return walk(root, 0).trim();
}

function walk(node: PMNode, depth: number): string {
  if (!node) return "";
  if (node.type === "text") return node.text ?? "";
  if (node.type === "chip") {
    const label = (node.attrs?.label as string | undefined) ?? "";
    return label;
  }

  const children = (node.content ?? []).map(c => walk(c, depth + 1)).join("");

  switch (node.type) {
    case "heading": {
      const level = typeof node.attrs?.level === "number" ? (node.attrs.level as number) : 1;
      const hash = "#".repeat(Math.max(1, Math.min(level, 6)));
      return `\n\n${hash} ${children}\n\n`;
    }
    case "paragraph":
      return `${children}\n\n`;
    case "bulletList":
    case "bullet_list":
    case "orderedList":
    case "ordered_list":
      return `${children}\n`;
    case "listItem":
    case "list_item":
      return `- ${children.trim()}\n`;
    case "hardBreak":
    case "hard_break":
      return "\n";
    default:
      return children;
  }
}