lms-sb/src/lib/md-to-tiptap.ts

import { unified } from "unified";
import remarkParse from "remark-parse";

// ── Types ─────────────────────────────────────────────────────────────────────

type Mark = { type: string; attrs?: Record<string, unknown> };

type TipTapNode = {
  type: string;
  attrs?: Record<string, unknown>;
  content?: TipTapNode[];
  text?: string;
  marks?: Mark[];
};

// eslint-disable-next-line @typescript-eslint/no-explicit-any
type MdastNode = Record<string, any>;

// ── Inline converter ──────────────────────────────────────────────────────────

function convertInline(nodes: MdastNode[], marks: Mark[] = []): TipTapNode[] {
  const result: TipTapNode[] = [];

  for (const node of nodes) {
    switch (node.type) {
      case "text": {
        if (!node.value) break;
        const n: TipTapNode = { type: "text", text: node.value as string };
        if (marks.length) n.marks = marks;
        result.push(n);
        break;
      }

      case "strong":
        result.push(...convertInline(node.children, [...marks, { type: "bold" }]));
        break;

      case "emphasis":
        result.push(...convertInline(node.children, [...marks, { type: "italic" }]));
        break;

      case "delete":
        result.push(...convertInline(node.children, [...marks, { type: "strike" }]));
        break;

      case "inlineCode":
        result.push({
          type: "text",
          text: node.value as string,
          marks: [...marks, { type: "code" }],
        });
        break;

      case "link": {
        const linkMark: Mark = {
          type: "link",
          attrs: {
            href: node.url ?? "#",
            target: "_blank",
            rel: "noopener noreferrer",
            class: null,
          },
        };
        result.push(...convertInline(node.children, [...marks, linkMark]));
        break;
      }

      case "image":
        // Only HTTP/HTTPS images — local Obsidian paths are not resolvable here
        if (typeof node.url === "string" && /^https?:\/\//.test(node.url)) {
          result.push({
            type: "image",
            attrs: {
              src: node.url,
              alt: node.alt ?? null,
              title: node.title ?? null,
            },
          });
        }
        break;

      case "break":
        result.push({ type: "hardBreak" });
        break;

      default:
        // Try to extract content from unknown inline nodes
        if (Array.isArray(node.children)) {
          result.push(...convertInline(node.children, marks));
        } else if (node.value) {
          const n: TipTapNode = { type: "text", text: node.value as string };
          if (marks.length) n.marks = marks;
          result.push(n);
        }
    }
  }

  return result;
}

// ── Block converter ───────────────────────────────────────────────────────────

function convertBlock(nodes: MdastNode[]): TipTapNode[] {
  const result: TipTapNode[] = [];

  for (const node of nodes) {
    switch (node.type) {
      case "paragraph": {
        const content = convertInline(node.children ?? []);
        result.push(content.length ? { type: "paragraph", content } : { type: "paragraph" });
        break;
      }

      case "heading": {
        const content = convertInline(node.children ?? []);
        result.push({
          type: "heading",
          attrs: { level: node.depth ?? 2 },
          content,
        });
        break;
      }

      case "blockquote": {
        const content = convertBlock(node.children ?? []);
        result.push({ type: "blockquote", content });
        break;
      }

      case "code": {
        result.push({
          type: "codeBlock",
          attrs: { language: node.lang ?? null },
          content: [{ type: "text", text: node.value as string }],
        });
        break;
      }

      case "list": {
        const listType = node.ordered ? "orderedList" : "bulletList";
        const items = (node.children as MdastNode[]).map((item) => ({
          type: "listItem",
          content: convertBlock(item.children ?? []),
        }));
        const listNode: TipTapNode = { type: listType, content: items };
        if (node.ordered) listNode.attrs = { start: (node.start as number) ?? 1 };
        result.push(listNode);
        break;
      }

      case "thematicBreak":
        result.push({ type: "horizontalRule" });
        break;

      // Skip html, definitions, footnotes, etc.
    }
  }

  return result;
}

// ── Public API ────────────────────────────────────────────────────────────────

/**
 * Converts a Markdown string to a TipTap/ProseMirror JSON document.
 * Handles: headings, paragraphs, bold, italic, strike, inline code,
 * code blocks, blockquotes, lists (nested), links, images (HTTP only),
 * horizontal rules, hard breaks.
 *
 * Obsidian-specific syntax (![[wikilink]], [[link]]) is silently ignored
 * since local file paths are not available during server-side import.
 */
export function mdToTiptap(markdown: string): object {
  // Strip Obsidian wikilinks: [[link]] → plain text, ![[image]] → removed
  const cleaned = markdown
    .replace(/!\[\[([^\]]+)\]\]/g, "") // remove ![[image]] embeds
    .replace(/\[\[([^\]|]+)(?:\|([^\]]+))?\]\]/g, (_m, target, alias) => alias ?? target); // [[link|alias]] → alias or target

  const tree = unified().use(remarkParse).parse(cleaned);
  // eslint-disable-next-line @typescript-eslint/no-explicit-any
  const content = convertBlock((tree as any).children ?? []);

  if (content.length === 0) {
    content.push({ type: "paragraph" });
  }

  return { type: "doc", content };
}