@book.dev/sdk 1.66.0 → 1.68.0

package/dist/markdownImport.js

@@ -0,0 +1,348 @@
+/**
+ * **Markdown / GFM → blocks importer.** Turns a CommonMark + GFM source string
+ * into the format-agnostic {@link ImportedDoc} IR (see `import.ts`), so
+ * {@link importDoc} can land it through the existing data paths. A parser's only
+ * job is this transform; it never touches the store, the CRDT, or the wire.
+ *
+ * **Parser choice — `marked`.** It is pure JS with zero runtime dependencies,
+ * ships first-class GFM (tables, task lists, strikethrough, autolinks), and —
+ * crucially — exposes a structured token AST via {@link lexer} (block tokens
+ * carry their inline children in `.tokens`). That lets us map tokens → blocks
+ * directly, with NO DOM. The UI's `htmlToRuns`/`htmlToBlocks` are deliberately
+ * *not* reused: they are `document`-bound (they return stripped plain text when
+ * `document` is undefined) and live in `@book.dev/ui`, which depends on the SDK —
+ * importing them here would invert that dependency. We build runs from marked's
+ * inline tokens instead.
+ *
+ * **Block mapping** (shapes cited from `@book.dev/ui`'s `blockeditor/model.ts`):
+ *
+ *  | Markdown / GFM            | Block (`ImportedBlock`)                         |
+ *  | ------------------------- | ----------------------------------------------- |
+ *  | `# … ######` heading      | `heading` · `props.level` 1–3 (clamped, as editor)|
+ *  | paragraph                 | `paragraph`                                      |
+ *  | `-`/`*`/`+` list          | `list` · `props.kind:'bullet'`                   |
+ *  | `1.` ordered list         | `list` · `props.kind:'number'`                   |
+ *  | nested list item          | flat `list`/`todo` · `props.indent` = depth      |
+ *  | `- [ ]` / `- [x]` task     | `todo` · `props.checked`                         |
+ *  | `>` blockquote            | `quote` (block children folded to its rich text) |
+ *  | ```` ``` ```` fenced code  | `code` · `props.language`                        |
+ *  | GFM table                 | `table` → `row` → `cell` children                |
+ *  | `---` thematic break      | `divider`                                        |
+ *  | `**`/`*`/`~~`/`` ` ``/link | inline `runs` (`{b,i,s,c,a}`)                     |
+ *  | `![alt](src)` image       | {@link imagePlaceholderBlock} (block) / linked run (cell)|
+ *  | raw HTML / unknown        | degrades to a `paragraph` — never dropped        |
+ *
+ * **Images.** An image never vanishes: in body content it becomes the import
+ * core's {@link imagePlaceholderBlock} (preserving `src` + `alt` + `title`); in
+ * a table cell — which can hold only inline text — it becomes a link run to the
+ * `src` carrying the `alt`. **Unknown / unsupported constructs** degrade to a
+ * paragraph of their text, honoring the converter contract that import content
+ * is "never silently dropped" (`model.ts`).
+ *
+ * **Front-matter.** A leading `---` YAML block is read for `title:` (used as the
+ * page title); the remaining lines are passed through verbatim as a `yaml` code
+ * block rather than dropped.
+ */
+import { lexer } from 'marked';
+import { imagePlaceholderBlock, } from './import';
+const DEFAULT_TITLE = 'Imported document';
+const hasAttrs = (a) => Object.keys(a).length > 0;
+const sameAttrs = (a, b) => JSON.stringify(a ?? {}) === JSON.stringify(b ?? {});
+/** Append text, merging into the previous run when the formatting is identical. */
+function pushRun(out, text, attrs) {
+    if (text === '')
+        return;
+    const run = hasAttrs(attrs) ? { t: text, a: { ...attrs } } : { t: text };
+    const prev = out[out.length - 1];
+    if (prev && sameAttrs(prev.a, run.a))
+        prev.t += run.t;
+    else
+        out.push(run);
+}
+const assetOf = (img) => ({
+    kind: 'image',
+    ref: img.href,
+    ...(img.text ? { alt: img.text } : {}),
+    ...(img.title ? { title: img.title } : {}),
+});
+/** Fold a list of marked inline tokens into rich runs under the given attrs. */
+function walkInline(tokens, attrs, out, ctx) {
+    if (!tokens)
+        return;
+    for (const tok of tokens) {
+        switch (tok.type) {
+            case 'text':
+            case 'escape': {
+                const t = tok;
+                if (t.tokens && t.tokens.length > 0)
+                    walkInline(t.tokens, attrs, out, ctx);
+                else
+                    pushRun(out, t.text ?? t.raw ?? '', attrs);
+                break;
+            }
+            case 'strong':
+                walkInline(tok.tokens, { ...attrs, b: true }, out, ctx);
+                break;
+            case 'em':
+                walkInline(tok.tokens, { ...attrs, i: true }, out, ctx);
+                break;
+            case 'del':
+                walkInline(tok.tokens, { ...attrs, s: true }, out, ctx);
+                break;
+            case 'codespan':
+                pushRun(out, tok.text, { ...attrs, c: true });
+                break;
+            case 'br':
+                pushRun(out, '\n', {});
+                break;
+            case 'link': {
+                const link = tok;
+                const linked = { ...attrs, a: link.href };
+                if (link.tokens && link.tokens.length > 0)
+                    walkInline(link.tokens, linked, out, ctx);
+                else
+                    pushRun(out, link.text ?? link.href, linked);
+                break;
+            }
+            case 'image': {
+                const img = tok;
+                if (ctx.inlineImages)
+                    pushRun(out, img.text || img.href, { ...attrs, a: img.href });
+                else
+                    ctx.images.push(assetOf(img));
+                break;
+            }
+            case 'checkbox':
+                // Task-list marker — handled via the list item's `task`/`checked` fields.
+                break;
+            case 'html':
+                // Inline raw HTML: degrade to its text, never drop.
+                pushRun(out, tok.text ?? tok.raw ?? '', attrs);
+                break;
+            default: {
+                // Any inline token we don't special-case: keep its text/raw verbatim.
+                const g = tok;
+                if (g.tokens && g.tokens.length > 0)
+                    walkInline(g.tokens, attrs, out, ctx);
+                else
+                    pushRun(out, g.text ?? g.raw ?? '', attrs);
+            }
+        }
+    }
+}
+/** Inline tokens → `{runs, images}`; `inlineImages` linkifies images in place. */
+function inlineRuns(tokens, inlineImages = false) {
+    const ctx = { images: [], inlineImages };
+    const runs = [];
+    walkInline(tokens, {}, runs, ctx);
+    return { runs, images: ctx.images };
+}
+const plainText = (runs) => runs.map((r) => r.t).join('');
+// ── Block tokens → ImportedBlock[] ───────────────────────────────────────────
+/** Text-bearing blocks whose runs fold into a parent quote's rich text. */
+const QUOTE_FOLDABLE = new Set(['paragraph', 'heading', 'list', 'todo', 'quote']);
+/** Emit list items as flat `list`/`todo` blocks, recursing nested lists by depth. */
+function emitList(list, depth, out) {
+    const kind = list.ordered ? 'number' : 'bullet';
+    for (const item of list.items) {
+        const runs = [];
+        const images = [];
+        const trailing = [];
+        const nested = [];
+        for (const child of item.tokens) {
+            if (child.type === 'checkbox') {
+                // The task marker — its state is carried by `item.task`/`item.checked`.
+                continue;
+            }
+            else if (child.type === 'list') {
+                nested.push(child);
+            }
+            else if (child.type === 'text' || child.type === 'paragraph') {
+                const r = inlineRuns(child.tokens);
+                runs.push(...r.runs);
+                images.push(...r.images);
+            }
+            else {
+                // Block content nested in a list item (code, blockquote, table…): keep
+                // it as a following block rather than dropping it.
+                tokensToBlocks([child], depth, trailing);
+            }
+        }
+        const isTodo = item.task === true;
+        const props = {};
+        if (isTodo) {
+            if (item.checked)
+                props.checked = true;
+        }
+        else {
+            props.kind = kind;
+        }
+        // Clamp the indent: the editor's CSS only defines `.obe-indent-1..4`, so a
+        // deeper nesting would render flat — pin it to the deepest defined level.
+        if (depth > 0)
+            props.indent = Math.min(4, depth);
+        out.push({ type: isTodo ? 'todo' : 'list', text: runs, ...(Object.keys(props).length > 0 ? { props } : {}) });
+        for (const img of images)
+            out.push(imagePlaceholderBlock(img));
+        out.push(...trailing);
+        for (const nl of nested)
+            emitList(nl, depth + 1, out);
+    }
+}
+/** Map a GFM table token to a `table` → `row` → `cell` block tree. */
+function emitTable(table) {
+    const rowOf = (cells) => ({
+        type: 'row',
+        children: cells.map((cell) => ({ type: 'cell', text: inlineRuns(cell.tokens, true).runs })),
+    });
+    return {
+        type: 'table',
+        props: { header: true },
+        children: [rowOf(table.header), ...table.rows.map(rowOf)],
+    };
+}
+/** Fold a blockquote's child blocks into a single `quote`, keeping the rest after. */
+function emitBlockquote(bq, out) {
+    const inner = [];
+    tokensToBlocks(bq.tokens, 0, inner);
+    const quoteRuns = [];
+    const trailing = [];
+    for (const b of inner) {
+        if (QUOTE_FOLDABLE.has(b.type) && b.text) {
+            if (quoteRuns.length > 0)
+                quoteRuns.push({ t: '\n' });
+            quoteRuns.push(...b.text);
+        }
+        else {
+            trailing.push(b);
+        }
+    }
+    out.push({ type: 'quote', text: quoteRuns });
+    out.push(...trailing);
+}
+/** Walk top-level (or nested) marked block tokens, appending blocks to `out`. */
+function tokensToBlocks(tokens, depth, out) {
+    for (const tok of tokens) {
+        switch (tok.type) {
+            case 'space':
+            case 'def':
+                break;
+            case 'heading': {
+                const h = tok;
+                const { runs, images } = inlineRuns(h.tokens);
+                out.push({ type: 'heading', text: runs, props: { level: Math.min(3, Math.max(1, h.depth)) } });
+                for (const img of images)
+                    out.push(imagePlaceholderBlock(img));
+                break;
+            }
+            case 'paragraph':
+            case 'text': {
+                const p = tok;
+                const { runs, images } = inlineRuns(p.tokens ?? undefined);
+                const text = plainText(runs);
+                if (text.trim() !== '')
+                    out.push({ type: 'paragraph', text: runs });
+                for (const img of images)
+                    out.push(imagePlaceholderBlock(img));
+                break;
+            }
+            case 'list':
+                emitList(tok, depth, out);
+                break;
+            case 'blockquote':
+                emitBlockquote(tok, out);
+                break;
+            case 'code': {
+                const c = tok;
+                const language = c.lang?.trim().split(/\s+/)[0];
+                out.push({ type: 'code', text: [{ t: c.text }], ...(language ? { props: { language } } : {}) });
+                break;
+            }
+            case 'table':
+                out.push(emitTable(tok));
+                break;
+            case 'hr':
+                out.push({ type: 'divider' });
+                break;
+            case 'html': {
+                // Raw HTML block — degrade to a paragraph carrying the markup verbatim.
+                const raw = (tok.text ?? tok.raw ?? '').trim();
+                if (raw !== '')
+                    out.push({ type: 'paragraph', text: [{ t: raw }] });
+                break;
+            }
+            default: {
+                // Unknown / unsupported construct: never drop — degrade to its text.
+                const g = tok;
+                const raw = (g.text ?? g.raw ?? '').trim();
+                if (raw !== '')
+                    out.push({ type: 'paragraph', text: [{ t: raw }] });
+            }
+        }
+    }
+}
+// ── Front-matter ─────────────────────────────────────────────────────────────
+const FRONT_MATTER_RE = /^\uFEFF?---[ \t]*\r?\n([\s\S]*?)\r?\n---[ \t]*(?:\r?\n|$)/;
+/** Split a leading `---` YAML block off the source; returns `{title, rest, body}`. */
+function splitFrontMatter(markdown) {
+    const m = FRONT_MATTER_RE.exec(markdown);
+    if (!m)
+        return { rest: '', body: markdown.replace(/^\uFEFF/, '') };
+    const lines = m[1].split(/\r?\n/);
+    let title;
+    const rest = [];
+    for (const line of lines) {
+        const kv = /^(\w[\w-]*)\s*:\s*(.*)$/.exec(line);
+        if (title === undefined && kv && kv[1].toLowerCase() === 'title') {
+            title = kv[2].trim().replace(/^["']|["']$/g, '');
+        }
+        else {
+            rest.push(line);
+        }
+    }
+    return { title, rest: rest.join('\n').trim(), body: markdown.slice(m[0].length) };
+}
+// ── Public API ───────────────────────────────────────────────────────────────
+/**
+ * Parse a Markdown/GFM string into {@link ImportedBlock}s. Pure and DOM-free, so
+ * it is directly unit-testable and reusable wherever only the block list is
+ * wanted (e.g. importing a fragment into an existing page). Honors the
+ * "never silently drop content" contract: unknown constructs degrade to text,
+ * images become placeholders.
+ */
+export function markdownToBlocks(markdown) {
+    const tokens = lexer(markdown ?? '', { gfm: true });
+    const blocks = [];
+    tokensToBlocks(tokens, 0, blocks);
+    return blocks;
+}
+/**
+ * Parse a full Markdown/GFM document into an {@link ImportedDoc} (a single page)
+ * ready for {@link importDoc}. The page title is resolved as:
+ *
+ *  1. `opts.title` (explicit), else
+ *  2. front-matter `title:`, else
+ *  3. a leading heading promoted to the title (consumed from the body) when
+ *     `opts.firstHeadingAsTitle !== false`, else
+ *  4. `opts.defaultTitle` (default `'Imported document'`).
+ *
+ * Any non-title front-matter is preserved as a leading `yaml` code block.
+ */
+export function markdownToImportedDoc(markdown, opts = {}) {
+    const { title: fmTitle, rest, body } = splitFrontMatter(markdown ?? '');
+    const blocks = markdownToBlocks(body);
+    let title = opts.title ?? fmTitle;
+    if (title === undefined && opts.firstHeadingAsTitle !== false && blocks[0]?.type === 'heading') {
+        const headingText = plainText(blocks[0].text ?? []).trim();
+        if (headingText !== '') {
+            title = headingText;
+            blocks.shift();
+        }
+    }
+    title = title?.trim() || opts.defaultTitle || DEFAULT_TITLE;
+    // Preserve any leftover front-matter rather than dropping it.
+    if (rest !== '')
+        blocks.unshift({ type: 'code', text: [{ t: rest }], props: { language: 'yaml' } });
+    return { pages: [{ title, blocks }] };
+}
+//# sourceMappingURL=markdownImport.js.map

package/dist/markdownImport.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"markdownImport.js","sourceRoot":"","sources":["../src/markdownImport.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4CG;AAEH,OAAO,EAAC,KAAK,EAA0B,MAAM,QAAQ,CAAC;AACtD,OAAO,EACL,qBAAqB,GAMtB,MAAM,UAAU,CAAC;AAelB,MAAM,aAAa,GAAG,mBAAmB,CAAC;AAe1C,MAAM,QAAQ,GAAG,CAAC,CAAoB,EAAW,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;AAE9E,MAAM,SAAS,GAAG,CAAC,CAAgC,EAAE,CAAgC,EAAW,EAAE,CAChG,IAAI,CAAC,SAAS,CAAC,CAAC,IAAI,EAAE,CAAC,KAAK,IAAI,CAAC,SAAS,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;AAEtD,mFAAmF;AACnF,SAAS,OAAO,CAAC,GAAoB,EAAE,IAAY,EAAE,KAAwB;IAC3E,IAAI,IAAI,KAAK,EAAE;QAAE,OAAO;IACxB,MAAM,GAAG,GAAkB,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAC,CAAC,EAAE,IAAI,EAAE,CAAC,EAAE,EAAC,GAAG,KAAK,EAAC,EAAC,CAAC,CAAC,CAAC,EAAC,CAAC,EAAE,IAAI,EAAC,CAAC;IAClF,MAAM,IAAI,GAAG,GAAG,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IACjC,IAAI,IAAI,IAAI,SAAS,CAAC,IAAI,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC;QAAE,IAAI,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC;;QACjD,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AACrB,CAAC;AAED,MAAM,OAAO,GAAG,CAAC,GAAiB,EAAiB,EAAE,CAAC,CAAC;IACrD,IAAI,EAAE,OAAO;IACb,GAAG,EAAE,GAAG,CAAC,IAAI;IACb,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAC,GAAG,EAAE,GAAG,CAAC,IAAI,EAAC,CAAC,CAAC,CAAC,EAAE,CAAC;IACpC,GAAG,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAC,KAAK,EAAE,GAAG,CAAC,KAAK,EAAC,CAAC,CAAC,CAAC,EAAE,CAAC;CACzC,CAAC,CAAC;AAEH,gFAAgF;AAChF,SAAS,UAAU,CAAC,MAA2B,EAAE,KAAwB,EAAE,GAAoB,EAAE,GAAc;IAC7G,IAAI,CAAC,MAAM;QAAE,OAAO;IACpB,KAAK,MAAM,GAAG,IAAI,MAAM,EAAE,CAAC;QACzB,QAAQ,GAAG,CAAC,IAAI,EAAE,CAAC;YACnB,KAAK,MAAM,CAAC;YACZ,KAAK,QAAQ,CAAC,CAAC,CAAC;gBACd,MAAM,CAAC,GAAG,GAAkB,CAAC;gBAC7B,IAAI,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC;oBAAE,UAAU,CAAC,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;;oBACtE,OAAO,CAAC,GAAG,EAAE,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,GAAG,IAAI,EAAE,EAAE,KAAK,CAAC,CAAC;gBAChD,MAAM;YACR,CAAC;YACD,KAAK,QAAQ;gBACX,UAAU,CAAE,GAAqB,CAAC,MAAM,EAAE,EAAC,GAAG,KAAK,EAAE,CAAC,EAAE,IAAI,EAAC,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;gBACzE,MAAM;YACR,KAAK,IAAI;gBACP,UAAU,CAAE,GAAiB,CAAC,MAAM,EAAE,EAAC,GAAG,KAAK,EAAE,CAAC,EAAE,IAAI,EAAC,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;gBACrE,MAAM;YACR,KAAK,KAAK;gBACR,UAAU,CAAE,GAAkB,CAAC,MAAM,EAAE,EAAC,GAAG,KAAK,EAAE,CAAC,EAAE,IAAI,EAAC,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;gBACtE,MAAM;YACR,KAAK,UAAU;gBACb,OAAO,CAAC,GAAG,EAAG,GAAuB,CAAC,IAAI,EAAE,EAAC,GAAG,KAAK,EAAE,CAAC,EAAE,IAAI,EAAC,CAAC,CAAC;gBACjE,MAAM;YACR,KAAK,IAAI;gBACP,OAAO,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE,CAAC,CAAC;gBACvB,MAAM;YACR,KAAK,MAAM,CAAC,CAAC,CAAC;gBACZ,MAAM,IAAI,GAAG,GAAkB,CAAC;gBAChC,MAAM,MAAM,GAAsB,EAAC,GAAG,KAAK,EAAE,CAAC,EAAE,IAAI,CAAC,IAAI,EAAC,CAAC;gBAC3D,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC;oBAAE,UAAU,CAAC,IAAI,CAAC,MAAM,EAAE,MAAM,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;;oBAChF,OAAO,CAAC,GAAG,EAAE,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;gBAClD,MAAM;YACR,CAAC;YACD,KAAK,OAAO,CAAC,CAAC,CAAC;gBACb,MAAM,GAAG,GAAG,GAAmB,CAAC;gBAChC,IAAI,GAAG,CAAC,YAAY;oBAAE,OAAO,CAAC,GAAG,EAAE,GAAG,CAAC,IAAI,IAAI,GAAG,CAAC,IAAI,EAAE,EAAC,GAAG,KAAK,EAAE,CAAC,EAAE,GAAG,CAAC,IAAI,EAAC,CAAC,CAAC;;oBAC7E,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC;gBACnC,MAAM;YACR,CAAC;YACD,KAAK,UAAU;gBACb,0EAA0E;gBAC1E,MAAM;YACR,KAAK,MAAM;gBACT,oDAAoD;gBACpD,OAAO,CAAC,GAAG,EAAG,GAAmB,CAAC,IAAI,IAAK,GAAmB,CAAC,GAAG,IAAI,EAAE,EAAE,KAAK,CAAC,CAAC;gBACjF,MAAM;YACR,OAAO,CAAC,CAAC,CAAC;gBACR,sEAAsE;gBACtE,MAAM,CAAC,GAAG,GAAqB,CAAC;gBAChC,IAAI,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC;oBAAE,UAAU,CAAC,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;;oBACtE,OAAO,CAAC,GAAG,EAAG,CAAC,CAAC,IAAe,IAAI,CAAC,CAAC,GAAG,IAAI,EAAE,EAAE,KAAK,CAAC,CAAC;YAC9D,CAAC;QACD,CAAC;IACH,CAAC;AACH,CAAC;AAED,kFAAkF;AAClF,SAAS,UAAU,CAAC,MAA2B,EAAE,YAAY,GAAG,KAAK;IACnE,MAAM,GAAG,GAAc,EAAC,MAAM,EAAE,EAAE,EAAE,YAAY,EAAC,CAAC;IAClD,MAAM,IAAI,GAAoB,EAAE,CAAC;IACjC,UAAU,CAAC,MAAM,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,CAAC,CAAC;IAClC,OAAO,EAAC,IAAI,EAAE,MAAM,EAAE,GAAG,CAAC,MAAM,EAAC,CAAC;AACpC,CAAC;AAED,MAAM,SAAS,GAAG,CAAC,IAAqB,EAAU,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AAEnF,gFAAgF;AAEhF,2EAA2E;AAC3E,MAAM,cAAc,GAAG,IAAI,GAAG,CAAC,CAAC,WAAW,EAAE,SAAS,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,CAAC;AAElF,qFAAqF;AACrF,SAAS,QAAQ,CAAC,IAAiB,EAAE,KAAa,EAAE,GAAoB;IACtE,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,QAAQ,CAAC;IAChD,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,KAAK,EAAE,CAAC;QAC9B,MAAM,IAAI,GAAoB,EAAE,CAAC;QACjC,MAAM,MAAM,GAAoB,EAAE,CAAC;QACnC,MAAM,QAAQ,GAAoB,EAAE,CAAC;QACrC,MAAM,MAAM,GAAkB,EAAE,CAAC;QACjC,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;YAChC,IAAI,KAAK,CAAC,IAAI,KAAK,UAAU,EAAE,CAAC;gBAC9B,wEAAwE;gBACxE,SAAS;YACX,CAAC;iBAAM,IAAI,KAAK,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;gBACjC,MAAM,CAAC,IAAI,CAAC,KAAoB,CAAC,CAAC;YACpC,CAAC;iBAAM,IAAI,KAAK,CAAC,IAAI,KAAK,MAAM,IAAI,KAAK,CAAC,IAAI,KAAK,WAAW,EAAE,CAAC;gBAC/D,MAAM,CAAC,GAAG,UAAU,CAAE,KAAwC,CAAC,MAAM,CAAC,CAAC;gBACvE,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC;gBACrB,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC;YAC3B,CAAC;iBAAM,CAAC;gBACN,uEAAuE;gBACvE,mDAAmD;gBACnD,cAAc,CAAC,CAAC,KAAK,CAAC,EAAE,KAAK,EAAE,QAAQ,CAAC,CAAC;YAC3C,CAAC;QACH,CAAC;QACD,MAAM,MAAM,GAAG,IAAI,CAAC,IAAI,KAAK,IAAI,CAAC;QAClC,MAAM,KAAK,GAA4B,EAAE,CAAC;QAC1C,IAAI,MAAM,EAAE,CAAC;YACX,IAAI,IAAI,CAAC,OAAO;gBAAE,KAAK,CAAC,OAAO,GAAG,IAAI,CAAC;QACzC,CAAC;aAAM,CAAC;YACN,KAAK,CAAC,IAAI,GAAG,IAAI,CAAC;QACpB,CAAC;QACD,2EAA2E;QAC3E,0EAA0E;QAC1E,IAAI,KAAK,GAAG,CAAC;YAAE,KAAK,CAAC,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;QACjD,GAAG,CAAC,IAAI,CAAC,EAAC,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,EAAC,KAAK,EAAC,CAAC,CAAC,CAAC,EAAE,CAAC,EAAC,CAAC,CAAC;QAC1G,KAAK,MAAM,GAAG,IAAI,MAAM;YAAE,GAAG,CAAC,IAAI,CAAC,qBAAqB,CAAC,GAAG,CAAC,CAAC,CAAC;QAC/D,GAAG,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC,CAAC;QACtB,KAAK,MAAM,EAAE,IAAI,MAAM;YAAE,QAAQ,CAAC,EAAE,EAAE,KAAK,GAAG,CAAC,EAAE,GAAG,CAAC,CAAC;IACxD,CAAC;AACH,CAAC;AAED,sEAAsE;AACtE,SAAS,SAAS,CAAC,KAAmB;IACpC,MAAM,KAAK,GAAG,CAAC,KAAyB,EAAiB,EAAE,CAAC,CAAC;QAC3D,IAAI,EAAE,KAAK;QACX,QAAQ,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,EAAC,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,UAAU,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC,IAAI,EAAC,CAAC,CAAC;KAC1F,CAAC,CAAC;IACH,OAAO;QACL,IAAI,EAAE,OAAO;QACb,KAAK,EAAE,EAAC,MAAM,EAAE,IAAI,EAAC;QACrB,QAAQ,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,GAAG,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;KAC1D,CAAC;AACJ,CAAC;AAED,sFAAsF;AACtF,SAAS,cAAc,CAAC,EAAqB,EAAE,GAAoB;IACjE,MAAM,KAAK,GAAoB,EAAE,CAAC;IAClC,cAAc,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,EAAE,KAAK,CAAC,CAAC;IACpC,MAAM,SAAS,GAAoB,EAAE,CAAC;IACtC,MAAM,QAAQ,GAAoB,EAAE,CAAC;IACrC,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;QACtB,IAAI,cAAc,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;YACzC,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC;gBAAE,SAAS,CAAC,IAAI,CAAC,EAAC,CAAC,EAAE,IAAI,EAAC,CAAC,CAAC;YACpD,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC;QAC5B,CAAC;aAAM,CAAC;YACN,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QACnB,CAAC;IACH,CAAC;IACD,GAAG,CAAC,IAAI,CAAC,EAAC,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAC,CAAC,CAAC;IAC3C,GAAG,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC,CAAC;AACxB,CAAC;AAED,iFAAiF;AACjF,SAAS,cAAc,CAAC,MAAe,EAAE,KAAa,EAAE,GAAoB;IAC1E,KAAK,MAAM,GAAG,IAAI,MAAM,EAAE,CAAC;QACzB,QAAQ,GAAG,CAAC,IAAI,EAAE,CAAC;YACnB,KAAK,OAAO,CAAC;YACb,KAAK,KAAK;gBACR,MAAM;YACR,KAAK,SAAS,CAAC,CAAC,CAAC;gBACf,MAAM,CAAC,GAAG,GAAqB,CAAC;gBAChC,MAAM,EAAC,IAAI,EAAE,MAAM,EAAC,GAAG,UAAU,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC;gBAC5C,GAAG,CAAC,IAAI,CAAC,EAAC,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,EAAC,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,KAAK,CAAC,CAAC,EAAC,EAAC,CAAC,CAAC;gBAC3F,KAAK,MAAM,GAAG,IAAI,MAAM;oBAAE,GAAG,CAAC,IAAI,CAAC,qBAAqB,CAAC,GAAG,CAAC,CAAC,CAAC;gBAC/D,MAAM;YACR,CAAC;YACD,KAAK,WAAW,CAAC;YACjB,KAAK,MAAM,CAAC,CAAC,CAAC;gBACZ,MAAM,CAAC,GAAG,GAAqC,CAAC;gBAChD,MAAM,EAAC,IAAI,EAAE,MAAM,EAAC,GAAG,UAAU,CAAC,CAAC,CAAC,MAAM,IAAI,SAAS,CAAC,CAAC;gBACzD,MAAM,IAAI,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC;gBAC7B,IAAI,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE;oBAAE,GAAG,CAAC,IAAI,CAAC,EAAC,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,IAAI,EAAC,CAAC,CAAC;gBAClE,KAAK,MAAM,GAAG,IAAI,MAAM;oBAAE,GAAG,CAAC,IAAI,CAAC,qBAAqB,CAAC,GAAG,CAAC,CAAC,CAAC;gBAC/D,MAAM;YACR,CAAC;YACD,KAAK,MAAM;gBACT,QAAQ,CAAC,GAAkB,EAAE,KAAK,EAAE,GAAG,CAAC,CAAC;gBACzC,MAAM;YACR,KAAK,YAAY;gBACf,cAAc,CAAC,GAAwB,EAAE,GAAG,CAAC,CAAC;gBAC9C,MAAM;YACR,KAAK,MAAM,CAAC,CAAC,CAAC;gBACZ,MAAM,CAAC,GAAG,GAAkB,CAAC;gBAC7B,MAAM,QAAQ,GAAG,CAAC,CAAC,IAAI,EAAE,IAAI,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;gBAChD,GAAG,CAAC,IAAI,CAAC,EAAC,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC,EAAC,CAAC,EAAE,CAAC,CAAC,IAAI,EAAC,CAAC,EAAE,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAC,KAAK,EAAE,EAAC,QAAQ,EAAC,EAAC,CAAC,CAAC,CAAC,EAAE,CAAC,EAAC,CAAC,CAAC;gBACxF,MAAM;YACR,CAAC;YACD,KAAK,OAAO;gBACV,GAAG,CAAC,IAAI,CAAC,SAAS,CAAC,GAAmB,CAAC,CAAC,CAAC;gBACzC,MAAM;YACR,KAAK,IAAI;gBACP,GAAG,CAAC,IAAI,CAAC,EAAC,IAAI,EAAE,SAAS,EAAC,CAAC,CAAC;gBAC5B,MAAM;YACR,KAAK,MAAM,CAAC,CAAC,CAAC;gBACZ,wEAAwE;gBACxE,MAAM,GAAG,GAAG,CAAE,GAAmB,CAAC,IAAI,IAAK,GAAmB,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;gBACjF,IAAI,GAAG,KAAK,EAAE;oBAAE,GAAG,CAAC,IAAI,CAAC,EAAC,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC,EAAC,CAAC,EAAE,GAAG,EAAC,CAAC,EAAC,CAAC,CAAC;gBAChE,MAAM;YACR,CAAC;YACD,OAAO,CAAC,CAAC,CAAC;gBACR,qEAAqE;gBACrE,MAAM,CAAC,GAAG,GAAqB,CAAC;gBAChC,MAAM,GAAG,GAAG,CAAE,CAAC,CAAC,IAAe,IAAI,CAAC,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;gBACvD,IAAI,GAAG,KAAK,EAAE;oBAAE,GAAG,CAAC,IAAI,CAAC,EAAC,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC,EAAC,CAAC,EAAE,GAAG,EAAC,CAAC,EAAC,CAAC,CAAC;YAClE,CAAC;QACD,CAAC;IACH,CAAC;AACH,CAAC;AAED,gFAAgF;AAEhF,MAAM,eAAe,GAAG,2DAA2D,CAAC;AAEpF,sFAAsF;AACtF,SAAS,gBAAgB,CAAC,QAAgB;IACxC,MAAM,CAAC,GAAG,eAAe,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IACzC,IAAI,CAAC,CAAC;QAAE,OAAO,EAAC,IAAI,EAAE,EAAE,EAAE,IAAI,EAAE,QAAQ,CAAC,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,EAAC,CAAC;IACjE,MAAM,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;IAClC,IAAI,KAAyB,CAAC;IAC9B,MAAM,IAAI,GAAa,EAAE,CAAC;IAC1B,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,EAAE,GAAG,yBAAyB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAChD,IAAI,KAAK,KAAK,SAAS,IAAI,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,KAAK,OAAO,EAAE,CAAC;YACjE,KAAK,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,OAAO,CAAC,cAAc,EAAE,EAAE,CAAC,CAAC;QACnD,CAAC;aAAM,CAAC;YACN,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAClB,CAAC;IACH,CAAC;IACD,OAAO,EAAC,KAAK,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,IAAI,EAAE,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,EAAC,CAAC;AAClF,CAAC;AAED,gFAAgF;AAEhF;;;;;;GAMG;AACH,MAAM,UAAU,gBAAgB,CAAC,QAAgB;IAC/C,MAAM,MAAM,GAAG,KAAK,CAAC,QAAQ,IAAI,EAAE,EAAE,EAAC,GAAG,EAAE,IAAI,EAAC,CAAC,CAAC;IAClD,MAAM,MAAM,GAAoB,EAAE,CAAC;IACnC,cAAc,CAAC,MAAM,EAAE,CAAC,EAAE,MAAM,CAAC,CAAC;IAClC,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,qBAAqB,CAAC,QAAgB,EAAE,OAA8B,EAAE;IACtF,MAAM,EAAC,KAAK,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAC,GAAG,gBAAgB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC;IACtE,MAAM,MAAM,GAAG,gBAAgB,CAAC,IAAI,CAAC,CAAC;IAEtC,IAAI,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,OAAO,CAAC;IAClC,IAAI,KAAK,KAAK,SAAS,IAAI,IAAI,CAAC,mBAAmB,KAAK,KAAK,IAAI,MAAM,CAAC,CAAC,CAAC,EAAE,IAAI,KAAK,SAAS,EAAE,CAAC;QAC/F,MAAM,WAAW,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;QAC3D,IAAI,WAAW,KAAK,EAAE,EAAE,CAAC;YACvB,KAAK,GAAG,WAAW,CAAC;YACpB,MAAM,CAAC,KAAK,EAAE,CAAC;QACjB,CAAC;IACH,CAAC;IACD,KAAK,GAAG,KAAK,EAAE,IAAI,EAAE,IAAI,IAAI,CAAC,YAAY,IAAI,aAAa,CAAC;IAE5D,8DAA8D;IAC9D,IAAI,IAAI,KAAK,EAAE;QAAE,MAAM,CAAC,OAAO,CAAC,EAAC,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC,EAAC,CAAC,EAAE,IAAI,EAAC,CAAC,EAAE,KAAK,EAAE,EAAC,QAAQ,EAAE,MAAM,EAAC,EAAC,CAAC,CAAC;IAE9F,OAAO,EAAC,KAAK,EAAE,CAAC,EAAC,KAAK,EAAE,MAAM,EAAC,CAAC,EAAC,CAAC;AACpC,CAAC"}

package/dist/notionImport.d.ts

@@ -0,0 +1,58 @@
+/**
+ * **Notion "Markdown & CSV" export → import IR.** Ingests the zip Notion produces
+ * for *Export → Markdown & CSV* and emits the format-agnostic {@link ImportedDoc}
+ * (see `import.ts`), so {@link importDoc} can land the whole tree through the
+ * existing data paths (Strategy B — a bundle the server id-remaps, link-rewrites,
+ * and name-dedups as one graph).
+ *
+ * **Export shape.** Notion writes a forest of folders + files; every page/database
+ * filename is suffixed with the 32-hex Notion id (dashes stripped):
+ *
+ * ```
+ *  Export-<uuid>/                         ← transparent wrapper (any non-page dir)
+ *    Launch Plan <id>.md                  ← a page
+ *    Launch Plan <id>/                     ← its child pages live here
+ *      Sub Note <id>.md
+ *    Team Tasks <id>.csv                  ← a database (schema + rows)
+ *    Team Tasks <id>/                      ← the database's row pages (bodies)
+ *      Ship v1 <id>.md
+ * ```
+ *
+ *  - **Folder nesting → the page tree.** A `.md` at `A/B/C.md` whose sibling
+ *    folder `A/B/C/` holds more `.md`/`.csv` becomes an {@link ImportedPage} with
+ *    those as `children`. Titles drop the hash suffix.
+ *  - **Each `.md` → blocks** via the merged Markdown importer
+ *    ({@link markdownToImportedDoc}, T2) — reused verbatim, not reparsed.
+ *  - **A `.csv` → an {@link ImportedDatabase}.** Column property types are inferred
+ *    from the cell values (text / number / checkbox / date / select / multi_select
+ *    / url / email — a comma-bearing cell ⇒ multi_select); each row links to its
+ *    `.md` page in the sibling folder for the row body. **Scalar properties only**
+ *    — a CSV cannot encode relation/rollup/formula targets, so those degrade to a
+ *    plain scalar (never a dangling `relation`), the v1 owner decision.
+ *  - **Internal links** (the hash-suffixed relative `.md`/`.csv` path Notion
+ *    writes) resolve to an `@`-mention run when the target was imported; an
+ *    unresolved internal link keeps its visible text (never dropped). External
+ *    `http(s)`/`mailto:` links and file/asset refs are preserved as links.
+ *  - **Icon** — a leading emoji on the page title is lifted to `sys_icon`.
+ *  - **Images/files** — handled by the Markdown importer as the import core's
+ *    {@link imagePlaceholderBlock} / a link run that preserves the ref.
+ *  - **Toggles** — Notion flattens these to a heading/list in Markdown; the
+ *    content survives via T2 (we do not reconstruct an `accordion`, whose
+ *    structure does not survive the export).
+ *
+ * **Never silently drop.** Unknown constructs degrade to text (T2's contract);
+ * any `.md`/`.csv` the tree walk somehow misses is appended as a top-level page;
+ * a row with no matching body still lands with its properties.
+ */
+import { type ImportedDoc } from './import';
+/**
+ * RFC-4180-ish CSV → a matrix of rows. Handles quoted fields containing commas
+ * and newlines, doubled `""` escapes, and CRLF — everything Notion emits.
+ */
+export declare function parseCsv(text: string): string[][];
+/**
+ * Parse a Notion *Markdown & CSV* export zip into an {@link ImportedDoc} ready for
+ * {@link importDoc}. Pure (no client, no I/O beyond the in-memory unzip), so it is
+ * directly unit-testable. Throws only when the bytes are not a readable zip.
+ */
+export declare function notionExportToImportedDoc(zipBytes: Uint8Array): ImportedDoc;