first commit

packages/gutenberg-to-portable-text/src/transformers/embed.ts

@@ -0,0 +1,142 @@
+/**
+ * Transformers for WordPress embed blocks
+ */
+
+import type { BlockTransformer } from "../types.js";
+import { attrString } from "../types.js";
+
+// Regex patterns for embed parsing
+const IFRAME_SRC_PATTERN = /<iframe[^>]+src=["']([^"']+)["']/i;
+const VIDEO_SRC_PATTERN = /<video[^>]+src=["']([^"']+)["']/i;
+const VIDEO_SOURCE_PATTERN = /<source[^>]+src=["']([^"']+)["']/i;
+const AUDIO_SRC_PATTERN = /<audio[^>]+src=["']([^"']+)["']/i;
+const AUDIO_SOURCE_PATTERN = /<source[^>]+src=["']([^"']+)["']/i;
+
+/**
+ * core/embed and variants → embed block
+ */
+export const embed: BlockTransformer = (block, _options, context) => {
+	const url = attrString(block.attrs, "url");
+	const providerSlug = attrString(block.attrs, "providerNameSlug");
+
+	// Extract iframe src if present
+	const iframeMatch = block.innerHTML.match(IFRAME_SRC_PATTERN);
+	const iframeSrc = iframeMatch?.[1];
+
+	return [
+		{
+			_type: "embed",
+			_key: context.generateKey(),
+			url: url || iframeSrc || "",
+			provider: providerSlug || detectProvider(url || iframeSrc || ""),
+			html: block.innerHTML.trim() || undefined,
+		},
+	];
+};
+
+/**
+ * core-embed/youtube → embed block
+ */
+export const youtube: BlockTransformer = (block, options, context) => {
+	return embed(block, options, context);
+};
+
+/**
+ * core-embed/twitter → embed block
+ */
+export const twitter: BlockTransformer = (block, options, context) => {
+	return embed(block, options, context);
+};
+
+/**
+ * core-embed/vimeo → embed block
+ */
+export const vimeo: BlockTransformer = (block, options, context) => {
+	return embed(block, options, context);
+};
+
+/**
+ * core/video → embed block (self-hosted video)
+ */
+export const video: BlockTransformer = (block, _options, context) => {
+	const src = attrString(block.attrs, "src");
+
+	// Extract from video tag if not in attrs
+	const videoMatch = block.innerHTML.match(VIDEO_SRC_PATTERN);
+	const sourceMatch = block.innerHTML.match(VIDEO_SOURCE_PATTERN);
+	const videoSrc = src || videoMatch?.[1] || sourceMatch?.[1];
+
+	return [
+		{
+			_type: "embed",
+			_key: context.generateKey(),
+			url: videoSrc || "",
+			provider: "video",
+			html: block.innerHTML.trim() || undefined,
+		},
+	];
+};
+
+/**
+ * core/audio → embed block (self-hosted audio)
+ */
+export const audio: BlockTransformer = (block, _options, context) => {
+	const src = attrString(block.attrs, "src");
+
+	// Extract from audio tag if not in attrs
+	const audioMatch = block.innerHTML.match(AUDIO_SRC_PATTERN);
+	const sourceMatch = block.innerHTML.match(AUDIO_SOURCE_PATTERN);
+	const audioSrc = src || audioMatch?.[1] || sourceMatch?.[1];
+
+	return [
+		{
+			_type: "embed",
+			_key: context.generateKey(),
+			url: audioSrc || "",
+			provider: "audio",
+			html: block.innerHTML.trim() || undefined,
+		},
+	];
+};
+
+/**
+ * Detect embed provider from URL
+ */
+function detectProvider(url: string): string | undefined {
+	if (!url) return undefined;
+
+	const urlLower = url.toLowerCase();
+
+	if (urlLower.includes("youtube.com") || urlLower.includes("youtu.be")) {
+		return "youtube";
+	}
+	if (urlLower.includes("vimeo.com")) {
+		return "vimeo";
+	}
+	if (urlLower.includes("twitter.com") || urlLower.includes("x.com")) {
+		return "twitter";
+	}
+	if (urlLower.includes("instagram.com")) {
+		return "instagram";
+	}
+	if (urlLower.includes("facebook.com")) {
+		return "facebook";
+	}
+	if (urlLower.includes("tiktok.com")) {
+		return "tiktok";
+	}
+	if (urlLower.includes("spotify.com")) {
+		return "spotify";
+	}
+	if (urlLower.includes("soundcloud.com")) {
+		return "soundcloud";
+	}
+	if (urlLower.includes("codepen.io")) {
+		return "codepen";
+	}
+	if (urlLower.includes("gist.github.com")) {
+		return "gist";
+	}
+
+	return undefined;
+}

packages/gutenberg-to-portable-text/src/transformers/index.ts

@@ -0,0 +1,115 @@
+/**
+ * Block transformers registry
+ */
+
+import type { BlockTransformer, PortableTextBlock } from "../types.js";
+import * as core from "./core.js";
+import * as embed from "./embed.js";
+
+/**
+ * Default block transformers for core WordPress blocks
+ */
+export const defaultTransformers: Record<string, BlockTransformer> = {
+	// Text blocks
+	"core/paragraph": core.paragraph,
+	"core/heading": core.heading,
+	"core/list": core.list,
+	"core/quote": core.quote,
+	"core/code": core.code,
+	"core/preformatted": core.preformatted,
+	"core/pullquote": core.pullquote,
+	"core/verse": core.verse,
+
+	// Media blocks
+	"core/image": core.image,
+	"core/gallery": core.gallery,
+	"core/file": core.file,
+	"core/media-text": core.mediaText,
+	"core/cover": core.cover,
+
+	// Layout blocks
+	"core/columns": core.columns,
+	"core/group": core.group,
+	"core/separator": core.separator,
+	"core/spacer": core.separator,
+	"core/table": core.table,
+	"core/buttons": core.buttons,
+	"core/button": core.button,
+
+	// Structural blocks
+	"core/more": core.more,
+	"core/nextpage": core.nextpage,
+
+	// Pass-through blocks (preserve as HTML)
+	"core/html": core.html,
+	"core/shortcode": core.shortcode,
+
+	// Embed blocks
+	"core/embed": embed.embed,
+	"core/video": embed.video,
+	"core/audio": embed.audio,
+
+	// Legacy embed block names (WP < 5.6)
+	"core-embed/youtube": embed.youtube,
+	"core-embed/twitter": embed.twitter,
+	"core-embed/vimeo": embed.vimeo,
+	"core-embed/facebook": embed.embed,
+	"core-embed/instagram": embed.embed,
+	"core-embed/soundcloud": embed.embed,
+	"core-embed/spotify": embed.embed,
+};
+
+/**
+ * Fallback transformer for unknown blocks
+ * Stores the original HTML for manual review
+ */
+export const fallbackTransformer: BlockTransformer = (
+	block,
+	_options,
+	context,
+): PortableTextBlock[] => {
+	// Skip completely empty blocks
+	if (!block.innerHTML.trim() && block.innerBlocks.length === 0) {
+		return [];
+	}
+
+	// If it has inner blocks, try to transform those
+	if (block.innerBlocks.length > 0) {
+		return context.transformBlocks(block.innerBlocks);
+	}
+
+	// Store as HTML fallback
+	return [
+		{
+			_type: "htmlBlock",
+			_key: context.generateKey(),
+			html: block.innerHTML,
+			originalBlockName: block.blockName,
+			originalAttrs: Object.keys(block.attrs).length > 0 ? block.attrs : undefined,
+		},
+	];
+};
+
+/**
+ * Get transformer for a block
+ */
+export function getTransformer(
+	blockName: string | null,
+	customTransformers?: Record<string, BlockTransformer>,
+): BlockTransformer {
+	if (!blockName) {
+		return fallbackTransformer;
+	}
+
+	// Check custom transformers first
+	if (customTransformers?.[blockName]) {
+		return customTransformers[blockName];
+	}
+
+	// Check default transformers
+	if (defaultTransformers[blockName]) {
+		return defaultTransformers[blockName];
+	}
+
+	return fallbackTransformer;
+}