@dxos/plugin-magazine 0.0.0

package/src/util/text.ts

@@ -0,0 +1,98 @@
+//
+// Copyright 2026 DXOS.org
+//
+
+/**
+ * Lightweight HTML helpers used by the Magazine curation flow.
+ * Regex-based; deliberately avoids a parser dependency. Good enough for
+ * snippet preview and image discovery; not a content extractor.
+ */
+
+const DEFAULT_SNIPPET_LENGTH = 280;
+
+/** Decode numeric (decimal + hex) and a handful of common named HTML entities. */
+export const decodeEntities = (input: string): string =>
+  input
+    .replace(/&#x([0-9a-fA-F]+);/g, (_, hex: string) => {
+      const cp = Number.parseInt(hex, 16);
+      return Number.isFinite(cp) ? String.fromCodePoint(cp) : '';
+    })
+    .replace(/&#(\d+);/g, (_, dec: string) => {
+      const cp = Number.parseInt(dec, 10);
+      return Number.isFinite(cp) ? String.fromCodePoint(cp) : '';
+    })
+    .replace(/ /gi, ' ')
+    .replace(/&/gi, '&')
+    .replace(/"/gi, '"')
+    .replace(/'/gi, "'")
+    .replace(/&lt;/gi, '<')
+    .replace(/&gt;/gi, '>');
+
+/**
+ * Strip HTML tags and decode common entities to produce plain text.
+ * Block-level tag boundaries (paragraphs, divs, headings, list items, line breaks)
+ * are converted to newlines so paragraph structure survives the stripping.
+ *
+ * Decodes entities up front so entity-encoded HTML (`&lt;p&gt;...&lt;/p&gt;` — common in
+ * RSS `<description>` payloads, especially under fast-xml-parser stopNodes) gets stripped
+ * the same as raw HTML. Without that, the encoded tags would survive stripping and only
+ * decode at the end, leaking literal `<p>` into the output.
+ */
+export const stripHtml = (html: string): string => {
+  if (!html) {
+    return '';
+  }
+
+  const decoded = decodeEntities(html);
+  // Remove script/style blocks entirely.
+  const withoutScripts = decoded
+    .replace(/<script\b[^<]*(?:(?!<\/script>)<[^<]*)*<\/script>/gi, ' ')
+    .replace(/<style\b[^<]*(?:(?!<\/style>)<[^<]*)*<\/style>/gi, ' ');
+  // Insert paragraph breaks at common block-level boundaries before stripping tags.
+  const withParagraphBreaks = withoutScripts
+    .replace(/<\s*br\s*\/?\s*>/gi, '\n')
+    .replace(/<\/\s*(?:p|div|section|article|header|footer|aside|h[1-6]|li|ul|ol|blockquote|pre|tr)\s*>/gi, '\n\n');
+  const withoutTags = withParagraphBreaks.replace(/<[^>]+>/g, '');
+  return withoutTags
+    .replace(/[\t\f\v ]+/g, ' ') // Collapse horizontal whitespace, preserve newlines.
+    .replace(/ *\n */g, '\n') // Trim spaces around newlines.
+    .replace(/\n{3,}/g, '\n\n') // Collapse 3+ blank lines to a single blank line.
+    .trim();
+};
+
+/** Produce a snippet of approximately `length` characters, cut on a word boundary. */
+export const makeSnippet = (text: string, length: number = DEFAULT_SNIPPET_LENGTH): string => {
+  if (!text) {
+    return '';
+  }
+  if (text.length <= length) {
+    return text;
+  }
+  const clipped = text.slice(0, length);
+  const lastSpace = clipped.lastIndexOf(' ');
+  const base = lastSpace > length * 0.6 ? clipped.slice(0, lastSpace) : clipped;
+  return `${base.trimEnd()}…`;
+};
+
+/** Extract image URLs from HTML. og:image (meta property) first, then <img src="...">. */
+export const extractImageUrls = (html: string): string[] => {
+  if (!html) {
+    return [];
+  }
+  const urls: string[] = [];
+  // Attributes may appear in either order on <meta>, so try both.
+  const ogMatch =
+    html.match(/<meta[^>]+property=["']og:image["'][^>]+content=["']([^"']+)["']/i) ??
+    html.match(/<meta[^>]+content=["']([^"']+)["'][^>]+property=["']og:image["']/i);
+  if (ogMatch?.[1]) {
+    urls.push(ogMatch[1]);
+  }
+  const imgRegex = /<img[^>]+src=["']([^"']+)["']/gi;
+  let match: RegExpExecArray | null;
+  while ((match = imgRegex.exec(html)) != null) {
+    if (match[1] && !urls.includes(match[1])) {
+      urls.push(match[1]);
+    }
+  }
+  return urls;
+};

package/src/vite-env.d.ts

@@ -0,0 +1,10 @@
+/// <reference types="vite/client" />
+
+//
+// Copyright 2026 DXOS.org
+//
+
+declare module '*.mdl?raw' {
+  const content: string;
+  export default content;
+}