v0.3.2

src/content/discovery.ts

@@ -0,0 +1,154 @@
+import type {
+  ImageCandidate,
+  PageCandidate,
+  PageDisclosureCandidate,
+  TextCandidate
+} from '../shared/types';
+import { discoverHeuristics } from './heuristics';
+
+const IMAGE_ID_ATTRIBUTE = 'data-slopdetect-image-id';
+const MAX_TEXT_BLOCKS = 8;
+const MIN_TEXT_CHARACTERS = 300;
+
+function isVisible(element: Element): boolean {
+  const rect = element.getBoundingClientRect();
+  const style = window.getComputedStyle(element);
+  return rect.width > 0 && rect.height > 0 && style.visibility !== 'hidden' && style.display !== 'none';
+}
+
+function makeImageCandidate(element: Element, url: string, alt = ''): ImageCandidate | undefined {
+  if (!url || !isVisible(element)) return undefined;
+  const rect = element.getBoundingClientRect();
+  if (rect.width < 16 || rect.height < 16) return undefined;
+  const existingId = element.getAttribute(IMAGE_ID_ATTRIBUTE);
+  const id = existingId || crypto.randomUUID();
+  element.setAttribute(IMAGE_ID_ATTRIBUTE, id);
+  return {
+    id,
+    url: new URL(url, document.baseURI).href,
+    alt: alt.trim().slice(0, 200),
+    width: Math.round(rect.width),
+    height: Math.round(rect.height)
+  };
+}
+
+function discoverImages(): ImageCandidate[] {
+  const candidates: ImageCandidate[] = [];
+  const seenUrls = new Set<string>();
+
+  for (const image of document.querySelectorAll('img')) {
+    const candidate = makeImageCandidate(image, image.currentSrc || image.src, image.alt);
+    if (candidate && !seenUrls.has(candidate.url)) {
+      candidates.push(candidate);
+      seenUrls.add(candidate.url);
+    }
+  }
+
+  for (const image of document.querySelectorAll('svg image')) {
+    const href = image.getAttribute('href') || image.getAttribute('xlink:href') || '';
+    const candidate = makeImageCandidate(image, href, image.getAttribute('aria-label') || 'SVG image');
+    if (candidate && !seenUrls.has(candidate.url)) {
+      candidates.push(candidate);
+      seenUrls.add(candidate.url);
+    }
+  }
+
+  const elements = [...document.querySelectorAll('body *')].slice(0, 2000);
+  for (const element of elements) {
+    const background = window.getComputedStyle(element).backgroundImage;
+    const match = /^url\(["']?(.*?)["']?\)$/.exec(background);
+    if (!match?.[1]) continue;
+    const candidate = makeImageCandidate(element, match[1], element.getAttribute('aria-label') || 'Background image');
+    if (candidate && !seenUrls.has(candidate.url)) {
+      candidates.push(candidate);
+      seenUrls.add(candidate.url);
+    }
+  }
+
+  return candidates;
+}
+
+export function estimateTokens(text: string): number {
+  return text.trim() ? Math.ceil(text.trim().split(/\s+/).length * 1.3) : 0;
+}
+
+function discoverText(): TextCandidate[] {
+  const selectors = 'article p, main p, [role="main"] p, article li, main li, blockquote';
+  const values = [...document.querySelectorAll<HTMLElement>(selectors)]
+    .filter(isVisible)
+    .map((element) => element.innerText.trim())
+    .filter((text) => text.length >= MIN_TEXT_CHARACTERS);
+
+  const unique = [...new Set(values)].sort((left, right) => right.length - left.length).slice(0, MAX_TEXT_BLOCKS);
+  if (unique.length === 0) {
+    const root = document.querySelector<HTMLElement>('article, main, [role="main"]') || document.body;
+    const combined = root?.innerText.trim().slice(0, 24_000) || '';
+    if (combined.length >= MIN_TEXT_CHARACTERS) unique.push(combined);
+  }
+
+  return unique.map((text) => ({ id: crypto.randomUUID(), text, tokenEstimate: estimateTokens(text) }));
+}
+
+function pushDisclosure(
+  output: PageDisclosureCandidate[],
+  source: string,
+  value: string | null | undefined
+): void {
+  const trimmed = value?.trim();
+  if (trimmed) output.push({ source, value: trimmed.slice(0, 2000) });
+}
+
+function discoverDisclosures(): PageDisclosureCandidate[] {
+  const candidates: PageDisclosureCandidate[] = [];
+  const interestingMeta = /ai|artificial|synthetic|provenance|credential|generator|content-origin/i;
+
+  for (const meta of document.querySelectorAll<HTMLMetaElement>('meta[name], meta[property]')) {
+    const name = meta.name || meta.getAttribute('property') || '';
+    if (interestingMeta.test(name)) {
+      pushDisclosure(candidates, `meta:${name}`, meta.content);
+    }
+  }
+
+  for (const link of document.querySelectorAll<HTMLLinkElement>('link[rel]')) {
+    if (/credential|provenance|c2pa|manifest/i.test(link.rel)) {
+      pushDisclosure(candidates, `link:${link.rel}`, link.href);
+    }
+  }
+
+  for (const script of document.querySelectorAll<HTMLScriptElement>('script[type="application/ld+json"]')) {
+    if (/"(?:aiGenerated|isAIGenerated|contentOrigin|digitalSourceType)"\s*:\s*(?:true|"[^"]*(?:trainedAlgorithmicMedia|synthetic|AI)[^"]*")/i.test(script.textContent || '')) {
+      pushDisclosure(candidates, 'structured-data', script.textContent);
+    }
+  }
+
+  const labelSelectors = 'figcaption, small, [class*="disclosure" i], [class*="provenance" i], [class*="credit" i], [aria-label*="AI-generated" i]';
+  for (const element of document.querySelectorAll<HTMLElement>(labelSelectors)) {
+    const value = (element.innerText || element.getAttribute('aria-label') || '').trim();
+    if (value.length <= 300 && /\b(?:AI[- ]generated|generated by AI|artificially generated|created with AI|synthetic media)\b/i.test(value)) {
+      pushDisclosure(candidates, 'visible-label', value);
+    }
+  }
+
+  for (const element of document.querySelectorAll<HTMLElement>('[data-ai-generated], [data-content-origin]')) {
+    pushDisclosure(
+      candidates,
+      element.hasAttribute('data-ai-generated') ? 'data-ai-generated' : 'data-content-origin',
+      element.getAttribute('data-ai-generated') || element.getAttribute('data-content-origin')
+    );
+  }
+
+  return candidates;
+}
+
+export function discoverPage(): PageCandidate {
+  return {
+    url: location.href,
+    title: document.title,
+    images: discoverImages(),
+    texts: discoverText(),
+    disclosures: discoverDisclosures(),
+    heuristics: discoverHeuristics()
+  };
+}
+
+export { IMAGE_ID_ATTRIBUTE };