deepline 0.3.89 → 0.3.90

package/dist/bundling-sources/shared_libs/plays/share/seo-quality.ts

@@ -0,0 +1,181 @@
+import type { PublishedPlayPagePackage } from './types';
+
+// SEO quality is deterministic authoring policy, not a web-app adapter.
+
+export type PlayPageSeoQualitySignal =
+  | 'unique_title'
+  | 'unique_description'
+  | 'required_body_sections'
+  | 'concrete_inputs'
+  | 'concrete_outputs'
+  | 'workflow_structure'
+  | 'provider_or_tool_count'
+  | 'related_links'
+  | 'json_ld_ready'
+  | 'canonical'
+  | 'h1'
+  | 'no_placeholder_copy';
+
+export type PlayPageSeoQualityResult = {
+  score: number;
+  maxScore: number;
+  passed: boolean;
+  passedSignals: PlayPageSeoQualitySignal[];
+  failedSignals: PlayPageSeoQualitySignal[];
+};
+
+export const PLAY_PAGE_REQUIRED_BODY_SECTIONS = [
+  'What this workflow does',
+  'Inputs you need',
+  'Outputs you get',
+  'How it works',
+  'Cost/latency when available',
+  'Common edits',
+  'Related workflows',
+  'Run or fork with Deepline CLI',
+] as const;
+
+const MIN_SCORE_TO_INDEX = 10;
+const PLACEHOLDER_PATTERNS = [
+  /\blorem ipsum\b/i,
+  /\btbd\b/i,
+  /\bcoming soon\b/i,
+  /\bplaceholder\b/i,
+  /\bTODO\b/,
+  /\bexample workflow\b/i,
+  /\bgeneric workflow\b/i,
+];
+
+function cleanWords(value: string): Set<string> {
+  return new Set(
+    value
+      .toLowerCase()
+      .replace(/[^a-z0-9\s]/g, ' ')
+      .split(/\s+/)
+      .filter((word) => word.length >= 4),
+  );
+}
+
+function overlapRatio(a: string, b: string): number {
+  const left = cleanWords(a);
+  const right = cleanWords(b);
+  if (left.size === 0 || right.size === 0) return 1;
+  let overlap = 0;
+  for (const word of left) {
+    if (right.has(word)) overlap += 1;
+  }
+  return overlap / Math.min(left.size, right.size);
+}
+
+function hasConcreteFields(fields: string[] | undefined): boolean {
+  const concrete = (fields ?? [])
+    .map((field) => field.trim())
+    .filter(Boolean)
+    .filter(
+      (field) =>
+        !/^(run parameters|input rows?|output table|results?)$/i.test(field),
+    );
+  return concrete.length > 0;
+}
+
+function hasNoPlaceholderCopy(pkg: PublishedPlayPagePackage): boolean {
+  const searchable = [
+    pkg.copy.title,
+    pkg.copy.shortDescription,
+    pkg.copy.jtbd,
+    pkg.copy.llmSummary,
+    pkg.copy.seoTitle,
+    pkg.copy.seoDescription,
+    ...pkg.copy.whenToUse,
+    ...pkg.copy.inputs,
+    ...pkg.copy.outputs,
+  ].join('\n');
+  return !PLACEHOLDER_PATTERNS.some((pattern) => pattern.test(searchable));
+}
+
+export function evaluatePlayPageSeoQuality(input: {
+  pkg: PublishedPlayPagePackage;
+  relatedLinkCount: number;
+}): PlayPageSeoQualityResult {
+  const { pkg, relatedLinkCount } = input;
+  const checks: Array<[PlayPageSeoQualitySignal, boolean]> = [
+    [
+      'unique_title',
+      pkg.copy.seoTitle.trim().length >= 18 &&
+        pkg.copy.title.trim().length > 0 &&
+        overlapRatio(pkg.copy.seoTitle, pkg.copy.seoDescription) < 0.82,
+    ],
+    [
+      'unique_description',
+      pkg.copy.seoDescription.trim().length >= 80 &&
+        pkg.copy.seoDescription.trim().length <= 170 &&
+        overlapRatio(pkg.copy.seoDescription, pkg.copy.shortDescription) < 0.95,
+    ],
+    ['required_body_sections', PLAY_PAGE_REQUIRED_BODY_SECTIONS.length === 8],
+    [
+      'concrete_inputs',
+      hasConcreteFields(pkg.copy.inputs) ||
+        hasConcreteFields(pkg.revision.inputSchema?.fields),
+    ],
+    [
+      'concrete_outputs',
+      hasConcreteFields(pkg.copy.outputs) ||
+        hasConcreteFields(pkg.revision.outputSchema?.fields),
+    ],
+    [
+      'workflow_structure',
+      pkg.revision.staticPipelineSummary.stepCount > 0 &&
+        pkg.revision.staticPipelineSummary.previewNodes.length > 0,
+    ],
+    [
+      'provider_or_tool_count',
+      pkg.revision.staticPipelineSummary.providerCount > 0 ||
+        pkg.revision.staticPipelineSummary.stepCount >= 2,
+    ],
+    ['related_links', relatedLinkCount >= 1],
+    [
+      'json_ld_ready',
+      pkg.copy.llmSummary.trim().length > 0 &&
+        pkg.copy.agentPrompt.trim().length > 0,
+    ],
+    [
+      'canonical',
+      /^\/p\/[^/]+\/[^/]+\/v\/\d+$/.test(pkg.page.canonicalPath) &&
+        pkg.page.canonicalPath.includes(
+          encodeURIComponent(pkg.definition.name),
+        ),
+    ],
+    ['h1', pkg.copy.title.trim().length > 0],
+    ['no_placeholder_copy', hasNoPlaceholderCopy(pkg)],
+  ];
+
+  const passedSignals = checks
+    .filter(([, passed]) => passed)
+    .map(([signal]) => signal);
+  const failedSignals = checks
+    .filter(([, passed]) => !passed)
+    .map(([signal]) => signal);
+
+  return {
+    score: passedSignals.length,
+    maxScore: checks.length,
+    passed:
+      passedSignals.length >= MIN_SCORE_TO_INDEX && failedSignals.length === 0,
+    passedSignals,
+    failedSignals,
+  };
+}
+
+export function applyPlayPageSeoQuality(input: {
+  pkg: PublishedPlayPagePackage;
+  relatedLinkCount: number;
+}): PublishedPlayPagePackage {
+  const result = evaluatePlayPageSeoQuality(input);
+  return {
+    ...input.pkg,
+    page: {
+      ...input.pkg.page,
+      seoIndexing: result.passed ? input.pkg.page.seoIndexing : 'noindex',
+    },
+  };
+}