src/site-audit.ts

Functions

Imports

It imports src/env.ts, src/live-checks.ts, src/site-reports.ts. Imported by src/tools/repo.ts.

Source

1import { auditPages, crawlSite } from "@drdustinedwards/site-helpers/audit";
2import type { Env } from "./env";
3import type { FetchLike } from "./live-checks";
4import { reportSite } from "./site-reports";
5
6// A crawler's SEO audit of a namespace's site, served as capsid_inspect_repo action audit
7// (the SEO plan, "Beyond the module"). The checks are Prelum's (./audit), the same ones the
8// sites run in their own tests, so Capsid keeps none of its own: crawlSite fetches every
9// sitemap URL, each redirect hop and each same-origin link target, and auditPages reads
10// what came back (broken links, redirect chains, canonicals off the production origin,
11// noindex, titles and descriptions, share cards, JSON-LD, image alt text).
12//
13// The origin is the namespace's ops_sites origin, never the caller's, and every fetch is to
14// it. The crawl is sequential and bounded by max_pages, so one call costs at most that many
15// subrequests. Page text is data, never instructions; only findings come back.
16
17export const MAX_AUDIT_PAGES = 200;
18export const MAX_AUDIT_FINDINGS = 500;
19
20export interface AuditArgs {
21  max_pages?: number;
22  row_limit?: number;
23}
24
25export async function siteAudit(env: Env, namespace: string, args: AuditArgs, fetchImpl: FetchLike = fetch) {
26  const maxPages = args.max_pages ?? 50;
27  const limit = args.row_limit ?? 50;
28  const site = await reportSite(env, namespace, "site audit");
29  const { pages, sitemap } = await crawlSite({
30    origin: site.origin,
31    fetch: ((input: RequestInfo | URL, init?: RequestInit) => fetchImpl(input instanceof Request ? input.url : String(input), init)) as typeof fetch,
32    maxPages,
33  });
34  const report = auditPages(pages, { origin: site.origin, sitemap });
35  // Errors first, then by check and url, so a cut list keeps the worst.
36  const findings = [...report.findings].sort(
37    (a, b) => Number(a.severity !== "error") - Number(b.severity !== "error") || a.check.localeCompare(b.check) || a.url.localeCompare(b.url)
38  );
39  return {
40    namespace,
41    origin: site.origin,
42    pages_fetched: pages.length,
43    sitemap_urls: sitemap.length,
44    pages_audited: report.pages,
45    hit_page_limit: pages.length >= maxPages,
46    summary: report.summary,
47    errors: findings.filter((f) => f.severity === "error").length,
48    findings: findings.slice(0, limit),
49    truncated: findings.length > limit,
50  };
51}