spb/cancerindex
Public
1import 'server-only';2import { SITE_URL, NAV, FOOTER_NAV } from '@/lib/site';3import { cancerSlugsForSitemap, countActiveCancers } from '@/lib/queries/cancers';4import { geneSymbolsForSitemap, countGenes } from '@/lib/queries/genomics';5import { drugSlugsForSitemap, countDrugs } from '@/lib/queries/drugs';6import { trialNctForSitemap, countTrials } from '@/lib/queries/trials';7import { listSources } from '@/lib/queries/sources';8import { listMetrics } from '@/lib/queries/rankings';9import { toDate } from '@/lib/format';1011/**12 * Chunked sitemaps: chunk 0 = static pages + sources + rankings; then one chunk per 20k entities of13 * each kind (cancers, genes, drugs, trials) so the index stays under the 50k-URL limit per file.14 */15export const PAGE = 20_000;16type Kind = 'cancer' | 'gene' | 'drug' | 'trial';17export interface Chunk {18 id: number;19 kind: Kind;20 offset: number;21}22export interface Entry {23 url: string;24 lastModified: Date;25 changeFrequency: 'daily' | 'weekly';26 priority: number;27}2829export async function sitemapChunks(): Promise<Chunk[]> {30 const [c, g, d, t] = await Promise.all([countActiveCancers(), countGenes(), countDrugs(), countTrials()]);31 const out: Chunk[] = [];32 let id = 1;33 for (const [kind, n] of [34 ['cancer', c],35 ['gene', g],36 ['drug', d],37 ['trial', t],38 ] as Array<[Kind, number]>) {39 for (let off = 0; off < Math.max(n, 1); off += PAGE) out.push({ id: id++, kind, offset: off });40 }41 return out;42}4344export async function sitemapEntries(id: number): Promise<Entry[]> {45 const now = new Date();46 if (id === 0) {47 const [sources, metrics] = await Promise.all([listSources(), listMetrics()]);48 return [49 { url: SITE_URL, lastModified: now, changeFrequency: 'daily', priority: 1 },50 ...NAV.map((n) => ({ url: `${SITE_URL}${n.href}`, lastModified: now, changeFrequency: 'daily' as const, priority: 0.8 })),51 ...FOOTER_NAV.filter((f) => !NAV.some((n) => n.href === f.href)).map((n) => ({ url: `${SITE_URL}${n.href}`, lastModified: now, changeFrequency: 'weekly' as const, priority: 0.5 })),52 ...sources.map((s) => ({ url: `${SITE_URL}/source/${s.slug}`, lastModified: toDate(s.updated_at) ?? now, changeFrequency: 'weekly' as const, priority: 0.6 })),53 ...metrics.filter((m) => m.snapshot_count > 0).map((m) => ({ url: `${SITE_URL}/rankings/${m.slug}`, lastModified: now, changeFrequency: 'daily' as const, priority: 0.7 })),54 ];55 }56 const chunk = (await sitemapChunks()).find((c) => c.id === id);57 if (!chunk) return [];58 switch (chunk.kind) {59 case 'cancer':60 return (await cancerSlugsForSitemap(chunk.offset, PAGE)).map((r) => ({ url: `${SITE_URL}/cancer/${r.slug}`, lastModified: toDate(r.updated_at) ?? now, changeFrequency: 'weekly' as const, priority: 0.7 }));61 case 'gene':62 return (await geneSymbolsForSitemap(chunk.offset, PAGE)).map((r) => ({ url: `${SITE_URL}/gene/${r.symbol}`, lastModified: toDate(r.updated_at) ?? now, changeFrequency: 'weekly' as const, priority: 0.5 }));63 case 'drug':64 return (await drugSlugsForSitemap(chunk.offset, PAGE)).map((r) => ({ url: `${SITE_URL}/drug/${r.slug}`, lastModified: toDate(r.updated_at) ?? now, changeFrequency: 'weekly' as const, priority: 0.5 }));65 case 'trial':66 return (await trialNctForSitemap(chunk.offset, PAGE)).map((r) => ({ url: `${SITE_URL}/trial/${r.nct_id}`, lastModified: toDate(r.updated_at) ?? now, changeFrequency: 'weekly' as const, priority: 0.4 }));67 }68}6970const esc = (s: string) => s.replace(/&/g, '&').replace(/</g, '<').replace(/>/g, '>').replace(/"/g, '"');7172export function urlsetXml(entries: Entry[]): string {73 return `<?xml version="1.0" encoding="UTF-8"?>\n<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n${entries.map((e) => ` <url><loc>${esc(e.url)}</loc><lastmod>${e.lastModified.toISOString()}</lastmod><changefreq>${e.changeFrequency}</changefreq><priority>${e.priority}</priority></url>`).join('\n')}\n</urlset>\n`;74}7576export function indexXml(ids: number[]): string {77 const now = new Date().toISOString();78 return `<?xml version="1.0" encoding="UTF-8"?>\n<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n${ids.map((id) => ` <sitemap><loc>${SITE_URL}/sitemap/${id}</loc><lastmod>${now}</lastmod></sitemap>`).join('\n')}\n</sitemapindex>\n`;79}80