fix: support running on subpaths

Fixes #196

packages/cli/src/util.ts

@@ -26,7 +26,7 @@ export async function validateHost(resolvedConfig: ResolvedUserConfig) {
       // change the URL to the redirect one, make sure it's not to a file (i.e /index.php)
       if (redirected && redirectUrl && !redirectUrl.includes('.')) {
         logger.success(`Request to site \`${site}\` redirected to \`${redirectUrl}\`, using that as the site.`)
-        resolvedConfig.site = normaliseHost(redirectUrl)
+        resolvedConfig.site = normaliseHost(redirectUrl).toString()
       }
       else {
         logger.success(`Successfully connected to \`${site}\`. (Status: \`${response.status}\`).`)

packages/core/src/puppeteer/tasks/html.ts

@@ -3,7 +3,7 @@ import fs from 'fs-extra'
 import type { CheerioAPI } from 'cheerio'
 import cheerio from 'cheerio'
 import type { Page } from 'puppeteer-core'
-import { $URL, withoutTrailingSlash } from 'ufo'
+import { withoutTrailingSlash } from 'ufo'
 import chalk from 'chalk'
 import type { HTMLExtractPayload, PuppeteerTask } from '../../types'
 import { useUnlighthouse } from '../../unlighthouse'
@@ -122,7 +122,7 @@ export const inspectHtmlTask: PuppeteerTask = async (props) => {
     if (response.redirected) {
       // strip any protocols from the url
       const siteHost = runtimeSettings.siteUrl.host.split(':')[0]
-      const redirectHost = new $URL(response.redirected).host.split(':')[0]
+      const redirectHost = new URL(response.redirected).host.split(':')[0]
       // allow subdomains
       if (siteHost !== redirectHost && !redirectHost.endsWith(`.${siteHost}`)) {
         routeReport.tasks.inspectHtmlTask = 'ignore'

packages/core/src/resolveConfig.ts

@@ -41,7 +41,17 @@ export const resolveUserConfig: (userConfig: UserConfig) => Promise<ResolvedUser
   // it's possible we don't know the site at runtime
   if (config.site) {
     // normalise site
-    config.site = normaliseHost(config.site)
+    const siteUrl = normaliseHost(config.site)
+    if (siteUrl.pathname !== '/' && siteUrl.pathname !== '') {
+      logger.warn('You are providing a site with a path, disabling sitemap, robots and dynamic sampling.')
+      config.scanner.sitemap = false
+      config.scanner.robotsTxt = false
+      config.scanner.dynamicSampling = false
+      config.site = siteUrl.toString()
+    }
+    else {
+      config.site = siteUrl.origin
+    }
   }
   if (config.lighthouseOptions) {
     if (config.lighthouseOptions.onlyCategories?.length) {
@@ -76,7 +86,8 @@ export const resolveUserConfig: (userConfig: UserConfig) => Promise<ResolvedUser
         uploadThroughputKbps: 750,
         cpuSlowdownMultiplier: 1,
       }
-    } else if (!config.site || config.site.includes('localhost') || config.scanner?.throttle === false) {
+    }
+    else if (!config.site || config.site.includes('localhost') || config.scanner?.throttle === false) {
       config.lighthouseOptions.throttlingMethod = 'provided'
       config.lighthouseOptions.throttling = {
         rttMs: 0,

packages/core/src/router/util.ts

@@ -1,5 +1,5 @@
 import { basename } from 'node:path'
-import { $URL, hasProtocol, isRelative, withBase, withLeadingSlash } from 'ufo'
+import { hasProtocol, isRelative, withBase, withLeadingSlash } from 'ufo'
 import type { NormalisedRoute } from '../types'
 import { hashPathName, trimSlashes } from '../util'
 import { useUnlighthouse } from '../unlighthouse'
@@ -10,7 +10,7 @@ export function isScanOrigin(url: string): boolean {
 
   const { runtimeSettings } = useUnlighthouse()
 
-  const $url = new $URL(url)
+  const $url = new URL(url)
   if ($url.hostname === runtimeSettings.siteUrl.hostname)
     return true
   // allow subdomains
@@ -32,7 +32,7 @@ export function normaliseRoute(url: string): NormalisedRoute {
     url = withBase(url, runtimeSettings.siteUrl.origin)
   }
 
-  const $url = new $URL(url)
+  const $url = new URL(url)
   // make sure we start with a leading slash
   const path = withLeadingSlash($url.pathname)
 

packages/core/src/types.ts

@@ -1,6 +1,6 @@
 import type http from 'node:http'
 import type https from 'node:https'
-import type { $URL, QueryObject } from 'ufo'
+import type { QueryObject } from 'ufo'
 import type LH from 'lighthouse/types/lh'
 import type { LaunchOptions, Page, PuppeteerNodeLaunchOptions } from 'puppeteer-core'
 import type { Hookable, NestedHooks } from 'hookable'
@@ -38,7 +38,7 @@ export interface NormalisedRoute {
   id: string
   path: string
   url: string
-  $url: $URL
+  $url: URL
   definition: RouteDefinition
   /**
    * A runtime path that the route was discovered from, useful if the route is a 404 and we want to know what directed
@@ -517,7 +517,7 @@ export interface RuntimeSettings {
   /**
    * A URL instance of the site for easier use of the host.
    */
-  siteUrl: $URL
+  siteUrl: URL
   /**
    * The URL of the server running the API and client.
    */

packages/core/src/unlighthouse.ts

@@ -213,9 +213,9 @@ export async function createUnlighthouse(userConfig: UserConfig, provider?: Prov
 
   ctx.setSiteUrl = async (url: string) => {
     const site = normaliseHost(url)
-    ctx.runtimeSettings.siteUrl = new $URL(site)
+    ctx.runtimeSettings.siteUrl = site
 
-    logger.debug(`Setting Unlighthouse Site URL [Site: ${site}]`)
+    logger.debug(`Setting Unlighthouse Site URL [Site: ${site.toString()}]`)
 
     const outputPath = join(
       resolvedConfig.outputPath,
@@ -224,15 +224,16 @@ export async function createUnlighthouse(userConfig: UserConfig, provider?: Prov
       runtimeSettings.configCacheKey || '',
     )
 
-    ctx.resolvedConfig.site = site
+    if (!ctx.resolvedConfig.site)
+      ctx.resolvedConfig.site = site.toString()
     ctx.runtimeSettings.outputPath = outputPath
     ctx.runtimeSettings.generatedClientPath = outputPath
 
-    await hooks.callHook('site-changed', site)
+    await hooks.callHook('site-changed', ctx.resolvedConfig.site)
   }
 
   ctx.setServerContext = async ({ url, server, app }) => {
-    const $server = new $URL(url)
+    const $server = new URL(url)
 
     logger.debug(`Setting Unlighthouse Server Context [Server: ${$server}]`)
 
@@ -298,7 +299,7 @@ export async function createUnlighthouse(userConfig: UserConfig, provider?: Prov
 
     ctx.routes = await resolveReportableRoutes()
     logger.debug('Resolved reportable routes', ctx.routes.length)
-    await createBroadcastingEvents()
+    createBroadcastingEvents()
     worker.queueRoutes(ctx.routes)
 
     if (provider?.name !== 'ci') {

packages/core/src/util.ts

@@ -74,8 +74,7 @@ export function normaliseHost(host: string) {
     host = `http${host.startsWith('localhost') ? '' : 's'}://${host}`
   host = host.includes('.') ? host : withTrailingSlash(host)
   // strip pathname from host
-  const url = new URL(host)
-  return `${url.protocol}//${url.host}/`
+  return new URL(host)
 }
 
 /**