diff options
Diffstat (limited to 'castle-search/src/engines.ts')
| -rw-r--r-- | castle-search/src/engines.ts | 13 |
1 files changed, 12 insertions, 1 deletions
diff --git a/castle-search/src/engines.ts b/castle-search/src/engines.ts index 76716c6..fb07649 100644 --- a/castle-search/src/engines.ts +++ b/castle-search/src/engines.ts @@ -45,6 +45,12 @@ export type ExtractionConfig = { mode: "items" | "headings"; /** Results container candidates, first match wins. */ roots: string[]; + /** + * Present only on the engine's own "nothing matched" page. Without it a page + * that genuinely has zero hits is indistinguishable from broken markup, and + * the error blames the extractor for what is just an over-narrow query. + */ + emptyMarker?: string; /** "items" mode: one result. */ item?: string; /** Anchor carrying the outbound link, relative to the item. */ @@ -124,6 +130,9 @@ const DUCKDUCKGO: EngineDefinition = { extraction: { mode: "items", roots: [".results", "#links"], + // Seen 2026-11-06: `<div class="result results_links no-results">` — it is + // itself a `.result`, which is why the extractor finds a container and no hits. + emptyMarker: ".no-results", item: ".result", link: "a.result__a[href]", title: "a.result__a", @@ -187,7 +196,9 @@ const BRAVE: EngineDefinition = { }, extraction: { mode: "items", - roots: ["#results"], + // `#results` is gone from the live page (2026-11-06); `#main` and `main` + // wrap the results. Left first so the older layout still matches. + roots: ["#results", "#main", "main"], // data-type="web" excludes the AI summariser, video and news cards, which // share the .snippet class but are not ranked web results. item: '.snippet[data-type="web"]', |