tablefacts 0.1.0 → 0.3.0

package/types/research/lib/search.d.mts

@@ -0,0 +1,58 @@
+/** Unwraps DuckDuckGo's `//duckduckgo.com/l/?uddg=<url>` redirect to the real result URL. */
+export declare function resultUrl(href: any): string | null;
+/** Pulls { url, title, snippet } out of DuckDuckGo's HTML result page (no DOM parser). */
+export declare function parseSearchResults(html: any): {
+    url: string;
+    title: string;
+    snippet: string;
+}[];
+/**
+ * Sorts results into the links the pipeline can follow. `website` is the best
+ * non-directory result whose title/host looks like the restaurant's name; a
+ * result that only matches the place is kept as a weaker candidate.
+ */
+export declare function classifyResults(results: any, { name }?: {
+    name?: string | undefined;
+}): {
+    website: any;
+    instagram: never;
+    tripadvisor: never;
+    facebook: never;
+    tiktok: never;
+    mapsUrl: never;
+    hubs: never[];
+    candidates: {
+        url: any;
+        title: any;
+    }[];
+};
+/**
+ * Runs the search for a restaurant and returns its candidate links. Tries every
+ * endpoint until one returns results; if all answer but none has results, the
+ * empty result is returned so the caller still says "no results" rather than
+ * "failed". Throws only when no endpoint answers.
+ */
+export declare function searchWeb({ name, location }: {
+    location?: string | undefined;
+    name: any;
+}): Promise<{
+    website: any;
+    instagram: never;
+    tripadvisor: never;
+    facebook: never;
+    tiktok: never;
+    mapsUrl: never;
+    hubs: never[];
+    candidates: {
+        url: any;
+        title: any;
+    }[];
+    source: string;
+    url: string;
+    query: string;
+    results: {
+        url: string;
+        title: string;
+        snippet: string;
+    }[];
+}>;

package/types/research/lib/social.d.mts

@@ -1,34 +1,31 @@
 /**
- * Public Instagram profile metadata. Without a login the page only exposes the
- * share card: "1,234 Followers, 56 Following, 789 Posts - Name (@handle) on
- * Instagram: "bio"". The bio often carries the phone, the area and a hub link.
+ * Public Instagram profile metadata, trying the public surfaces in turn: the
+ * share card, the web profile API, oEmbed, and finally the bio a search engine
+ * indexed (`snippet`). Only honest requests are made; a walled profile is
+ * reported as blocked, never worked around. Returns `blocked` only when none of
+ * them had anything.
  */
-export declare function readInstagram(handle: any): Promise<{
-    handle: any;
-    url: string;
-    blocked: any;
-    displayName?: undefined;
-    followers?: undefined;
-    posts?: undefined;
-    bio?: undefined;
-    phones?: undefined;
-    links?: undefined;
-    image?: undefined;
-} | {
-    blocked?: undefined;
-    handle: any;
-    url: string;
-    displayName: any;
-    followers: any;
-    posts: any;
-    bio: any;
-    phones: any[];
-    links: any;
-    image: any;
-}>;
-/** TripAdvisor: schema.org JSON-LD when the page loads; DataDome often blocks it. */
+export declare function readInstagram(handle: any, { snippet }?: {
+    snippet?: string | undefined;
+}): Promise<any>;
+/**
+ * Name and location out of a TripAdvisor restaurant URL slug, e.g.
+ * "...-Reviews-Maki_Bar_Medellin-Medellin_Antioquia_Department.html" gives
+ * { name: "Maki Bar Medellin", location: "Medellin Antioquia Department" }. The
+ * city/region boundary is ambiguous for multi-word cities, so the whole location
+ * is kept and the caller treats it as a hint, never as a confirmed locality.
+ */
+export declare function parseTripadvisorUrl(raw: any): {
+    name: string;
+    location: string;
+} | null;
+/** TripAdvisor: schema.org JSON-LD when the page loads; DataDome often blocks it. The URL slug is always kept. */
 export declare function readTripadvisor(url: any, { browser }?: {}): Promise<{
     url: any;
+    slug: {
+        name: string;
+        location: string;
+    } | null;
     blocked: string;
     jsonld?: undefined;
     description?: undefined;
@@ -37,6 +34,10 @@ export declare function readTripadvisor(url: any, { browser }?: {}): Promise<{
 } | {
     blocked?: undefined;
     url: any;
+    slug: {
+        name: string;
+        location: string;
+    } | null;
     jsonld: {
         name: any;
         telephone: any;
@@ -68,23 +69,23 @@ export declare function readTripadvisor(url: any, { browser }?: {}): Promise<{
         } | null;
     } | null;
     description: any;
-    hoursText: any;
+    hoursText: string[];
     images: any[];
 }>;
 /** A link-in-bio hub (Linktree, Beacons, bio.link): its links, classified like a website's. */
 export declare function readHub(url: any, { browser }?: {}): Promise<{
-    links?: undefined;
     images?: undefined;
     url: any;
     blocked: string;
     title?: undefined;
     meta?: undefined;
+    links?: undefined;
     anchors?: undefined;
     logos?: undefined;
 } | {
     blocked?: undefined;
     url: any;
-    title: any;
+    title: string;
     meta: {
         siteName: any;
         title: any;
@@ -108,7 +109,7 @@ export declare function readHub(url: any, { browser }?: {}): Promise<{
     };
     anchors: {
         href: string | null;
-        text: any;
+        text: string;
     }[];
     logos: {
         url: string | null;

package/types/research/lib/util.d.mts

@@ -2,6 +2,11 @@ import { fold } from "../../lib/text.mjs";
 export { fold };
 export declare const BROWSER_UA = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/130.0 Safari/537.36 cannario-research";
 export declare const sleep: (ms: any) => Promise<any>;
+/** Decodes the HTML entities a page's attributes or text may hold, named and numeric. */
+export declare const decodeHtml: (s: any) => string;
+export declare const IG_RESERVED: Set<string>;
+/** The profile handle in an instagram.com URL, or "" for a post, reel or other non-profile path. */
+export declare const instagramHandle: (url: any) => string;
 /** Runs `fn(item, index)` over `items` with at most `size` in flight; results keep the order of `items`. */
 export declare function mapPool(items: any, size: any, fn: any): Promise<any[]>;
 export declare const MAX_BODY_BYTES: number;

package/types/research/lib/website.d.mts

@@ -1,7 +1,7 @@
 /** Pulls every fact a page holds. `base` resolves relative links. */
-export declare function analyzeHtml(html: any, base: any, lines?: any): {
+export declare function analyzeHtml(html: any, base: any, lines?: string[]): {
     url: any;
-    title: any;
+    title: string;
     lang: any;
     meta: {
         siteName: any;
@@ -59,13 +59,13 @@ export declare function analyzeHtml(html: any, base: any, lines?: any): {
         url: string | null;
         kind: any;
     }[];
-    textLength: any;
-    headings: any[];
-    paragraphs: any;
-    hoursText: any;
+    textLength: number;
+    headings: string[];
+    paragraphs: string[];
+    hoursText: string[];
     anchors: {
         href: string | null;
-        text: any;
+        text: string;
     }[];
 };
 /**
@@ -86,7 +86,7 @@ export declare function readPage(url: any, { renderJs, browser }?: {
     blocked?: undefined;
     page: {
         url: any;
-        title: any;
+        title: string;
         lang: any;
         meta: {
             siteName: any;
@@ -144,13 +144,13 @@ export declare function readPage(url: any, { renderJs, browser }?: {
             url: string | null;
             kind: any;
         }[];
-        textLength: any;
-        headings: any[];
-        paragraphs: any;
-        hoursText: any;
+        textLength: number;
+        headings: string[];
+        paragraphs: string[];
+        hoursText: string[];
         anchors: {
             href: string | null;
-            text: any;
+            text: string;
         }[];
         rendered: boolean;
     };
@@ -158,7 +158,7 @@ export declare function readPage(url: any, { renderJs, browser }?: {
     blocked?: undefined;
     page: {
         url: any;
-        title: any;
+        title: string;
         lang: any;
         meta: {
             siteName: any;
@@ -216,13 +216,13 @@ export declare function readPage(url: any, { renderJs, browser }?: {
             url: string | null;
             kind: any;
         }[];
-        textLength: any;
-        headings: any[];
-        paragraphs: any;
-        hoursText: any;
+        textLength: number;
+        headings: string[];
+        paragraphs: string[];
+        hoursText: string[];
         anchors: {
             href: string | null;
-            text: any;
+            text: string;
         }[];
     };
 }>;
@@ -234,7 +234,7 @@ export declare function scrapeSite(startUrl: any, { renderJs, maxPages, browser
     url: any;
     pages: {
         url: any;
-        title: any;
+        title: string;
         rendered: boolean;
     }[];
     lang: any;