@fulldotdev/scan 0.1.0

package/dist/engine/html.d.ts

@@ -0,0 +1,307 @@
+import { type HttpResult } from "./types.js";
+export declare function resolveLink(value: string | undefined, base: string): string | null;
+export declare function markdownLinks(body: string, base: string): {
+    text: string;
+    url: string | null;
+}[];
+export declare const loadMoreText: RegExp;
+export declare function parseAmount(raw: string): number | null;
+export declare function visiblePrices(text: string): {
+    raw: string;
+    currency: string;
+    amount: number | null;
+}[];
+export declare function extractPage(response: HttpResult, representation?: "original-response-html" | "rendered-dom"): {
+    http: {
+        url: string;
+        finalUrl: string;
+        status: number | null;
+        headers: Record<string, string>;
+        redirects: {
+            url: string;
+            status: number;
+            location: string;
+        }[];
+        bytes: number;
+        durationMs: number;
+        fetchedAt: string;
+        outcome: "ok" | "http-error" | "inconclusive" | "blocked" | "error";
+        error?: string;
+        truncated?: boolean;
+        attempts: number;
+    };
+    base: string;
+    title: string;
+    titles: string[];
+    descriptions: string[];
+    language: string | null;
+    meta: {
+        name: string | undefined;
+        property: string | undefined;
+        httpEquiv: string | undefined;
+        content: string;
+    }[];
+    declarations: {
+        rel: string;
+        href: string | null;
+        rawHref: string | undefined;
+        type: string | undefined;
+        hreflang: string | undefined;
+        sizes: string | undefined;
+        fetchpriority: string | undefined;
+    }[];
+    links: {
+        href: string | null;
+        rawHref: string;
+        text: string;
+        name: string;
+        rel: string;
+        target: string;
+        hreflang: string | undefined;
+        fragment: string | null;
+        location: string;
+        pagination: true | undefined;
+    }[];
+    images: {
+        src: string | null;
+        alt: string | null;
+        width: string | undefined;
+        height: string | undefined;
+        loading: string | undefined;
+        fetchpriority: string | undefined;
+        srcset: string | undefined;
+        sources: {
+            srcset: string | undefined;
+            sizes: string | undefined;
+            type: string | undefined;
+            media: string | undefined;
+        }[];
+        sizes: string | undefined;
+    }[];
+    scripts: {
+        src: string | null;
+        type: string | undefined;
+        async: boolean;
+        defer: boolean;
+    }[];
+    styles: {
+        rel: string;
+        href: string | null;
+        rawHref: string | undefined;
+        type: string | undefined;
+        hreflang: string | undefined;
+        sizes: string | undefined;
+        fetchpriority: string | undefined;
+    }[];
+    forms: {
+        index: number;
+        id: string | undefined;
+        action: string | null;
+        rawAction: string | null;
+        method: string;
+        noValidate: boolean;
+        enctype: string;
+        controls: {
+            tag: string;
+            id: string | undefined;
+            name: string | undefined;
+            type: string;
+            labels: string[];
+            ariaLabel: string | undefined;
+            referencedLabels: string[];
+            hasAccessibleName: boolean;
+            required: boolean;
+            disabled: boolean;
+            autocomplete: string | undefined;
+            pattern: string | undefined;
+            min: string | undefined;
+            max: string | undefined;
+            minLength: string | undefined;
+            maxLength: string | undefined;
+            actionOverride: string | null;
+            methodOverride: string | undefined;
+        }[];
+        insecureDestination: boolean;
+        verification: string;
+    }[];
+    structuredData: ({
+        index: number;
+        raw: string;
+        data: any;
+        parseValid: boolean;
+        structuralErrors: string[];
+        validationScope: string;
+        error?: undefined;
+    } | {
+        data?: undefined;
+        structuralErrors?: undefined;
+        validationScope?: undefined;
+        index: number;
+        raw: string;
+        parseValid: boolean;
+        error: string;
+    })[];
+    microdata: {
+        tag: string;
+        type: string | undefined;
+        property: string | undefined;
+        id: string | undefined;
+        content: string;
+    }[];
+    rdfa: {
+        tag: string;
+        type: string | undefined;
+        property: string | undefined;
+        resource: string | undefined;
+        content: string;
+    }[];
+    headings: {
+        level: number;
+        text: string;
+    }[];
+    ids: (string | undefined)[];
+    canonicals: {
+        rel: string;
+        href: string | null;
+        rawHref: string | undefined;
+        type: string | undefined;
+        hreflang: string | undefined;
+        sizes: string | undefined;
+        fetchpriority: string | undefined;
+    }[];
+    hreflangs: {
+        rel: string;
+        href: string | null;
+        rawHref: string | undefined;
+        type: string | undefined;
+        hreflang: string | undefined;
+        sizes: string | undefined;
+        fetchpriority: string | undefined;
+    }[];
+    social: {
+        name: string | undefined;
+        property: string | undefined;
+        httpEquiv: string | undefined;
+        content: string;
+    }[];
+    bodyText: string;
+    mainText: string;
+    wordCount: number;
+    readability: {
+        paragraphs: number;
+        longestParagraphWords: number;
+        paragraphsOver150: number;
+        sentences: number;
+        longSentenceShare: number;
+    };
+    textHash: string;
+    htmlHash: string;
+    robots: {
+        name: string | undefined;
+        property: string | undefined;
+        httpEquiv: string | undefined;
+        content: string;
+    }[];
+    xRobotsTag: string;
+    technologies: {
+        name: string;
+        category: string;
+        source: string;
+        evidence: string;
+        confidence: string;
+        version: null;
+    }[];
+    pagination: {
+        next: string | null;
+        prev: string | null;
+        numbered: string[];
+        source: string;
+    } | null;
+    loadMore: {
+        tag: string;
+        text: string;
+        href: string | null;
+    }[];
+    prices: {
+        raw: string;
+        currency: string;
+        amount: number | null;
+    }[];
+    availability: {
+        inStock: boolean;
+        outOfStock: boolean;
+        inStockText: string | null;
+        outOfStockText: string | null;
+    };
+    contacts: {
+        phones: string[];
+        emails: string[];
+    };
+    truncated: boolean;
+    representation: "original-response-html" | "rendered-dom";
+};
+export type Page = ReturnType<typeof extractPage>;
+export declare function compareRendered(original: Page, rendered: Page): {
+    title: {
+        original: string;
+        rendered: string;
+        same: boolean;
+    };
+    canonical: {
+        original: (string | null)[];
+        rendered: (string | null)[];
+        same: boolean;
+    };
+    robots: {
+        original: string;
+        rendered: string;
+        same: boolean;
+    };
+    language: {
+        original: string | null;
+        rendered: string | null;
+        same: boolean;
+    };
+    headings: {
+        original: number;
+        rendered: number;
+        added: string[];
+        removed: string[];
+    };
+    links: {
+        original: number;
+        rendered: number;
+        added: string[];
+        removed: string[];
+    };
+    structuredData: {
+        original: string[];
+        rendered: string[];
+        addedTypes: string[];
+        removedTypes: string[];
+    };
+    text: {
+        originalWords: number;
+        renderedWords: number;
+        same: boolean;
+        ratio: number | null;
+    };
+    scope: string;
+};
+export declare function structuredTypes(data: any, depth?: number): string[];
+export declare function detectTechnologies(html: string, headers: Record<string, string>): {
+    name: string;
+    category: string;
+    source: string;
+    evidence: string;
+    confidence: string;
+    version: null;
+}[];
+export declare function isHtml(response: HttpResult): boolean;
+export declare function markdownEvidence(body: string, contentType: string | undefined): {
+    usable: boolean;
+    looksHtml: boolean;
+    looksMarkdown: boolean;
+};
+export declare function textTokens(s: string): Set<string>;
+export declare function textSimilarity(a: string, b: string): number;