import { HTTPClient } from "../../http/HTTPClient";
import { ProviderError } from "../../models/errors";
import { SearchOptions } from "../../types/search";
import { ImageSearchResult } from "../models";
import { BaseImageProvider } from "./BaseImageProvider";

export const BING_BASE_URL = "https://www.bing.com";
export const BING_IMAGE_SEARCH_PATH = "/images/search";
export const BING_REFERER = "https://www.bing.com/";

interface BingTileJSON {
  murl?: string;
  purl?: string;
  t?: string;
  turl?: string;
}

export interface BingImageProviderOptions {
  httpClient?: HTTPClient;
  httpTimeoutSeconds?: number;
}

/**
 * Decode common HTML entities present in Bing tile JSON attribute strings.
 */
export function decodeHTMLEntities(encoded: string): string {
  return encoded
    .replace(/"/g, '"')
    .replace(/&/g, "&")
    .replace(/'/g, "'")
    .replace(/'/g, "'")
    .replace(/&lt;/g, "<")
    .replace(/&gt;/g, ">");
}

/**
 * Lightweight Bing Image Search provider implementation.
 * Queries Bing HTML image search endpoints (`/images/search`) and extracts tile metadata
 * embedded in `<a class="iusc" m="...">` attributes.
 * Requires no API keys.
 */
export class BingImageProvider extends BaseImageProvider {
  public static readonly PROVIDER_NAME = "bing";

  constructor(options: BingImageProviderOptions = {}) {
    super({
      providerName: BingImageProvider.PROVIDER_NAME,
      httpClient: options.httpClient,
      httpTimeoutSeconds: options.httpTimeoutSeconds,
    });
  }

  /**
   * Execute image search using Bing HTML search endpoint.
   *
   * @param query Non-empty search query string.
   * @param options Search options (numResults, extra kwargs).
   * @returns Array of normalized ImageSearchResult objects.
   * @throws ProviderError if HTTP request fails.
   */
  public async searchImages(
    query: string,
    options: SearchOptions = {}
  ): Promise<ImageSearchResult[]> {
    if (!query || !query.trim()) {
      return [];
    }

    const cleanQuery = query.trim();
    const numResults = options.numResults ?? 10;

    const url = new URL(BING_IMAGE_SEARCH_PATH, BING_BASE_URL);
    url.searchParams.append("q", cleanQuery);
    url.searchParams.append("form", "HDRSC2");
    url.searchParams.append("adlt", "off");

    const headers: Record<string, string> = {
      Referer: BING_REFERER,
      Accept: "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,*/*;q=0.8",
      "Accept-Language": "en-US,en;q=0.9",
    };

    let response;
    try {
      response = await this._httpClient.get<string>(url.toString(), { headers });
    } catch (exc: unknown) {
      const err = exc instanceof Error ? exc : new Error(String(exc));
      throw new ProviderError(`Bing image search HTTP request failed: ${err.message}`);
    }

    if (response.statusCode !== 200 || !response.text) {
      throw new ProviderError(
        `Bing image search failed with status code ${response.statusCode}`
      );
    }

    const html = response.text;
    const tiles = this.parseBingHtml(html);

    return tiles.slice(0, numResults);
  }

  /**
   * Parse HTML response and extract deduplicated image tile metadata.
   */
  public parseBingHtml(html: string): ImageSearchResult[] {
    const results: ImageSearchResult[] = [];
    const seen = new Set<string>();

    // Flexible regex matching for <a ... class="iusc" ... m="..."> or <a ... m="..." ... class="iusc">
    const iuscRegex = /<a\s+[^>]*?class=["']?[^"']*iusc[^"']*["']?[^>]*?m=(["'])(.*?)\1[^>]*>/gi;
    const iuscRegexAlt = /<a\s+[^>]*?m=(["'])(.*?)\1[^>]*?class=["']?[^"']*iusc[^"']*["']?[^>]*>/gi;

    const matches: string[] = [];
    let match: RegExpExecArray | null;

    while ((match = iuscRegex.exec(html)) !== null) {
      matches.push(match[2]);
    }
    while ((match = iuscRegexAlt.exec(html)) !== null) {
      matches.push(match[2]);
    }

    for (const rawM of matches) {
      if (!rawM || !rawM.trim()) continue;

      const decoded = decodeHTMLEntities(rawM);
      let tileJson: BingTileJSON;
      try {
        tileJson = JSON.parse(decoded) as BingTileJSON;
      } catch {
        continue;
      }

      if (!tileJson || !tileJson.murl || typeof tileJson.murl !== "string") {
        continue;
      }

      const imageUrl = tileJson.murl.trim();

      // Ensure valid HTTP(S) URL
      if (!imageUrl.startsWith("http://") && !imageUrl.startsWith("https://")) {
        continue;
      }

      if (seen.has(imageUrl)) {
        continue;
      }
      seen.add(imageUrl);

      const sourceUrl = typeof tileJson.purl === "string" ? tileJson.purl.trim() : "";
      let sourceDomain: string | undefined = undefined;
      if (sourceUrl) {
        try {
          const parsed = new URL(sourceUrl);
          sourceDomain = parsed.hostname;
        } catch {
          sourceDomain = undefined;
        }
      }

      const title = typeof tileJson.t === "string" && tileJson.t.trim() ? tileJson.t.trim() : "";
      const thumbnailUrl = typeof tileJson.turl === "string" && tileJson.turl.trim() ? tileJson.turl.trim() : undefined;

      results.push({
        title,
        imageUrl,
        sourceUrl,
        sourceDomain,
        thumbnailUrl,
        raw: tileJson as unknown as Record<string, unknown>,
      });
    }

    return results;
  }
}

import { HTTPClient } from "../../http/HTTPClient";
import { ProviderError } from "../../models/errors";
import { SearchOptions } from "../../types/search";
import { ImageSearchResult } from "../models";
import { BaseImageProvider } from "./BaseImageProvider";

export const BING_BASE_URL = "https://www.bing.com";
export const BING_IMAGE_SEARCH_PATH = "/images/search";
export const BING_REFERER = "https://www.bing.com/";

interface BingTileJSON {
  murl?: string;
  purl?: string;
  t?: string;
  turl?: string;
}

export interface BingImageProviderOptions {
  httpClient?: HTTPClient;
  httpTimeoutSeconds?: number;
}

/**
 * Decode common HTML entities present in Bing tile JSON attribute strings.
 */
export function decodeHTMLEntities(encoded: string): string {
  return encoded
    .replace(/"/g, '"')
    .replace(/&/g, "&")
    .replace(/'/g, "'")
    .replace(/'/g, "'")
    .replace(/&lt;/g, "<")
    .replace(/&gt;/g, ">");
}

/**
 * Lightweight Bing Image Search provider implementation.
 * Queries Bing HTML image search endpoints (`/images/search`) and extracts tile metadata
 * embedded in `<a class="iusc" m="...">` attributes.
 * Requires no API keys.
 */
export class BingImageProvider extends BaseImageProvider {
  public static readonly PROVIDER_NAME = "bing";

  constructor(options: BingImageProviderOptions = {}) {
    super({
      providerName: BingImageProvider.PROVIDER_NAME,
      httpClient: options.httpClient,
      httpTimeoutSeconds: options.httpTimeoutSeconds,
    });
  }

  /**
   * Execute image search using Bing HTML search endpoint.
   *
   * @param query Non-empty search query string.
   * @param options Search options (numResults, extra kwargs).
   * @returns Array of normalized ImageSearchResult objects.
   * @throws ProviderError if HTTP request fails.
   */
  public async searchImages(
    query: string,
    options: SearchOptions = {}
  ): Promise<ImageSearchResult[]> {
    if (!query || !query.trim()) {
      return [];
    }

    const cleanQuery = query.trim();
    const numResults = options.numResults ?? 10;

    const url = new URL(BING_IMAGE_SEARCH_PATH, BING_BASE_URL);
    url.searchParams.append("q", cleanQuery);
    url.searchParams.append("form", "HDRSC2");
    url.searchParams.append("adlt", "off");

    const headers: Record<string, string> = {
      Referer: BING_REFERER,
      Accept: "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,*/*;q=0.8",
      "Accept-Language": "en-US,en;q=0.9",
    };

    let response;
    try {
      response = await this._httpClient.get<string>(url.toString(), { headers });
    } catch (exc: unknown) {
      const err = exc instanceof Error ? exc : new Error(String(exc));
      throw new ProviderError(`Bing image search HTTP request failed: ${err.message}`);
    }

    if (response.statusCode !== 200 || !response.text) {
      throw new ProviderError(
        `Bing image search failed with status code ${response.statusCode}`
      );
    }

    const html = response.text;
    const tiles = this.parseBingHtml(html);

    return tiles.slice(0, numResults);
  }

  /**
   * Parse HTML response and extract deduplicated image tile metadata.
   */
  public parseBingHtml(html: string): ImageSearchResult[] {
    const results: ImageSearchResult[] = [];
    const seen = new Set<string>();

    // Flexible regex matching for <a ... class="iusc" ... m="..."> or <a ... m="..." ... class="iusc">
    const iuscRegex = /<a\s+[^>]*?class=["']?[^"']*iusc[^"']*["']?[^>]*?m=(["'])(.*?)\1[^>]*>/gi;
    const iuscRegexAlt = /<a\s+[^>]*?m=(["'])(.*?)\1[^>]*?class=["']?[^"']*iusc[^"']*["']?[^>]*>/gi;

    const matches: string[] = [];
    let match: RegExpExecArray | null;

    while ((match = iuscRegex.exec(html)) !== null) {
      matches.push(match[2]);
    }
    while ((match = iuscRegexAlt.exec(html)) !== null) {
      matches.push(match[2]);
    }

    for (const rawM of matches) {
      if (!rawM || !rawM.trim()) continue;

      const decoded = decodeHTMLEntities(rawM);
      let tileJson: BingTileJSON;
      try {
        tileJson = JSON.parse(decoded) as BingTileJSON;
      } catch {
        continue;
      }

      if (!tileJson || !tileJson.murl || typeof tileJson.murl !== "string") {
        continue;
      }

      const imageUrl = tileJson.murl.trim();

      // Ensure valid HTTP(S) URL
      if (!imageUrl.startsWith("http://") && !imageUrl.startsWith("https://")) {
        continue;
      }

      if (seen.has(imageUrl)) {
        continue;
      }
      seen.add(imageUrl);

      const sourceUrl = typeof tileJson.purl === "string" ? tileJson.purl.trim() : "";
      let sourceDomain: string | undefined = undefined;
      if (sourceUrl) {
        try {
          const parsed = new URL(sourceUrl);
          sourceDomain = parsed.hostname;
        } catch {
          sourceDomain = undefined;
        }
      }

      const title = typeof tileJson.t === "string" && tileJson.t.trim() ? tileJson.t.trim() : "";
      const thumbnailUrl = typeof tileJson.turl === "string" && tileJson.turl.trim() ? tileJson.turl.trim() : undefined;

      results.push({
        title,
        imageUrl,
        sourceUrl,
        sourceDomain,
        thumbnailUrl,
        raw: tileJson as unknown as Record<string, unknown>,
      });
    }

    return results;
  }
}