src/discovery/lib/normalizer.ts

import { InsertCompany } from "../../shared/supabase/schema";
import { ScrapedCompany } from "./web-scraper";
import { LinkedInCompanyData } from "./linkedin-scraper";
import { SerperResult } from "../providers/serper";

/**
 * Normalizes raw data from multiple sources into a single canonical Company record.
 * Priority: LinkedIn > Website > Serper snippet
 */
export function normalizeCompany(
  serperResult: SerperResult,
  website: ScrapedCompany,
  linkedin: LinkedInCompanyData | null,
  region: string,
  source: string
): InsertCompany {
  const name =
    linkedin?.name ??
    website.name ??
    cleanTitle(serperResult.title);

  const description =
    linkedin?.description ??
    website.description ??
    serperResult.snippet;

  const employeeCount =
    linkedin?.employeeCount ??
    website.employeeCount ??
    null;

  const employeeRange =
    linkedin?.employeeRange ??
    website.employeeRange ??
    estimateRange(employeeCount);

  const industry =
    linkedin?.industry ??
    website.industry ??
    null;

  const country =
    linkedin?.headquarters
      ? extractCountry(linkedin.headquarters)
      : regionToCountry(region);

  const linkedinUrl =
    linkedin !== null
      ? extractLinkedInCompanyUrl(serperResult.link) ?? website.linkedinUrl
      : website.linkedinUrl;

  const growthSignals = buildGrowthSignals(website, linkedin);

  return {
    domain: website.domain,
    name: name ?? "Unknown",
    industry,
    employee_count: employeeCount,
    employee_range: employeeRange,
    description: description?.slice(0, 1000) ?? null,
    website_url: `https://${website.domain}`,
    linkedin_url: linkedinUrl ?? null,
    country,
    region,
    tech_stack: website.techStack,
    growth_signals: growthSignals,
    raw_data: {
      serper_title: serperResult.title,
      serper_snippet: serperResult.snippet,
      serper_link: serperResult.link,
    },
    source,
    status: "discovered",
  };
}

// ─── Helpers ─────────────────────────────────────────────────

function cleanTitle(title: string): string {
  return title
    .split(/[|\-–]/)[0]
    .replace(/\b(home|official|website|welcome to)\b/gi, "")
    .trim();
}

function estimateRange(count: number | null): string | null {
  if (!count) return null;
  if (count < 50) return "10-49";
  if (count < 100) return "50-99";
  if (count < 200) return "100-199";
  if (count < 500) return "200-499";
  if (count < 1000) return "500-999";
  return "1000+";
}

function extractCountry(headquarters: string): string | null {
  const parts = headquarters.split(",");
  return parts[parts.length - 1]?.trim() ?? null;
}

function regionToCountry(region: string): string {
  const map: Record<string, string> = {
    US: "United States", UK: "United Kingdom",
    AU: "Australia", UAE: "United Arab Emirates",
    SA: "Saudi Arabia", SG: "Singapore",
  };
  return map[region] ?? region;
}

function extractLinkedInCompanyUrl(url: string): string | null {
  const match = url.match(/https?:\/\/(www\.)?linkedin\.com\/company\/[^/?#]+/);
  return match ? match[0] : null;
}

function buildGrowthSignals(
  website: ScrapedCompany,
  linkedin: LinkedInCompanyData | null
): object[] {
  const signals: object[] = [];

  // AI-related job postings
  website.jobPostings
    .filter((j) => j.hasAiSignal)
    .forEach((j) => {
      signals.push({
        type: "job_posting",
        content: j.title,
        source_url: j.url,
        ai_related: true,
        detected_at: new Date().toISOString(),
      });
    });

  // Recent LinkedIn posts
  (linkedin?.recentPosts ?? []).forEach((post) => {
    signals.push({
      type: "social_post",
      content: post.slice(0, 200),
      ai_related: /automat|ai\b|machine learning|digital/i.test(post),
      detected_at: new Date().toISOString(),
    });
  });

  return signals.slice(0, 10); // cap at 10 signals
}