pitchfast/lib/lead-discovery-run.ts

import {
  normalizePhone,
  normalizeText,
  getUsableContactEmail,
  type GooglePlaceCandidate,
} from "./lead-discovery-google";
import type { Id } from "../convex/_generated/dataModel";

type AgentRunLike = {
  status: string;
  updatedAt?: number;
};

type LeadDiscoveryCounterInput = {
  leadsFound: number;
  leadsCreated: number;
  errors: number;
};

type LeadDiscoveryContactInput = {
  usableEmail?: string | null;
};

export type LeadDiscoveryContactStatus =
  | "new"
  | "missing_contact"
  | "audit_ready"
  | "outreach_ready"
  | "contacted"
  | "replied"
  | "do_not_contact";

type WebsiteEnrichmentScheduleInput = {
  websiteUrl?: string | null;
  websiteDomain?: string | null;
  contactStatus: LeadDiscoveryContactStatus;
};

export type LeadDiscoveryPriority = "high" | "medium" | "low" | "defer" | "blocked";

type LeadDiscoveryPriorityInput = {
  isBlacklisted?: boolean;
  isDuplicate?: boolean;
  hasWebsite?: boolean;
  hasWebsiteSignal?: boolean;
};

type LeadDiscoveryLeadRecordInput<TCampaignId extends string, TRunId extends string> = {
  campaignId: TCampaignId;
  runId: TRunId;
  niche: string;
  postalCode: string;
  candidate: GooglePlaceCandidate;
  now: number;
};

function optionalString(value: string | null | undefined) {
  return value && value.trim().length > 0 ? value : undefined;
}

function optionalNumber(value: number | null) {
  return typeof value === "number" && Number.isFinite(value)
    ? value
    : undefined;
}

export const PENDING_AGENT_RUN_GRACE_MS = 10 * 60 * 1000;

export function isStalePendingAgentRun(run: AgentRunLike, now: number) {
  const updatedAt = typeof run.updatedAt === "number" ? run.updatedAt : 0;

  return (
    run.status === "pending" &&
    updatedAt > 0 &&
    now - updatedAt > PENDING_AGENT_RUN_GRACE_MS
  );
}

export function canStartAgentRun(runs: AgentRunLike[], now = Date.now()) {
  return !runs.some((run) => {
    if (run.status === "running") {
      return true;
    }

    return run.status === "pending" && !isStalePendingAgentRun(run, now);
  });
}

export function buildLeadDiscoveryCounters(input: LeadDiscoveryCounterInput) {
  return {
    leadsFound: input.leadsFound,
    leadsCreated: input.leadsCreated,
    auditsCreated: 0,
    outreachPrepared: 0,
    errors: input.errors,
  };
}

export function getLeadDiscoveryContactStatus(
  input: LeadDiscoveryContactInput,
) {
  if (input.usableEmail) {
    return "new";
  }

  return "missing_contact";
}

export function shouldScheduleWebsiteEnrichment(
  input: WebsiteEnrichmentScheduleInput,
) {
  const hasWebsiteData =
    optionalString(input.websiteUrl) !== undefined ||
    optionalString(input.websiteDomain) !== undefined;

  return input.contactStatus === "missing_contact" && hasWebsiteData;
}

export function buildLeadDiscoveryLeadRecord<
  TCampaignId extends string,
  TRunId extends string,
>(input: LeadDiscoveryLeadRecordInput<TCampaignId, TRunId>) {
  type LeadDiscoveryDuplicateStatus =
    | "unchecked"
    | "unique"
    | "possible_duplicate"
    | "duplicate";

  const usableEmail = getUsableContactEmail(input.candidate);

  const lead: {
    campaignId: TCampaignId;
    discoveryRunId: TRunId;
    companyName: string;
    niche: string;
    address: string;
    postalCode: string;
    googlePlaceId: string;
    googleMapsUrl?: string;
    googlePrimaryType?: string;
    googleTypes: string[];
    googleRating?: number;
    googleUserRatingCount?: number;
    googleBusinessStatus?: string;
    sourceProvider: "google_places" | "local_business_data";
    sourceBusinessId?: string;
    sourceFetchedAt: number;
    websiteUrl?: string;
    websiteDomain?: string;
    phone?: string;
    normalizedGooglePlaceId?: string;
    normalizedSourceBusinessId?: string;
    normalizedEmail?: string;
    normalizedPhone?: string;
    normalizedCompanyName?: string;
    normalizedAddress?: string;
    email?: string;
    emailSource?: string;
    contactPerson?: string;
    priorityReason?: string;
    duplicateReason?: string;
    duplicateOfLeadId?: Id<"leads">;
    blacklistReason?: string;
    priority: LeadDiscoveryPriority;
    contactStatus: "new" | "missing_contact";
    duplicateStatus: LeadDiscoveryDuplicateStatus;
    blacklistStatus: "clear";
    createdAt: number;
    updatedAt: number;
  } = {
    campaignId: input.campaignId,
    discoveryRunId: input.runId,
    companyName: input.candidate.businessName,
    niche: input.niche,
    address: input.candidate.address,
    postalCode: input.postalCode,
    googlePlaceId: input.candidate.placeId,
    googleTypes: input.candidate.googleTypes,
    sourceProvider: input.candidate.sourceProvider,
    sourceFetchedAt: input.candidate.sourceFetchedAt,
    priority: "medium",
    contactStatus: getLeadDiscoveryContactStatus({
      usableEmail: usableEmail?.email,
    }),
    duplicateStatus: "unique",
    blacklistStatus: "clear",
    createdAt: input.now,
    updatedAt: input.now,
  };

  const googleMapsUrl = optionalString(input.candidate.googleMapsUrl);
  const googlePrimaryType = optionalString(input.candidate.googlePrimaryType);
  const googleRating = optionalNumber(input.candidate.rating);
  const googleUserRatingCount = optionalNumber(input.candidate.userRatingCount);
  const googleBusinessStatus = optionalString(input.candidate.businessStatus);
  const sourceBusinessId = optionalString(input.candidate.sourceBusinessId);
  const websiteUrl = optionalString(input.candidate.websiteUrl);
  const websiteDomain = optionalString(input.candidate.websiteDomain);
  const phone = optionalString(input.candidate.phone);
  const normalizedPhone = normalizePhone(phone);
  const normalizedCompanyName = normalizeText(input.candidate.businessName);
  const normalizedAddress = normalizeText(input.candidate.address);

  if (normalizedCompanyName !== "") {
    lead.normalizedCompanyName = normalizedCompanyName;
  }

  if (normalizedAddress !== "") {
    lead.normalizedAddress = normalizedAddress;
  }

  if (normalizedPhone !== "") {
    lead.normalizedPhone = normalizedPhone;
  }

  if (googleMapsUrl !== undefined) {
    lead.googleMapsUrl = googleMapsUrl;
  }
  if (googlePrimaryType !== undefined) {
    lead.googlePrimaryType = googlePrimaryType;
  }
  if (googleRating !== undefined) {
    lead.googleRating = googleRating;
  }
  if (googleUserRatingCount !== undefined) {
    lead.googleUserRatingCount = googleUserRatingCount;
  }
  if (googleBusinessStatus !== undefined) {
    lead.googleBusinessStatus = googleBusinessStatus;
  }
  if (sourceBusinessId !== undefined) {
    lead.sourceBusinessId = sourceBusinessId;
  }
  if (websiteUrl !== undefined) {
    lead.websiteUrl = websiteUrl;
  }
  if (websiteDomain !== undefined) {
    lead.websiteDomain = websiteDomain;
  }
  if (phone !== undefined) {
    lead.phone = phone;
  }
  if (usableEmail) {
    lead.normalizedEmail = usableEmail.email;
    lead.email = usableEmail.email;
    if (usableEmail.emailSource !== null) {
      lead.emailSource = usableEmail.emailSource;
    }
    if (usableEmail.contactPerson !== null) {
      lead.contactPerson = usableEmail.contactPerson;
    }
  } else {
    lead.contactStatus = "missing_contact";
  }

  return lead;
}

export function getLeadDiscoveryPriority(
  input: LeadDiscoveryPriorityInput,
): { priority: LeadDiscoveryPriority; reason: string } {
  if (input.isBlacklisted) {
    return {
      priority: "blocked",
      reason: "Lead ist auf der Sperrliste.",
    };
  }

  if (input.isDuplicate) {
    return {
      priority: "defer",
      reason: "Dublettenprüfung oder Reviewpause.",
    };
  }

  if (!input.hasWebsite) {
    return {
      priority: "high",
      reason: "Kein Website-Indikator vorhanden.",
    };
  }

  if (input.hasWebsiteSignal) {
    return {
      priority: "low",
      reason: "Website vorhanden: geringer Kontaktaufwand.",
    };
  }

  return {
    priority: "medium",
    reason: "Standardpriorität.",
  };
}