craftvia/src/lib/imports/extraction.ts

import { z } from "zod";
import type { ExtractedField, WorkOrderExtraction } from "@/server/ai/providers";

/**
 * Client-safe extraction model for the order import (spec §9.4/§9.5).
 *
 * - `EXTRACTION_FIELDS`: canonical field order (review mask, corrections, tests).
 * - `workOrderExtractionSchema`: Zod validation of provider output (null → undefined for sub-objects).
 * - `extractionJsonSchema()`: JSON schema for Claude structured outputs (all keys required,
 *   nullable via anyOf, additionalProperties:false — the structured-output subset).
 * - `StoredExtraction`: shape persisted in ImportJob.extraction.
 */

export const EXTRACTION_VERSION = "craftvia-import-v1";

export const EXTRACTION_FIELDS = [
  "orderNumber",
  "offerNumber",
  "customerNumber",
  "companyName",
  "customerFirstName",
  "customerLastName",
  "customerAddress",
  "siteName",
  "siteAddress",
  "contactName",
  "phone",
  "email",
  "orderDate",
  "documentDate",
  "plannedStart",
  "plannedEnd",
  "title",
  "description",
  "positions",
  "notes",
  "totalAmount",
  "references",
] as const satisfies readonly (keyof WorkOrderExtraction)[];

export type ExtractionFieldKey = (typeof EXTRACTION_FIELDS)[number];

export type Address = { street?: string; houseNumber?: string; postalCode?: string; city?: string; country?: string };
export type Position = { position?: string; name: string; articleNumber?: string; quantity?: number; unit?: string; isMaterial?: boolean };

const confidence = z.number().min(0).max(1).catch(0);
const source = z.string().nullish().transform((v) => v ?? undefined);
const optStr = z.string().nullish().transform((v) => (v == null || v.trim() === "" ? undefined : v.trim()));

function field<T extends z.ZodTypeAny>(value: T) {
  return z.object({ value: value.nullable().catch(null), confidence, source });
}

const addressSchema = z.object({
  street: optStr,
  houseNumber: optStr,
  postalCode: optStr,
  city: optStr,
  country: optStr,
});

const positionSchema = z.object({
  position: optStr,
  name: z.string().min(1),
  articleNumber: optStr,
  quantity: z.number().nullish().transform((v) => (v == null || Number.isNaN(v) ? undefined : v)),
  unit: optStr,
  isMaterial: z.boolean().nullish().transform((v) => v ?? undefined),
});

const str = z.string().transform((v) => v.trim());

export const workOrderExtractionSchema = z.object({
  orderNumber: field(str),
  offerNumber: field(str),
  customerNumber: field(str),
  companyName: field(str),
  customerFirstName: field(str),
  customerLastName: field(str),
  customerAddress: field(addressSchema),
  siteName: field(str),
  siteAddress: field(addressSchema),
  contactName: field(str),
  phone: field(str),
  email: field(str),
  orderDate: field(str),
  documentDate: field(str),
  plannedStart: field(str),
  plannedEnd: field(str),
  title: field(str),
  description: field(str),
  positions: field(z.array(positionSchema)),
  notes: field(str),
  totalAmount: field(z.number()),
  references: field(z.array(z.string())),
});

/** Parse arbitrary provider output into a WorkOrderExtraction (throws on structurally invalid input). */
export function parseExtraction(raw: unknown): WorkOrderExtraction {
  const parsed = workOrderExtractionSchema.parse(raw);
  // Empty strings count as "not found".
  const out = {} as Record<string, ExtractedField<unknown>>;
  for (const key of EXTRACTION_FIELDS) {
    const f = parsed[key] as ExtractedField<unknown>;
    const empty = f.value === null || (typeof f.value === "string" && f.value === "");
    out[key] = { value: empty ? null : f.value, confidence: empty ? 0 : f.confidence, ...(f.source ? { source: f.source } : {}) };
  }
  return out as unknown as WorkOrderExtraction;
}

/** Extraction with every field empty (no provider / manual entry). */
export function emptyExtraction(): WorkOrderExtraction {
  const out = {} as Record<string, ExtractedField<unknown>>;
  for (const key of EXTRACTION_FIELDS) out[key] = { value: null, confidence: 0 };
  return out as unknown as WorkOrderExtraction;
}

// ---------- JSON schema for structured outputs ----------

type Json = Record<string, unknown>;
const nullable = (schema: Json): Json => ({ anyOf: [schema, { type: "null" }] });
const obj = (properties: Record<string, Json>): Json => ({
  type: "object",
  properties,
  required: Object.keys(properties),
  additionalProperties: false,
});

const addressJson = obj({
  street: nullable({ type: "string" }),
  houseNumber: nullable({ type: "string" }),
  postalCode: nullable({ type: "string" }),
  city: nullable({ type: "string" }),
  country: nullable({ type: "string", description: "ISO 3166-1 alpha-2, e.g. DE" }),
});

const positionJson = obj({
  position: nullable({ type: "string" }),
  name: { type: "string" },
  articleNumber: nullable({ type: "string" }),
  quantity: nullable({ type: "number" }),
  unit: nullable({ type: "string" }),
  isMaterial: nullable({ type: "boolean" }),
});

const fieldJson = (value: Json, description: string): Json => ({
  ...obj({
    value: nullable(value),
    confidence: { type: "number", description: "0..1 — how certain the value is correct" },
    source: nullable({ type: "string", description: "short verbatim snippet from the document" }),
  }),
  description,
});

const FIELD_DESCRIPTIONS: Record<ExtractionFieldKey, [Json, string]> = {
  orderNumber: [{ type: "string" }, "Order / confirmation number of the issuer (Auftragsnummer)"],
  offerNumber: [{ type: "string" }, "Offer number (Angebotsnummer)"],
  customerNumber: [{ type: "string" }, "Customer number (Kundennummer)"],
  companyName: [{ type: "string" }, "Company name of the customer (the ordering party, NOT the craft business)"],
  customerFirstName: [{ type: "string" }, "First name if the customer is a private person"],
  customerLastName: [{ type: "string" }, "Last name if the customer is a private person"],
  customerAddress: [addressJson, "Billing / postal address of the customer"],
  siteName: [{ type: "string" }, "Name of the site / construction site (Objekt, Baustelle)"],
  siteAddress: [addressJson, "Address where the work is carried out"],
  contactName: [{ type: "string" }, "Contact person"],
  phone: [{ type: "string" }, "Phone number of the customer or contact"],
  email: [{ type: "string" }, "E-mail address of the customer or contact"],
  orderDate: [{ type: "string" }, "Order date, ISO 8601 (YYYY-MM-DD)"],
  documentDate: [{ type: "string" }, "Document date, ISO 8601 (YYYY-MM-DD)"],
  plannedStart: [{ type: "string" }, "Planned start of execution, ISO 8601 (YYYY-MM-DD)"],
  plannedEnd: [{ type: "string" }, "Planned end of execution, ISO 8601 (YYYY-MM-DD)"],
  title: [{ type: "string" }, "Short title of the job (max. 80 characters)"],
  description: [{ type: "string" }, "Description of services (Leistungsbeschreibung)"],
  positions: [{ type: "array", items: positionJson }, "Line items with quantities; isMaterial=true for material/articles"],
  notes: [{ type: "string" }, "Special notes (access, deadlines, safety)"],
  totalAmount: [{ type: "number" }, "Total gross amount in EUR as a number"],
  references: [{ type: "array", items: { type: "string" } }, "Other references (project no., purchase order no.)"],
};

/** JSON schema of the complete provider answer: full text + structured fields. */
export function extractionJsonSchema(): Json {
  const fields: Record<string, Json> = {};
  for (const key of EXTRACTION_FIELDS) {
    const [value, description] = FIELD_DESCRIPTIONS[key];
    fields[key] = fieldJson(value, description);
  }
  return obj({
    text: { type: "string", description: "Full recognised text of the document (reading order)" },
    extraction: obj(fields),
  });
}

// ---------- persisted shape ----------

export type PlausibilityHintCode =
  | "date_invalid"
  | "date_implausible"
  | "postal_code_invalid"
  | "email_invalid"
  | "phone_invalid"
  | "end_before_start"
  | "manual_entry";

export type PlausibilityHint = { field: ExtractionFieldKey | null; code: PlausibilityHintCode };

export type DuplicateCandidate = { customerId: string; score: number; reasons: string[] };
export type SiteCandidate = { siteId: string; customerId: string; name: string; score: number; reasons: string[] };

export type StoredExtraction = {
  fields: WorkOrderExtraction;
  hints: PlausibilityHint[];
  siteCandidates: SiteCandidate[];
};

export function readStoredExtraction(raw: unknown): StoredExtraction {
  const r = (raw ?? {}) as Partial<StoredExtraction>;
  let fields: WorkOrderExtraction;
  try {
    fields = r.fields ? parseExtraction(r.fields) : emptyExtraction();
  } catch {
    fields = emptyExtraction();
  }
  return {
    fields,
    hints: Array.isArray(r.hints) ? r.hints : [],
    siteCandidates: Array.isArray(r.siteCandidates) ? r.siteCandidates : [],
  };
}

/** Fields below this confidence are marked "unsicher" in the review mask. */
export const LOW_CONFIDENCE = 0.8;