feat(etl): контролен код по Булстат за идентичност по ЕИК

packages/db/src/competition-sql.test.ts

@@ -31,10 +31,10 @@ function readScript(dbPath: string, path: string): void {
 const FIXTURE = `
 INSERT INTO authorities (id, name, bulstat, type_group) VALUES
   ('auth:A', 'Институция А', '100000001', 'община'),
-  ('auth:B', 'Институция Б', '100000002', 'община');
+  ('auth:B', 'Институция Б', '100000019', 'община');
 INSERT INTO bidders (id, name, bulstat, eik_normalized, eik_valid, kind) VALUES
-  ('eik:X', 'Фирма Х', '200000001', '200000001', 1, 'company'),
-  ('eik:Y', 'Фирма У', '200000002', '200000002', 1, 'company');
+  ('eik:X', 'Фирма Х', '200000002', '200000002', 1, 'company'),
+  ('eik:Y', 'Фирма У', '200000015', '200000015', 1, 'company');
 INSERT INTO tenders (id, source_id, title, authority_id, cpv_code, procedure_type, status) VALUES
   ('t:A', 'UNP-A', 'Поръчка А', 'auth:A', '45', 'открита процедура', 'awarded'),
   ('t:B', 'UNP-B', 'Поръчка Б', 'auth:B', '45', 'открита процедура', 'awarded');

packages/db/src/contractor-identity-sql.test.ts

@@ -0,0 +1,132 @@
+/// <reference types="node" />
+import { readFileSync } from 'node:fs';
+import { dirname, resolve } from 'node:path';
+import { DatabaseSync } from 'node:sqlite';
+import { fileURLToPath } from 'node:url';
+import { describe, expect, it } from 'vitest';
+
+const root = resolve(dirname(fileURLToPath(import.meta.url)), '../../..');
+const schema = readFileSync(resolve(root, 'packages/db/migrations/0000_init.sql'), 'utf8');
+const staging = readFileSync(resolve(root, 'scripts/work-staging-schema.sql'), 'utf8');
+const normalize = readFileSync(resolve(root, 'scripts/normalize-raw.sql'), 'utf8');
+const precompute = readFileSync(resolve(root, 'scripts/precompute.sql'), 'utf8');
+const refresh = readFileSync(resolve(root, 'scripts/refresh-slice.sql'), 'utf8');
+
+const seed = `
+INSERT INTO raw_contracts
+  (source, fetched_at, unp, contract_number, contract_date, published_at,
+   authority_eik, authority_name, procurement_subject, estimated_value,
+   signing_value, currency, contractor_eik, contractor_name)
+VALUES
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-VALID', 'C-VALID', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Валиден ЕИК', 100, 100, 'EUR',
+   '103267194', 'ТИПО ООД'),
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-TYPO', 'C-TYPO', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Грешна контролна цифра', 200, 200, 'EUR',
+   '103267195', 'ТИПО ООД'),
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-EMPTY', 'C-EMPTY', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Празно име', 300, 300, 'EUR',
+   '103267195', ''),
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-NULL', 'C-NULL', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Липсваща самоличност', 400, 400, 'EUR',
+   NULL, NULL),
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-FOLD-1', 'C-FOLD-1', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Кавички и регистър 1', 10, 10, 'EUR',
+   '', '„Строй Инвест" ЕООД'),
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-FOLD-2', 'C-FOLD-2', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Кавички и регистър 2', 10, 10, 'EUR',
+   '', '"СТРОЙ ИНВЕСТ" ЕООД'),
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-FOLD-3', 'C-FOLD-3', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Кавички и интервали', 10, 10, 'EUR',
+   '', '«строй   инвест» ЕООД'),
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-DASH-EN', 'C-DASH-EN', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Тире 1', 10, 10, 'EUR',
+   '', '„Марица–Изток"'),
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-DASH-ASCII', 'C-DASH-ASCII', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Тире 2', 10, 10, 'EUR',
+   '', '„Марица-Изток"'),
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-GUARD-HYPHEN', 'C-GUARD-HYPHEN', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Пази тирето', 10, 10, 'EUR',
+   '', '„Марица-Изток"'),
+  ('eop:contracts:test', '2026-07-18T00:00:00Z', 'UNP-GUARD-SPACE', 'C-GUARD-SPACE', '2026-07-01',
+   '2026-07-02', '123456786', 'Тестов възложител', 'Пази интервала', 10, 10, 'EUR',
+   '', '„Марица Изток"');
+`;
+
+function build(path: 'normalize' | 'refresh'): DatabaseSync {
+  const db = new DatabaseSync(':memory:');
+  db.exec(schema);
+  db.exec(staging);
+  db.exec(seed);
+  if (path === 'normalize') {
+    db.exec(normalize);
+    db.exec(precompute);
+  } else {
+    db.exec(refresh);
+  }
+  return db;
+}
+
+describe.each(['normalize', 'refresh'] as const)('%s contractor identity', (path) => {
+  it('preserves every eligible contract across valid, typo, empty, and NULL identities', () => {
+    const db = build(path);
+    try {
+      const contracts = db
+        .prepare('SELECT contract_number, bidder_id FROM contracts ORDER BY contract_number')
+        .all() as { contract_number: string; bidder_id: string }[];
+
+      expect(contracts).toEqual([
+        { contract_number: 'C-DASH-ASCII', bidder_id: 'name:МАРИЦА-ИЗТОК' },
+        { contract_number: 'C-DASH-EN', bidder_id: 'name:МАРИЦА-ИЗТОК' },
+        { contract_number: 'C-EMPTY', bidder_id: 'unknown:анонимен' },
+        { contract_number: 'C-FOLD-1', bidder_id: 'name:СТРОЙ ИНВЕСТ ЕООД' },
+        { contract_number: 'C-FOLD-2', bidder_id: 'name:СТРОЙ ИНВЕСТ ЕООД' },
+        { contract_number: 'C-FOLD-3', bidder_id: 'name:СТРОЙ ИНВЕСТ ЕООД' },
+        { contract_number: 'C-GUARD-HYPHEN', bidder_id: 'name:МАРИЦА-ИЗТОК' },
+        { contract_number: 'C-GUARD-SPACE', bidder_id: 'name:МАРИЦА ИЗТОК' },
+        { contract_number: 'C-NULL', bidder_id: 'unknown:анонимен' },
+        { contract_number: 'C-TYPO', bidder_id: 'name:ТИПО ООД' },
+        { contract_number: 'C-VALID', bidder_id: 'eik:103267194' },
+      ]);
+
+      expect(
+        db
+          .prepare(
+            `SELECT id, name, bulstat, eik_normalized, eik_valid, is_consortium, kind
+             FROM bidders WHERE id = 'unknown:анонимен'`,
+          )
+          .get(),
+      ).toEqual({
+        id: 'unknown:анонимен',
+        name: 'Неизвестен изпълнител',
+        bulstat: null,
+        eik_normalized: null,
+        eik_valid: 0,
+        is_consortium: 0,
+        kind: 'unknown',
+      });
+
+      const sums = db
+        .prepare(
+          `SELECT
+             (SELECT SUM(signing_value) FROM raw_contracts
+              WHERE source LIKE 'eop:%' OR source LIKE 'ocds:%') AS raw_total,
+             (SELECT SUM(amount_eur) FROM contracts) AS contract_total,
+             (SELECT SUM(won_eur) FROM company_totals) AS company_total`,
+        )
+        .get() as { raw_total: number; contract_total: number; company_total: number };
+      expect(sums).toEqual({ raw_total: 1070, contract_total: 1070, company_total: 1070 });
+
+      expect(
+        db
+          .prepare(
+            `SELECT COUNT(*) AS n FROM search_index
+             WHERE kind = 'company' AND ref = 'unknown:анонимен'`,
+          )
+          .get(),
+      ).toEqual({ n: 0 });
+    } finally {
+      db.close();
+    }
+  });
+});

packages/db/src/etl-entity-canonicalization-sql.test.ts

@@ -134,7 +134,7 @@ describe('ETL entity canonicalization through real SQL scripts', () => {
   });
 
   it('uses the bidder-name mode for EIK- and name-keyed bidders in both ETL paths', () => {
-    const bidderEik = '400000001';
+    const bidderEik = '400000004';
 
     for (const [label, scriptPath] of etlPaths) {
       withEtlDb(label, (dbPath) => {

packages/db/src/queries/companies.test.ts

@@ -4,11 +4,11 @@ import type { CompanyTotalsRow } from './rows';
 
 const filteredRows: (CompanyTotalsRow & { sort_value: number })[] = [
   {
-    bidder_id: 'eik:111111111',
+    bidder_id: 'eik:111111113',
     name: 'Филтрирана фирма',
     kind: 'company',
     ownership_kind: null,
-    eik: '111111111',
+    eik: '111111113',
     eik_valid: 1,
     settlement: 'София',
     won_eur: 1000,
@@ -25,11 +25,11 @@ const filteredRows: (CompanyTotalsRow & { sort_value: number })[] = [
 const unfilteredRows: (CompanyTotalsRow & { sort_value: number })[] = [
   ...filteredRows,
   {
-    bidder_id: 'eik:999999999',
+    bidder_id: 'eik:999999995',
     name: 'Нефилтрирана фирма',
     kind: 'company',
     ownership_kind: null,
-    eik: '999999999',
+    eik: '999999995',
     eik_valid: 1,
     settlement: 'Пловдив',
     won_eur: 900,
@@ -93,7 +93,7 @@ describe('streamCompaniesCsv', () => {
       .map((line) => line.split(',')[0]);
 
     expect(csvEiks).toEqual(list.items.map((item) => item.eik));
-    expect(csvEiks).toEqual(['111111111']);
+    expect(csvEiks).toEqual(['111111113']);
   });
 
   it('exports fewer rows for a sector filter than for the unfiltered corpus', async () => {
@@ -105,6 +105,26 @@ describe('streamCompaniesCsv', () => {
     expect(countRows(filtered)).toBeLessThan(countRows(unfiltered));
     expect(countRows(filtered)).toBe(1);
   });
+
+  it('excludes the unknown identity bucket from the list and CSV', async () => {
+    const db = fakeDb();
+    const sql: string[] = [];
+    const real = db.prepare.bind(db);
+    db.prepare = ((query: string) => {
+      sql.push(query);
+      return real(query);
+    }) as typeof db.prepare;
+
+    await listCompanies(db, {});
+    await streamCompaniesCsv(db, {}).text();
+
+    expect(sql.filter((query) => query.includes('company_totals'))).not.toHaveLength(0);
+    expect(
+      sql
+        .filter((query) => query.includes('company_totals'))
+        .every((query) => query.includes("kind <> 'unknown'")),
+    ).toBe(true);
+  });
 });
 
 describe('prototype-key params (untrusted query values)', () => {

packages/db/src/queries/companies.ts

@@ -109,7 +109,7 @@ function source(p: CompanyListParams): { from: string; params: unknown[] } {
  * Keep consumed filter keys in sync with COMPANY_FILTER_KEYS and companyFilterSignature().
  */
 function entityWhere(p: CompanyListParams): { sql: string; params: unknown[] } {
-  const where: string[] = [];
+  const where: string[] = [`kind <> 'unknown'`];
   const params: unknown[] = [];
   if (p.kinds?.length && p.kinds.length < 2) {
     where.push(`kind = ?`);

packages/db/src/queries/contracts-filter-sql.test.ts

@@ -26,14 +26,14 @@ const FIXTURE = `
 INSERT INTO authorities (id, name, bulstat, type_group) VALUES
   ('auth:100000001', 'Институция А', '100000001', 'община');
 INSERT INTO bidders (id, name, bulstat, eik_normalized, eik_valid, kind) VALUES
-  ('eik:200000001', 'Фирма Х', '200000001', '200000001', 1, 'company');
+  ('eik:200000002', 'Фирма Х', '200000002', '200000002', 1, 'company');
 INSERT INTO tenders (id, source_id, title, authority_id, cpv_code, procedure_type, status) VALUES
   ('t:A', 'UNP-A', 'Поръчка А', 'auth:100000001', '45000000', 'открита процедура', 'awarded');
 INSERT INTO contracts (id, tender_id, bidder_id, amount, currency, signed_at, bids_received, value_flag, amount_eur) VALUES
-  ('c:NULL', 't:A', 'eik:200000001', 100, 'EUR', '2024-01-01', NULL, 'ok', 100),
-  ('c:ZERO', 't:A', 'eik:200000001', 200, 'EUR', '2024-01-02', 0,    'ok', 200),
-  ('c:ONE',  't:A', 'eik:200000001', 300, 'EUR', '2024-01-03', 1,    'ok', 300),
-  ('c:TWO',  't:A', 'eik:200000001', 400, 'EUR', '2024-01-04', 2,    'ok', 400);
+  ('c:NULL', 't:A', 'eik:200000002', 100, 'EUR', '2024-01-01', NULL, 'ok', 100),
+  ('c:ZERO', 't:A', 'eik:200000002', 200, 'EUR', '2024-01-02', 0,    'ok', 200),
+  ('c:ONE',  't:A', 'eik:200000002', 300, 'EUR', '2024-01-03', 1,    'ok', 300),
+  ('c:TWO',  't:A', 'eik:200000002', 400, 'EUR', '2024-01-04', 2,    'ok', 400);
 `;
 
 /** Minimal D1Database facade over node:sqlite — enough for the query layer's prepare/bind/all/first. */

packages/db/src/queries/contracts.test.ts

@@ -21,9 +21,9 @@ const contractRow = {
   unp: 'UNP-1',
   cpv_code: '45000000',
   eu_funded: 0,
-  authority_id: 'auth:123456789',
+  authority_id: 'auth:123456786',
   authority_name: 'Authority',
-  bidder_id: 'eik:111111111',
+  bidder_id: 'eik:111111113',
   bidder_name: 'Bidder',
   bidder_kind: 'company' as const,
   procedure_type: 'Открита процедура',

packages/db/src/queries/details.test.ts

@@ -41,14 +41,14 @@ const baseContractRow = {
   tender_fx_rate: null as number | null,
   start_date: null,
   end_date: null,
-  authority_id: 'auth:123456789',
+  authority_id: 'auth:123456786',
   authority_name: 'Authority',
   authority_type_group: 'ministry',
   authority_settlement: 'Sofia',
-  bidder_id: 'eik:111111111',
+  bidder_id: 'eik:111111113',
   bidder_name: 'Bidder',
   bidder_kind: 'company' as const,
-  bidder_eik: '111111111',
+  bidder_eik: '111111113',
   bidder_settlement: 'Sofia',
 };
 
@@ -102,7 +102,7 @@ describe('getContract', () => {
           estimated_fx_rate: null,
           bidder_name: 'Bidder',
           bidder_kind: 'company',
-          bidder_id: 'eik:111111111',
+          bidder_id: 'eik:111111113',
         },
         {
           lot_id: 'lot:UNP-1:2',
@@ -147,7 +147,7 @@ describe('getContract', () => {
           estimated_fx_rate: 0.9,
           bidder_name: 'Bidder',
           bidder_kind: 'company',
-          bidder_id: 'eik:111111111',
+          bidder_id: 'eik:111111113',
         },
         {
           lot_id: 'lot:UNP-1:2',

packages/db/src/queries/details.ts

@@ -248,7 +248,7 @@ export async function getAuthority(
       .prepare(
         `SELECT c.bidder_id, b.name, b.kind, SUM(c.amount_eur) AS won, COUNT(*) AS n
          FROM contracts c JOIN tenders t ON t.id = c.tender_id JOIN bidders b ON b.id = c.bidder_id
-         WHERE t.authority_id = ? AND c.amount_eur IS NOT NULL
+         WHERE t.authority_id = ? AND c.amount_eur IS NOT NULL AND b.kind <> 'unknown'
          GROUP BY c.bidder_id ORDER BY won DESC LIMIT 7`,
       )
       .bind(authorityId)

packages/db/src/queries/home.test.ts

@@ -61,9 +61,10 @@ const totalsRow = {
   refreshed_at: '2024-06-02T10:00:00Z',
 };
 
-function fakeDb(totals: typeof totalsRow | null): D1Database {
+function fakeDb(totals: typeof totalsRow | null, capture?: string[]): D1Database {
   return {
     prepare(sql: string) {
+      capture?.push(sql);
       return {
         bind() {
           return this;
@@ -121,6 +122,15 @@ describe('getHomeData', () => {
     expect(data.topMunicipalities).toHaveLength(1);
   });
 
+  it('excludes the unknown identity bucket from top companies', async () => {
+    const sql: string[] = [];
+    await getHomeData(fakeDb(totalsRow, sql));
+
+    expect(sql.find((query) => query.includes('FROM company_totals'))).toContain(
+      "WHERE kind <> 'unknown'",
+    );
+  });
+
   it('includes single-offer contract lists', async () => {
     const data = await getHomeData(fakeDb(totalsRow));
 

packages/db/src/queries/home.ts

@@ -53,7 +53,9 @@ export async function getHomeData(db: D1Database): Promise<HomeData> {
   const [companies, ministries, municipalities, recentSingleOffer, topSingleOffer, singleOfferRow] =
     await Promise.all([
       db
-        .prepare(`SELECT * FROM company_totals ORDER BY won_eur DESC, bidder_id LIMIT 10`)
+        .prepare(
+          `SELECT * FROM company_totals WHERE kind <> 'unknown' ORDER BY won_eur DESC, bidder_id LIMIT 10`,
+        )
         .all<CompanyTotalsRow>(),
       db
         .prepare(

packages/db/src/queries/identity.test.ts

@@ -23,7 +23,11 @@ describe('company slug', () => {
     expect(bidderIdFromSlug(slug)).toBe(id);
   });
   it('round-trips a 13-digit ЕИК', () => {
-    expect(bidderIdFromSlug(companySlug('eik:1234567890123'))).toBe('eik:1234567890123');
+    expect(bidderIdFromSlug(companySlug('eik:8316417910124'))).toBe('eik:8316417910124');
+  });
+  it('resolves the unknown bidder linked from contract pages', () => {
+    expect(companySlug('unknown:анонимен')).toBe('unknown:анонимен');
+    expect(bidderIdFromSlug('unknown:анонимен')).toBe('unknown:анонимен');
   });
 });
 

packages/db/src/queries/identity.ts

@@ -21,7 +21,8 @@ function b64urlDecode(s: string): string {
 
 const EIK_RE = /^\d{9}(\d{4})?$/;
 
-/** bidder id → `/companies/:slug` segment. Valid ЕИК → the digits; name-keyed → `n` + base64url(name). */
+/** bidder id → `/companies/:slug` segment. Valid ЕИК → digits; name-keyed → `n` + base64url(name);
+ * the labelled unknown bucket stays literal so contract-page links remain resolvable. */
 export function companySlug(bidderId: string): string {
   if (bidderId.startsWith('eik:')) return bidderId.slice(4);
   if (bidderId.startsWith('name:')) return 'n' + b64urlEncode(bidderId.slice(5));
@@ -31,6 +32,7 @@ export function companySlug(bidderId: string): string {
 /** `/companies/:slug` segment → bidder id, or null if it cannot be decoded. */
 export function bidderIdFromSlug(slug: string): string | null {
   if (EIK_RE.test(slug)) return 'eik:' + slug;
+  if (slug === 'unknown:анонимен') return slug;
   if (slug.startsWith('n')) {
     try {
       return 'name:' + b64urlDecode(slug.slice(1));