[vakifbank-tr] Fix bugs with pdf parser (#783)

zenmoney · Nov 11, 2024 · e1eec02 · e1eec02
1 parent a4f70d0
commit e1eec02
Show file tree

Hide file tree

Showing 3 changed files with 135 additions and 71 deletions.
diff --git a/src/plugins/vakifbank-tr/__tests__/statementTransaction.test.ts b/src/plugins/vakifbank-tr/__tests__/statementTransaction.test.ts
@@ -1,4 +1,5 @@
 import { AccountType } from '../../../types/zenmoney'
+import { extractInstrument } from '../api'
 
 import { convertVakifPdfStatementTransaction } from '../converters'
 import { TransactionWithId, VakifStatementTransaction } from '../models'
@@ -170,3 +171,8 @@ const testCases: Array<[VakifStatementTransaction[], TransactionWithId[]]> = [
 it.each(testCases)('converts transaction parsed from pdf statement', (rawTransaction: VakifStatementTransaction[], transaction: TransactionWithId[]) => {
   expect(convertVakifPdfStatementTransaction('1234567890', rawTransaction)).toEqual(transaction)
 })
+
+it('can extract account instrument code', () => {
+  expect(extractInstrument(' Account Type:VADESİZ USDNickname: ')).toEqual('USD')
+  expect(extractInstrument(' Account Type:VADESİZ TLNickname: ')).toEqual('TL')
+})
diff --git a/src/plugins/vakifbank-tr/api.ts b/src/plugins/vakifbank-tr/api.ts
@@ -5,21 +5,16 @@ import { parsePdfFromBlob } from './pdfToStr'
 import { parseDateAndTimeFromPdfText, parseDateFromPdfText, parseFormattedNumber } from './converters'
 
 export async function parsePdfVakifStatement (): Promise<null | Array<{ account: VakifStatementAccount, transactions: VakifStatementTransaction[] }>> {
-  const blob = await ZenMoney.pickDocuments(['application/pdf'], true)
-  if (blob == null || !blob.length) {
-    throw new TemporaryError('Выберите один или несколько файлов в формате .pdf')
-  }
-  for (const { size, type } of blob) {
-    if (type !== 'application/pdf') {
-      throw new TemporaryError('Выписка должна быть в расширении .pdf')
-    } else if (size >= 1000 * 1000) {
-      throw new TemporaryError('Максимальный размер файла - 1 мб')
-    }
-  }
+  const blob = await getPdfDocuments()
+  validateDocuments(blob)
   const pdfStrings = await parsePdfFromBlob({ blob })
+  return parsePdfStatements(pdfStrings)
+}
+
+export function parsePdfStatements (pdfStrings: string[]): Array<{ account: VakifStatementAccount, transactions: VakifStatementTransaction[] }> {
   const result = []
   for (const textItem of pdfStrings) {
-    if (!/www.vakifbank.com.tr/i.test(textItem)) {
+    if (!isVakifBankStatement(textItem)) {
       throw new TemporaryError('Похоже, это не выписка VakifBank')
     }
     try {
@@ -32,12 +27,34 @@ export async function parsePdfVakifStatement (): Promise<null | Array<{ account:
   return result
 }
 
+function isVakifBankStatement (text: string): boolean {
+  return /www.vakifbank.com.tr/i.test(text)
+}
+
+export function validateDocuments (blob: Blob[]): void {
+  for (const { size, type } of blob) {
+    if (type !== 'application/pdf') {
+      throw new TemporaryError('Выписка должна быть в расширении .pdf')
+    } else if (size >= 1000 * 1000) {
+      throw new TemporaryError('Максимальный размер файла - 1 мб')
+    }
+  }
+}
+
+async function getPdfDocuments (): Promise<Blob[]> {
+  const blob = await ZenMoney.pickDocuments(['application/pdf'], true)
+  if (blob == null || !blob.length) {
+    throw new TemporaryError('Выберите один или несколько файлов в формате .pdf')
+  }
+  return blob
+}
+
 export function parseSinglePdfString (text: string, statementUid?: string): { account: VakifStatementAccount, transactions: VakifStatementTransaction[] } {
   const balanceAmount = extractBalance(text)
   const rawAccount: VakifStatementAccount = {
     balance: balanceAmount,
     id: extractAccountId(text),
-    instrument: 'TL',
+    instrument: extractInstrument(text),
     title: 'Vakifbank *' + extractAccountId(text).slice(-4),
     date: extractStatementDate(text)
   }
@@ -49,6 +66,11 @@ export function parseSinglePdfString (text: string, statementUid?: string): { ac
   return parsedContent
 }
 
+export function extractInstrument (text: string): string {
+  const match = text.match(/Account Type:VADESİZ\s+([A-Z]{2,3})Nickname:/)
+  return match ? match[1] : 'TL' // Default to 'TL' if no match found
+}
+
 function extractAccountId (text: string): string {
   const accountNumRegex = /Account No: (\d{17})/
   const match = accountNumRegex.exec(text)

diff --git a/src/plugins/vakifbank-tr/converters.ts b/src/plugins/vakifbank-tr/converters.ts
@@ -1,8 +1,11 @@
 import { groupBy, maxBy, minBy } from 'lodash'
 
-import { Amount, AccountType, NonParsedMerchant, AccountOrCard } from '../../types/zenmoney'
+import { Amount, AccountType, NonParsedMerchant, AccountOrCard, Transaction } from '../../types/zenmoney'
 import { TransactionWithId, VakifStatementAccount, VakifStatementTransaction } from './models'
 
+const MERCHANT_TITLE_REGEX = /ISLEM NO :\s*(\d{4})?-(.*?)\s\s(.*?)\*\*\*\*/
+const MERCHANT_MCC_REGEX = /Mcc: (\d{4})/
+
 export function parseFormattedNumber (value: string): number {
   return parseFloat(value.replace(/\./g, '').replace(',', '.'))
 }
@@ -32,68 +35,101 @@ export function convertVakifPdfStatementTransaction (accountId: string, rawTrans
 
   const chunks = chunksByStatementUid(rawTransaction)
   for (const transactions of chunks) {
-    if (transactions.length === 1 || transactions.length === 2) {
-      const mainTransaction = maxBy(transactions, x => Math.abs(parseFormattedNumber(x.amount)))
-      const feeTransaction = transactions.length === 1 ? null : minBy(transactions, x => Math.abs(parseFormattedNumber(x.amount)))
-      const transaction = mainTransaction
-      if (transaction == null) { throw new Error('InvalidState') }
-      let merchant: NonParsedMerchant | null = null
-      let comment: string | null = null
-
-      const merchanTitleRegEx = /ISLEM NO :\s*(\d{4})?-(.*?)\s\s(.*?)\*\*\*\*/
-      const merchantMccRegEx = /Mcc: (\d{4})/
-      const merchanTitleMatch = merchanTitleRegEx.exec(transaction.description2 ?? '')
-      const merchantMccMatch = merchantMccRegEx.exec(transaction.description2 ?? '')
-
-      if (merchanTitleMatch != null || merchantMccMatch != null) {
-        merchant = {
-          location: null,
-          fullTitle: merchanTitleMatch?.[2] ?? '',
-          mcc: merchantMccMatch == null ? null : parseInt(merchantMccMatch[1], 10)
-        }
-      } else {
-        comment = (transaction.description1 ?? '') + ': ' + (transaction.description2 ?? '')
-      }
-
-      result.push({
-        statementUid: transaction.statementUid,
-        transaction: {
-          comment,
-          date: new Date(transaction.date),
-          hold: false,
-          merchant,
-          movements: [
-            {
-              account: { id: accountId },
-              fee: feeTransaction == null ? 0 : parseFormattedNumber(feeTransaction.amount),
-              id: transaction.statementUid,
-              sum: parseFormattedNumber(transaction.amount),
-              invoice: null
-            }
-          ]
-        }
-      })
-      if (transaction.description1 === 'ATM Withdrawal') {
-        result[0].transaction.comment = transaction.description2
-        result[0].transaction.movements.push({
-          account: {
-            company: null,
-            instrument: 'TL',
-            syncIds: null,
-            type: AccountType.cash
-          },
-          fee: 0,
-          id: transaction.statementUid,
-          sum: parseFormattedNumber(transaction.amount) * -1,
-          invoice: null
-        })
-      }
-    }
+    if (transactions.length !== 1 && transactions.length !== 2) continue
+
+    const transaction = buildTransaction(accountId, transactions)
+    result.push(transaction)
   }
 
   return result
 }
 
+function buildTransaction (accountId: string, transactions: VakifStatementTransaction[]): TransactionWithId {
+  const mainTransaction = getMainTransaction(transactions)
+  const feeAmount = getFeeAmount(transactions)
+  const merchant = extractMerchantInfo(mainTransaction)
+  const comment = merchant ? null : `${mainTransaction.description1 ?? ''}: ${mainTransaction.description2 ?? ''}`
+
+  const transaction: Transaction = {
+    comment,
+    date: new Date(mainTransaction.date),
+    hold: false,
+    merchant,
+    movements: [
+      createMainMovement(accountId, mainTransaction, feeAmount)
+    ]
+  }
+
+  if (mainTransaction.description1 === 'ATM Withdrawal' || mainTransaction.description1 === 'ATM QR Withdrawal') {
+    transaction.comment = mainTransaction.description2
+    transaction.movements.push(createOppositeMovement(mainTransaction, AccountType.cash))
+  }
+
+  return {
+    statementUid: mainTransaction.statementUid,
+    transaction
+  }
+}
+
+function getMainTransaction (transactions: VakifStatementTransaction[]): VakifStatementTransaction {
+  const mainTransaction = maxBy(transactions, x => Math.abs(parseFormattedNumber(x.amount)))
+  if (!mainTransaction) throw new Error('InvalidState')
+  return mainTransaction
+}
+
+function getFeeAmount (transactions: VakifStatementTransaction[]): number {
+  if (transactions.length !== 2) return 0
+
+  const feeTransaction = minBy(transactions, x => Math.abs(parseFormattedNumber(x.amount)))
+  return feeTransaction ? parseFormattedNumber(feeTransaction.amount) : 0
+}
+
+function createMainMovement (
+  accountId: string,
+  transaction: VakifStatementTransaction,
+  feeAmount: number
+): Transaction['movements'][number] {
+  return {
+    account: { id: accountId },
+    fee: feeAmount,
+    id: transaction.statementUid,
+    sum: parseFormattedNumber(transaction.amount),
+    invoice: null
+  }
+}
+
+function createOppositeMovement (
+  transaction: VakifStatementTransaction,
+  accountType: AccountType
+): Transaction['movements'][number] {
+  return {
+    account: {
+      company: null,
+      instrument: 'TL',
+      syncIds: null,
+      type: accountType
+    },
+    fee: 0,
+    id: transaction.statementUid,
+    sum: parseFormattedNumber(transaction.amount) * -1,
+    invoice: null
+  }
+}
+
+function extractMerchantInfo (transaction: VakifStatementTransaction): NonParsedMerchant | null {
+  const merchanTitleMatch = MERCHANT_TITLE_REGEX.exec(transaction.description2 ?? '')
+  const merchantMccMatch = MERCHANT_MCC_REGEX.exec(transaction.description2 ?? '')
+
+  if (merchanTitleMatch != null || merchantMccMatch != null) {
+    return {
+      location: null,
+      fullTitle: merchanTitleMatch?.[2] ?? '',
+      mcc: merchantMccMatch ? parseInt(merchantMccMatch[1], 10) : null
+    }
+  }
+  return null
+}
+
 export function convertPdfStatementAccount (rawAccount: VakifStatementAccount): AccountOrCard {
   const account: AccountOrCard = {
     id: rawAccount.id,