extras_de_cont 1.2.0 → 1.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6dde609d769123f6ba26d34fc0efa79b82ad91b18f6f84d725362480f6c9ef3a
-  data.tar.gz: 236b56eb1c323f3dafc2278135424f5cd51edb336fdd45462c0c73a1e1db83af
+  metadata.gz: df78cdd18c6c97610fd739d7154922d9d5cafe9d8460c54104429698d93d0479
+  data.tar.gz: 8ea76ba23c3c32f5a4d39c7610abdb3dbef95a8c96cb831592316c81165bd610
 SHA512:
-  metadata.gz: 48a679a6514c868d895d68ba56ed821e4528b7b48d9acb90638f16a5d723b2736e7e67afd6fdd85481ec1c9980318e16bc823091f564e6bc1963c554bc505e6c
-  data.tar.gz: a31df94d256b541092714e9d54351cd785e46dd0e58120c593699733cb9f90530548bcce9040d5cdfa36c421d775c80335f95a19f24711918549e454c1fee147
+  metadata.gz: ccfbbd9f393fe5de67d966edbdfae2b50e2cc23a40d8ec5def3e532176648e5243b08e22a7387423ecce688c81e0e0f9dd78a5b493678f620b079ee0e63b5ab5
+  data.tar.gz: 44b5c55785d2b93c0c85797826a90fc56733c9fa2dc60e494e61ee2e3041857498d32603cf410ac9c70998f463971b72134f4a9d8cd986374249713c3fe8a124

data/extras_de_cont.gemspec

@@ -2,7 +2,7 @@
 
 Gem::Specification.new do |s|
   s.name = "extras_de_cont"
-  s.version = "1.2.0"
+  s.version = "1.3.0"
   s.licenses = ["GPLv3"]
   s.summary = "A simple library which helps you extract transactions from a PDF bank statement."
   s.description = <<~TEXT

data/lib/extras_de_cont/rules/ing.rb

@@ -0,0 +1,174 @@
+# frozen_string_literal: true
+
+require "date"
+require "extras_de_cont/transaction"
+
+module ExtrasDeCont
+  module Rules
+    class Ing < Rules::Base
+      ROMANIAN_MONTHS = {
+        "ianuarie" => 1, "februarie" => 2, "martie" => 3, "aprilie" => 4,
+        "mai" => 5, "iunie" => 6, "iulie" => 7, "august" => 8,
+        "septembrie" => 9, "octombrie" => 10, "noiembrie" => 11, "decembrie" => 12
+      }.freeze
+
+      RO_MONTH_NAMES = ROMANIAN_MONTHS.keys.freeze
+      DATE_PATTERN = /\b(\d{1,2})\s+(#{RO_MONTH_NAMES.join("|")})\s+(\d{4})\b/i
+      DATE_PREFIX = /\A\s*#{DATE_PATTERN}/
+
+      TABLE_HEADER_PATTERN = /Data\s+Detalii tranzactie\s+Debit\s+Credit/
+
+      AMOUNT_PATTERN = /\d{1,3}(?:\.\d{3})*,\d{2}/
+
+      NOISE_PATTERNS = [
+        /\AExtras de cont\z/,
+        /\APentru perioada:/,
+        /\AValabil fara semnatura/,
+        /\AING Bank/,
+        /\ASediul:/,
+        /\ANr\. inregistrare/,
+        /\ACIF:/,
+        /\ATitular cont:/,
+        /\ACNP:/,
+        /\AStr\. /,
+        /\ATip cont:/,
+        /\ANumar cont:/,
+        /\AMoneda:/,
+        /\A\d{6},/,
+        /\ARoxana Petria/,
+        /\AAlexandra Ilie/,
+        /\AȘef Serviciu/,
+        /\ASef Serviciu/,
+        /\ASucursala/,
+        /\AÎN/,
+        /\AInformatii despre/,
+        /\Ape www\./,
+        /\d+\/\d+$/
+      ].freeze
+
+      def parse(text)
+        transactions = []
+        current_currency = nil
+        current_table = nil
+        above_lines = []
+        below_lines = []
+        date_line = nil
+
+        each_normalized_line(text) do |line|
+          if line.start_with?("Moneda:")
+            current_currency = line.split.last
+            next
+          end
+
+          if table_header?(line)
+            try_flush(date_line, above_lines, below_lines, current_table, current_currency, transactions)
+            current_table = extract_column_positions(line)
+            above_lines, below_lines, date_line = [], [], nil
+            next
+          end
+
+          if noise?(line)
+            try_flush(date_line, above_lines, below_lines, current_table, current_currency, transactions)
+            above_lines, below_lines, date_line = [], [], nil
+            next
+          end
+
+          next if current_table.nil?
+
+          if date_line?(line)
+            try_flush(date_line, above_lines, below_lines, current_table, current_currency, transactions)
+            date_line = line
+            below_lines = []
+            next
+          end
+
+          if date_line
+            below_lines << line
+          else
+            above_lines << line
+          end
+        end
+
+        try_flush(date_line, above_lines, below_lines, current_table, current_currency, transactions)
+        transactions
+      end
+
+      private
+
+      def each_normalized_line(text)
+        text.each_line do |line|
+          normalized = line.tr("\u00A0", " ").strip
+          next if normalized.empty?
+          yield normalized
+        end
+      end
+
+      def table_header?(line)
+        line.match?(TABLE_HEADER_PATTERN)
+      end
+
+      def date_line?(line)
+        line.match?(DATE_PREFIX)
+      end
+
+      def noise?(line)
+        NOISE_PATTERNS.any? { |pattern| line.match?(pattern) }
+      end
+
+      def extract_column_positions(line)
+        {
+          debit: line.index("Debit"),
+          credit: line.index("Credit")
+        }
+      end
+
+      def try_flush(date_line, above_lines, below_lines, table, currency, transactions)
+        return if date_line.nil? || table.nil?
+
+        transaction = build_transaction(date_line, above_lines, below_lines, table, currency)
+        transactions << transaction if transaction
+      end
+
+      def build_transaction(date_line, above_lines, below_lines, table, currency)
+        date_match = date_line.match(DATE_PREFIX)
+        return if date_match.nil?
+
+        amounts = date_line.to_enum(:scan, AMOUNT_PATTERN).map { Regexp.last_match }
+        return if amounts.empty?
+
+        transaction_amount_match = amounts.last
+        description_start = date_match.end(0)
+        description_end = transaction_amount_match.begin(0)
+        main_description = date_line[description_start...description_end].to_s.strip
+
+        amount_string = transaction_amount_match[0]
+        amount = parse_amount(amount_string)
+        midpoint = (table[:debit] + table[:credit]) / 2
+        amount = -amount if transaction_amount_match.begin(0) < midpoint
+
+        description = build_description(main_description, above_lines, below_lines)
+
+        Transaction.new(
+          parse_date(date_match[1].to_i, date_match[2], date_match[3].to_i),
+          description,
+          amount,
+          currency
+        )
+      end
+
+      def parse_date(day, month_name, year)
+        month = ROMANIAN_MONTHS[month_name.downcase]
+        Date.new(year, month, day)
+      end
+
+      def parse_amount(value)
+        value.delete(".").sub(",", ".").to_f
+      end
+
+      def build_description(main_desc, above_lines, below_lines)
+        parts = [*above_lines.map(&:strip), main_desc, *below_lines.map(&:strip)]
+        parts.reject(&:empty?).join(" | ")
+      end
+    end
+  end
+end

data/lib/extras_de_cont.rb

@@ -3,6 +3,7 @@
 require "extras_de_cont/parser"
 require "extras_de_cont/rules/base"
 require "extras_de_cont/rules/brd"
+require "extras_de_cont/rules/ing"
 require "extras_de_cont/rules/revolut"
 require "extras_de_cont/rules/unicredit"
 
@@ -11,6 +12,7 @@ module ExtrasDeCont
   # Map of supported banks (symbol → rule class)
   BANK_RULES = {
     brd: Rules::Brd,
+    ing: Rules::Ing,
     revolut: Rules::Revolut,
     unicredit: Rules::UniCredit
   }.freeze

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: extras_de_cont
 version: !ruby/object:Gem::Version
-  version: 1.2.0
+  version: 1.3.0
 platform: ruby
 authors:
 - Denis Nutiu
@@ -40,6 +40,7 @@ files:
 - lib/extras_de_cont/parser.rb
 - lib/extras_de_cont/rules/base.rb
 - lib/extras_de_cont/rules/brd.rb
+- lib/extras_de_cont/rules/ing.rb
 - lib/extras_de_cont/rules/revolut.rb
 - lib/extras_de_cont/rules/unicredit.rb
 - lib/extras_de_cont/transaction.rb