extras_de_cont 1.0.1 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 69bb4ccabcbeca26046ace09db26a3b6e207e5ccb269f0572373472ae6ce8cd4
-  data.tar.gz: c249a2485737903eb4cc42c26ddcd41f17d4aa653d23546e0ede16e93c6ed8ee
+  metadata.gz: '029f7b904cb76faa331a686ec02f81a0fad9ffce39e0bea1c05e0ed95a6bb34c'
+  data.tar.gz: c98d9dc06a00db1cface2c7ea7b58f563502c2fa58da4481d2a928e99fe08766
 SHA512:
-  metadata.gz: b347db52d2b09495de4f60edd64897440b548f8a010b3c6d732e17cfa2de684e2dbdd33c882766dbc1d6ce60419dbd7b99bc7e0213e8e913b7dec922c3b2cbfd
-  data.tar.gz: 7945a8150853d19eab987edbbe1e5d99b6682ec35068faabb6083bc43fcfd62643c0b21a5bb970641cdc21011be3a2aace847acf4e022765aca899dd63ff4908
+  metadata.gz: cd20be09910a3ca8ad2a13bc07ce619d76950e080e3bb927a2f3dee0f24441f173ad842437dfc9f8196b24579d5976519c5558031aa8cd33e93cd5263e229705
+  data.tar.gz: 27c5cda6efa0043af432ac596c7d2ff736dd3f26c520f3cd2ea79a45506eb41a48d5c9ca236432d1f9999e73f9d8c11dabc201e4ee23cab82f08e9bc7a351b6b

data/extras_de_cont.gemspec

@@ -2,51 +2,14 @@
 
 Gem::Specification.new do |s|
   s.name = "extras_de_cont"
-  s.version = "1.0.1"
+  s.version = "1.1.0"
   s.licenses = ["GPLv3"]
   s.summary = "A simple library which helps you extract transactions from a PDF bank statement."
   s.description = <<~TEXT
     A simple library which helps you extract transactions from a PDF bank statement.
     Fine tuned for Romanian bank statements.
 
-    # ExtrasDeCont
-
-    A ruby gem for extracting bank statements from PDFs.
-
-    ## Simple usage
-
-    Create a PDF parser and print the extracted text:
-
-    ```ruby
-    require "bundler/setup"
-    require "extras_de_cont"
-
-    parser = ExtrasDeCont::Parser.new("/home/dnutiu/Documents/tranzactii_revolut.pdf")
-    puts parser.text
-    ```
-
-    Or, extract all the transactions from a Revolut Bank statement PDF:
-
-    ```ruby
-    transactions = ExtrasDeCont.parse(file, bank: :revolut)
-
-    transactions.each do |t|
-      puts "\#{t.date}, \#{t.description}, \#{t.amount}, \#{t.currency}"
-    end
-    ```
-
-    Or use the included entrypoint:
-
-    ```bash
-    bundle exec ruby -Ilib bin/main /home/dnutiu/Documents/tranzactii_revolut.pdf
-    ```
-
-    Run the Revolut parser test with:
-
-    ```bash
-    ruby -Ilib:test test/revolut_rule_test.rb
-    ```
-
+    Repository: https://github.com/dnutiu/extras-de-cont
   TEXT
   s.authors = ["Denis Nutiu"]
   s.email = "dnutiu@nuculabs.dev"

data/lib/extras_de_cont/rules/revolut.rb

@@ -11,24 +11,45 @@ module ExtrasDeCont
         "Pending from ",
         "Account transactions from ",
         "Reverted from ",
-        "Deposit transactions from "
+        "Deposit transactions from ",
+        "Transactions from "
       ].freeze
 
       DOCUMENT_NOISE_HEADERS = [
+        "Account statement",
         "Balance summary",
         "The balance on your statement might differ",
+        "There were no transactions during this period",
+        "Transaction types",
+        "Your funds are held and protected by a licensed bank",
         "Report lost or stolen card",
         "+",
         "Get help directly in app",
+        "Get help directly In app",
         "Scan the QR code",
         "RON Statement",
+        " Statement",
         "Generated on the ",
         "Revolut Bank UAB",
         "© "
       ].freeze
 
-      DATE_PREFIX = /\A(?<date>[A-Z][a-z]{2} \d{1,2}, \d{4})\b/
-      AMOUNT = /-?\d[\d,]*\.\d{2} [A-Z]{3}/
+      CURRENCY_SYMBOLS = {
+        "$" => "USD",
+        "€" => "EUR",
+        "£" => "GBP",
+        "zł" => "PLN",
+        "Kč" => "CZK",
+        "Ft" => "HUF",
+        "лв" => "BGN",
+        "₺" => "TRY",
+        "₴" => "UAH"
+      }.freeze
+      DATE_FORMATS = ["%b %e, %Y", "%e %b %Y"].freeze
+      DATE_PREFIX = /\A(?<date>(?:[A-Z][a-z]{2} \d{1,2}, \d{4}|\d{1,2} [A-Z][a-z]{2} \d{4}))\b/
+      NUMBER = /-?(?:\d{1,3}(?:[ ,]\d{3})+|\d+)\.\d{2}/
+      CURRENCY_SYMBOL = Regexp.union(CURRENCY_SYMBOLS.keys.sort_by { |symbol| -symbol.length })
+      AMOUNT = /(?:#{NUMBER} [A-Z]{3}|#{CURRENCY_SYMBOL}#{NUMBER}|#{NUMBER} ?#{CURRENCY_SYMBOL})/
 
       def parse(text)
         transactions = []
@@ -105,16 +126,41 @@ module ExtrasDeCont
           parse_date(match[:date]),
           description,
           amount,
-          amount_string.split.last
+          parse_currency(amount_string)
         )
       end
 
       def parse_date(value)
-        Date.strptime(value, "%b %e, %Y")
+        DATE_FORMATS.each do |format|
+          return Date.strptime(value, format)
+        rescue Date::Error
+          next
+        end
       end
 
       def parse_amount(value)
-        value.split.first.delete(",").to_f
+        numeric_value(value).delete(", ").to_f
+      end
+
+      def parse_currency(value)
+        symbol = currency_symbol(value)
+        return CURRENCY_SYMBOLS.fetch(symbol) if symbol
+
+        value.split.last
+      end
+
+      def numeric_value(value)
+        symbol = currency_symbol(value)
+        return value.delete_prefix(symbol) if symbol && value.start_with?(symbol)
+        return value.delete_suffix(symbol).strip if symbol
+
+        value.sub(/\s+[A-Z]{3}\z/, "")
+      end
+
+      def currency_symbol(value)
+        CURRENCY_SYMBOLS.keys.find do |symbol|
+          value.start_with?(symbol) || value.end_with?(symbol)
+        end
       end
 
       def section_header?(line)
@@ -147,11 +193,12 @@ module ExtrasDeCont
         amount_matches = row.to_enum(:scan, AMOUNT).map { Regexp.last_match }
         return if amount_matches.empty?
 
-        transaction_match = if table.fetch(:has_balance)
-                              amount_matches[-2] if amount_matches.length > 1
-                            else
-                              amount_matches[-1]
-                            end
+        transaction_match =
+          if table.fetch(:has_balance)
+            amount_matches[-2] if amount_matches.length > 1
+          else
+            amount_matches[-1]
+          end
         return if transaction_match.nil?
 
         description = row[date_match.end(0)...transaction_match.begin(0)].to_s.strip

data/lib/extras_de_cont/rules/unicredit.rb

@@ -1,8 +1,205 @@
 # frozen_string_literal: true
 
+require "date"
+require "extras_de_cont/transaction"
+
 module ExtrasDeCont
   module Rules
     class UniCredit < 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+Descriere\s+Debit\s+Credit\s+Sold/
+
+      SECTION_HEADERS = [
+        "TRANZACȚII",
+        "SUMAR CONT",
+        "EXTRAS DE CONT"
+      ].freeze
+
+      NOISE_PATTERNS = [
+        /\AUniCredit Bank S\.A\./,
+        /\ABulevardul/,
+        /\ASector \d/,
+        /\ATel:/,
+        /\AEmail:/,
+        /\Aunicredit\.ro/,
+        /\ACapital social:/,
+        /\APrezentul extras/,
+        /\AFondurile disponibile/,
+        /\APentru mai multe/,
+        /\ANUME CLIENT:/,
+        /\AADRESA:/,
+        /\ASUCURSALA:/,
+        /\ADATA EXTRAS CONT/,
+        /\APERIOADA/,
+        /\ATIP CONT:/,
+        /\AIBAN:/,
+        /\AMONEDA:/,
+        /\AOperator de date/,
+        /\ASold inițial/,
+        /\ASold final/,
+        /\AOperator de date cu/
+      ].freeze
+
+      NEW_TRANSACTION_MARKERS = [
+        /\A\+CMS CLT-/,
+        /\A\+GPP/,
+        /\APlata electronica/,
+        /\APlata Instant/,
+        /\AIncasare Instant/,
+        /\ATransfer electronic/
+      ].freeze
+
+      AMOUNT_PATTERN = /\d{1,3}(?:[.,]\d{3})*\.\d{2}/
+      CURRENCY_FROM_HEADER = /Sold\(([A-Z]{3})\)/
+
+      def parse(text)
+        transactions = []
+        current_currency = nil
+        current_table = nil
+        above_lines = []
+        below_lines = []
+        date_line = nil
+
+        each_normalized_line(text) do |line|
+          if (m = line.match(CURRENCY_FROM_HEADER))
+            current_currency = m[1]
+          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) || section_header?(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
+            if new_transaction_marker?(line)
+              try_flush(date_line, above_lines, below_lines, current_table, current_currency, transactions)
+              date_line, below_lines = nil, []
+              above_lines = [line]
+            else
+              below_lines << line
+            end
+          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 section_header?(line)
+        SECTION_HEADERS.any? { |header| line == header }
+      end
+
+      def new_transaction_marker?(line)
+        NEW_TRANSACTION_MARKERS.any? { |pattern| line.match?(pattern) }
+      end
+
+      def extract_column_positions(line)
+        {
+          debit: line.index("Debit"),
+          credit: line.index("Credit"),
+          sold: line.index("Sold")
+        }
+      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.size < 2
+
+        transaction_amount_match = amounts[-2]
+        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 = amount_string.delete(", ").to_f
+        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 || extract_currency_from_header(date_line)
+        )
+      end
+
+      def parse_date(day, month_name, year)
+        month = ROMANIAN_MONTHS[month_name.downcase]
+        Date.new(year, month, day)
+      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
+
+      def extract_currency_from_header(date_line)
+        m = date_line.match(CURRENCY_FROM_HEADER)
+        m ? m[1] : nil
+      end
     end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: extras_de_cont
 version: !ruby/object:Gem::Version
-  version: 1.0.1
+  version: 1.1.0
 platform: ruby
 authors:
 - Denis Nutiu
@@ -23,48 +23,11 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '2.15'
-description: |+
+description: |
   A simple library which helps you extract transactions from a PDF bank statement.
   Fine tuned for Romanian bank statements.
 
-  # ExtrasDeCont
-
-  A ruby gem for extracting bank statements from PDFs.
-
-  ## Simple usage
-
-  Create a PDF parser and print the extracted text:
-
-  ```ruby
-  require "bundler/setup"
-  require "extras_de_cont"
-
-  parser = ExtrasDeCont::Parser.new("/home/dnutiu/Documents/tranzactii_revolut.pdf")
-  puts parser.text
-  ```
-
-  Or, extract all the transactions from a Revolut Bank statement PDF:
-
-  ```ruby
-  transactions = ExtrasDeCont.parse(file, bank: :revolut)
-
-  transactions.each do |t|
-    puts "#{t.date}, #{t.description}, #{t.amount}, #{t.currency}"
-  end
-  ```
-
-  Or use the included entrypoint:
-
-  ```bash
-  bundle exec ruby -Ilib bin/main /home/dnutiu/Documents/tranzactii_revolut.pdf
-  ```
-
-  Run the Revolut parser test with:
-
-  ```bash
-  ruby -Ilib:test test/revolut_rule_test.rb
-  ```
-
+  Repository: https://github.com/dnutiu/extras-de-cont
 email: dnutiu@nuculabs.dev
 executables: []
 extensions: []
@@ -103,4 +66,3 @@ rubygems_version: 4.0.10
 specification_version: 4
 summary: A simple library which helps you extract transactions from a PDF bank statement.
 test_files: []
-...