ng-bank-parser 0.1.2

data/lib/ng-bank-parser/parsers/firstbank-pdf-parser.rb

@@ -0,0 +1,81 @@
+require 'pdf-reader'
+require_relative 'firstbank-pdf-parser/helpers'
+
+
+module NgBankParser
+	class FirstbankPdf
+		extend FirstbankPdfHelpers
+
+		@@transactions = []
+		
+		def self.parse(path, password = nil)
+			accepted_formats = [".pdf"];
+			unless accepted_formats.include? File.extname(path)
+				return error_message 'Invalid file format'
+			end
+
+			if has_encryption? path
+				if password
+					unless get_unlocked_pdf? path, password
+						return error_message 'Password supplied for decryption is invalid.'
+					end
+				else
+					return error_message 'PDF File is encrypted and no password was supplied'
+				end
+			end
+
+			unless contains_account_data?
+				return error_message 'Unable to read account details'
+			end
+
+			if contains_transactions_table?
+				extract_transactions(clean(get_raw_transactions))
+				data = {}
+				data[:bank_name] = 'First Bank'
+				data[:account_number] = get_account_number
+				data[:account_name] = get_account_name
+				data[:from_date] = get_from_date
+				data[:to_date] = get_to_date
+				data[:transactions] = @@transactions
+				send_response data
+			else
+				return error_message 'Could not find any transactions'
+			end
+
+		end
+
+
+		private
+
+		def self.extract_transactions(jagged_array = [[]])
+			jagged_array.each do |array|
+				if is_transaction_row? array
+					transaction = {}
+					transaction[:ref] = ''
+					transaction[:date] = Date.strptime(array[0], '%d-%b-%y')
+					transaction[:remarks] = array[1]
+					transaction[:amount] = array[3].delete(',').to_f
+					transaction[:balance] = array[4].delete(',').to_f
+					if transaction[:balance].to_i > get_last_balance
+						transaction[:type] = 'credit'
+						update_last_balance transaction[:balance]
+					else
+						transaction[:type] = 'debit'
+						update_last_balance transaction[:balance]
+					end
+					@@transactions << transaction
+				else
+					@@transactions.last[:remarks] += array[0] if @@transactions
+				end
+			end
+		end
+
+
+		def self.clean(jagged_array = [[]])
+			jagged_array.reject! do |array|
+				is_row_invalid? array
+			end
+		end
+
+	end
+end

data/lib/ng-bank-parser/parsers/firstbank-pdf-parser/helpers.rb

@@ -0,0 +1,172 @@
+require 'pdf-reader'
+require 'date'
+require 'open-uri'
+require_relative 'statement_utils'
+require_relative '../../pdf-unlocker.rb'
+
+module NgBankParser
+	module FirstbankPdfHelpers
+		include StatementUtils
+		
+		@@pdf_reader = nil
+		@@raw_transactions = [[]]
+
+		def has_encryption? path
+			begin
+				@@pdf_reader = PDF::Reader.new(path)
+				false
+			rescue PDF::Reader::EncryptedPDFError
+				true
+			end
+		end
+
+
+		def get_unlocked_pdf? path, password
+			response = PDFUnlocker.new(File.new(path), password).unlocked_pdf
+			return false unless response
+			if response.include? 'Unlock Failed'
+				return false
+			else
+				pseudo_file = StringIO.new
+				pseudo_file.write(response)
+				@@pdf_reader = PDF::Reader.new(pseudo_file)
+				return true
+			end
+		end
+
+
+		def get_raw_transactions
+			@@raw_transactions
+		end
+
+
+		def get_transaction_data
+			pages = get_pages @@pdf_reader
+			pages.each do |page|
+				page_text = get_page_text page
+				index = get_transaction_table_index page_text
+				unless index == -1
+					add_to_transactions page_text[index..-1]
+				end
+			end
+		end
+
+
+		def get_account_data
+			lines = get_first_page_text @@pdf_reader
+	  		lines.each do |line|
+				if line[0].start_with? 'Account No:'
+			 		set_account_number line
+			  		set_last_balance line
+				elsif line[0].start_with? 'Account Name:'
+			  		set_account_name line
+				elsif line[0].start_with? 'For the Period of:'
+			  		set_statement_period line
+				end
+			end
+		end
+
+
+		def get_account_number
+			@@account_number
+		end
+
+
+		def get_account_name
+			@@account_name
+		end
+
+
+		def get_last_balance
+			@@last_balance.to_i
+		end
+
+		def get_from_date
+			Date.strptime(@@from_date.strip,"%d-%b-%Y")
+		end
+
+
+		def get_to_date
+			Date.strptime(@@to_date.strip,"%d-%b-%Y")
+		end
+
+
+		def contains_transactions_table?
+			get_transaction_data
+			@@raw_transactions
+		end
+
+
+		def contains_account_data?
+			get_account_data
+			@@account_name && @@account_number && @@last_balance && @@statement_period
+		end
+
+
+		def set_account_number line
+			@@account_number = line[1] unless line[1].blank?
+		end
+
+
+		def set_account_name line
+			@@account_name = line[1] unless line[1].blank?
+		end
+
+
+		def set_last_balance line
+			@@last_balance = line[2] unless line[1].blank?
+		end
+
+
+		def update_last_balance balance
+			@@last_balance = balance
+		end
+
+
+		def set_statement_period line
+			unless line[1].blank?
+				@@statement_period = line[1].split('to')
+				@@from_date, @@to_date = @@statement_period
+			end
+		end
+
+
+		def is_transaction_row? row
+			 row[0] =~ /(\d\d-[a-zA-Z]{3}-\d\d)/
+		end
+
+
+		def is_row_invalid? row
+			row.length == 0 ||
+			row[0].start_with?('END OF STATEMENT') ||
+			row[0] == ('Balance B/F') ||
+			row[0].start_with?('Page')
+		end
+
+
+		def error_message msg
+			return {
+					status: 0,
+					message: msg
+				}
+		end
+
+
+		def send_response data
+			return {
+					status: 1,
+					data: data
+				}
+		end
+
+
+		private
+
+		def add_to_transactions lines
+			lines.each do |line|
+				@@raw_transactions << line.strip.split(/\s\s+/)
+			end
+		end
+
+	end
+end

data/lib/ng-bank-parser/parsers/firstbank-pdf-parser/statement_utils.rb

@@ -0,0 +1,51 @@
+module NgBankParser
+  module StatementUtils
+
+    def get_first_page_text reader
+      lines = reader.pages.first.text.remove_empty_lines.lines #lines without the spaces
+      lines.map{ |line| split_on_2_or_more_spaces(line) }
+    end
+
+
+    def get_all_text reader
+      all_lines = []
+      reader.pages.each do |page|
+        lines_of_page = page.text.remove_empty_lines.lines
+        all_lines += lines_of_page
+      end
+      all_lines
+    end
+
+
+    def get_page_text page_text
+      page_text.text.remove_empty_lines.lines
+    end
+
+
+    def get_pages reader
+      reader.pages
+    end
+
+
+    def get_transaction_table_index lines
+      lines_in_file = lines.map{ |line| split_on_2_or_more_spaces(line) }
+      lines_in_file.each_with_index do |line, index|
+        if line[0] == 'TransDate'
+            return index + 1
+        end
+      end
+      return -1 #no transactions found on page
+    end
+
+
+    private
+
+    def split_on_2_or_more_spaces str
+      str.strip.split(/\s\s+/)
+    end
+
+    def remove_empty_lines
+      self.gsub /^$\n/, ''
+    end
+  end
+end

data/lib/ng-bank-parser/parsers/gtb-excel-parser.rb

@@ -0,0 +1,66 @@
+require_relative 'gtb-excel-parser/helpers'
+
+module NgBankParser
+	class GtbExcel
+		extend GtbExcelHelpers
+
+		class << self
+			def parse(path, password = nil)
+				accepted_formats = [".xls",".xlsx"];
+				unless accepted_formats.include? File.extname(path)
+					return error_message("Invalid file format")
+				end
+
+				file = read_file_contents(path)
+				if (file[:type] == "html")
+					html_parse(file[:contents])
+				elsif (file[:type] == "xls")
+					xls_parse(file[:contents])
+				else
+					return error_message("Could not parse this file")
+				end
+			end
+
+			def html_parse(file)
+				data = {}
+
+		        data[:transactions] = get_transactions_from_html(file)
+		        data[:account_number] = file.css("#lblAcctNo").text().return_first_number
+		        data[:from_date] = file.css("#lblPeriod1").text().convert_string_to_date
+		        data[:to_date] = file.css("#lblPeriod2").text().convert_string_to_date
+		        data[:account_name] = file.css("#lblAcctName").text()
+		        data[:bank_name] = "Guaranty Trust Bank"
+
+		       	send_response(data)
+			end
+
+			def xls_parse(file)
+				data = {}
+
+				data[:transactions] = get_transactions_from_excel(file)
+		        data[:account_number] = file.row(10)[0].return_first_number
+		        date_strings = file.row(14)[0].get_date_strings
+		        data[:from_date] = date_strings[0].convert_string_to_date
+		        data[:to_date] = date_strings[1].convert_string_to_date
+		        data[:account_name] = file.row(5)[0]
+		        data[:bank_name] = "Guaranty Trust Bank"
+
+				send_response(data)
+			end
+
+			def error_message(text)
+				return { 
+					status: 0, 
+					message: text
+				}
+			end
+
+			def send_response(data)
+				return {
+					status: 1,
+					data: data
+				}
+			end
+		end
+	end
+end

data/lib/ng-bank-parser/parsers/gtb-excel-parser/helpers.rb

@@ -0,0 +1,121 @@
+require 'nokogiri'
+require 'roo'
+require 'date'
+
+class Hash
+  def without(*keys)
+    cpy = self.dup
+    keys.each { |key| cpy.delete(key) }
+    cpy
+  end
+end
+
+class String
+	def convert_to_number
+		self.to_s.scan(/\b-?[\d.]+/).join.to_f
+	end
+	def return_first_number
+		self.scan(/\d+/)[0]
+	end
+	def convert_string_to_date
+		date_string = self.scan(/.....\d*..\d{4}/)[0]
+		Date.strptime(date_string,"%d/%b/%Y")
+	end
+	def get_date_strings
+		self.scan(/.....\d*..\d{4}/)
+	end
+end
+
+
+module GtbExcelHelpers
+	def read_file_contents(path)
+		if (File.extname(path) == '.xls')
+			contents = Nokogiri::HTML(open(path))
+			if has_transactions_table(contents)
+				{type: "html", contents: contents}
+			else 
+				{ type: "unknown" }
+			end
+		else
+			contents = Roo::Excelx.new(path)
+			{type: "xls", contents: contents}
+		end
+	end
+
+	def has_transactions_table(contents)
+		contents.css("#dgtrans")		
+	end
+
+	def get_transactions_from_html(file)
+		extract_transaction_rows_from_html(file)
+		convert_html_rows_to_transactions
+	end
+
+	def extract_transaction_rows_from_html(file)
+		@rows = file.xpath('//table[@id="dgtrans"]/tr') # Get transaction table
+		@rows.shift # Remove header row
+	end
+
+	def convert_html_rows_to_transactions
+		transactions = @rows.collect do |row|
+			transaction = {}
+			[
+				[:date, 'td[1]/text()'], #date
+				[:ref, 'td[2]/text()'], #ref
+				[:debit, 'td[4]/text()'], #debit
+				[:credit, 'td[5]/text()'], #credit
+				[:balance, 'td[6]/text()'], #balance
+				[:remarks, 'td[7]/text()'], #remarks
+			].each do |column_name, xpath|
+				integer_columns = [:debit, :credit, :balance]
+				column_value = row.at_xpath(xpath).text()
+
+				# If it's an integer field convert to number
+				if integer_columns.include?(column_name)
+					column_value = column_value.convert_to_number;
+				end
+				
+				transaction[column_name] = column_value				
+				transaction[:date] =  Date.strptime(column_value,"%d-%b-%Y") if column_name == :date
+			end
+
+			filter_debit_or_credit(transaction)
+        end
+        filter_invalid(transactions)
+	end
+
+	def filter_invalid(transactions)
+		transactions.select do |row|
+			is_valid_transaction(row)
+		end
+	end
+
+	def filter_debit_or_credit(transaction)
+		if (transaction[:debit].nil? || transaction[:debit] == 0)
+			transaction[:type] = "credit"
+			transaction[:amount] = transaction[:credit]
+		else
+			transaction[:type] = "debit"
+			transaction[:amount] = transaction[:debit]
+		end
+
+		# Remove credit and debit keys
+    	transaction.without(:debit, :credit)
+	end
+
+	def get_transactions_from_excel(file)
+		transactions = []
+        file.each(date: 'Trans Date', ref: 'Reference', debit: 'Debit', credit: 'Credit', balance: 'Balance', remarks: 'Remarks') do |row|
+  			if is_valid_transaction(row)
+  				transaction = filter_debit_or_credit(row)
+  				transaction[:ref] = transaction[:ref].to_s
+  				transactions << transaction
+  			end
+		end
+		transactions
+	end
+
+	def is_valid_transaction(row)
+		return row[:date].is_a?(Date) && (row[:amount] || row[:credit] || row[:debit]).is_a?(Float) && row[:balance].is_a?(Float) && !row[:remarks].empty?
+	end
+end