banktools-se 2.1.0 → 2.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 74491115cff66b4bf05d2d842ebc5e244580c253
-  data.tar.gz: 51f456d3da82641e22dc0c236c84b0fdf9b1b66e
+  metadata.gz: 1c5c3cd7492d627e0bd743ad56328b7d97daad3a
+  data.tar.gz: e83b11dbab0e65165fcbcf2ed490770624023ea9
 SHA512:
-  metadata.gz: b28bb0e72521ab8641fa2545002a4a86e6add6856d526c6a98f681002265aa3a4b8f531959c330beba6552fc6138fbd45f598f8fb9023092d9474b91ce491669
-  data.tar.gz: b61fb6be0c11bf1c784dec49d58a4ed199a9adfc244fe119aa2152e2e60c198fe70ca3d74a057bc6ac3985eff4fab3c5aac0a40b6681d8d2715a0a8a2ec1624a
+  metadata.gz: 173c075e3289cd074802e1b0530b2cfa4dcf11a087bec1bc6f1582252ab44c67d785ebf239e77e14930ec3db646d820098bded4fed99646e136b30fccfed7061
+  data.tar.gz: 67767a7305bb6425c53200c8e664e9635ce2162c319f70772bd3747a00a104589c874d78429a25769802468a1097f910ca77c533ee388b40cac2ce48cb249173

data/README.markdown

@@ -45,8 +45,10 @@ Inspired by [iulianu/iban-tools](https://github.com/iulianu/iban-tools). Please
 
     # This feature is intended to try to find all OCR numbers in a noisy bank statement string.
     # By design it may find too many numbers (e.g. valid substrings of other numbers), so you should check results against actual outstanding invoices.
-    BankTools::SE::OCR.find_all_in_string("OCR1230 and ref4564 and 789")  # => [ "1230", "4564" ]
+    # By default, it excludes OCRs shorter than 4 digits, but this limit can be specified per below.
+    BankTools::SE::OCR.find_all_in_string("OCR1230 and ref4564 and 7890")  # => [ "1230", "4564" ]
     BankTools::SE::OCR.find_all_in_string("1230 and 123067", length_digit: true, pad: "0")  # => [ "123067" ]
+    BankTools::SE::OCR.find_all_in_string("00 and 18", min_length: 2)  # => [ "00", "18" ]
 
     # Plusgiro
 

data/lib/banktools-se/ocr.rb

@@ -5,6 +5,7 @@ module BankTools
     class OCR
       class InvalidOCR < StandardError; end
       class OverlongOCR < InvalidOCR; end
+      class TooShortOCR < InvalidOCR; end
       class BadPadding < InvalidOCR; end
       class BadLengthDigit < InvalidOCR; end
       class BadChecksum < InvalidOCR; end
@@ -28,23 +29,24 @@ module BankTools
 
         length = number_with_ocr.length
         if length > MAX_LENGTH
-          raise OverlongOCR, "Bankgiro OCR must be #{MIN_LENGTH} - #{MAX_LENGTH} characters (this one would be #{length} characters)"
+          raise OverlongOCR, "OCR must be #{MIN_LENGTH} - #{MAX_LENGTH} characters (this one would be #{length} characters)"
         end
 
         number_with_ocr
       end
 
-      def self.to_number(number, length_digit: false, pad: "")
-        number = number.to_s
+      def self.to_number(ocr, length_digit: false, pad: "")
+        ocr = ocr.to_s
         should_have_length_digit = length_digit
         strip_padding = pad.to_s
 
-        raise MustBeNumeric unless number.match(/\A\d+\z/)
-        raise BadChecksum unless Utils.valid_luhn?(number)
+        raise MustBeNumeric unless ocr.match(/\A\d+\z/)
+        raise BadChecksum unless Utils.valid_luhn?(ocr)
+        raise TooShortOCR if ocr.length < MIN_LENGTH
 
         if should_have_length_digit
-          length_digit = number[-2]
-          last_digit_of_actual_length = number.length.to_s[-1]
+          length_digit = ocr[-2]
+          last_digit_of_actual_length = ocr.length.to_s[-1]
           raise BadLengthDigit if length_digit != last_digit_of_actual_length
         end
 
@@ -54,16 +56,23 @@ module BankTools
         if strip_padding.length > 0
           expected_padding_end = -digits_to_chop - 1
           expected_padding_start = expected_padding_end - strip_padding.length + 1
-          raise BadPadding if number[expected_padding_start..expected_padding_end] != strip_padding
+          raise BadPadding if ocr[expected_padding_start..expected_padding_end] != strip_padding
         end
 
         digits_to_chop += strip_padding.length
 
-        number[0...-digits_to_chop]
+        ocr[0...-digits_to_chop]
       end
 
-      def self.find_all_in_string(string, length_digit: false, pad: "")
-        string.scan(/\d+/).select { |candidate|
+      def self.find_all_in_string(string, length_digit: false, pad: "", min_length: 4)
+        expanded_string = string + " " + string.gsub("\n", "") + " " + string.gsub(";", "")
+
+        numbers = expanded_string.scan(/\d+/)
+
+        expanded_numbers = with_numbers_found_by_removing_prefix_and_postfix(numbers).
+          reject { |n| n.length < min_length }
+
+        expanded_numbers.select { |candidate|
           begin
             to_number(candidate, length_digit: length_digit, pad: pad)
             true
@@ -72,6 +81,20 @@ module BankTools
           end
         }.uniq
       end
+
+      private
+
+      private_class_method \
+        def self.with_numbers_found_by_removing_prefix_and_postfix(numbers)
+          numbers + numbers.flat_map { |number|
+            0.upto(number.size).flat_map { |i|
+              [
+                number[0...i],
+                number[i...number.size],
+              ]
+            }
+          }
+      end
     end
   end
 end

data/lib/banktools-se/version.rb

@@ -1,5 +1,5 @@
 module BankTools
   module SE
-    VERSION = "2.1.0"
+    VERSION = "2.3.0"
   end
 end

data/spec/ocr_spec.rb

@@ -46,6 +46,11 @@ describe BankTools::SE::OCR do
       BankTools::SE::OCR.to_number("1234567890037", length_digit: true, pad: "0").should eq "1234567890"
     end
 
+    it "raises if the given number is too short to be a valid OCR" do
+      expect { BankTools::SE::OCR.to_number("0") }.to raise_error(BankTools::SE::OCR::TooShortOCR)
+      expect { BankTools::SE::OCR.to_number("00") }.not_to raise_error
+    end
+
     it "raises if checksum is wrong" do
       expect { BankTools::SE::OCR.to_number("1231") }.to raise_error(BankTools::SE::OCR::BadChecksum)
     end
@@ -72,7 +77,7 @@ describe BankTools::SE::OCR do
       expect(BankTools::SE::OCR.find_all_in_string("1230 1234 4564")).to eq [ "1230", "4564" ]
     end
 
-    it "requires OCRs to comply with length_digit and pad options" do
+    it "requires OCRs to comply with the specified length_digit and pad options" do
       string = "1230 4564 123067 456061"
       expect(BankTools::SE::OCR.find_all_in_string(string)).to eq [ "1230", "4564", "123067", "456061" ]
       expect(BankTools::SE::OCR.find_all_in_string(string, length_digit: true, pad: "0")).to eq [ "123067", "456061" ]
@@ -82,6 +87,39 @@ describe BankTools::SE::OCR do
       expect(BankTools::SE::OCR.find_all_in_string("x1230x")).to eq [ "1230" ]
     end
 
+    it "handles OCR numbers both separated and split by newlines" do
+      expect(BankTools::SE::OCR.find_all_in_string("1230\n4564")).to include "1230", "4564", "12304564"
+      expect(BankTools::SE::OCR.find_all_in_string("45\n64")).to eq [ "4564" ]
+    end
+
+    it "handles OCR numbers both separated and split by semicolons" do
+      expect(BankTools::SE::OCR.find_all_in_string("1230;4564")).to include "1230", "4564", "12304564"
+      expect(BankTools::SE::OCR.find_all_in_string("45;64")).to eq [ "4564" ]
+    end
+
+    it "handles numbers smushed together" do
+      # "Ref 1: 1230" with characters gone missing.
+      expect(BankTools::SE::OCR.find_all_in_string("REF 11230")).to include "1230"
+
+      # Two OCRs without separation.
+      expect(BankTools::SE::OCR.find_all_in_string("12304564")).to include "1230", "4564"
+
+      # Amount smushed into OCR.
+      expect(BankTools::SE::OCR.find_all_in_string("EHRENKRONAAUFTR: EUR 17,183188720001 PAYMENT")).to include "3188720001"
+
+      # OCR smushed into item ID.
+      string = "Referenznummer 3201675000187604. HISTORISTISCHER SALONTISCH."
+      expect(BankTools::SE::OCR.find_all_in_string(string)).to include "3201675000"
+    end
+
+    it "lets you configure the accepted OCR min_length" do
+      expect(BankTools::SE::OCR.find_all_in_string("12304564")).to eq [ "12304564", "04564", "1230", "4564" ]
+      expect(BankTools::SE::OCR.find_all_in_string("12304564", min_length: 6)).to eq [ "12304564" ]
+
+      expect(BankTools::SE::OCR.find_all_in_string("1234")).to eq []
+      expect(BankTools::SE::OCR.find_all_in_string("1234", min_length: 2)).to eq [ "34" ]
+    end
+
     it "excludes duplicates" do
       expect(BankTools::SE::OCR.find_all_in_string("1230 1230 4564")).to eq [ "1230", "4564" ]
     end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: banktools-se
 version: !ruby/object:Gem::Version
-  version: 2.1.0
+  version: 2.3.0
 platform: ruby
 authors:
 - Henrik Nyh