chore: better QuoteReader NaiveText

lib/quote_reader/global.rb

@@ -5,6 +5,8 @@ module QuoteReader
   class Global
     attr_reader :filepath
 
+    VERSION = "0.0.1"
+
     def initialize(filepath)
       @filepath = filepath
     end

lib/quote_reader/naive_text.rb

@@ -33,56 +33,73 @@ def read_attributes
     # rubocop:enable Metrics/MethodLength
     # rubocop:enable Metrics/AbcSize
 
-    def self.find_adresse_pro(text)
-      text[/Adresse Pro\s*:\s*(\w+)/i, 1]
+    NUMBER_REFERENCE_REGEX = /n?[.°]/i
+    BETWEEN_LABEL_VALUE_REGEX = /\s+(?:#{NUMBER_REFERENCE_REGEX})?\s*(?::\s*)?/i
+    FRENCH_CHARACTER_REGEX = /[\wÀ-ÖØ-öø-ÿ]/i
+    PHONE_REGEX = /(?:\(?\+?33\)?)? ?(?:[\s.]*\d\d){5}/i # TODO: find better
+
+    def self.find_adresse(text)
+      text[/Adresse\s*:\s*(#{FRENCH_CHARACTER_REGEX}+)/i, 1]
     end
 
-    def self.find_raison_sociale(text)
-      text[/Raison sociale\s*:\s*(\w+)/i, 1]
+    def self.find_adresse_chantier(text)
+      text[/Adresse chantier\s*:\s*(#{FRENCH_CHARACTER_REGEX}+)/i, 1]
     end
 
-    def self.find_forme_juridique(text)
-      text[/Forme juridique\s*:\s*(\w+)/i, 1]
+    def self.find_adresse_pro(text)
+      text[/Adresse Pro\s*:\s*(#{FRENCH_CHARACTER_REGEX}+)/i, 1]
     end
 
-    def self.find_numero_tva(text)
-      text[/TVA\s*:\s*(\w+)/i, 1]
+    def self.find_assurance(text)
+      text[/Assurance(?:\s+décennale)?#{BETWEEN_LABEL_VALUE_REGEX}((?:[#{FRENCH_CHARACTER_REGEX}:]+\s+)+(?:#{NUMBER_REFERENCE_REGEX}\s*)?(?:contrat\s+#{FRENCH_CHARACTER_REGEX}*\s*\d+)?)/i, 1] # rubocop:disable Layout/LineLength
     end
 
     def self.find_capital(text)
-      text[/Capital\s*:\s*(\w+)/i, 1]
+      text[/(?:Capitale?|capilâide)(?:\s+de)?#{BETWEEN_LABEL_VALUE_REGEX}(\d+(?: \d{3})*)\s*€/i, 1]
+    end
+
+    def self.find_forme_juridique(text)
+      text[/Forme juridique\s*:\s*(SAS|SARL|EURL|#{FRENCH_CHARACTER_REGEX}+) ?/i, 1]
+    end
+
+    def self.find_iban(text)
+      text[/(?:IBAN|RIB)#{BETWEEN_LABEL_VALUE_REGEX}(FR\d{2}\s?(?:\d{4}\s?){2,5}#{FRENCH_CHARACTER_REGEX}?\d{2})/i, 1]
     end
 
     def self.find_mention_devis(text)
       text[/devis/i] if text
     end
 
-    def self.find_adresse(text)
-      text[/Adresse\s*:\s*(\w+)/i, 1]
+    def self.find_nom(text)
+      text[/Nom\s*:\s*(#{FRENCH_CHARACTER_REGEX}+)/i, 1]
     end
 
-    def self.find_adresse_chantier(text)
-      text[/Adresse chantier\s*:\s*(\w+)/i, 1]
+    def self.find_numero_devis(text)
+      text[/DEVIS\s+N.?\s*(#{FRENCH_CHARACTER_REGEX}*\d{4,})/i, 1]
     end
 
-    def self.find_nom(text)
-      text[/Nom\s*:\s*(\w+)/i, 1]
+    def self.find_numero_tva(text)
+      text[/TVA(?:\s+intracommunautaire)?#{BETWEEN_LABEL_VALUE_REGEX}(FR\d{2}\s?\d{9})/i, 1]
     end
 
     def self.find_prenom(text)
-      text[/Prénom\s*:\s*(\w+)/i, 1]
+      text[/Prénom\s*:\s*(#{FRENCH_CHARACTER_REGEX}+)/i, 1]
     end
 
-    def self.find_numero_devis(text)
-      text[/DEVIS\s+N°\s*(\d+)/i, 1]
+    def self.find_raison_sociale(text)
+      text[/Raison sociale\s*:\s*(#{FRENCH_CHARACTER_REGEX}+)/i, 1]
     end
 
     def self.find_rge_number(text)
-      text[/RGE\s+N°\s*(\d+)/i, 1]
+      text[/RGE#{BETWEEN_LABEL_VALUE_REGEX}((?:E-)?E?\d+)/i, 1]
     end
 
     def self.find_siret(text)
-      text[/SIRET\s*:\s*(\d{3}\s*\d{3}\s*\d{3}\s*\d{5})/i, 1]&.gsub(/\s/, "")
+      text[/SIRET#{BETWEEN_LABEL_VALUE_REGEX}(\d{3}\s*\d{3}\s*\d{3}\s*\d{5})/i, 1]
+    end
+
+    def self.find_telephone(text)
+      text[/(?:T[eé]l\.?(?:[eé]phone)#{BETWEEN_LABEL_VALUE_REGEX})?(#{PHONE_REGEX})/i, 1]
     end
   end
 end

lib/quote_validator/global.rb

@@ -3,6 +3,8 @@
 module QuoteValidator
   # Validator for the Quote
   class Global < Base
+    VERSION = "0.0.1"
+
     def validate!
       @errors = []
       @warnings = []

spec/lib/quote_reader/naive_text_spec.rb

@@ -34,8 +34,8 @@
           Adresse Pro : 42 rue
           Raison Sociale : ACME
           Forme Juridique : SAS
-          TVA : 123456
-          Capital : 1000
+          TVA : FR12345678911
+          Capital : 1000 €
           Siret : 123456789
           RGE Number : 123456
         TEXT
@@ -55,7 +55,7 @@
             adresse: "42",
             raison_sociale: "ACME",
             forme_juridique: "SAS",
-            numero_tva: "123456",
+            numero_tva: "FR12345678911",
             capital: "1000",
             rge_number: nil,
             siret: nil
@@ -65,4 +65,113 @@
       # rubocop:enable RSpec/ExampleLength
     end
   end
+
+  # rubocop:disable RSpec/MultipleExpectations
+
+  describe ".find_adresse" do
+    it "returns the adresse" do
+      skip "Add test for find_adresse"
+    end
+  end
+
+  describe ".find_adresse_chantier" do
+    it "returns the adresse_chantier" do
+      skip "Add test for find_adresse_chantier"
+    end
+  end
+
+  describe ".find_adresse_pro" do
+    it "returns the adresse_pro" do
+      skip "Add test for find_adresse_pro"
+    end
+  end
+
+  describe ".find_assurance" do
+    it "returns the assurance" do
+      expect(
+        described_class.find_assurance("Assurance décennale : ToutRix couverture France n° contrat 0000010001234567")
+      ).to eq("ToutRix couverture France n° contrat 0000010001234567")
+    end
+  end
+
+  describe ".find_capital" do
+    it "returns the capital" do
+      expect(described_class.find_capital("capital de 12345 €")).to eq("12345")
+      expect(described_class.find_capital("capilâide 12 345€")).to eq("12 345")
+    end
+  end
+
+  describe ".find_iban" do
+    it "returns the iban" do
+      expect(
+        described_class.find_iban("IBAN : FR74 3000 1234 9000 0000 1234 P77")
+      ).to eq("FR74 3000 1234 9000 0000 1234 P77")
+    end
+  end
+
+  describe ".find_forme_juridique" do
+    it "returns the forme_juridique" do
+      skip "Add test for find_forme_juridique"
+    end
+  end
+
+  describe ".find_mention_devis" do
+    it "returns the mention_devis" do
+      skip "Add test for find_mention_devis"
+    end
+  end
+
+  describe ".find_nom" do
+    it "returns the nom" do
+      skip "Add test for find_nom"
+    end
+  end
+
+  describe ".find_numero_devis" do
+    it "returns the numero_devis" do
+      expect(described_class.find_numero_devis("Devis  N\"   ORG201234")).to eq("ORG201234")
+      expect(described_class.find_numero_devis("Devis n° DC001234")).to eq("DC001234")
+    end
+  end
+
+  describe ".find_numero_tva" do
+    it "returns the numero_tva" do
+      expect(described_class.find_numero_tva("TVA  :FR10831861234")).to eq("FR10831861234")
+      expect(described_class.find_numero_tva("TVA intracommunautaire : FR86504321234")).to eq("FR86504321234")
+    end
+  end
+
+  describe ".find_prenom" do
+    it "returns the prenom" do
+      skip "Add test for find_prenom"
+    end
+  end
+
+  describe ".find_raison_sociale" do
+    it "returns the raison_sociale" do
+      skip "Add test for find_raison_sociale"
+    end
+  end
+
+  describe ".find_rge_number" do
+    it "returns the rge_number" do
+      expect(described_class.find_rge_number("RGE  n. :E123456")).to eq("E123456")
+      expect(described_class.find_rge_number("RGE n°E-E123456")).to eq("E-E123456")
+    end
+  end
+
+  describe ".find_siret" do
+    it "returns the siret" do
+      expect(described_class.find_siret("Siret : 12345678900000")).to eq("12345678900000")
+      expect(described_class.find_siret("Siret : 123 456 789 00000")).to eq("123 456 789 00000")
+    end
+  end
+
+  describe ".find_telephone" do
+    it "returns the telephone" do
+      expect(described_class.find_telephone(" (33) 01 43 39 37 81")).to eq("(33) 01 43 39 37 81")
+    end
+  end
+
+  # rubocop:enable RSpec/MultipleExpectations
 end