justiz 0.1.3 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 0d196d5dbc4f9c497f5a6340ddd7377cf416dff6
-  data.tar.gz: 84a09a957eb241e514241357e52aee7376093ac3
+  metadata.gz: 2956b90485c84af57cd711724282b36debf63a3c
+  data.tar.gz: 09412689d63cb8fd117e9f06b7bb9f7a3fcc49b6
 SHA512:
-  metadata.gz: 244790715df8b1b00ec551430edbdf7a1872d6a0588a73240ff6bbd83d29494a7afe751416907690cc1188368f2947b827c5c8a760158a355924485ff175efd6
-  data.tar.gz: 3090fd312cedff09deb3600fe326a4c1b0973df6ec0b144ebfec38672d08ad6e2bc855268503a27f47c6c9c6868c1a6067541a242b8faf54868b82ff180939e3
+  metadata.gz: bf3287ebafd6d08a41a00e0871eea16e4fbd1e0f3931730177d3e0ed995fd6a4c67a32ddc3949554f5dd06859807f2e309ba1e00c236d51e2e7fbf4f885e1d28
+  data.tar.gz: 3c93f26946552ad11888666f775c3e2658d44b330bda2e4057f21dda5ab3d657dcbbafb19a5151423e5484da8065226a841671413dd223b2860c25e418ac688d

data/justiz.gemspec

@@ -10,7 +10,7 @@ Gem::Specification.new do |spec|
   spec.email         = ["mikep@quake.net"]
   spec.description   = %q{Extracts contact data.}
   spec.summary       = %q{Extract contact data from http://www.justizadressen.nrw.de/}
-  spec.homepage      = ""
+  spec.homepage      = "https://github.com/mike-park/justiz"
   spec.license       = "MIT"
 
   spec.files         = `git ls-files`.split($/)

data/lib/justiz/contact.rb

@@ -6,6 +6,16 @@ module Justiz
     FIELDS = [:court, :location, :post, :phone, :fax, :justiz_id, :url, :email]
     attr_accessor :attributes, *FIELDS
 
+    # remove from court if present
+    COURT_NOISE = [
+        ' - Insolvenzabteilung -',
+        ' - Handels- und Genossenschaftsregister -',
+        ' - Vereinsregister -',
+        ' - Partnerschaftsregister -',
+        ' - Zwangsversteigerung/-verwaltung -',
+        ' - Familiengericht -'
+    ]
+
     def initialize(attributes = {})
       self.attributes = attributes
     end
@@ -16,11 +26,16 @@ module Justiz
       end
     end
 
+    def court=(name)
+      @court = remove_court_noise(name)
+    end
+
     def id
       # too many duplicates
       #[court, justiz_id].compact.join("")
-      # currently no duplicates
-      [court, email].compact.join("")
+      # with noise removed also too many duplicates
+      #[court, email].compact.join("")
+      digest
     end
 
     def location_address
@@ -38,5 +53,17 @@ module Justiz
       end
       Digest.hexencode(sha256.digest)
     end
+
+    def to_s
+      inspect
+    end
+
+    private
+
+    def remove_court_noise(court)
+      court = court.dup
+      COURT_NOISE.each { |noise| court.sub!(/#{noise}$/, '') }
+      court
+    end
   end
 end

data/lib/justiz/scraper/courts.rb

@@ -11,9 +11,10 @@ module Justiz
       end
 
       def contacts
-        states.keys.map do |state|
+        contacts = states.keys.map do |state|
           contacts_for(state)
         end.flatten.compact
+        uniq_contacts(contacts)
       end
 
       def contacts_for(state)
@@ -21,17 +22,23 @@ module Justiz
         return page.contacts unless page.limit_warning?
 
         # do each type separately hoping to avoid limit warning
-        court_types.keys.map do |court_type|
+        contacts = court_types.keys.map do |court_type|
           contacts_of_type(court_type, state)
-        end.flatten.compact.uniq
+        end.flatten.compact
+        uniq_contacts(contacts)
       end
 
       def contacts_of_type(type, state)
-        load_page(type, state, with_warning: true).contacts
+        contacts = load_page(type, state, with_warning: true).contacts
+        uniq_contacts(contacts)
       end
 
       private
 
+      def uniq_contacts(contacts)
+        contacts.uniq {|contact| contact.digest }
+      end
+
       def home_page
         @home_page ||= Page.new(agent.get('http://www.justizadressen.nrw.de/og.php?MD=nrw'))
       end
@@ -69,7 +76,7 @@ module Justiz
       end
 
       def contacts
-        @contacts ||= parse_contacts
+        @contacts ||= parse_contacts.uniq
       end
 
       def parse_contacts

data/lib/justiz/version.rb

@@ -1,3 +1,3 @@
 module Justiz
-  VERSION = "0.1.3"
+  VERSION = "0.2.0"
 end

data/spec/lib/contact_spec.rb

@@ -34,4 +34,16 @@ describe Justiz::Contact do
     digest = "76ef09d0c7d0078015df7a948cf0352c00f6451dab354389b21895a50d89a4a8"
     expect(contact.digest).to eq(digest)
   end
+
+  it "should remove - xxx - from court" do
+    court = "Something#{Justiz::Contact::COURT_NOISE.first}"
+    contact = Justiz::Contact.new(court: court)
+    expect(contact.court).to eq('Something')
+  end
+
+  it "should not remove - xxx - from court" do
+    court = 'Something - Not be removed -'
+    contact = Justiz::Contact.new(court: court)
+    expect(contact.court).to eq(court)
+  end
 end

data/spec/lib/scraper/courts_spec.rb

@@ -73,7 +73,7 @@ describe Justiz::Scraper::Courts do
         VCR.use_cassette 'courts/all_nrw' do
           contacts = subject.contacts_for('NRW')
           #ap contacts
-          expect(contacts.count).to eq(513)
+          expect(contacts.count).to eq(377)
         end
       end
     end
@@ -83,16 +83,16 @@ describe Justiz::Scraper::Courts do
         VCR.use_cassette 'courts/all_all' do
           original = {
               "BRD" => "Bundesgerichte/-behörden 12",
-              "BW" => "Baden-Württemberg 430",
-              "BAY" => "Bayern 348",
+              "BW" => "Baden-Württemberg 284",
+              "BAY" => "Bayern 263",
               "B" => "Berlin 38",
               "BRA" => "Brandenburg 64",
               "BRE" => "Bremen 19",
               "HH" => "Hamburg 35",
               "HES" => "Hessen 115",
               "MV" => "Mecklenburg-Vorpommern 55",
-              "NS" => "Niedersachsen 305",
-              "NRW" => "Nordrhein-Westfalen 513",
+              "NS" => "Niedersachsen 262",
+              "NRW" => "Nordrhein-Westfalen 377",
               "RPF" => "Rheinland-Pfalz 101",
               "SAA" => "Saarland 32",
               "SAC" => "Sachsen 79",
@@ -108,7 +108,7 @@ describe Justiz::Scraper::Courts do
             total += count
           end
           expect(states).to eq(original)
-          expect(total).to eq(2336)
+          expect(total).to eq(1926)
         end
       end
     end
@@ -200,6 +200,13 @@ describe Justiz::Scraper::Courts do
         expect(contact.post_address).to be_a(Justiz::Address)
       end
     end
+
+    it "should rename court" do
+      VCR.use_cassette 'courts/zvg_nrw' do
+        contact = subject.contacts_of_type('ZVG', 'NRW').first
+        expect(contact.court).to_not match(/Zwangsversteigerung/)
+      end
+    end
   end
 end