slaw 0.5.1 → 0.6.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 26e7c12c3e421410a6be3b18c19f589915e920fb
-  data.tar.gz: e8c1e77994e30c194b42bcca11bbc3b4eace6f76
+  metadata.gz: e5ab33b04df46f9d2d087f7a22ce9c0e7d12278e
+  data.tar.gz: 276c7b704a30c435d0dfbce2e74c34cd2e902b51
 SHA512:
-  metadata.gz: 3e229100f80879b9646135b9c9f6fab2c543a20cd89ed6b021a74164fb92874b876451571bc9552d3150a2953a00be8d927b63c3e7ad780dec800751bff086c3
-  data.tar.gz: c8d2d73f08b67a2535816d9b5ae8f8abfb7cd2627f3aef1688a986bd4e6ba1e71a01cb1c53c0a5b66d9f836191abef6c89e8f43767a12d8b949d329e635c1043
+  metadata.gz: 7632de6a68c70b2ce44a4854ec4bcb35867f269447b8c339286056301ef57d73548edabddd5a0ea963d4bc19c66b37d571beb1f8f635d7dc533ca32e1f57bd35
+  data.tar.gz: 6e4a1d0778b6616f6956fba51a1b604514a036f188d89232984d7f57dd5f75f3bad2b031507d9d2eeeb1f470d9b85fef063d055719e97853df7d8665f4715a1a

data/bin/slaw

@@ -8,7 +8,7 @@ class SlawCLI < Thor
 
   class_option :verbose, type: :boolean, desc: "Display log output on stderr"
 
-  desc "parse FILE", "parse FILE into Akoma Ntoso XML"
+  desc "parse FILE", "Parse FILE into Akoma Ntoso XML"
   option :input, enum: ['text', 'pdf'], desc: "Type of input if it can't be determined automatically"
   option :pdftotext, desc: "Location of the pdftotext binary if not in PATH"
   option :definitions, type: :boolean, desc: "Find and link definitions (this can be slow). Default: false"
@@ -36,6 +36,12 @@ class SlawCLI < Thor
     puts act.to_xml(indent: 2)
   end
 
+  map %w(--version) => :__print_version
+  desc "--version", "Print slaw version info"
+  def __print_version
+    say "slaw #{Slaw::VERSION}"
+  end
+
   no_commands do
     def logging
       logger = Log4r::Logger.new('Slaw')

data/lib/slaw/parse/builder.rb

@@ -154,7 +154,9 @@ module Slaw
                 doc.xpath('//a:component/a:doc[@name="schedules"]//a:heading/text()', a: NS)
 
         nodes.each do |heading|
-          heading.content = heading.content.downcase.gsub(/^\w/) { $&.upcase }
+          if !(heading.content =~ /[a-z]/)
+            heading.content = heading.content.downcase.gsub(/^\w/) { $&.upcase }
+          end
         end
       end
 

data/lib/slaw/version.rb

@@ -1,3 +1,3 @@
 module Slaw
-  VERSION = "0.5.1"
+  VERSION = "0.6.0"
 end

data/lib/slaw/za/act_nodes.rb

@@ -3,14 +3,17 @@ module Slaw
     module Act
       class Act < Treetop::Runtime::SyntaxNode
         FRBR_URI = '/za/act/1980/01'
+        WORK_URI = FRBR_URI
+        EXPRESSION_URI = "#{FRBR_URI}/eng@"
+        MANIFESTATION_URI = EXPRESSION_URI
 
         def to_xml(b)
           b.act(contains: "originalVersion") { |b|
             write_meta(b)
             write_preamble(b)
             write_body(b)
-            write_schedules(b)
           }
+          write_schedules(b)
         end
 
         def write_meta(b)
@@ -34,23 +37,23 @@ module Slaw
           b.identification(source: "#slaw") { |b|
             # use stub values so that we can generate a validating document
             b.FRBRWork { |b|
-              b.FRBRthis(value: "#{FRBR_URI}/main")
-              b.FRBRuri(value: '/za/act/locale/1980/name')
+              b.FRBRthis(value: "#{WORK_URI}/main")
+              b.FRBRuri(value: WORK_URI)
               b.FRBRalias(value: 'Short Title')
               b.FRBRdate(date: '1980-01-01', name: 'Generation')
               b.FRBRauthor(href: '#council', as: '#author')
               b.FRBRcountry(value: 'za')
             }
             b.FRBRExpression { |b|
-              b.FRBRthis(value: '/za/act/locale/1980/name/main/eng@')
-              b.FRBRuri(value: '/za/act/locale/1980/name/eng@')
+              b.FRBRthis(value: "#{EXPRESSION_URI}/main")
+              b.FRBRuri(value: EXPRESSION_URI)
               b.FRBRdate(date: '1980-01-01', name: 'Generation')
               b.FRBRauthor(href: '#council', as: '#author')
               b.FRBRlanguage(language: 'eng')
             }
             b.FRBRManifestation { |b|
-              b.FRBRthis(value: '/za/act/locale/1980/name/main/eng@')
-              b.FRBRuri(value: '/za/act/locale/1980/name/eng@')
+              b.FRBRthis(value: "#{MANIFESTATION_URI}/main")
+              b.FRBRuri(value: MANIFESTATION_URI)
               b.FRBRdate(date: Time.now.strftime('%Y-%m-%d'), name: 'Generation')
               b.FRBRauthor(href: '#slaw', as: '#author')
             }
@@ -58,11 +61,7 @@ module Slaw
         end
 
         def write_preamble(b)
-          if preamble.text_value != ""
-            b.preamble { |b|
-              preamble.to_xml(b)
-            }
-          end
+          preamble.to_xml(b)
         end
 
         def write_body(b)
@@ -78,11 +77,15 @@ module Slaw
 
       class Preamble < Treetop::Runtime::SyntaxNode
         def to_xml(b)
-          statements.elements.each { |e|
-            if not (e.content.text_value =~ /^preamble/i)
-              b.p(e.content.text_value)
-            end
-          }
+          if text_value != ""
+            b.preamble { |b|
+              statements.elements.each { |e|
+                if not (e.content.text_value =~ /^preamble/i)
+                  b.p(e.content.text_value)
+                end
+              }
+            }
+          end
         end
       end
 
@@ -315,37 +318,43 @@ module Slaw
           return if schedules.elements.empty?
 
           b.components { |b| 
-            b.component(id: 'component-0') { |b|
-              b.doc(name: 'schedules') { |b|
-                b.meta { |b| 
-                  b.identification(source: "#slaw") { |b|
-                    b.FRBRWork { |b|
-                      b.FRBRthis(value: '/za/act/locale/1980/name/main/schedules')
-                      b.FRBRuri(value: '/za/act/locale/1980/name/schedules')
-                      b.FRBRdate(date: '1980-01-01', name: 'Generation')
-                      b.FRBRauthor(href: '#council', as: '#author')
-                      b.FRBRcountry(value: 'za')
-                    }
-                    b.FRBRExpression { |b|
-                      b.FRBRthis(value: '/za/act/locale/1980/name/main//schedules/eng@')
-                      b.FRBRuri(value: '/za/act/locale/1980/name/schedules/eng@')
-                      b.FRBRdate(date: '1980-01-01', name: 'Generation')
-                      b.FRBRauthor(href: '#council', as: '#author')
-                      b.FRBRlanguage(language: 'eng')
-                    }
-                    b.FRBRManifestation { |b|
-                      b.FRBRthis(value: '/za/act/locale/1980/name/main/schedules/eng@')
-                      b.FRBRuri(value: '/za/act/locale/1980/name/schedules/eng@')
-                      b.FRBRdate(date: Time.now.strftime('%Y-%m-%d'), name: 'Generation')
-                      b.FRBRauthor(href: '#slaw', as: '#author')
-                    }
-                  }
-                }
+            schedules.elements.each_with_index { |e, i| write_schedule(e, i+1, b) }
+          }
+        end
 
-                b.mainBody { |b|
-                  schedules.elements.each_with_index { |e, i| e.to_xml(b, i) }
+        def write_schedule(element, i, b)
+          # component name
+          comp = "schedule#{i}"
+
+          b.component(id: "component-#{i}") { |b|
+            b.doc(name: "schedule#{i}") { |b|
+              b.meta { |b|
+                b.identification(source: "#slaw") { |b|
+                  b.FRBRWork { |b|
+                    b.FRBRthis(value: "#{Act::WORK_URI}/#{comp}")
+                    b.FRBRuri(value: Act::WORK_URI)
+                    b.FRBRalias(value: element.alias)
+                    b.FRBRdate(date: '1980-01-01', name: 'Generation')
+                    b.FRBRauthor(href: '#council', as: '#author')
+                    b.FRBRcountry(value: 'za')
+                  }
+                  b.FRBRExpression { |b|
+                    b.FRBRthis(value: "#{Act::EXPRESSION_URI}/#{comp}")
+                    b.FRBRuri(value: Act::EXPRESSION_URI)
+                    b.FRBRdate(date: '1980-01-01', name: 'Generation')
+                    b.FRBRauthor(href: '#council', as: '#author')
+                    b.FRBRlanguage(language: 'eng')
+                  }
+                  b.FRBRManifestation { |b|
+                    b.FRBRthis(value: "#{Act::MANIFESTATION_URI}/#{comp}")
+                    b.FRBRuri(value: Act::MANIFESTATION_URI)
+                    b.FRBRdate(date: Time.now.strftime('%Y-%m-%d'), name: 'Generation')
+                    b.FRBRauthor(href: '#slaw', as: '#author')
+                  }
                 }
               }
+
+              b.mainBody { |b| element.to_xml(b, i) }
             }
           }
         end
@@ -357,6 +366,14 @@ module Slaw
           return (n && !n.empty?) ? n : nil
         end
 
+        def alias
+          if num
+            "Schedule #{num}"
+          else
+            "Schedule"
+          end
+        end
+
         def heading
           if schedule_heading.schedule_title.respond_to? :content
             schedule_heading.schedule_title.content.text_value
@@ -366,21 +383,15 @@ module Slaw
         end
 
         def to_xml(b, i)
-          n = num
-          id = if n
-                 "schedule-#{n}"
-               else
-                 "schedules"
-               end
-
-          b.chapter(id: id) { |b|
-            b.num(num) if num
-            b.heading(heading) if heading
+          n = num.nil? ? i : num
+          id = "schedule-#{n}"
 
-            b.section(id: id + ".section-0") { |b|
-              b.content { |b|
-                statements.elements.each { |e| b.p(e.content.text_value) }
-              }
+          # there is no good AKN hierarchy container for schedules, so we
+          # just use article because we don't use it anywhere else.
+          b.article(id: id) { |b|
+            b.heading(heading) if heading
+            b.content { |b|
+              statements.elements.each { |e| b.p(e.content.text_value) }
             }
           }
         end

data/spec/parse/builder_spec.rb

@@ -541,6 +541,44 @@ XML
             </blockList>
           </content>
         </subsection>
+XML
+      )
+    end
+  end
+
+  describe '#normalise_headings' do
+    it 'should normalise ALL CAPS headings' do
+      doc = xml2doc(section(<<XML
+          <heading>DEFINITIONS FOR A.B.C.</heading>
+          <content>
+            <p></p>
+          </content>
+XML
+      ))
+      subject.normalise_headings(doc)
+      doc.to_s.should == section(<<XML
+        <heading>Definitions for a.b.c.</heading>
+        <content>
+          <p/>
+        </content>
+XML
+      )
+    end
+
+    it 'should not normalise normal headings' do
+      doc = xml2doc(section(<<XML
+          <heading>Definitions for A.B.C.</heading>
+          <content>
+            <p></p>
+          </content>
+XML
+      ))
+      subject.normalise_headings(doc)
+      doc.to_s.should == section(<<XML
+        <heading>Definitions for A.B.C.</heading>
+        <content>
+          <p/>
+        </content>
 XML
       )
     end

data/spec/za/act_spec.rb

@@ -364,5 +364,166 @@ EOS
       sched.statements.elements[0].content.text_value.should == "Baz"
       sched.statements.elements[1].content.text_value.should == "Boom"
     end
+
+    it 'should serialise many schedules correctly' do
+      node = parse :schedules, <<EOS
+Schedule "1"
+A Title
+1. Foo
+2. Bar
+Schedule 2
+Another Title
+Baz
+Boom
+EOS
+
+      s = ""
+      builder = ::Builder::XmlMarkup.new(indent: 2, target: s)
+
+      node.to_xml(builder)
+
+      today = Time.now.strftime('%Y-%m-%d')
+
+      s.should == <<EOS
+<components>
+  <component id="component-1">
+    <doc name="schedule1">
+      <meta>
+        <identification source="#slaw">
+          <FRBRWork>
+            <FRBRthis value="/za/act/1980/01/schedule1"/>
+            <FRBRuri value="/za/act/1980/01"/>
+            <FRBRalias value="Schedule 1"/>
+            <FRBRdate date="1980-01-01" name="Generation"/>
+            <FRBRauthor href="#council" as="#author"/>
+            <FRBRcountry value="za"/>
+          </FRBRWork>
+          <FRBRExpression>
+            <FRBRthis value="/za/act/1980/01/eng@/schedule1"/>
+            <FRBRuri value="/za/act/1980/01/eng@"/>
+            <FRBRdate date="1980-01-01" name="Generation"/>
+            <FRBRauthor href="#council" as="#author"/>
+            <FRBRlanguage language="eng"/>
+          </FRBRExpression>
+          <FRBRManifestation>
+            <FRBRthis value="/za/act/1980/01/eng@/schedule1"/>
+            <FRBRuri value="/za/act/1980/01/eng@"/>
+            <FRBRdate date="#{today}" name="Generation"/>
+            <FRBRauthor href="#slaw" as="#author"/>
+          </FRBRManifestation>
+        </identification>
+      </meta>
+      <mainBody>
+        <article id="schedule-1">
+          <heading>A Title</heading>
+          <content>
+            <p>1. Foo</p>
+            <p>2. Bar</p>
+          </content>
+        </article>
+      </mainBody>
+    </doc>
+  </component>
+  <component id="component-2">
+    <doc name="schedule2">
+      <meta>
+        <identification source="#slaw">
+          <FRBRWork>
+            <FRBRthis value="/za/act/1980/01/schedule2"/>
+            <FRBRuri value="/za/act/1980/01"/>
+            <FRBRalias value="Schedule 2"/>
+            <FRBRdate date="1980-01-01" name="Generation"/>
+            <FRBRauthor href="#council" as="#author"/>
+            <FRBRcountry value="za"/>
+          </FRBRWork>
+          <FRBRExpression>
+            <FRBRthis value="/za/act/1980/01/eng@/schedule2"/>
+            <FRBRuri value="/za/act/1980/01/eng@"/>
+            <FRBRdate date="1980-01-01" name="Generation"/>
+            <FRBRauthor href="#council" as="#author"/>
+            <FRBRlanguage language="eng"/>
+          </FRBRExpression>
+          <FRBRManifestation>
+            <FRBRthis value="/za/act/1980/01/eng@/schedule2"/>
+            <FRBRuri value="/za/act/1980/01/eng@"/>
+            <FRBRdate date="#{today}" name="Generation"/>
+            <FRBRauthor href="#slaw" as="#author"/>
+          </FRBRManifestation>
+        </identification>
+      </meta>
+      <mainBody>
+        <article id="schedule-2">
+          <heading>Another Title</heading>
+          <content>
+            <p>Baz</p>
+            <p>Boom</p>
+          </content>
+        </article>
+      </mainBody>
+    </doc>
+  </component>
+</components>
+EOS
+
+    end
+
+    it 'should serialise a single schedule without a heading' do
+      node = parse :schedules, <<EOS
+Schedule "1"
+Other than as is set out hereinbelow, no signs other than locality bound signs, temporary signs including loose portable sign, estate agents signs, newspaper headline posters and posters (the erection of which must comply with the appropriate schedules pertinent thereto) shall be erected on Municipal owned land.
+1. Foo
+2. Bar
+EOS
+
+      s = ""
+      builder = ::Builder::XmlMarkup.new(indent: 2, target: s)
+
+      node.to_xml(builder)
+
+      today = Time.now.strftime('%Y-%m-%d')
+
+      s.should == <<EOS
+<components>
+  <component id="component-1">
+    <doc name="schedule1">
+      <meta>
+        <identification source="#slaw">
+          <FRBRWork>
+            <FRBRthis value="/za/act/1980/01/schedule1"/>
+            <FRBRuri value="/za/act/1980/01"/>
+            <FRBRalias value="Schedule 1"/>
+            <FRBRdate date="1980-01-01" name="Generation"/>
+            <FRBRauthor href="#council" as="#author"/>
+            <FRBRcountry value="za"/>
+          </FRBRWork>
+          <FRBRExpression>
+            <FRBRthis value="/za/act/1980/01/eng@/schedule1"/>
+            <FRBRuri value="/za/act/1980/01/eng@"/>
+            <FRBRdate date="1980-01-01" name="Generation"/>
+            <FRBRauthor href="#council" as="#author"/>
+            <FRBRlanguage language="eng"/>
+          </FRBRExpression>
+          <FRBRManifestation>
+            <FRBRthis value="/za/act/1980/01/eng@/schedule1"/>
+            <FRBRuri value="/za/act/1980/01/eng@"/>
+            <FRBRdate date="#{today}" name="Generation"/>
+            <FRBRauthor href="#slaw" as="#author"/>
+          </FRBRManifestation>
+        </identification>
+      </meta>
+      <mainBody>
+        <article id="schedule-1">
+          <content>
+            <p>Other than as is set out hereinbelow, no signs other than locality bound signs, temporary signs including loose portable sign, estate agents signs, newspaper headline posters and posters (the erection of which must comply with the appropriate schedules pertinent thereto) shall be erected on Municipal owned land.</p>
+            <p>1. Foo</p>
+            <p>2. Bar</p>
+          </content>
+        </article>
+      </mainBody>
+    </doc>
+  </component>
+</components>
+EOS
+    end
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: slaw
 version: !ruby/object:Gem::Version
-  version: 0.5.1
+  version: 0.6.0
 platform: ruby
 authors:
 - Greg Kempe
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-03-05 00:00:00.000000000 Z
+date: 2015-04-28 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler