slaw 4.2.0 → 5.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 872b9abcfdb880f255e429e07a51f465a86b1ce640639146066220f954baed9d
-  data.tar.gz: 5cc6f53df26d3893959d056dca4df32fa2328820975b0d3d1f5b02259b7e2bf5
+  metadata.gz: 5ed924bc3ba2524cc7c23bf29d69aac408195ae35348e45ca6e7f6d14a018620
+  data.tar.gz: 0ac2bc7e3a1089c56234cabaca35f8961a04db7960655f539412a21293c7ea09
 SHA512:
-  metadata.gz: ac9b7e1c43cfcabff3035c510f102cf7e0b43148234527ff0f6f093efaf43ef197197d3be07fb492449f96f95629b104ec5bfd33f2b1437302531f81ad8f77cb
-  data.tar.gz: 0ddb2245be02b53d08fc03bca1be7616a1cd100518686aa555c2a2fdc341408fdb0f59208685b07d15520c01461a116aefadbdc39ef52d3ab152a41288e3815a
+  metadata.gz: c43f632d52c957e236a117e33d228093f9c41387fca0743b62efa7195f327492a68fb23c215b985ace37f7c89582ad77ad4aecede444da0ace2fd0335437389c
+  data.tar.gz: cc262fe65f346c9bf4b2d959121ec76c091e3aafd1f5f4d8e55e074082c8436f2fe50db5b777b33c7a8c12690fb3fc42f9aa60839fb5ff6733907254e82b0349

data/README.md

@@ -81,6 +81,11 @@ You can create your own grammar by creating a gem that provides these files and
 
 ## Changelog
 
+### 5.0.0 (25 Oct 2019)
+
+* Schedules have a new grammar to make it easier for users to understand headings and subheadings.
+* The way schedule IDs are generated has been simplified.
+
 ### 4.2.0 (7 Sept 2019)
 
 * BODY is allowed to be empty

data/lib/slaw/grammars/schedules.treetop

@@ -20,9 +20,32 @@ module Slaw
       end
 
       rule schedule_title
-        space? schedule_title_prefix space? "\""? num:alphanums? "\""? [ \t:.-]* title:inline_items?
+        new_schedule_title / legacy_schedule_title
+      end
+
+      rule new_schedule_title
+        space? 'SCHEDULE'
+        eol
+        heading:schedule_heading?
+        subheading:schedule_subheading?
+        <ScheduleTitle>
+      end
+
+      rule schedule_heading
+        'HEADING' space? title:inline_items? eol
+      end
+
+      rule schedule_subheading
+        'SUBHEADING' space? title:inline_items? eol
+      end
+
+      # The legacy schedule mechanism is retained because it matches PDF imports well.
+      # Subsequent edits will use the new schedule format.
+      rule legacy_schedule_title
+        space? schedule_title_prefix [ \t:.-]* heading:inline_items?
         subheading:(newline space? inline_items)?
         eol
+        <LegacyScheduleTitle>
       end
 
       rule schedule_title_prefix

data/lib/slaw/grammars/schedules_nodes.rb

@@ -18,72 +18,104 @@ module Slaw
         end
       end
 
-      class Schedule < Treetop::Runtime::SyntaxNode
-        def num
-          n = schedule_title.num.text_value
-          return (n && !n.empty?) ? n : nil
+      class ScheduleTitle < Treetop::Runtime::SyntaxNode
+        def heading_text
+          if heading.empty? or heading.title.empty?
+            nil
+          else
+            heading.title.elements
+              .select { |x| x.instance_of? ::Slaw::Grammars::Inlines::InlineItem }
+              .map { |x| x.text_value }
+              .join('')
+              .strip
+          end
         end
 
-        def alias
-          if not schedule_title.title.text_value.blank?
-            # plain-text elements only
-            schedule_title.title.elements.select { |x| x.instance_of? ::Slaw::Grammars::Inlines::InlineItem }.map { |x| x.text_value }.join('').strip
-          elsif num
-            "Schedule #{num}"
+        def to_xml(b, idprefix=nil, heading_text)
+          if not heading.empty? and not heading.title.empty?
+            b.heading { |b| heading.title.to_xml(b, idprefix) }
           else
-            "Schedule"
+            b.heading(heading_text)
+          end
+
+          if not subheading.empty? and not subheading.title.empty?
+            b.subheading { |b| subheading.title.to_xml(b, idprefix) }
           end
         end
+      end
 
-        def title
-          if not schedule_title.title.text_value.blank?
-            schedule_title.title
-          else
+      class LegacyScheduleTitle < Treetop::Runtime::SyntaxNode
+        def heading_plain_text
+          heading.elements
+            .select { |x| x.instance_of? ::Slaw::Grammars::Inlines::InlineItem }
+            .map { |x| x.text_value }
+            .join('')
+            .strip
+        end
+
+        def heading_text
+          if heading.empty?
             nil
+          else
+            text = self.heading_plain_text
+
+            # change legacy titles that are just numbers (eg. the "1" in "Schedule 1")
+            # to "Schedule 1"
+            text = "Schedule #{text}" if /^\d+$/.match?(text)
+
+            text
           end
         end
 
-        def subheading
-          if not schedule_title.subheading.text_value.blank?
-            schedule_title.subheading.inline_items
+        def to_xml(b, idprefix=nil, heading_text)
+          if not heading.empty? and not (/^\d+$/.match?(self.heading_plain_text))
+            b.heading { |b| heading.to_xml(b, idprefix) }
           else
-            nil
+            b.heading(heading_text)
+          end
+
+          if not subheading.empty?
+            b.subheading { |b| subheading.inline_items.to_xml(b, idprefix) }
           end
         end
+      end
+
+      class Schedule < Treetop::Runtime::SyntaxNode
+        def schedule_id(heading_text, i)
+          heading_text.downcase().strip().gsub(/[^a-z0-9]/i, '').gsub(/ +/, '')
+        end
 
         def to_xml(b, idprefix=nil, i=1)
-          if num
-            n = num
-            component = "schedule#{n}"
-          else
-            n = i
-            # make a component name from the schedule title
-            component = self.alias.downcase().strip().gsub(/[^a-z0-9]/i, '').gsub(/ +/, '')
+          heading_text = self.schedule_title.heading_text
+          if not heading_text
+            heading_text = "Schedule"
+            heading_text << " #{i}" if i > 1
           end
 
-          id = "#{idprefix}#{component}"
+          # the schedule id is derived from the heading
+          schedule_id = self.schedule_id(heading_text, i)
 
-          b.component(id: "component-#{id}") { |b|
-            b.doc_(name: component) { |b|
+          b.component(id: "component-#{schedule_id}") { |b|
+            b.doc_(name: schedule_id) { |b|
               b.meta { |b|
                 b.identification(source: "#slaw") { |b|
                   b.FRBRWork { |b|
-                    b.FRBRthis(value: "#{WORK_URI}/#{component}")
+                    b.FRBRthis(value: "#{WORK_URI}/#{schedule_id}")
                     b.FRBRuri(value: WORK_URI)
-                    b.FRBRalias(value: self.alias)
+                    b.FRBRalias(value: heading_text)
                     b.FRBRdate(date: '1980-01-01', name: 'Generation')
                     b.FRBRauthor(href: '#council')
                     b.FRBRcountry(value: 'za')
                   }
                   b.FRBRExpression { |b|
-                    b.FRBRthis(value: "#{EXPRESSION_URI}/#{component}")
+                    b.FRBRthis(value: "#{EXPRESSION_URI}/#{schedule_id}")
                     b.FRBRuri(value: EXPRESSION_URI)
                     b.FRBRdate(date: '1980-01-01', name: 'Generation')
                     b.FRBRauthor(href: '#council')
                     b.FRBRlanguage(language: 'eng')
                   }
                   b.FRBRManifestation { |b|
-                    b.FRBRthis(value: "#{MANIFESTATION_URI}/#{component}")
+                    b.FRBRthis(value: "#{MANIFESTATION_URI}/#{schedule_id}")
                     b.FRBRuri(value: MANIFESTATION_URI)
                     b.FRBRdate(date: Time.now.strftime('%Y-%m-%d'), name: 'Generation')
                     b.FRBRauthor(href: '#slaw')
@@ -92,17 +124,12 @@ module Slaw
               }
 
               b.mainBody { |b| 
-                idprefix = "#{id}."
+                idprefix = "#{schedule_id}."
 
                 # there is no good AKN hierarchy container for schedules, so we
                 # use hcontainer instead
-                b.hcontainer(id: id, name: "schedule") { |b|
-                  if title and title.elements
-                    b.heading { |b| title.to_xml(b, idprefix) }
-                  else
-                    b.heading(self.alias)
-                  end
-                  b.subheading { |b| subheading.to_xml(b, idprefix) } if subheading
+                b.hcontainer(id: schedule_id, name: "schedule") { |b|
+                  schedule_title.to_xml(b, idprefix, heading_text)
                   body.children.elements.each_with_index { |e| e.to_xml(b, idprefix, i) } if body.is_a? Body
                 }
               }

data/lib/slaw/grammars/za/act_text.xsl

@@ -24,6 +24,8 @@
                   starts-with($prefix, 'CHAPTER ') or
                   starts-with($prefix, 'PART ') or
                   starts-with($prefix, 'SCHEDULE ') or
+                  starts-with($prefix, 'HEADING ') or
+                  starts-with($prefix, 'SUBHEADING ') or
                   starts-with($prefix, 'LONGTITLE ') or
                   starts-with($prefix, 'CROSSHEADING ') or
                   starts-with($prefix, '{|') or
@@ -151,13 +153,14 @@
 
 
   <!-- components/schedules -->
-  <!-- new-style schedules, "article" elements -->
+  <!-- new-style schedules, "hcontainer" elements -->
   <xsl:template match="a:hcontainer[@name='schedule']">
-    <xsl:text>SCHEDULE - </xsl:text>
+    <xsl:text>SCHEDULE&#10;HEADING </xsl:text>
     <xsl:apply-templates select="a:heading" />
     <xsl:text>&#10;</xsl:text>
 
     <xsl:if test="a:subheading">
+      <xsl:text>SUBHEADING </xsl:text>
       <xsl:apply-templates select="a:subheading" />
       <xsl:text>&#10;</xsl:text>
     </xsl:if>
@@ -169,11 +172,12 @@
 
   <!-- old-style schedules, "article" elements -->
   <xsl:template match="a:doc/a:mainBody/a:article">
-    <xsl:text>SCHEDULE - </xsl:text>
+    <xsl:text>SCHEDULE&#10;HEADING </xsl:text>
     <xsl:value-of select="../../a:meta/a:identification/a:FRBRWork/a:FRBRalias/@value" />
     <xsl:text>&#10;</xsl:text>
 
     <xsl:if test="a:heading">
+      <xsl:text>SUBHEADING </xsl:text>
       <xsl:apply-templates select="a:heading" />
       <xsl:text>&#10;</xsl:text>
     </xsl:if>

data/lib/slaw/version.rb

@@ -1,3 +1,3 @@
 module Slaw
-  VERSION = "4.2.0"
+  VERSION = "5.0.0"
 end

data/spec/generator_spec.rb

@@ -54,6 +54,8 @@ Some content.
             <p>participation</p>
             <p>Schedule 2 ignored</p>
             <p>Schedules</p>
+            <p>HEADING x</p>
+            <p>SUBHEADING x</p>
             <p>BODY not escaped</p>
             <p>BODY</p>
             <p>PREAMBLE not escaped</p>
@@ -88,6 +90,10 @@ participation
 
 Schedules
 
+\HEADING x
+
+\SUBHEADING x
+
 BODY not escaped
 
 \BODY
@@ -242,8 +248,9 @@ three
       doc = subject.generate_from_text(<<EOS
 1. Something
 
-Schedule - First Schedule [[remark]]
-Subheading [[another]]
+SCHEDULE
+HEADING First Schedule [[remark]]
+SUBHEADING Subheading [[another]]
 
 Subject to approval in terms of this By-Law.
 EOS
@@ -253,8 +260,9 @@ EOS
 
 1. Something
 
-SCHEDULE - First Schedule [[remark]]
-Subheading [[another]]
+SCHEDULE
+HEADING First Schedule [[remark]]
+SUBHEADING Subheading [[another]]
 
 Subject to approval in terms of this By-Law.