slaw 3.2.0 → 3.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4411a620b2f2ff5cad3942e1944d344a9e6282e654df6313a78a99e654a91347
-  data.tar.gz: '0789f05372eda817a52d97ba55ec4481f77d69557908709e3d4117b57c5d822b'
+  metadata.gz: 71eaedcb879cb296c11690f4d885135caf60282724350fac3d8e500a61605c32
+  data.tar.gz: 26bc65de89b73bf06b5084c7f233c70c18ff940296e1721e5247d51e39d0b251
 SHA512:
-  metadata.gz: 9b0da1d5da9fb240a3efac4cd280fc567fc58979f219186c3b8719b13c3a3d2ebba5c20abc939e84b187539fa8a3d4ad1c5c7431b421902c0a7ae831dff51fda
-  data.tar.gz: be10a3d7c7ca33d81490b4c9c126a1bf2d6db2a9ec008bd989af4d3fb9485b55d74f560cf561b08b41669babd85831568be6f47937ee8b748ec6d4be02805712
+  metadata.gz: 83ac8241f21659e0f678ef7222015491ba18e86228d83687b1cbd8e40f57c93339b05f74884b2c727522fc7093a408497135db7dd72e27fd5b59c93146b78620
+  data.tar.gz: 59db7ddb266e6d029a8370d623b6482b1c67fb479a0a2b6fcda2cf7d3612a7314946a82165c8fe8a5b397e59b09c9704b68bbc02f9ed0cc6747ad3db4c363d0a

data/.rspec

@@ -0,0 +1 @@
+--require spec_helper

data/README.md

@@ -81,6 +81,11 @@ You can create your own grammar by creating a gem that provides these files and
 
 ## Changelog
 
+### 3.3.0 (1 May 2019)
+
+* Only renest annotated blocklists
+* Table grammar uses additional rules and permits whitespace
+
 ### 3.2.0 (22 April 2019)
 
 * Permit inline content in chapter, part and section headings

data/lib/slaw/grammars/tables.treetop

@@ -14,38 +14,64 @@ module Slaw
       # lines, which we do support.
 
       rule table
-        space? '{|' eol
+        table_start
         table_body
-        '|}' eol
+        table_end
         <Table>
       end
 
+      rule table_start
+        space? '{|' eol
+      end
+
+      rule table_end
+        space? '|}' eol
+      end
+
       rule table_body
         (table_row / table_cell)*
       end
 
       rule table_row
-        '|-' space? eol
+        space? '|-' space? eol
       end
 
       rule table_cell
         # don't match end-of-table
-        !'|}'
-        [!|] attribs:table_attribs? space?
-        # first content line, then multiple lines
-        content:(line:table_line (![!|] space? line:table_line)*)
+        !table_end
+
+        # td (|) or th (!) cell marker with attributes
+        table_cell_start attribs:table_attribs? space?
+
+        # cell's first content line, then multiple lines
+        content:(line:table_line (!table_cell_start space? line:table_line)*)
         <TableCell>
       end
 
+      # td (|) or th (!) cell marker
+      rule table_cell_start
+        space? [!|]
+
+        {
+          def th?
+            elements[1].text_value == '!'
+          end
+        }
+      end
+
       rule table_line
         inline_items:inline_items? eol
         <TableLine>
       end
 
+      # foo=one bar=two |
       rule table_attribs
         space? attribs:(table_attrib+) '|'
       end
 
+      # foo=bar
+      # foo="bar"
+      # foo='bar'
       rule table_attrib
         name:([a-z_-]+) '=' value:(
                                    ('"' (!'"' .)* '"') /

data/lib/slaw/grammars/tables_nodes.rb

@@ -33,7 +33,7 @@ module Slaw
 
       class TableCell < Treetop::Runtime::SyntaxNode
         def to_xml(b, idprefix)
-          tag = text_value[0] == '!' ? 'th' : 'td'
+          tag = table_cell_start.th? ? 'th' : 'td'
 
           attrs = {}
           if not attribs.empty?

data/lib/slaw/grammars/za/act_nodes.rb

@@ -316,7 +316,7 @@ module Slaw
             id = idprefix + "list#{i}"
             idprefix = id + '.'
 
-            b.blockList(id: id) { |b|
+            b.blockList(id: id, renest: true) { |b|
               b.listIntroduction { |b| yield b } if block_given?
 
               elements.each { |e| e.to_xml(b, idprefix) }

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

@@ -10,7 +10,7 @@
   <xsl:template name="escape">
     <xsl:param name="value"/>
 
-    <xsl:variable name="prefix" select="translate(substring($value, 1, 10), 'abcdefghijklmnopqrstuvwxyz', 'ABCDEFGHIJKLMNOPQRSTUVWXYZ')" />
+    <xsl:variable name="prefix" select="translate(substring($value, 1, 13), 'abcdefghijklmnopqrstuvwxyz', 'ABCDEFGHIJKLMNOPQRSTUVWXYZ')" />
     <!-- '(' is considered special, so translate numbers into '(' so we can find and escape them -->
     <xsl:variable name="numprefix" select="translate(substring($value, 1, 3), '1234567890', '((((((((((')" />
 
@@ -21,6 +21,8 @@
                   starts-with($prefix, 'CHAPTER ') or
                   starts-with($prefix, 'PART ') or
                   starts-with($prefix, 'SCHEDULE ') or
+                  starts-with($prefix, 'LONGTITLE ') or
+                  starts-with($prefix, 'CROSSHEADING ') or
                   starts-with($prefix, '{|') or
                   starts-with($numprefix, '(')">
       <xsl:text>\</xsl:text>

data/lib/slaw/parse/blocklists.rb

@@ -3,10 +3,11 @@ module Slaw
     module Blocklists
       include Slaw::Namespace
 
-      # Correctly re-nest nested block lists. We do this by identifying the
-      # numbering format of each item in the list and comparing it with the
-      # surrounding elements. When the numbering format changes, we start
-      # a new nested list.
+      # Correctly re-nest nested block lists that are tagged with the "renest" attribute.
+      #
+      # We do this by identifying the numbering format of each item in the list
+      # and comparing it with the surrounding elements. When the numbering
+      # format changes, we start a new nested list.
       #
       # We make sure to handle special cases such as `(i)` coming between
       # `(h)` and `(j)` versus being at the start of a `(i), (ii), (iii)`
@@ -34,7 +35,8 @@ module Slaw
       #
       # @param doc [Nokogiri::XML::Document] the document
       def self.nest_blocklists(doc)
-        doc.xpath('//a:blockList', a: NS).each do |blocklist|
+        doc.xpath('//a:blockList[@renest]', a: NS).each do |blocklist|
+          blocklist.remove_attribute('renest')
           items = blocklist.xpath('a:item', a: NS)
           nest_blocklist_items(items.to_a, guess_number_format(items.first), nil, nil) unless items.empty?
         end

data/lib/slaw/version.rb

@@ -1,3 +1,3 @@
 module Slaw
-  VERSION = "3.2.0"
+  VERSION = "3.3.0"
 end

data/spec/generator_spec.rb

@@ -119,7 +119,7 @@ PREFACE not escaped
         <heading>Section</heading>
         <paragraph id="section-9.paragraph-0">
           <content>
-            <blockList id="section-9.paragraph-0.list1">
+            <blockList id="section-9.paragraph-0.list1" renest="true">
               <listIntroduction>(2) A special meeting <remark>[ foo ]</remark>:</listIntroduction>
               <item id="section-9.paragraph-0.list1.a">
                 <num>(a)</num>

data/spec/parse/builder_spec.rb

@@ -10,7 +10,7 @@ describe Slaw::Parse::Builder do
   describe '#adjust_blocklists' do
     it 'should nest simple blocks' do
       doc = xml2doc(subsection(<<XML
-            <blockList id="section-10.1.lst0">
+            <blockList id="section-10.1.lst0" renest="true">
               <item id="section-10.1.lst0.a">
                 <num>(a)</num>
                 <p>foo</p>
@@ -79,7 +79,7 @@ XML
 
     it 'should jump back up a level' do
       doc = xml2doc(subsection(<<XML
-            <blockList id="section-10.1.lst0">
+            <blockList id="section-10.1.lst0" renest="true">
               <item id="section-10.1.lst0.a">
                 <num>(a)</num>
                 <p>foo</p>
@@ -130,7 +130,7 @@ XML
 
     it 'should handle (i) correctly' do
       doc = xml2doc(subsection(<<XML
-            <blockList id="section-10.1.lst0">
+            <blockList id="section-10.1.lst0" renest="true">
               <item id="section-10.1.lst0.h">
                 <num>(h)</num>
                 <p>foo</p>
@@ -171,7 +171,7 @@ XML
 
     it 'should handle (u) (v) and (x) correctly' do
       doc = xml2doc(subsection(<<XML
-            <blockList id="section-10.1.lst0">
+            <blockList id="section-10.1.lst0" renest="true">
               <item id="section-10.1.lst0.t">
                 <num>(t)</num>
                 <p>foo</p>
@@ -229,7 +229,7 @@ XML
 
     it 'should handle (j) correctly' do
       doc = xml2doc(subsection(<<XML
-              <blockList id="section-28.3.list2">
+              <blockList id="section-28.3.list2" renest="true">
                 <item id="section-28.3.list2.g">
                   <num>(g)</num>
                   <p>all <term refersTo="#term-memorial_work" id="trm381">memorial work</term> up to 150 mm in thickness must be securely attached to the base;</p>
@@ -303,7 +303,7 @@ XML
     # -------------------------------------------------------------------------
     it 'should handle (I) correctly' do
       doc = xml2doc(subsection(<<XML
-              <blockList id="section-28.3.list2">
+              <blockList id="section-28.3.list2" renest="true">
                 <item id="section-28.3.list2.g">
                   <num>(g)</num>
                   <p>all memorial work up to 150 mm in thickness must be securely attached to the base;</p>
@@ -445,7 +445,7 @@ XML
 
     it 'should treat (AA) after (z) a sublist' do
       doc = xml2doc(subsection(<<XML
-        <blockList id="list0">
+        <blockList id="list0" renest="true">
           <item id="list0.y">
             <num>(y)</num>
             <p>foo</p>
@@ -496,7 +496,7 @@ XML
 
     it 'should handle deeply nested lists' do
       doc = xml2doc(subsection(<<XML
-        <blockList id="list0">
+        <blockList id="list0" renest="true">
           <item id="list0.a">
             <num>(a)</num>
             <p>foo</p>
@@ -599,7 +599,7 @@ XML
 
     it 'should jump back up a level when finding (i) near (h)' do
       doc = xml2doc(subsection(<<XML
-            <blockList id="section-10.1.lst0">
+            <blockList id="section-10.1.lst0" renest="true">
               <item id="section-10.1.lst0.h">
                 <num>(h)</num>
                 <p>foo</p>
@@ -650,7 +650,7 @@ XML
 
     it 'should handle dotted numbers correctly' do
       doc = xml2doc(subsection(<<XML
-            <blockList id="section-9.subsection-2.list2">
+            <blockList id="section-9.subsection-2.list2" renest="true">
               <item id="section-9.subsection-2.list2.9.2.1">
                 <num>9.2.1</num>
                 <p>is incapable of trading because of an illness, provided that:</p>

data/spec/spec_helper.rb

@@ -1 +1,6 @@
 require 'xml_helpers'
+
+RSpec.configure do |config|
+  # use old-style foo.should == 'foo' syntax
+  config.expect_with(:rspec) { |c| c.syntax = :should }
+end

data/spec/za/act_block_spec.rb

@@ -85,7 +85,7 @@ EOS
   <paragraph id="paragraph0">
     <content>
       <p>Some content before the section</p>
-      <blockList id="paragraph0.list1">
+      <blockList id="paragraph0.list1" renest="true">
         <item id="paragraph0.list1.a">
           <num>(a)</num>
           <p>foo</p>
@@ -654,7 +654,7 @@ EOS
   <num>(2)</num>
   <content>
     <p>title</p>
-    <blockList id="2.list1">
+    <blockList id="2.list1" renest="true">
       <item id="2.list1.a">
         <num>(a)</num>
         <p>one</p>
@@ -687,7 +687,7 @@ EOS
       to_xml(node, "", 1).should == '<subsection id="1">
   <num>(1)</num>
   <content>
-    <blockList id="1.list0">
+    <blockList id="1.list0" renest="true">
       <item id="1.list0.a">
         <num>(a)</num>
         <p>one</p>
@@ -719,7 +719,7 @@ EOS
       to_xml(node, "", 1).should == '<subsection id="1">
   <num>(1)</num>
   <content>
-    <blockList id="1.list0">
+    <blockList id="1.list0" renest="true">
       <item id="1.list0.a">
         <num>(a)</num>
         <p>one</p>
@@ -778,7 +778,7 @@ EOS
   <num>(1)</num>
   <content>
     <p>here\'s my really cool list,</p>
-    <blockList id="1.list1">
+    <blockList id="1.list1" renest="true">
       <item id="1.list1.a">
         <num>(a)</num>
         <p>one</p>
@@ -812,7 +812,7 @@ EOS
   <num>(1)</num>
   <content>
     <p>here\'s my really cool list,</p>
-    <blockList id="1.list1">
+    <blockList id="1.list1" renest="true">
       <item id="1.list1.a">
         <num>(a)</num>
         <p/>
@@ -854,7 +854,7 @@ EOS
   <num>9.9</num>
   <content>
     <p>foo</p>
-    <blockList id="9.9.list1">
+    <blockList id="9.9.list1" renest="true">
       <item id="9.9.list1.9.9.1">
         <num>9.9.1</num>
         <p>item1</p>
@@ -888,7 +888,7 @@ EOS
   <num>(1)</num>
   <content>
     <p>a list</p>
-    <blockList id="1.list1">
+    <blockList id="1.list1" renest="true">
       <item id="1.list1.a">
         <num>(a)</num>
         <p>item 1</p>
@@ -899,7 +899,7 @@ EOS
       </item>
     </blockList>
     <p>some text</p>
-    <blockList id="1.list3">
+    <blockList id="1.list3" renest="true">
       <item id="1.list3.c">
         <num>(c)</num>
         <p>item 3</p>
@@ -1346,7 +1346,7 @@ EOS
   <heading>Section</heading>
   <paragraph id="section-1.paragraph0">
     <content>
-      <blockList id="section-1.paragraph0.list0">
+      <blockList id="section-1.paragraph0.list0" renest="true">
         <item id="section-1.paragraph0.list0.a">
           <num>(a)</num>
           <p>first</p>
@@ -1424,7 +1424,7 @@ EOS
   <paragraph id="section-1.paragraph0">
     <content>
       <p>naked statement (c) blah</p>
-      <blockList id="section-1.paragraph0.list1">
+      <blockList id="section-1.paragraph0.list1" renest="true">
         <item id="section-1.paragraph0.list1.a">
           <num>(a)</num>
           <p>foo</p>
@@ -1466,7 +1466,7 @@ EOS
     <num>(2)</num>
     <content>
       <p>Schedule 1</p>
-      <blockList id="section-1.2.list1">
+      <blockList id="section-1.2.list1" renest="true">
         <item id="section-1.2.list1.a">
           <num>(a)</num>
           <p>Part 1</p>
@@ -1538,7 +1538,7 @@ EOS
     <num>3.1</num>
     <content>
       <p>Informal trading may include, amongst others:-</p>
-      <blockList id="section-3.3.1.list1">
+      <blockList id="section-3.3.1.list1" renest="true">
         <item id="section-3.3.1.list1.3.1.1">
           <num>3.1.1</num>
           <p>street trading;</p>

data/spec/za/act_inline_spec.rb

@@ -125,7 +125,7 @@ EOS
     <num>(1)</num>
     <content>
       <p>something</p>
-      <blockList id="section-1.1.list1">
+      <blockList id="section-1.1.list1" renest="true">
         <item id="section-1.1.list1.a">
           <num>(a)</num>
           <p>with a remark <remark status="editorial">[Section 1 amended by Act 23 of 2004]</remark></p>
@@ -285,7 +285,7 @@ EOS
 EOS
       to_xml(node, "").should == '<paragraph id="paragraph0">
   <content>
-    <blockList id="paragraph0.list0">
+    <blockList id="paragraph0.list0" renest="true">
       <item id="paragraph0.list0.2.18.1">
         <num>2.18.1</num>
         <p>a traffic officer appointed in terms of section 3 of the Road Traffic <ref href="/za/act/1989/29">Act, No. 29 of 1989</ref> or section 3A of the National Road Traffic <ref href="/za/act/1996/93">Act No. 93 of 1996</ref> as the case may be;</p>

data/spec/za/act_table_spec.rb

@@ -164,6 +164,50 @@ EOS
 </table>'
     end
 
+    it 'should allow whitespace at start of table rows' do
+      node = parse :table, <<EOS
+    {|
+    ! foo
+ three
+  |-
+  | four
+    |}
+EOS
+
+      to_xml(node, "prefix.").should == '<table id="prefix.table0">
+  <tr>
+    <th>
+      <p>foo<eol/>three</p>
+    </th>
+  </tr>
+  <tr>
+    <td>
+      <p>four</p>
+    </td>
+  </tr>
+</table>'
+    end
+
+    it 'should tolerate lines that aren\'t really ending lines' do
+      node = parse :table, <<EOS
+{|
+| cell
+|} another cell
+|}
+EOS
+
+      to_xml(node, "prefix.").should == '<table id="prefix.table0">
+  <tr>
+    <td>
+      <p>cell</p>
+    </td>
+    <td>
+      <p>} another cell</p>
+    </td>
+  </tr>
+</table>'
+    end
+
     it 'should parse a table in a section' do
       node = parse :section, <<EOS
 10. A section title

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: slaw
 version: !ruby/object:Gem::Version
-  version: 3.2.0
+  version: 3.3.0
 platform: ruby
 authors:
 - Greg Kempe
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-04-22 00:00:00.000000000 Z
+date: 2019-05-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -118,6 +118,7 @@ extensions: []
 extra_rdoc_files: []
 files:
 - ".gitignore"
+- ".rspec"
 - ".travis.yml"
 - Gemfile
 - LICENSE.txt