[hotfix] fixed tag_with_html_regexp_array in case of span styles

Author: shlag3n

lib/doc_template/tags/base_tag.rb

@@ -24,14 +24,14 @@ def self.tag_with_html_regexp
       #
       # There can be situations when Google Document is exported
       # in a broken way:
-      #   <span>[imag</span><span>e: OP.LE.L1.004</span><span>]
+      #   <span>[imag</span><span style='font-family:"Montserrat"'>e: OP.LE.L1.004</span><span>]
       # In this case we need to guess the part which should be substituted
       #
       # As a result for the tag with name `image` will be the following
       #   [
-      #     /\[[^\]]*image[[^:,;.]]*(<\/span><span\s?[^>]>)?:?\s?[^\]]*\]/i,
-      #     /\[[^\]]*imag[[^:,;.]]*(<\/span><span\s?[^>]>)?e:?\s?[^\]]*\]/i,
-      #     /\[[^\]]*ima[[^:,;.]]*(<\/span><span\s?[^>]>)?ge:?\s?[^\]]*\]/i
+      #     /\[[^\]]*image[[^:,;.]]*(<\/span>\s*<span\s?[^>]*>)?:?\s?[^\]]*\]/i,
+      #     /\[[^\]]*imag[[^:,;.]]*(<\/span>\s*<span\s?[^>]*>)?e:?\s?[^\]]*\]/i,
+      #     /\[[^\]]*ima[[^:,;.]]*(<\/span>\s*<span\s?[^>]*>)?ge:?\s?[^\]]*\]/i
       #   ]
       #
       # @param [Integer] min_char Minimum number of character by which we guess the tag
@@ -46,7 +46,7 @@ def self.tag_with_html_regexp_array(min_char = 3)
               first_part = tag_name[0..idx]
               last_part = tag_name[(idx + 1)..]
 
-              %r{\[[^\]]*#{first_part}[[^:,;.]]*(</span><span\s?[^>]>)?#{last_part}:?\s?[^\]]*\]}i
+              %r{\[[^\]]*#{first_part}[[^:,;.]]*(</span>\s*<span\s?[^>]*>)?#{last_part}:?\s?[^\]]*\]}i
             end
           end
       end

spec/lib/doc_template/tags/base_tag_spec.rb

@@ -8,6 +8,10 @@ class BaseSpecTag < DocTemplate::Tags::BaseTag
   def self.clean_tag_with_html_regexp
     @tag_with_html_regexp = nil
   end
+
+  def self.clean_tag_with_html_regexp_array
+    @tag_with_html_regexp_array = nil
+  end
 end
 
 describe DocTemplate::Tags::BaseTag do
@@ -33,10 +37,36 @@ def self.clean_tag_with_html_regexp
 
     context 'when TAG_NAME is not defined' do
       before { BaseSpecTag.send :remove_const, :TAG_NAME }
+      after { BaseSpecTag.const_set(:TAG_NAME, 'image') }
 
       it 'raises an error' do
         expect { subject }.to raise_error(StandardError, 'TAG_NAME is not specified')
       end
     end
+
+    describe '.tag_with_html_regexp_array' do
+      before { BaseSpecTag.clean_tag_with_html_regexp_array }
+
+      subject { BaseSpecTag.tag_with_html_regexp_array }
+
+      let(:content) do
+        <<~HTML
+          <p style='padding:0;font-size:11pt;font-family:"Arial";line-height:1.5;text-align:left'><span style='
+          vertical-align:baseline;font-size:11pt;font-family:"Cabin";'><span>[i</span><span
+          style='font-family:"Montserrat"'>mage:OP.PT.L13.014, 50, Esta imagen muestra ruedas de carretas que han sido
+          enterradas por la tierra movida por el viento durante una tormenta de polvo.] Los Estados Unidos tienen una
+          historia trágica con sedimentos en movimiento por el viento Durante la década de 1930, gran parte de los
+          Estados Unidos estaba pasando por una sequía.</span></span></p>
+        HTML
+      end
+
+      subject { BaseSpecTag.tag_with_html_regexp_array(1) }
+
+      it 'returns array of RegExp' do
+        expect(subject).to all(be_a(Regexp))
+        expect(subject.size).to eq 5
+        expect(subject.detect { _1.match(content) }).not_to be_nil
+      end
+    end
   end
 end