RubyGems - wovnrb - Versions diffs - 1.0.13 → 1.1.0 - Mend

wovnrb 1.0.13 → 1.1.0

Files changed (75) hide show

data/test/fixtures/unified_values/small_html/text_different_inline_each_other_expected.json ADDED Viewed

@@ -0,0 +1,22 @@
+[
+  {
+    "xpath": "/html/body/div/text()",
+    "srcs": [
+      "<span>",
+      "<a>",
+      "text value1",
+      "</a>",
+      "</span>"
+    ]
+  },
+  {
+    "xpath": "/html/body/div/text()[2]",
+    "srcs": [
+      "<strong>",
+      "<em>",
+      "text value1",
+      "</em>",
+      "</strong>"
+    ]
+  }
+]

data/test/fixtures/unified_values/small_html/text_in_svg_actual.html ADDED Viewed

@@ -0,0 +1,9 @@
+<html>
+<head></head>
+<body>
+<svg width="1px" height="1px" xmlns="http://www.w3.org/2000/svg">
+  <text x="1" y="1" font-size="30">text value!</text>
+  <use href="#test-sym" x="0" y="1" width="1" height="1"/>
+</svg>
+</body>
+</html>

data/test/fixtures/unified_values/small_html/text_in_svg_expected.json ADDED Viewed

@@ -0,0 +1,8 @@
+[
+  {
+    "xpath": "/html/body/svg/text/text()",
+    "srcs": [
+      "text value!"
+    ]
+  }
+]

data/test/fixtures/unified_values/small_html/text_with_html_entity_actual.html ADDED Viewed

@@ -0,0 +1,6 @@
+<html>
+<head></head>
+<body>
+<p>This is &lt;a&gt; text value!</p>
+</body>
+</html>

data/test/fixtures/unified_values/small_html/text_with_html_entity_expected.json ADDED Viewed

@@ -0,0 +1,8 @@
+[
+  {
+    "xpath": "/html/body/p/text()",
+    "srcs": [
+      "This is &lt;a&gt; text value!"
+    ]
+  }
+]

data/test/fixtures/unified_values/small_html/unknown_or_custom_tag_actual.html ADDED Viewed

@@ -0,0 +1,12 @@
+<html>
+<head></head>
+<body>
+<div>
+  <span>
+    <a href="#">text value1</a>
+    <unknown class="emoticon">text value2</unknown>
+    text value3
+  </span>
+</div>
+</body>
+</html>

data/test/fixtures/unified_values/small_html/unknown_or_custom_tag_expected.json ADDED Viewed

@@ -0,0 +1,24 @@
+[
+  {
+    "xpath": "/html/body/div/text()",
+    "srcs": [
+      "<span>",
+      "<a>",
+      "text value1",
+      "</a>"
+    ]
+  },
+  {
+    "xpath": "/html/body/div/span/unknown/text()",
+    "srcs": [
+      "text value2"
+    ]
+  },
+  {
+    "xpath": "/html/body/div/text()[2]",
+    "srcs": [
+      "text value3",
+      "</span>"
+    ]
+  }
+]

data/test/fixtures/unified_values/small_html/unnecessay_top_end_tag_actual.html ADDED Viewed

@@ -0,0 +1,12 @@
+<html>
+<head></head>
+<body>
+<li>
+  <a href="#">
+    <span><span><img src='sample.jpeg'></span></span>
+    <span>text value1</span>
+    <span>text value2</span>
+  </a>
+</li>
+</body>
+</html>

data/test/fixtures/unified_values/small_html/unnecessay_top_end_tag_expected.json ADDED Viewed

@@ -0,0 +1,14 @@
+[
+  {
+    "xpath": "/html/body/li/text()[2]",
+    "srcs": [
+      "<span>",
+      "text value1",
+      "</span>",
+      "<span>",
+      "text value2",
+      "</span>",
+      "</a>"
+    ]
+  }
+]

data/test/fixtures/unified_values/small_html/wovn_ignore_actual.html ADDED Viewed

@@ -0,0 +1,10 @@
+<html>
+<head></head>
+<body>
+<div>
+  <span>
+    This is a <b wovn-ignore>complex</b> text value!
+  </span>
+</div>
+</body>
+</html>

data/test/fixtures/unified_values/small_html/wovn_ignore_expected.json ADDED Viewed

@@ -0,0 +1,13 @@
+[
+  {
+    "xpath": "/html/body/div/text()",
+    "srcs": [
+      "<span>",
+      "This is a",
+      "<b wovn-ignore>",
+      "</b>",
+      "text value!",
+      "</span>"
+    ]
+  }
+]

data/test/lib/html_replacers/unified_values/dst_swapping_targets_creator_test.rb ADDED Viewed

@@ -0,0 +1,137 @@
+require 'test_helper'
+module Wovnrb
+  module UnifiedValues
+    class DstSwappingTargetsCreatorTest < WovnMiniTest
+      def test_run
+        text_index = {
+          '' => {
+            'en' => [
+              { 'data' => 'a<a>b</a>c' }
+            ]
+          }
+        }
+        DstSwappingTargetsCreator.new(text_index).run!
+        assert_equal(%w[a b c], text_index['']['en'][0]['swapping_targets'])
+      end
+      def test_run_with_data_with_spaces
+        text_index = {
+          '' => {
+            'en' => [
+              { 'data' => ' a <a> b </a> c ' }
+            ]
+          }
+        }
+        DstSwappingTargetsCreator.new(text_index).run!
+        assert_equal([' a ', ' b ', ' c '], text_index['']['en'][0]['swapping_targets'])
+      end
+      def test_run_with_data_stated_by_tag
+        text_index = {
+          '' => {
+            'en' => [
+              { 'data' => '<a>b</a>c' }
+            ]
+          }
+        }
+        DstSwappingTargetsCreator.new(text_index).run!
+        assert_equal(['', 'b', 'c'], text_index['']['en'][0]['swapping_targets'])
+      end
+      def test_run_with_data_ended_by_tag
+        text_index = {
+          '' => {
+            'en' => [
+              { 'data' => 'a<a>b</a>' }
+            ]
+          }
+        }
+        DstSwappingTargetsCreator.new(text_index).run!
+        assert_equal(['a', 'b', ''], text_index['']['en'][0]['swapping_targets'])
+      end
+      def test_run_with_data_with_no_content_inside_tag
+        text_index = {
+          '' => {
+            'en' => [
+              { 'data' => 'a<a></a>c' }
+            ]
+          }
+        }
+        DstSwappingTargetsCreator.new(text_index).run!
+        assert_equal(['a', '', 'c'], text_index['']['en'][0]['swapping_targets'])
+      end
+      def test_run_with_data_with_tag_only
+        text_index = {
+          '' => {
+            'en' => [
+              { 'data' => '<a></a>' }
+            ]
+          }
+        }
+        DstSwappingTargetsCreator.new(text_index).run!
+        assert_equal(['', '', ''], text_index['']['en'][0]['swapping_targets'])
+      end
+      def test_run_with_data_without_tag
+        text_index = {
+          '' => {
+            'en' => [
+              { 'data' => 'a' }
+            ]
+          }
+        }
+        DstSwappingTargetsCreator.new(text_index).run!
+        assert_equal(['a'], text_index['']['en'][0]['swapping_targets'])
+      end
+      def test_run_with_data_with_wovn_ignore
+        text_index = {
+          '' => {
+            'en' => [
+              { 'data' => 'a<a wovn-ignore>b</a>c' }
+            ]
+          }
+        }
+        DstSwappingTargetsCreator.new(text_index).run!
+        assert_equal(%w[a c], text_index['']['en'][0]['swapping_targets'])
+      end
+      def test_run_with_data_with_closing_tag
+        text_index = {
+          '' => {
+            'en' => [
+              { 'data' => 'a<br>bc' }
+            ]
+          }
+        }
+        DstSwappingTargetsCreator.new(text_index).run!
+        assert_equal(%w[a bc], text_index['']['en'][0]['swapping_targets'])
+      end
+      def test_run_with_data_with_both_closing_tag_and_no_closing_tag
+        text_index = {
+          '' => {
+            'en' => [
+              { 'data' => 'a<a>b<br>c</a>d' }
+            ]
+          }
+        }
+        DstSwappingTargetsCreator.new(text_index).run!
+        assert_equal(%w[a b c d], text_index['']['en'][0]['swapping_targets'])
+      end
+    end
+  end
+end

data/test/lib/html_replacers/unified_values/element_category_test.rb ADDED Viewed

@@ -0,0 +1,49 @@
+require 'test_helper'
+module Wovnrb
+  module UnifiedValues
+    class ElementCategoryTest < WovnMiniTest
+      def test_contents
+        assert_equal(116, ElementCategory::CONTENT_TYPES.size)
+      end
+      def test_inline_elements
+        expected_inlines = %w[a abbr b bdi bdo button cite code data dfn em i kbd label legend mark meter option q rb rp rt rtc s samp small span strong sub sup time u var]
+        assert_same_elements(expected_inlines, ElementCategory::INLINE_ELEMENTS.to_a)
+      end
+      def test_empty_elements
+        expected_inlines = %w[br param source track wbr input]
+        assert_same_elements(expected_inlines, ElementCategory::EMPTY_ELEMENTS.to_a)
+      end
+      def test_ignore_elements
+        expected_inlines = %w[area audio canvas embed iframe img map meta object picture video]
+        assert_same_elements(expected_inlines, ElementCategory::IGNORE_ELEMENTS.to_a)
+      end
+      def test_skip_elements
+        expected_inlines = %w[base link noscript script style template]
+        assert_same_elements(expected_inlines, ElementCategory::SKIP_ELEMENTS.to_a)
+      end
+      def test_skip_elements_without_attributes
+        expected_inlines = %w[textarea]
+        assert_same_elements(expected_inlines, ElementCategory::SKIP_ELEMENTS_WITHOUT_ATTRIBUTES.to_a)
+      end
+      def test_block_elements
+        expected_inlines = %w[address article aside bb blockquote body caption col colgroup datalist dd del details dialog div dl dt fieldset figcaption figure footer form h1 h2 h3 h4 h5 h6 head header hgroup hr html ins li main menu nav ol optgroup output p pre progress ruby section select slot summary svg table tbody td tfoot th thead title tr ul]
+        assert_same_elements(expected_inlines, ElementCategory::BLOCK_ELEMENTS.to_a)
+      end
+      def test_no_duplication
+        elements = ElementCategory::INLINE_ELEMENTS + ElementCategory::EMPTY_ELEMENTS + ElementCategory::IGNORE_ELEMENTS + ElementCategory::SKIP_ELEMENTS + ElementCategory::SKIP_ELEMENTS_WITHOUT_ATTRIBUTES + ElementCategory::BLOCK_ELEMENTS
+        element_size_sum = ElementCategory::INLINE_ELEMENTS.size + ElementCategory::EMPTY_ELEMENTS.size + ElementCategory::IGNORE_ELEMENTS.size + ElementCategory::SKIP_ELEMENTS.size + ElementCategory::SKIP_ELEMENTS_WITHOUT_ATTRIBUTES.size + ElementCategory::BLOCK_ELEMENTS.size
+        assert_equal(elements.size, ElementCategory::CONTENT_TYPES.size)
+        assert_equal(element_size_sum, ElementCategory::CONTENT_TYPES.size)
+      end
+    end
+  end
+end

data/test/lib/html_replacers/unified_values/node_swapping_targets_creator_test.rb ADDED Viewed

@@ -0,0 +1,137 @@
+require 'test_helper'
+module Wovnrb
+  module UnifiedValues
+    class NodeSwappingTargetsCreatorTest < WovnMiniTest
+      def test_run
+        html = 'a<a>b</a>c'
+        result = TextScraper.new(Set.new).run(Nokogiri::HTML5(html))
+        nodes_info = [
+            {
+                nodes: result.first[:nodes]
+            }
+        ]
+        NodeSwappingTargetsCreator.new(nodes_info).run!
+        assert_equal(%w[a b c], nodes_info[0][:swapping_targets].map(&:to_s))
+      end
+      def test_run_with_data_stated_by_tag
+        html = '<a>b</a>c'
+        result = TextScraper.new(Set.new).run(Nokogiri::HTML5(html))
+        nodes = result.first[:nodes]
+        nodes_info = [
+            {
+                nodes: nodes
+            }
+        ]
+        NodeSwappingTargetsCreator.new(nodes_info).run!
+        dummy = NodeSwappingTargetsCreator.new('').create_dummy_empty_text_node(next_node: nodes[0])
+        assert_equal(['', 'b', 'c'], nodes_info[0][:swapping_targets].map(&:to_s))
+        text_for_dummy = 'a'
+        dummy.content = text_for_dummy
+        assert_equal(text_for_dummy, nodes[0].previous.to_s)
+      end
+      def test_run_with_data_ended_by_tag
+        html = 'a<a>b</a>'
+        result = TextScraper.new(Set.new).run(Nokogiri::HTML5(html))
+        nodes = result.first[:nodes]
+        nodes_info = [
+            {
+                nodes: nodes
+            }
+        ]
+        NodeSwappingTargetsCreator.new(nodes_info).run!
+        dummy = NodeSwappingTargetsCreator.new('').create_dummy_empty_text_node(previous_node: nodes[-1])
+        assert_equal(['a', 'b', ''], nodes_info[0][:swapping_targets].map(&:to_s))
+        text_for_dummy = 'a'
+        dummy.content = text_for_dummy
+        assert_equal(text_for_dummy, nodes[-1].next.to_s)
+      end
+      def test_run_with_data_with_no_content_inside_tag
+        html = 'a<a></a>c'
+        result = TextScraper.new(Set.new).run(Nokogiri::HTML5(html))
+        nodes = result.first[:nodes]
+        nodes_info = [
+            {
+                nodes: nodes
+            }
+        ]
+        NodeSwappingTargetsCreator.new(nodes_info).run!
+        dummy = NodeSwappingTargetsCreator.new('').create_dummy_empty_text_node(next_node: nodes[2])
+        assert_equal(['a', '', 'c'], nodes_info[0][:swapping_targets].map(&:to_s))
+        text_for_dummy = 'a'
+        dummy.content = text_for_dummy
+        assert_equal(text_for_dummy, nodes[2].previous.to_s)
+      end
+      def test_run_with_data_without_tag
+        html = 'a'
+        result = TextScraper.new(Set.new).run(Nokogiri::HTML5(html))
+        nodes = result.first[:nodes]
+        nodes_info = [
+            {
+                nodes: nodes
+            }
+        ]
+        NodeSwappingTargetsCreator.new(nodes_info).run!
+        assert_equal(['a'], nodes_info[0][:swapping_targets].map(&:to_s))
+      end
+      def test_run_with_data_with_wovn_ignore
+        html = 'a<a wovn-ignore>b</a>c'
+        result = TextScraper.new(Set.new).run(Nokogiri::HTML5(html))
+        nodes_info = [
+            {
+                nodes: result.first[:nodes]
+            }
+        ]
+        NodeSwappingTargetsCreator.new(nodes_info).run!
+        assert_equal(%w[a c], nodes_info[0][:swapping_targets].map(&:to_s))
+      end
+      def test_run_with_data_with_closing_tag
+        html = 'a<br>bc'
+        result = TextScraper.new(Set.new).run(Nokogiri::HTML5(html))
+        nodes_info = [
+            {
+                nodes: result.first[:nodes]
+            }
+        ]
+        NodeSwappingTargetsCreator.new(nodes_info).run!
+        assert_equal(%w[a bc], nodes_info[0][:swapping_targets].map(&:to_s))
+      end
+      def test_run_with_data_with_both_closing_tag_and_no_closing_tag
+        html = 'a<a>b<br>c</a>d'
+        result = TextScraper.new(Set.new).run(Nokogiri::HTML5(html))
+        nodes_info = [
+            {
+                nodes: result.first[:nodes]
+            }
+        ]
+        NodeSwappingTargetsCreator.new(nodes_info).run!
+        assert_equal(%w[a b c d], nodes_info[0][:swapping_targets].map(&:to_s))
+      end
+    end
+  end
+end