Json ld in html

common/common.js

@@ -271,6 +271,6 @@ function unComment(doc, content) {
   return content
     .replace(/<!--/, '')
     .replace(/-->/, '')
-    .replace(/< !--/g, '<!--')
-    .replace(/-- >/g, '-->');
+    .replace(/< !\s*-\s*-/g, '<!--')
+    .replace(/-\s*- >/g, '-->');
 }

common/extract-examples.rb

@@ -319,13 +319,18 @@ def save_example(examples:, element:, title:, example_number:, error:, warn:)
           $stdout.write "F".colorize(:red)
           next
         end
+
+        # Get base from document, if present
+        html_base = doc.at_xpath('/html/head/base/@href')
+        ex[:base] = html_base.to_s if html_base
+
         script_content = doc.at_xpath(xpath)
         if script_content
           # Remove (faked) XML comments and unescape sequences
           content = script_content
             .inner_html
-            .sub(/^\s*< !--/, '')
-            .sub(/-- >\s*$/, '')
+            .sub(/^\s*< !\s*-\s*-/, '')
+            .sub(/-\s*- >\s*$/, '')
             .gsub(/&lt;/, '<')
         end
 
@@ -438,7 +443,15 @@ def save_example(examples:, element:, title:, example_number:, error:, warn:)
       # Set argument to referenced content to be parsed
       args[0] = if examples[ex[:result_for]][:ext] == 'html' && method == :expand
         # If we are expanding, and the reference is HTML, find the first script element.
-        doc = Nokogiri::HTML.parse(examples[ex[:result_for]][:content])
+        doc = Nokogiri::HTML.parse(
+          examples[ex[:result_for]][:content]
+          .sub(/^\s*< !\s*-\s*-/, '')
+          .sub(/-\s*- >\s*$/, ''))
+
+        # Get base from document, if present
+        html_base = doc.at_xpath('/html/head/base/@href')
+        options[:base] = html_base.to_s if html_base
+
         script_content = doc.at_xpath(xpath)
         unless script_content
           errors << "Example #{ex[:number]} at line #{ex[:line]} references example #{ex[:result_for].inspect} with no JSON-LD script element"
@@ -447,12 +460,13 @@ def save_example(examples:, element:, title:, example_number:, error:, warn:)
         end
         StringIO.new(script_content
           .inner_html
-          .sub(/^\s*< !--/, '')
-          .sub(/-- >\s*$/, '')
           .gsub(/&lt;/, '<'))
       elsif examples[ex[:result_for]][:ext] == 'html' && ex[:target]
         # Only use the targeted script
-        doc = Nokogiri::HTML.parse(examples[ex[:result_for]][:content])
+        doc = Nokogiri::HTML.parse(
+          examples[ex[:result_for]][:content]
+          .sub(/^\s*< !\s*-\s*-/, '')
+          .sub(/-\s*- >\s*$/, ''))
         script_content = doc.at_xpath(xpath)
         unless script_content
           errors << "Example #{ex[:number]} at line #{ex[:line]} references example #{ex[:result_for].inspect} with no JSON-LD script element"
@@ -461,8 +475,6 @@ def save_example(examples:, element:, title:, example_number:, error:, warn:)
         end
         StringIO.new(script_content
           .to_html
-          .sub(/^\s*< !--/, '')
-          .sub(/-- >\s*$/, '')
           .gsub(/&lt;/, '<'))
       else
         StringIO.new(examples[ex[:result_for]][:content])

tests/compact-manifest.jsonld

@@ -1130,6 +1130,42 @@
       "input": "compact/ep15-in.jsonld",
       "context": "compact/ep15-context.jsonld",
       "expect": "invalid container mapping"
+    }, {
+      "@id": "#th001",
+      "@type": ["jld:PositiveEvaluationTest", "jld:CompactTest"],
+      "name": "Compacts embedded JSON-LD script element",
+      "purpose": "Tests embedded JSON-LD in HTML",
+      "input": "compact/h001-in.html",
+      "context": "compact/h001-context.jsonld",
+      "expect": "compact/h001-out.jsonld",
+      "option": {"specVersion": "json-ld-1.1"}
+    }, {
+      "@id": "#th002",
+      "@type": ["jld:PositiveEvaluationTest", "jld:CompactTest"],
+      "name": "Compacts first embedded JSON-LD script element",
+      "purpose": "Tests embedded JSON-LD in HTML",
+      "input": "compact/h002-in.html",
+      "context": "compact/h002-context.jsonld",
+      "expect": "compact/h002-out.jsonld",
+      "option": {"specVersion": "json-ld-1.1"}
+    }, {
+      "@id": "#th003",
+      "@type": ["jld:PositiveEvaluationTest", "jld:CompactTest"],
+      "name": "Compacts targeted JSON-LD script element",
+      "purpose": "Tests embedded JSON-LD in HTML with fragment identifier",
+      "input": "compact/h003-in.html#second",
+      "context": "compact/h003-context.jsonld",
+      "expect": "compact/h003-out.jsonld",
+      "option": {"specVersion": "json-ld-1.1"}
+    }, {
+      "@id": "#th004",
+      "@type": ["jld:PositiveEvaluationTest", "jld:CompactTest"],
+      "name": "Compacts all embedded JSON-LD script elements with extractAllScripts option",
+      "purpose": "Tests embedded JSON-LD in HTML extracting all elements",
+      "input": "compact/h004-in.html",
+      "context": "compact/h004-context.jsonld",
+      "expect": "compact/h004-out.jsonld",
+      "option": {"specVersion": "json-ld-1.1", "extractAllScripts": true}
    }, {
       "@id": "#tm001",
       "@type": ["jld:PositiveEvaluationTest", "jld:CompactTest"],

tests/compact/h001-context.jsonld

@@ -0,0 +1,5 @@
+{
+  "@context": {
+    "foo": {"@id": "http://example.com/foo", "@container": "@list"}
+  }
+}

tests/compact/h001-in.html

@@ -0,0 +1,12 @@
+<html>
+  <head>
+    <script type="application/ld+json">
+    {
+      "@context": {
+        "foo": {"@id": "http://example.com/foo", "@container": "@list"}
+      },
+      "foo": [{"@value": "bar"}]
+    }
+    </script>
+  </head>
+</html>

tests/compact/h001-out.jsonld

@@ -0,0 +1,6 @@
+{
+  "@context": {
+    "foo": {"@id": "http://example.com/foo", "@container": "@list"}
+  },
+  "foo": ["bar"]
+}

tests/compact/h002-context.jsonld

@@ -0,0 +1,5 @@
+{
+  "@context": {
+    "foo": {"@id": "http://example.com/foo", "@container": "@list"}
+  }
+}

tests/compact/h002-in.html

@@ -0,0 +1,21 @@
+<html>
+  <head>
+    <script type="application/ld+json">
+    {
+      "@context": {
+        "foo": {"@id": "http://example.com/foo", "@container": "@list"}
+      },
+      "foo": [{"@value": "bar"}]
+    }
+    </script>
+    <script type="application/ld+json">
+    {
+      "@context": {"ex": "http://example.com/"},
+      "@graph": [
+        {"ex:foo": {"@value": "foo"}},
+        {"ex:bar": {"@value": "bar"}}
+      ]
+    }
+    </script>
+  </head>
+</html>

tests/compact/h002-out.jsonld

@@ -0,0 +1,6 @@
+{
+  "@context": {
+    "foo": {"@id": "http://example.com/foo", "@container": "@list"}
+  },
+  "foo": ["bar"]
+}

tests/compact/h003-context.jsonld

@@ -0,0 +1,3 @@
+{
+  "@context": {"ex": "http://example.com/"}
+}

tests/compact/h003-in.html

@@ -0,0 +1,21 @@
+<html>
+  <head>
+    <script id="first" type="application/ld+json">
+    {
+      "@context": {
+        "foo": {"@id": "http://example.com/foo", "@container": "@list"}
+      },
+      "foo": [{"@value": "bar"}]
+    }
+    </script>
+    <script id="second" type="application/ld+json">
+    {
+      "@context": {"ex": "http://example.com/"},
+      "@graph": [
+        {"ex:foo": {"@value": "foo"}},
+        {"ex:bar": {"@value": "bar"}}
+      ]
+    }
+    </script>
+  </head>
+</html>

tests/compact/h003-out.jsonld

@@ -0,0 +1,7 @@
+{
+  "@context": {"ex": "http://example.com/"},
+  "@graph": [
+    {"ex:foo": "foo"},
+    {"ex:bar": "bar"}
+  ]
+}

tests/compact/h004-context.jsonld

@@ -0,0 +1,6 @@
+{
+  "@context": {
+    "ex": "http://example.com/",
+    "foo": {"@id": "http://example.com/foo", "@container": "@list"}
+  }
+}

tests/compact/h004-in.html

@@ -0,0 +1,21 @@
+<html>
+  <head>
+    <script type="application/ld+json">
+    {
+      "@context": {
+        "foo": {"@id": "http://example.com/foo", "@container": "@list"}
+      },
+      "foo": [{"@value": "bar"}]
+    }
+    </script>
+    <script type="application/ld+json">
+    {
+      "@context": {"ex": "http://example.com/"},
+      "@graph": [
+        {"ex:foo": {"@value": "foo"}},
+        {"ex:bar": {"@value": "bar"}}
+      ]
+    }
+    </script>
+  </head>
+</html>

tests/compact/h004-out.jsonld

@@ -0,0 +1,15 @@
+{
+  "@context": {
+    "ex": "http://example.com/",
+    "foo": {"@id": "http://example.com/foo", "@container": "@list"}
+  },
+  "@graph": [
+    {"foo": ["bar"]},
+    {
+      "@graph": [
+        {"ex:foo": "foo"},
+        {"ex:bar": "bar"}
+      ]
+    }
+  ]
+}