Add attributes to start tags

html-parser-api/src/main/java/ru/noties/markwon/html/HtmlTag.java

@@ -4,6 +4,7 @@ import android.support.annotation.NonNull;
 import android.support.annotation.Nullable;
 
 import java.util.List;
+import java.util.Map;
 
 /**
  * @see Inline
@@ -32,6 +33,9 @@ public interface HtmlTag {
      */
     boolean isEmpty();
 
+    @NonNull
+    Map<String, String> attributes();
+
     /**
      * Represents <em>really</em> inline HTML tags (unline commonmark definitions)
      */

html-parser-impl/src/main/java/ru/noties/markwon/html/HtmlTagImpl.java

@@ -5,6 +5,7 @@ import android.support.annotation.Nullable;
 
 import java.util.Collections;
 import java.util.List;
+import java.util.Map;
 
 abstract class HtmlTagImpl implements HtmlTag {
 
@@ -12,11 +13,13 @@ abstract class HtmlTagImpl implements HtmlTag {
 
     final String name;
     final int start;
+    final Map<String, String> attributes;
     int end = NO_VALUE;
 
-    protected HtmlTagImpl(@NonNull String name, int start) {
+    protected HtmlTagImpl(@NonNull String name, int start, @NonNull Map<String, String> attributes) {
         this.name = name;
         this.start = start;
+        this.attributes = attributes;
     }
 
     @NonNull
@@ -40,16 +43,23 @@ abstract class HtmlTagImpl implements HtmlTag {
         return start == end;
     }
 
+    @NonNull
+    @Override
+    public Map<String, String> attributes() {
+        return attributes;
+    }
+
     boolean isClosed() {
         return end > NO_VALUE;
     }
 
     abstract void closeAt(int end);
 
+
     static class InlineImpl extends HtmlTagImpl implements Inline {
 
-        InlineImpl(@NonNull String name, int start) {
+        InlineImpl(@NonNull String name, int start, @NonNull Map<String, String> attributes) {
-            super(name, start);
+            super(name, start, attributes);
         }
 
         @Override
@@ -65,6 +75,7 @@ abstract class HtmlTagImpl implements HtmlTag {
                     "name='" + name + '\'' +
                     ", start=" + start +
                     ", end=" + end +
+                    ", attributes=" + attributes +
                     '}';
         }
     }
@@ -74,20 +85,28 @@ abstract class HtmlTagImpl implements HtmlTag {
         @NonNull
         static BlockImpl root() {
             //noinspection ConstantConditions
-            return new BlockImpl("", 0, null);
+            return new BlockImpl("", 0, null, null);
         }
 
         @NonNull
-        static BlockImpl create(@NonNull String name, int start, @NonNull BlockImpl parent) {
+        static BlockImpl create(
-            return new BlockImpl(name, start, parent);
+                @NonNull String name,
+                int start,
+                @NonNull Map<String, String> attributes,
+                @NonNull BlockImpl parent) {
+            return new BlockImpl(name, start, attributes, parent);
         }
 
         final BlockImpl parent;
         List<BlockImpl> children;
 
         @SuppressWarnings("NullableProblems")
-        BlockImpl(@NonNull String name, int start, @NonNull BlockImpl parent) {
+        BlockImpl(
-            super(name, start);
+                @NonNull String name,
+                int start,
+                @NonNull Map<String, String> attributes,
+                @NonNull BlockImpl parent) {
+            super(name, start, attributes);
             this.parent = parent;
         }
 
@@ -127,12 +146,24 @@ abstract class HtmlTagImpl implements HtmlTag {
             return (List<Block>) (List<? extends Block>) children;
         }
 
+        @NonNull
+        @Override
+        public Map<String, String> attributes() {
+            //noinspection ConstantConditions
+            if (attributes == null) {
+                throw new IllegalStateException("#attributes() getter was called on the root node " +
+                        "which should not be exposed outside internal usage");
+            }
+            return attributes;
+        }
+
         @Override
         public String toString() {
             return "BlockImpl{" +
                     "name='" + name + '\'' +
                     ", start=" + start +
                     ", end=" + end +
+                    ", attributes=" + attributes +
                     ", parent=" + (parent != null ? parent.name : null) +
                     ", children=" + children +
                     '}';

html-parser-impl/src/main/java/ru/noties/markwon/html/MarkwonHtmlParserImpl.java

@@ -7,14 +7,19 @@ import java.io.IOException;
 import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.Collections;
+import java.util.HashMap;
 import java.util.HashSet;
 import java.util.List;
+import java.util.Locale;
+import java.util.Map;
 import java.util.Set;
 
 import ru.noties.markwon.html.HtmlTag.Block;
 import ru.noties.markwon.html.HtmlTag.Inline;
 import ru.noties.markwon.html.HtmlTagImpl.BlockImpl;
 import ru.noties.markwon.html.HtmlTagImpl.InlineImpl;
+import ru.noties.markwon.html.jsoup.nodes.Attribute;
+import ru.noties.markwon.html.jsoup.nodes.Attributes;
 import ru.noties.markwon.html.jsoup.parser.CharacterReader;
 import ru.noties.markwon.html.jsoup.parser.ParseErrorList;
 import ru.noties.markwon.html.jsoup.parser.Token;
@@ -209,7 +214,7 @@ public class MarkwonHtmlParserImpl extends MarkwonHtmlParser {
 
         final String name = startTag.normalName;
 
-        final InlineImpl inline = new InlineImpl(name, output.length());
+        final InlineImpl inline = new InlineImpl(name, output.length(), extractAttributes(startTag));
 
         if (isVoidTag(name)
                 || startTag.selfClosing) {
@@ -271,7 +276,7 @@ public class MarkwonHtmlParserImpl extends MarkwonHtmlParser {
 
         final int start = output.length();
 
-        final BlockImpl block = BlockImpl.create(name, start, currentBlock);
+        final BlockImpl block = BlockImpl.create(name, start, extractAttributes(startTag), currentBlock);
 
         final boolean isVoid = isVoidTag(name) || startTag.selfClosing;
         if (isVoid) {
@@ -396,4 +401,25 @@ public class MarkwonHtmlParserImpl extends MarkwonHtmlParser {
             append(output, "\n");
         }
     }
+
+    @NonNull
+    protected static Map<String, String> extractAttributes(@NonNull Token.StartTag startTag) {
+
+        Map<String, String> map;
+
+        final Attributes attributes = startTag.attributes;
+        final int size = attributes.size();
+
+        if (size > 0) {
+            map = new HashMap<>(size);
+            for (Attribute attribute : attributes) {
+                map.put(attribute.getKey().toLowerCase(Locale.US), attribute.getValue());
+            }
+            map = Collections.unmodifiableMap(map);
+        } else {
+            map = Collections.emptyMap();
+        }
+
+        return map;
+    }
 }

html-parser-impl/src/main/java/ru/noties/markwon/html/jsoup/nodes/Attribute.java

@@ -137,13 +137,13 @@ public class Attribute implements Map.Entry<String, String>, Cloneable  {
 //        return new Attribute(unencodedKey, value, null); // parent will get set when Put
 //    }
 
-    protected boolean isDataAttribute() {
+//    protected boolean isDataAttribute() {
-        return isDataAttribute(key);
+//        return isDataAttribute(key);
-    }
+//    }
-
+//
-    protected static boolean isDataAttribute(String key) {
+//    protected static boolean isDataAttribute(String key) {
-        return key.startsWith(Attributes.dataPrefix) && key.length() > Attributes.dataPrefix.length();
+//        return key.startsWith(Attributes.dataPrefix) && key.length() > Attributes.dataPrefix.length();
-    }
+//    }
 
 //    /**
 //     * Collapsible if it's a boolean attribute and value is empty or same as name

html-parser-impl/src/main/java/ru/noties/markwon/html/jsoup/nodes/Attributes.java

@@ -28,7 +28,7 @@ import static ru.noties.markwon.html.jsoup.helper.Normalizer.lowerCase;
  * @author Jonathan Hedley, jonathan@hedley.net
  */
 public class Attributes implements Iterable<Attribute>, Cloneable {
-    protected static final String dataPrefix = "data-";
+//    protected static final String dataPrefix = "data-";
     private static final int InitialCapacity = 4; // todo - analyze Alexa 1MM sites, determine best setting
 
     // manages the key/val arrays
@@ -282,14 +282,14 @@ public class Attributes implements Iterable<Attribute>, Cloneable {
         return Collections.unmodifiableList(list);
     }
 
-    /**
+//    /**
-     * Retrieves a filtered view of attributes that are HTML5 custom data attributes; that is, attributes with keys
+//     * Retrieves a filtered view of attributes that are HTML5 custom data attributes; that is, attributes with keys
-     * starting with {@code data-}.
+//     * starting with {@code data-}.
-     * @return map of custom data attributes.
+//     * @return map of custom data attributes.
-     */
+//     */
-    public Map<String, String> dataset() {
+//    public Map<String, String> dataset() {
-        return new Dataset(this);
+//        return new Dataset(this);
-    }
+//    }
 
 //    /**
 //     Get the HTML representation of these attributes.
@@ -380,65 +380,65 @@ public class Attributes implements Iterable<Attribute>, Cloneable {
         }
     }
 
-    private static class Dataset extends AbstractMap<String, String> {
+//    private static class Dataset extends AbstractMap<String, String> {
-        private final Attributes attributes;
+//        private final Attributes attributes;
+//
+//        private Dataset(Attributes attributes) {
+//            this.attributes = attributes;
+//        }
+//
+//        @Override
+//        public Set<Entry<String, String>> entrySet() {
+//            return new EntrySet();
+//        }
+//
+//        @Override
+//        public String put(String key, String value) {
+//            String dataKey = dataKey(key);
+//            String oldValue = attributes.hasKey(dataKey) ? attributes.get(dataKey) : null;
+//            attributes.put(dataKey, value);
+//            return oldValue;
+//        }
+//
+//        private class EntrySet extends AbstractSet<Map.Entry<String, String>> {
+//
+//            @Override
+//            public Iterator<Map.Entry<String, String>> iterator() {
+//                return new DatasetIterator();
+//            }
+//
+//            @Override
+//            public int size() {
+//                int count = 0;
+//                Iterator iter = new DatasetIterator();
+//                while (iter.hasNext())
+//                    count++;
+//                return count;
+//            }
+//        }
+//
+//        private class DatasetIterator implements Iterator<Map.Entry<String, String>> {
+//            private Iterator<Attribute> attrIter = attributes.iterator();
+//            private Attribute attr;
+//            public boolean hasNext() {
+//                while (attrIter.hasNext()) {
+//                    attr = attrIter.next();
+//                    if (attr.isDataAttribute()) return true;
+//                }
+//                return false;
+//            }
+//
+//            public Entry<String, String> next() {
+//                return new Attribute(attr.getKey().substring(dataPrefix.length()), attr.getValue());
+//            }
+//
+//            public void remove() {
+//                attributes.remove(attr.getKey());
+//            }
+//        }
+//    }
 
-        private Dataset(Attributes attributes) {
+//    private static String dataKey(String key) {
-            this.attributes = attributes;
+//        return dataPrefix + key;
-        }
+//    }
-
-        @Override
-        public Set<Entry<String, String>> entrySet() {
-            return new EntrySet();
-        }
-
-        @Override
-        public String put(String key, String value) {
-            String dataKey = dataKey(key);
-            String oldValue = attributes.hasKey(dataKey) ? attributes.get(dataKey) : null;
-            attributes.put(dataKey, value);
-            return oldValue;
-        }
-
-        private class EntrySet extends AbstractSet<Map.Entry<String, String>> {
-
-            @Override
-            public Iterator<Map.Entry<String, String>> iterator() {
-                return new DatasetIterator();
-            }
-
-            @Override
-            public int size() {
-                int count = 0;
-                Iterator iter = new DatasetIterator();
-                while (iter.hasNext())
-                    count++;
-                return count;
-            }
-        }
-
-        private class DatasetIterator implements Iterator<Map.Entry<String, String>> {
-            private Iterator<Attribute> attrIter = attributes.iterator();
-            private Attribute attr;
-            public boolean hasNext() {
-                while (attrIter.hasNext()) {
-                    attr = attrIter.next();
-                    if (attr.isDataAttribute()) return true;
-                }
-                return false;
-            }
-
-            public Entry<String, String> next() {
-                return new Attribute(attr.getKey().substring(dataPrefix.length()), attr.getValue());
-            }
-
-            public void remove() {
-                attributes.remove(attr.getKey());
-            }
-        }
-    }
-
-    private static String dataKey(String key) {
-        return dataPrefix + key;
-    }
 }