recognize and parse Schema.org Markup

src/com/dom_distiller/client/SchemaOrgParser.java

Index: src/com/dom_distiller/client/SchemaOrgParser.java
diff --git a/src/com/dom_distiller/client/SchemaOrgParser.java b/src/com/dom_distiller/client/SchemaOrgParser.java
new file mode 100644
index 0000000000000000000000000000000000000000..31b71c807413473574a790eb38f6577a34efdd3f
--- /dev/null
+++ b/src/com/dom_distiller/client/SchemaOrgParser.java
@@ -0,0 +1,557 @@
+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+package com.dom_distiller.client;
+
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import com.google.gwt.dom.client.AnchorElement;
+import com.google.gwt.dom.client.Element;
+import com.google.gwt.dom.client.ImageElement;
+import com.google.gwt.dom.client.MetaElement;
+import com.google.gwt.dom.client.Node;
+import com.google.gwt.dom.client.NodeList;
+
+/**
+ * This class recognizes and parses schema.org markup tags, and returns the properties that matter
+ * to distilled content.
+ * Schema.org markup (http://schema.org) is based on the microdata format
+ * (http://www.whatwg.org/specs/web-apps/current-work/multipage/microdata.html).
+ * For the basic Schema.org Thing type, the basic properties are: name, url, description, image.
+ * In addition, for each type that we support, we also parse more specific properties:
+ * - Article: headline (i.e. title), publisher, copyright year, copyright holder, date published,
+ *            date modified, author, article section
+ * - ImageObject: headline (i.e. title), publisher, copyright year, copyright holder, content url,
+ *                encoding format, caption, representative of page, width, height
+ * - Person: family name, given name
+ * - Organization: legal name.
+ * The value of a Schema.Org property can be a Schema.Org type, i.e. embedded.  E.g., the author or
+ * publisher of article or publisher of image could be a Schema.Org Person or Organization type;
+ * in fact, this is the reason we support Person and Organization types.
+ */
+public class SchemaOrgParser implements MarkupParser.Parser {

[cjhopman, 2014/04/21 16:52:22]

Can we split this class into two parts:

1. A SchemaOrgParser (that is basically all the the parse() step and all that it
does)
2. A MarkupParser.Parser that has all the logic for the getXxx() calls and uses
(1)

[kuan, 2014/04/23 15:32:36]

Done.

+    private static final String NAME_PROP = "name";
+    private static final String URL_PROP = "url";
+    private static final String DESCRIPTION_PROP = "description";
+    private static final String IMAGE_PROP = "image";
+    private static final String HEADLINE_PROP = "headline";
+    private static final String PUBLISHER_PROP = "publisher";
+    private static final String COPYRIGHT_HOLDER_PROP = "copyrightHolder";
+    private static final String COPYRIGHT_YEAR_PROP = "copyrightYear";
+    private static final String CONTENT_URL_PROP = "contentUrl";
+    private static final String ENCODING_FORMAT_PROP = "encodingFormat";
+    private static final String CAPTION_PROP = "caption";
+    private static final String REPRESENTATIVE_PROP = "representativeOfPage";
+    private static final String WIDTH_PROP = "width";
+    private static final String HEIGHT_PROP = "height";
+    private static final String DATE_PUBLISHED_PROP = "datePublished";
+    private static final String DATE_MODIFIED_PROP = "dateModified";
+    private static final String AUTHOR_PROP = "author";
+    private static final String SECTION_PROP = "articleSection";
+    private static final String FAMILY_NAME_PROP = "familyName";
+    private static final String GIVEN_NAME_PROP = "givenName";
+    private static final String LEGAL_NAME_PROP = "legalName";
+    private static final String AUTHOR_REL = "author";
+
+    private enum Type {  // All these types are extended from Thing, directly or indirectly.
+        IMAGE,
+        ARTICLE,
+        PERSON,
+        ORGANIZATION,
+        UNSUPPORTED,
+    }
+
+    private static class ThingItem {
+        private final Type mType;
+        private final String[] mStringPropertyNames;
+        private final String[] mItemPropertyNames;
+        private final String[] mStringProperties;
+        private final ThingItem[] mItemProperties;
+
+        ThingItem(Type type, String[] stringPropertyNames, String[] itemPropertyNames) {
+            mType = type;
+            mStringPropertyNames = stringPropertyNames;
+            mItemPropertyNames = itemPropertyNames;
+            mStringProperties = new String[mStringPropertyNames.length];
+            mItemProperties = new ThingItem[mItemPropertyNames.length];
+        }
+
+        String toStringProperty() {
+            return "";
+        }
+
+        MarkupParser.Image getImage() {
+            // Use value of IMAGE_PROP to create a MarkupParser.Image.
+            String imageUrl = getStringProperty(IMAGE_PROP);
+            if (imageUrl.isEmpty()) return null;
+            MarkupParser.Image image = new MarkupParser.Image();
+            image.image = imageUrl;
+            image.url = imageUrl;
+            return image;
+        }
+
+        MarkupParser.Article getArticle() {
+            return null;
+        }
+
+        final boolean isSupported() { return mType != Type.UNSUPPORTED; }
+
+        final boolean isImageRepresentativeOfPage() {
+            String value = getStringProperty(REPRESENTATIVE_PROP);
+            return value.equalsIgnoreCase("true");
+        }
+
+        // Store |value| for property with |name|, unless the property already has a non-empty
+        // value, in which case, |value| will be ignored.  This means we only keep the first value.
+        final void putStringValue(String name, String value) {
+            for (int i = 0; i < mStringPropertyNames.length; i++) {
+                if (name.equals(mStringPropertyNames[i])) {
+                    String existing = mStringProperties[i];
+                    if (existing == null || existing.isEmpty()) mStringProperties[i] = value;
+                    break;
+                }
+            }
+        }
+
+        // Store |value| for property with |name|, unless the property already has a non-null value,
+        // in which case, |value| will be ignored.  This means we only keep the first value.
+        final void putItemValue(String name, ThingItem value) {
+            for (int i = 0; i < mItemPropertyNames.length; i++) {
+                if (name.equals(mItemPropertyNames[i])) {
+                    if (mItemProperties[i] == null) mItemProperties[i] = value;
+                    break;
+                }
+            }
+        }
+
+        final String getStringProperty(String name) {
+            // Check if property exists in |mStringProperties|.
+            for (int i = 0; i < mStringPropertyNames.length; i++) {
+                if (name.equals(mStringPropertyNames[i])) {
+                    String value = mStringProperties[i];
+                    if (value != null && !value.isEmpty()) return value;
+                    break;
+                }
+            }
+            // Otherwise, repeat for |mItemProperties|.
+            for (int i = 0; i < mItemPropertyNames.length; i++) {
+                if (!name.equals(mItemPropertyNames[i])) continue;
+                if (mItemProperties[i] != null) return mItemProperties[i].toStringProperty();
+                break;
+            }
+            return "";
+        }
+    }
+
+    private final List<ThingItem> mItemScopes = new ArrayList<ThingItem>();
+    private String mAuthorFromRel = "";
+    private static final Map<String, Type> sTypeUrls;
+    private static final Map<String, String[]> sTagAttributesMap;
+    private static final String[] sEmptyPropertyNames = {
+        // Intentionally empty, declared so that it's initialized statically.
+    };
+
+    static {
+        sTypeUrls = new HashMap<String, Type>();
+        sTypeUrls.put("http://schema.org/ImageObject", Type.IMAGE);
+        sTypeUrls.put("http://schema.org/Article", Type.ARTICLE);

[cjhopman, 2014/04/21 16:52:22]

We should probably recognize schema.org/NewsArticle and /BlobPosting as articles
too (and maybe all the other subtypes of article)

[kuan, 2014/04/23 15:32:36]

Done.

+        sTypeUrls.put("http://schema.org/Person", Type.PERSON);
+        sTypeUrls.put("http://schema.org/Organization", Type.ORGANIZATION);

[cjhopman, 2014/04/21 16:52:22]

There are a whole bunch of subtypes of Organization that we will ignore (though
a lot of them we probably don't care about).

[kuan, 2014/04/23 15:32:36]

i added more subtypes, but i'm not sure if they're enough or excess.

+
+        // The key for |sTagAttributesMap| is the tag name, while the entry value is an array of
+        // attributes in the specified tag from which to extract information:
+        // - 0th attribute: contains the value for the property specified in itemprop
+        // - 1st attribute: if available, contains the value for the author property.
+        sTagAttributesMap = new HashMap<String, String[]>();
+        sTagAttributesMap.put("IMG", new String[] { "SRC" });
+        sTagAttributesMap.put("AUDIO", new String[] { "SRC" });
+        sTagAttributesMap.put("EMBED", new String[] { "SRC" });
+        sTagAttributesMap.put("IFRAME", new String[] { "SRC" });
+        sTagAttributesMap.put("SOURCE", new String[] { "SRC" });
+        sTagAttributesMap.put("TRACK", new String[] { "SRC" });
+        sTagAttributesMap.put("VIDEO", new String[] { "SRC" });
+        sTagAttributesMap.put("A", new String[] { "HREF", "REL" });
+        sTagAttributesMap.put("LINK", new String[] { "HREF", "REL" });
+        sTagAttributesMap.put("AREA", new String[] { "HREF" });
+        sTagAttributesMap.put("META", new String[] { "CONTENT" });
+        sTagAttributesMap.put("TIME", new String[] { "DATETIME" });
+        sTagAttributesMap.put("OBJECT", new String[] { "DATA" });
+        sTagAttributesMap.put("DATA", new String[] { "VALUE" });
+        sTagAttributesMap.put("METER", new String[] { "VALUE" });
+    }
+
+    /**
+     * The object that extracts and verifies Schema.org markup tags from |root|.
+     */
+    public SchemaOrgParser(Element root) {
+        // TODO(kuan): Parsing all tags is pretty expensive, should we do so only lazily?
+        // If parse lazily, all get* methods will need to check for parsed state and, if necessary,
+        // parse before returning the requested properties.
+        // Note that the <html> element can also be the start of a Schema.org item, and hence needs
+        // to be parsed.
+        parse(root, null);
+    }
+
+    @Override
+    public String getTitle() {
+        String title = findStringProperty(HEADLINE_PROP);
+        if (title.isEmpty()) title = findStringProperty(NAME_PROP);

[cjhopman, 2014/04/21 16:52:22]

This seems to mean that we will get the first item with a name. I really don't
think we want to return the name of a person or image or organization as the
title. I think we should probably be getting the title only from the article
item.

[kuan, 2014/04/23 15:32:36]

Done.

+        return title;
+    }
+
+    @Override
+    public String getType() {
+        if (mItemScopes.isEmpty()) return "";
+        // Assume the type of the first item is the page type.
+        return mItemScopes.get(0).mType.toString();

[cjhopman, 2014/04/21 16:52:22]

Does schema.org specify that this is a good way to get a type of a page? I don't
think that they really have this concept and picking the first item doesn't seem
correct since it's just picking the marked up thing that happens to appear first
in the html. 

I think we either just have to always return nothing or maybe return some sort
of article type if there is an article item.

[kuan, 2014/04/23 15:32:36]

Done.  for the record, as per our off-line discussion: after this cl, i intend
to consolidate and standardize the types returned by the 3 parsers in
MarkupParser, added todo.

+    }
+
+    @Override
+    public String getUrl() {
+        return findStringProperty(URL_PROP);

[cjhopman, 2014/04/21 16:52:22]

I don't really understand what we expect the url from getUrl() to be.

I think the only thing that makes sense for this would be for it to be the url
of an article item (not a person/organization/etc).

[kuan, 2014/04/23 15:32:36]

Done.

+    }
+
+    @Override
+    public MarkupParser.Image[] getImages() {

[cjhopman, 2014/04/21 16:52:22]

We should be careful about what this returns, we only want to include images
that are part of the article. For example, we don't want this list to include a
picture of the author or organization.

The images that we know are part of the article would include:
the article item's image property
all image items with the article as their associatedArticle property

Things that we could probably assume are part of the article:
all image items

[kuan, 2014/04/23 15:32:36]

Done.

per our discussion off-line, i won't impl the associatedArticle property yet; it
requires pretty complicated parsing and resolution of itemref attribute.  so we
have the article item's image property and all the ImageObject's in this cl.

however, i did impl the "associatedMedia" and "encoding" properties of article,
which take precedence over the article's "image" property and forces ImageItem's
representativeOfPage property to true.  also added test.

+        if (mItemScopes.isEmpty()) return null;
+        List<MarkupParser.Image> images = new ArrayList<MarkupParser.Image>();
+        for (int i = 0; i < mItemScopes.size(); i++) {
+            ThingItem item = mItemScopes.get(i);
+            MarkupParser.Image image = item.getImage();
+            if (image != null) {
+                if (item.isImageRepresentativeOfPage()) {
+                    // Image should be the dominant, i.e. first, one.
+                    images.add(0, image);
+                } else {
+                    images.add(image);
+                }
+            }
+        }
+        if (images.isEmpty()) return null;
+        return images.toArray(new MarkupParser.Image[images.size()]);
+    }
+
+    @Override
+    public String getDescription() {
+        return findStringProperty(DESCRIPTION_PROP);

[cjhopman, 2014/04/21 16:52:22]

Again, this should probably only be the description from an article item. Same
for author and copyright.

[kuan, 2014/04/23 15:32:36]

Done.

+    }
+
+    @Override
+    public String getPublisher() {
+        return findStringProperty(PUBLISHER_PROP);

[cjhopman, 2014/04/21 16:52:22]

I think we would only want the publisher property from an article, not other
types (maybe we may also want to look at the copyrightHolder or other properties
of article)

[kuan, 2014/04/23 15:32:36]

Done.

+    }
+
+    @Override
+    public String getCopyright() {
+        if (mItemScopes.isEmpty()) return "";
+        // Returns a concatenated string of copyright year and copyright holder of the first item
+        // that has these properties, delimited by a whitespace.
+        String copyright = "";
+        for (int i = 0; i < mItemScopes.size() && copyright.isEmpty(); i++) {
+            ThingItem item = mItemScopes.get(i);
+            copyright = concat(item.getStringProperty(COPYRIGHT_YEAR_PROP),
+                               item.getStringProperty(COPYRIGHT_HOLDER_PROP));
+        }
+        return copyright.isEmpty() ? copyright : "Copyright " + copyright;
+    }
+
+    @Override
+    public String getAuthor() {
+        String author = findStringProperty(AUTHOR_PROP);
+        return author.isEmpty() ? mAuthorFromRel : author;
+    }
+
+    @Override
+    public MarkupParser.Article getArticle() {
+        if (mItemScopes.isEmpty()) return null;
+        // Returns the first article.
+        MarkupParser.Article article = null;
+        for (int i = 0; i < mItemScopes.size() && article == null; i++) {
+            article = mItemScopes.get(i).getArticle();
+        }
+        return article;
+    }
+
+    @Override
+    public boolean optOut() {
+        return false;
+    }
+
+    private void parse(Element e, ThingItem parentItem) {
+        ThingItem newItem = null;
+        boolean isItemScope = isItemScope(e);
+        // A non-null |parentItem| means we're currently parsing the elements for a schema.org type.
+        String[] propertyNames = parentItem != null ? getItemProp(e) : new String[0];
+
+        if (isItemScope) {
+            // The "itemscope" and "itemtype" attributes of |e| indicate the start of an item.
+            // Create the corresponding extended-ThingItem, and add it to the list if:
+            // 1) its type is supported, and
+            // 2) if the parent is an unsupported type, it's not an "itemprop" attribute of the
+            //    parent, based on the rule that an item is a top-level item if its element doesn't
+            //    have an itemprop attribute.
+            newItem = createItemForElement(e);
+            if (newItem != null && newItem.isSupported() &&
+                (parentItem == null || parentItem.isSupported() || propertyNames.length == 0)) {
+                mItemScopes.add(newItem);
+            }
+        }
+
+        // If parent is a supported type, parse the element for >= 1 properties in "itemprop"
+        // attribute.
+        if (propertyNames.length > 0 && parentItem.isSupported() &&
+            (newItem == null || newItem.isSupported())) {
+            for (int i = 0; i < propertyNames.length; i++) {
+                // If a new item was created above, the property value of this "itemprop" attribute
+                // is an embedded item, so add it to the parent item.
+                if (newItem != null) {
+                    parentItem.putItemValue(propertyNames[i], newItem);
+                } else {
+                   // Otherwise, extract the property value from the tag itself, and add it to the
+                   // parent item.
+                   parentItem.putStringValue(propertyNames[i], getPropertyValue(e));
+                }
+            }
+        }
+
+        // If <a> or <link> tags specify rel="author", extract it.

[cjhopman, 2014/04/25 20:52:34]

Where does this rel="author" stuff come from? I can't find anything about it on
schema.org or in the Microdata Dom API spec.

+        if (mAuthorFromRel.isEmpty()) mAuthorFromRel = getAuthorFromRelAttribute(e);
+
+        // Now, parse each child element recursively.
+        NodeList<Node> children = e.getChildNodes();
+        for (int i = 0; i < children.getLength(); i++) {
+            Node child = children.getItem(i);
+            if (child.getNodeType() != Node.ELEMENT_NODE) continue;
+            parse(Element.as(child), newItem != null ? newItem : parentItem);
+        }
+    }
+
+    private Type getItemType(Element e) {
+        // "itemtype" attribute is case-sensitive.
+        String type = e.getAttribute("ITEMTYPE");
+        return sTypeUrls.containsKey(type) ? sTypeUrls.get(type) : Type.UNSUPPORTED;
+    }
+
+    private ThingItem createItemForElement(Element e) {
+        ThingItem newItem = null;
+        Type type = getItemType(e);
+        switch (type) {
+            case IMAGE:
+                newItem = new ImageItem();
+                break;
+            case ARTICLE:
+                newItem = new ArticleItem();
+                break;
+            case PERSON:
+                newItem = new PersonItem();
+                break;
+            case ORGANIZATION:
+                newItem = new OrganizationItem();
+                break;
+            case UNSUPPORTED:
+                newItem = new UnsupportedItem();
+                break;
+            default:
+                return null;
+        }
+        return newItem;
+    }
+
+    // Returns the first item that has the requested property value.
+    private String findStringProperty(String name) {
+        if (mItemScopes.isEmpty()) return "";
+        for (int i = 0; i < mItemScopes.size(); i++) {
+            String value = mItemScopes.get(i).getStringProperty(name);
+            if (!value.isEmpty()) return value;
+        }
+        return "";
+    }
+
+    private static class ImageItem extends ThingItem {
+        private static final String[] sStringPropertyNames = {
+            NAME_PROP,
+            URL_PROP,
+            DESCRIPTION_PROP,
+            IMAGE_PROP,
+            HEADLINE_PROP,
+            PUBLISHER_PROP,
+            COPYRIGHT_HOLDER_PROP,
+            COPYRIGHT_YEAR_PROP,
+            CONTENT_URL_PROP,
+            ENCODING_FORMAT_PROP,
+            CAPTION_PROP,
+            REPRESENTATIVE_PROP,
+            WIDTH_PROP,
+            HEIGHT_PROP,
+        };
+
+        private static final String[] sItemPropertyNames = {
+            PUBLISHER_PROP,
+            COPYRIGHT_HOLDER_PROP,
+        };
+
+        ImageItem() {
+            super(Type.IMAGE, sStringPropertyNames, sItemPropertyNames);
+        }
+
+        @Override
+        MarkupParser.Image getImage() {
+            MarkupParser.Image image = new MarkupParser.Image();
+            String url = getStringProperty(CONTENT_URL_PROP);
+            image.image = !url.isEmpty() ? url : getStringProperty(NAME_PROP);
+            image.url = image.image;
+            image.type = getStringProperty(ENCODING_FORMAT_PROP);
+            image.caption = getStringProperty(CAPTION_PROP);
+            try {
+                image.width = Integer.parseInt(getStringProperty(WIDTH_PROP), 10);
+            } catch (Exception e) {
+            }
+            try {
+                image.height = Integer.parseInt(getStringProperty(HEIGHT_PROP), 10);
+            } catch (Exception e) {
+            }
+            return image;
+        }
+    }
+
+    private static class ArticleItem extends ThingItem {
+        private static final String[] sStringPropertyNames = {
+            NAME_PROP,
+            URL_PROP,
+            DESCRIPTION_PROP,
+            IMAGE_PROP,
+            HEADLINE_PROP,
+            PUBLISHER_PROP,
+            COPYRIGHT_HOLDER_PROP,
+            COPYRIGHT_YEAR_PROP,
+            DATE_MODIFIED_PROP,
+            DATE_PUBLISHED_PROP,
+            AUTHOR_PROP,
+            SECTION_PROP,
+        };
+
+        private static final String[] sItemPropertyNames = {
+            PUBLISHER_PROP,
+            COPYRIGHT_HOLDER_PROP,
+            AUTHOR_PROP,
+        };
+
+        ArticleItem() {
+            super(Type.ARTICLE, sStringPropertyNames, sItemPropertyNames);
+        }
+
+        @Override
+        MarkupParser.Article getArticle() {
+            MarkupParser.Article article = new MarkupParser.Article();
+            article.publishedTime = getStringProperty(DATE_PUBLISHED_PROP);
+            article.modifiedTime = getStringProperty(DATE_MODIFIED_PROP);
+            article.section = getStringProperty(SECTION_PROP);
+            String author = getStringProperty(AUTHOR_PROP);
+            article.authors = author.isEmpty() ? new String[0] : new String[] { author };
+            return article;
+        }
+    }
+
+    private static class PersonItem extends ThingItem {
+        private static final String[] sStringPropertyNames = {
+            NAME_PROP,
+            URL_PROP,
+            DESCRIPTION_PROP,
+            IMAGE_PROP,
+            FAMILY_NAME_PROP,
+            GIVEN_NAME_PROP,
+        };
+
+        PersonItem() {
+            super(Type.PERSON, sStringPropertyNames, sEmptyPropertyNames);
+        }
+
+        // Returns either the value of NAME_PROP, or concatenated values of GIVEN_NAME_PROP and
+        // FAILY_NAME_PROP delimited by a whitespace.

[cjhopman, 2014/04/21 16:52:22]

s/FAILY/FAMILY

[kuan, 2014/04/23 15:32:36]

Done.

+        @Override
+        String toStringProperty() {
+            String fullname = getStringProperty(NAME_PROP);
+            if (fullname.isEmpty()) {
+                fullname = concat(getStringProperty(GIVEN_NAME_PROP),
+                                  getStringProperty(FAMILY_NAME_PROP));
+            }
+            return fullname;
+        }
+    }
+
+    private static class OrganizationItem extends ThingItem {
+        private static final String[] sStringPropertyNames = {
+            NAME_PROP,
+            URL_PROP,
+            DESCRIPTION_PROP,
+            IMAGE_PROP,
+            LEGAL_NAME_PROP,
+        };
+
+        OrganizationItem() {
+            super(Type.ORGANIZATION, sStringPropertyNames, sEmptyPropertyNames);
+        }
+
+        // Returns either the value of NAME_PROP or LEGAL_NAME_PROP.
+        @Override
+        String toStringProperty() {
+            String name = getStringProperty(NAME_PROP);
+            if (name.isEmpty()) name = getStringProperty(LEGAL_NAME_PROP);
+            return name;
+        }
+    }
+
+    private static class UnsupportedItem extends ThingItem {
+        UnsupportedItem(){
+            super(Type.UNSUPPORTED, sEmptyPropertyNames, sEmptyPropertyNames);
+        }
+    }
+
+    private static boolean isItemScope(Element e) {
+        return e.hasAttribute("ITEMSCOPE") && e.hasAttribute("ITEMTYPE");
+    }
+
+    private static String[] getItemProp(Element e) {
+        // "itemprop" attribute is case-sensitive, and can have multiple properties.
+        String itemprop = e.getAttribute("ITEMPROP");
+        if (itemprop.isEmpty()) return new String[0];
+        String[] splits = StringUtil.split(itemprop, "\\s+");
+        return splits.length > 0 ? splits : new String[] { itemprop };
+    }
+
+    // Extracts the property value from |e|.  For some tags, the value is a specific attribute,
+    // while for others, it's the text between the start and end tags.
+    private static String getPropertyValue(Element e) {
+        String value = "";
+        String tagName = e.getTagName();
+        if (sTagAttributesMap.containsKey(tagName)) {
+            value = e.getAttribute(sTagAttributesMap.get(tagName)[0]);
+        }
+        if (value.isEmpty()) value = e.getInnerText();
+        return value;
+    }
+
+    // Extracts the author property from |e|'s "rel=author" attribute.
+    private static String getAuthorFromRelAttribute(Element e) {
+        String author = "";
+        String tagName = e.getTagName();
+        if (sTagAttributesMap.containsKey(tagName)) {
+            String[] attrs = sTagAttributesMap.get(tagName);
+            if (attrs.length > 1 && e.getAttribute(attrs[1]).equals(AUTHOR_REL)) {
+                author = e.getInnerText();
+            }
+        }
+        return author;
+    }
+
+    private static String concat(String first, String second) {
+        String concat = first;
+        if (!concat.isEmpty() && !second.isEmpty()) concat += " ";
+        concat += second;
+        return concat;
+    }
+}