Coverage Report

Coverage Report - org.crosswire.jsword.book.filter.thml.CustomHandler

Classes in this File

Line Coverage

Branch Coverage

Complexity

CustomHandler

0/67

0/36

5.167

 /**
  * Distribution License:
  * JSword is free software; you can redistribute it and/or modify it under
  * the terms of the GNU Lesser General Public License, version 2.1 or later
  * as published by the Free Software Foundation. This program is distributed
  * in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even
  * the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
  * See the GNU Lesser General Public License for more details.
  *
  * The License is available on the internet at:
  *      http://www.gnu.org/copyleft/lgpl.html
  * or by writing to:
  *      Free Software Foundation, Inc.
  *      59 Temple Place - Suite 330
  *      Boston, MA 02111-1307, USA
  *
  * © CrossWire Bible Society, 2005 - 2016
  *
  */
 package org.crosswire.jsword.book.filter.thml;
 
 import java.util.HashMap;
 import java.util.LinkedList;
 import java.util.Locale;
 import java.util.Map;
 
 import org.crosswire.jsword.book.Book;
 import org.crosswire.jsword.book.DataPolice;
 import org.crosswire.jsword.passage.Key;
 import org.jdom2.Content;
 import org.jdom2.Element;
 import org.jdom2.Text;
 import org.xml.sax.Attributes;
 import org.xml.sax.SAXException;
 import org.xml.sax.helpers.DefaultHandler;
 
 /**
  * To convert SAX events into OSIS events.
  * 
  * <p>
  * This is based upon the THML reference page:
  * <a href="http://www.ccel.org/ThML/ThML1.04.htm">http://www.ccel.org/ThML/ThML1.04.htm</a>
  * to work out what the tags meant.
  * 
  * @see gnu.lgpl.License The GNU Lesser General Public License for details.
  * @author Joe Walker
  */
 public class CustomHandler extends DefaultHandler {
     /**
      * Simple ctor
      * 
      * @param book the book
      * @param key the key
      */
     public CustomHandler(Book book, Key key) {
         this.book = book;
         this.key = key;
         this.stack = new LinkedList<Content>();
     }
 
     /* (non-Javadoc)
      * @see org.xml.sax.helpers.DefaultHandler#startElement(java.lang.String, java.lang.String, java.lang.String, org.xml.sax.Attributes)
      */
     @Override
     public void startElement(String uri, String localname, String qname, Attributes attrs) throws SAXException {
         Element ele = null;
 
         // If we are looking at the root element
         // then the stack is empty
         if (!stack.isEmpty()) {
             Object top = stack.getFirst();
 
             // If the element and its descendants are to be ignored
             // then there is a null element on the stack
             if (top == null) {
                 return;
             }
 
             // It might be a text element
             if (top instanceof Element) {
                 ele = (Element) top;
             }
         }
 
         Tag t = getTag(localname, qname);
 
         if (t != null) {
             stack.addFirst(t.processTag(book, key, ele, attrs));
         }
     }
 
     /* (non-Javadoc)
      * @see org.xml.sax.helpers.DefaultHandler#characters(char[], int, int)
      */
     @Override
     public void characters(char[] data, int offset, int length) {
         // what we are adding
         String text = new String(data, offset, length);
 
         if (stack.isEmpty()) {
             stack.addFirst(new Text(text));
             return;
         }
 
         // What we are adding to
         Content top = stack.getFirst();
 
         // If the element and its descendants are to be ignored
         // then there is a null element on the stack
         if (top == null) {
             return;
         }
 
         if (top instanceof Text) {
             ((Text) top).append(text);
             return;
         }
 
         if (top instanceof Element) {
             Element current = (Element) top;
 
             int size = current.getContentSize();
 
             // If the last element in the list is a string then we should add
             // this string on to the end of it rather than add a new list item
             // because (probably as an artifact of the HTML/XSL transform we get
             // a space inserted in the output even when 2 calls to this method
             // split a word.
             if (size > 0) {
                 Content last = current.getContent(size - 1);
                 if (last instanceof Text) {
                     ((Text) last).append(text);
                     return;
                 }
             }
             current.addContent(new Text(text));
         }
     }
 
     /* (non-Javadoc)
      * @see org.xml.sax.helpers.DefaultHandler#endElement(java.lang.String, java.lang.String, java.lang.String)
      */
     @Override
     public void endElement(String uri, String localname, String qname) {
         if (stack.isEmpty()) {
             return;
         }
         // When we are done processing an element we need to remove
         // it from the stack so that nothing more is attached to it.
         Content top = stack.removeFirst();
         if (top instanceof Element) {
             Element finished = (Element) top;
             Tag t = getTag(localname, qname);
 
             if (t != null) {
                 t.processContent(book, key, finished);
             }
 
             // If it was the last element then it was the root element
             // so save it
             if (stack.isEmpty()) {
                 rootElement = finished;
             }
         }
     }
 
     /**
      * @return the root element
      */
     public Element getRootElement() {
         return rootElement;
     }
 
     private Tag getTag(String localname, String qname) {
         // sometimes qname is empty e.g. on Android 2.1
         String name;
         if (qname != null && qname.length() > 0) {
             name = qname;
         } else {
             name = localname;
         }
 
         Tag t = TAG_MAP.get(name);
 
         // Some of the THML books are broken in that they use uppercase
         // element names, which the spec disallows, but we might as well
         // look out for them
         if (t == null) {
             t = TAG_MAP.get(name.toLowerCase(Locale.ENGLISH));
 
             if (t == null) {
                 DataPolice.report(book, key, "Unknown thml element: " + localname + " qname=" + name);
 
                 // Report on it only once and make sure the content is output.
                 t = new AnonymousTag(name);
                 TAG_MAP.put(name, t);
                 return t;
             }
 
             // DataPolice.report(book, key, "Wrong case used in thml element: " + name);
         }
         return t;
     }
 
     /**
      * When the document is parsed, this is the last element popped off the
      * stack.
      */
     private Element rootElement;
 
     /**
      * The book being parsed.
      */
     private Book book;
 
     /**
      * The book being parsed.
      */
     private Key key;
 
     /**
      * The stack of elements that we have created
      */
     private LinkedList<Content> stack;
 
     /**
      * The known tag types
      */
     private static final Map<String, Tag> TAG_MAP = new HashMap<String, Tag>();
 
     static {
         /*
          * ThML is based upon Voyager XHTML and all Voyager elements are
          * allowed. However not all elements make sense.
          */
         Tag[] tags = new Tag[] {
                 // The following are defined in Voyager xhtml 4.0
                 new ATag(),
                 new AbbrTag(),
                 new AliasTag("acronym", new AbbrTag()),
                 new AnonymousTag("address"),
                 new SkipTag("applet"),
                 new SkipTag("area"),
                 new BTag(),
                 new SkipTag("base"),
                 new SkipTag("basefont"),
                 new IgnoreTag("bdo"),
                 new BigTag(),
                 new BlockquoteTag(),
                 new IgnoreTag("body"),
                 new BrTag(),
                 new SkipTag("button"),
                 new AnonymousTag("caption"),
                 new CenterTag(),
                 new AnonymousTag("cite"),
                 new AnonymousTag("code"),
                 new SkipTag("col"),
                 new SkipTag("colgroup"),
                 new AliasTag("dd", new LiTag()),
                 new AnonymousTag("del"),
                 new AnonymousTag("dfn"),
                 new DivTag(),
                 new AliasTag("dl", new UlTag()),
                 new AliasTag("dt", new LiTag()),
                 new AliasTag("em", new ITag()),
                 new IgnoreTag("fieldset"),
                 new FontTag(),
                 new SkipTag("form"),
                 new SkipTag("frame"),
                 new SkipTag("frameset"),
                 new AliasTag("h1", new HTag(1)),
                 new AliasTag("h2", new HTag(2)),
                 new AliasTag("h3", new HTag(3)),
                 new AliasTag("h4", new HTag(4)),
                 new AliasTag("h5", new HTag(5)),
                 new AliasTag("h6", new HTag(6)),
                 new SkipTag("head"),
                 new HrTag(),
                 new IgnoreTag("html"),
                 new IgnoreTag("frameset"),
                 new ITag(),
                 new SkipTag("iframe"),
                 new ImgTag(),
                 new SkipTag("input"),
                 new AnonymousTag("ins"),
                 new AnonymousTag("kbd"),
                 new AnonymousTag("label"),
                 new AnonymousTag("legend"),
                 new LiTag(),
                 new SkipTag("link"),
                 new SkipTag("map"),
                 new SkipTag("meta"),
                 new SkipTag("noscript"),
                 new SkipTag("object"),
                 new OlTag(),
                 new SkipTag("optgroup"),
                 new SkipTag("option"),
                 new PTag(),
                 new SkipTag("param"),
                 new IgnoreTag("pre"),
                 new QTag(),
                 new RootTag(),
                 new STag(),
                 new AnonymousTag("samp"),
                 new SkipTag("script"),
                 new SkipTag("select"),
                 new SmallTag(),
                 new IgnoreTag("span"),
                 new AliasTag("strong", new BTag()),
                 new SkipTag("style"),
                 new SubTag(),
                 new SupTag(),
                 new SyncTag(),
                 new TableTag(),
                 new IgnoreTag("tbody"),
                 new TdTag(),
                 new IgnoreTag("tfoot"),
                 new SkipTag("textarea"),
                 new SkipTag("title"),
                 new IgnoreTag("thead"),
                 new ThTag(),
                 new TrTag(),
                 new TtTag(),
                 new UTag(),
                 new UlTag(),
                 new AnonymousTag("var"),
 
                 // ThML adds the following to Voyager
                 // Note: hymn.mod is not here nor are additional head&DC
                 // elements
                 new AnonymousTag("added"),
                 new AnonymousTag("attr"),
                 new AnonymousTag("argument"),
                 new CitationTag(),
                 new AnonymousTag("date"),
                 new AnonymousTag("deleted"),
                 new AnonymousTag("def"),
                 new AliasTag("div1", new DivTag(1)),
                 new AliasTag("div2", new DivTag(2)),
                 new AliasTag("div3", new DivTag(3)),
                 new AliasTag("div4", new DivTag(4)),
                 new AliasTag("div5", new DivTag(5)),
                 new AliasTag("div6", new DivTag(6)),
                 new ForeignTag(),
                 new AnonymousTag("index"),
                 new AnonymousTag("insertIndex"),
                 new AnonymousTag("glossary"),
                 new NoteTag(),
                 new NameTag(),
                 new PbTag(),
                 new AnonymousTag("scripCom"),
                 new AnonymousTag("scripContext"),
                 new ScripRefTag(),
                 new ScriptureTag(),
                 new TermTag(),
                 new AnonymousTag("unclear"),
                 new VerseTag(),
         };
         for (int i = 0; i < tags.length; i++) {
             Tag t = tags[i];
             String tagName = t.getTagName();
             TAG_MAP.put(tagName, t);
         }
     }
 
 }

1		/**
2		* Distribution License:
3		* JSword is free software; you can redistribute it and/or modify it under
4		* the terms of the GNU Lesser General Public License, version 2.1 or later
5		* as published by the Free Software Foundation. This program is distributed
6		* in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even
7		* the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
8		* See the GNU Lesser General Public License for more details.
9		*
10		* The License is available on the internet at:
11		* http://www.gnu.org/copyleft/lgpl.html
12		* or by writing to:
13		* Free Software Foundation, Inc.
14		* 59 Temple Place - Suite 330
15		* Boston, MA 02111-1307, USA
16		*
17		* © CrossWire Bible Society, 2005 - 2016
18		*
19		*/
20		package org.crosswire.jsword.book.filter.thml;
21
22		import java.util.HashMap;
23		import java.util.LinkedList;
24		import java.util.Locale;
25		import java.util.Map;
26
27		import org.crosswire.jsword.book.Book;
28		import org.crosswire.jsword.book.DataPolice;
29		import org.crosswire.jsword.passage.Key;
30		import org.jdom2.Content;
31		import org.jdom2.Element;
32		import org.jdom2.Text;
33		import org.xml.sax.Attributes;
34		import org.xml.sax.SAXException;
35		import org.xml.sax.helpers.DefaultHandler;
36
37		/**
38		* To convert SAX events into OSIS events.
39		*
40		* <p>
41		* This is based upon the THML reference page:
42		* <a href="http://www.ccel.org/ThML/ThML1.04.htm">http://www.ccel.org/ThML/ThML1.04.htm</a>
43		* to work out what the tags meant.
44		*
45		* @see gnu.lgpl.License The GNU Lesser General Public License for details.
46		* @author Joe Walker
47		*/
48		public class CustomHandler extends DefaultHandler {
49		/**
50		* Simple ctor
51		*
52		* @param book the book
53		* @param key the key
54		*/
55	0	public CustomHandler(Book book, Key key) {
56	0	this.book = book;
57	0	this.key = key;
58	0	this.stack = new LinkedList<Content>();
59	0	}
60
61		/* (non-Javadoc)
62		* @see org.xml.sax.helpers.DefaultHandler#startElement(java.lang.String, java.lang.String, java.lang.String, org.xml.sax.Attributes)
63		*/
64		@Override
65		public void startElement(String uri, String localname, String qname, Attributes attrs) throws SAXException {
66	0	Element ele = null;
67
68		// If we are looking at the root element
69		// then the stack is empty
70	0	if (!stack.isEmpty()) {
71	0	Object top = stack.getFirst();
72
73		// If the element and its descendants are to be ignored
74		// then there is a null element on the stack
75	0	if (top == null) {
76	0	return;
77		}
78
79		// It might be a text element
80	0	if (top instanceof Element) {
81	0	ele = (Element) top;
82		}
83		}
84
85	0	Tag t = getTag(localname, qname);
86
87	0	if (t != null) {
88	0	stack.addFirst(t.processTag(book, key, ele, attrs));
89		}
90	0	}
91
92		/* (non-Javadoc)
93		* @see org.xml.sax.helpers.DefaultHandler#characters(char[], int, int)
94		*/
95		@Override
96		public void characters(char[] data, int offset, int length) {
97		// what we are adding
98	0	String text = new String(data, offset, length);
99
100	0	if (stack.isEmpty()) {
101	0	stack.addFirst(new Text(text));
102	0	return;
103		}
104
105		// What we are adding to
106	0	Content top = stack.getFirst();
107
108		// If the element and its descendants are to be ignored
109		// then there is a null element on the stack
110	0	if (top == null) {
111	0	return;
112		}
113
114	0	if (top instanceof Text) {
115	0	((Text) top).append(text);
116	0	return;
117		}
118
119	0	if (top instanceof Element) {
120	0	Element current = (Element) top;
121
122	0	int size = current.getContentSize();
123
124		// If the last element in the list is a string then we should add
125		// this string on to the end of it rather than add a new list item
126		// because (probably as an artifact of the HTML/XSL transform we get
127		// a space inserted in the output even when 2 calls to this method
128		// split a word.
129	0	if (size > 0) {
130	0	Content last = current.getContent(size - 1);
131	0	if (last instanceof Text) {
132	0	((Text) last).append(text);
133	0	return;
134		}
135		}
136	0	current.addContent(new Text(text));
137		}
138	0	}
139
140		/* (non-Javadoc)
141		* @see org.xml.sax.helpers.DefaultHandler#endElement(java.lang.String, java.lang.String, java.lang.String)
142		*/
143		@Override
144		public void endElement(String uri, String localname, String qname) {
145	0	if (stack.isEmpty()) {
146	0	return;
147		}
148		// When we are done processing an element we need to remove
149		// it from the stack so that nothing more is attached to it.
150	0	Content top = stack.removeFirst();
151	0	if (top instanceof Element) {
152	0	Element finished = (Element) top;
153	0	Tag t = getTag(localname, qname);
154
155	0	if (t != null) {
156	0	t.processContent(book, key, finished);
157		}
158
159		// If it was the last element then it was the root element
160		// so save it
161	0	if (stack.isEmpty()) {
162	0	rootElement = finished;
163		}
164		}
165	0	}
166
167		/**
168		* @return the root element
169		*/
170		public Element getRootElement() {
171	0	return rootElement;
172		}
173
174		private Tag getTag(String localname, String qname) {
175		// sometimes qname is empty e.g. on Android 2.1
176		String name;
177	0	if (qname != null && qname.length() > 0) {
178	0	name = qname;
179		} else {
180	0	name = localname;
181		}
182
183	0	Tag t = TAG_MAP.get(name);
184
185		// Some of the THML books are broken in that they use uppercase
186		// element names, which the spec disallows, but we might as well
187		// look out for them
188	0	if (t == null) {
189	0	t = TAG_MAP.get(name.toLowerCase(Locale.ENGLISH));
190
191	0	if (t == null) {
192	0	DataPolice.report(book, key, "Unknown thml element: " + localname + " qname=" + name);
193
194		// Report on it only once and make sure the content is output.
195	0	t = new AnonymousTag(name);
196	0	TAG_MAP.put(name, t);
197	0	return t;
198		}
199
200		// DataPolice.report(book, key, "Wrong case used in thml element: " + name);
201		}
202	0	return t;
203		}
204
205		/**
206		* When the document is parsed, this is the last element popped off the
207		* stack.
208		*/
209		private Element rootElement;
210
211		/**
212		* The book being parsed.
213		*/
214		private Book book;
215
216		/**
217		* The book being parsed.
218		*/
219		private Key key;
220
221		/**
222		* The stack of elements that we have created
223		*/
224		private LinkedList<Content> stack;
225
226		/**
227		* The known tag types
228		*/
229	0	private static final Map<String, Tag> TAG_MAP = new HashMap<String, Tag>();
230
231		static {
232		/*
233		* ThML is based upon Voyager XHTML and all Voyager elements are
234		* allowed. However not all elements make sense.
235		*/
236	0	Tag[] tags = new Tag[] {
237		// The following are defined in Voyager xhtml 4.0
238		new ATag(),
239		new AbbrTag(),
240		new AliasTag("acronym", new AbbrTag()),
241		new AnonymousTag("address"),
242		new SkipTag("applet"),
243		new SkipTag("area"),
244		new BTag(),
245		new SkipTag("base"),
246		new SkipTag("basefont"),
247		new IgnoreTag("bdo"),
248		new BigTag(),
249		new BlockquoteTag(),
250		new IgnoreTag("body"),
251		new BrTag(),
252		new SkipTag("button"),
253		new AnonymousTag("caption"),
254		new CenterTag(),
255		new AnonymousTag("cite"),
256		new AnonymousTag("code"),
257		new SkipTag("col"),
258		new SkipTag("colgroup"),
259		new AliasTag("dd", new LiTag()),
260		new AnonymousTag("del"),
261		new AnonymousTag("dfn"),
262		new DivTag(),
263		new AliasTag("dl", new UlTag()),
264		new AliasTag("dt", new LiTag()),
265		new AliasTag("em", new ITag()),
266		new IgnoreTag("fieldset"),
267		new FontTag(),
268		new SkipTag("form"),
269		new SkipTag("frame"),
270		new SkipTag("frameset"),
271		new AliasTag("h1", new HTag(1)),
272		new AliasTag("h2", new HTag(2)),
273		new AliasTag("h3", new HTag(3)),
274		new AliasTag("h4", new HTag(4)),
275		new AliasTag("h5", new HTag(5)),
276		new AliasTag("h6", new HTag(6)),
277		new SkipTag("head"),
278		new HrTag(),
279		new IgnoreTag("html"),
280		new IgnoreTag("frameset"),
281		new ITag(),
282		new SkipTag("iframe"),
283		new ImgTag(),
284		new SkipTag("input"),
285		new AnonymousTag("ins"),
286		new AnonymousTag("kbd"),
287		new AnonymousTag("label"),
288		new AnonymousTag("legend"),
289		new LiTag(),
290		new SkipTag("link"),
291		new SkipTag("map"),
292		new SkipTag("meta"),
293		new SkipTag("noscript"),
294		new SkipTag("object"),
295		new OlTag(),
296		new SkipTag("optgroup"),
297		new SkipTag("option"),
298		new PTag(),
299		new SkipTag("param"),
300		new IgnoreTag("pre"),
301		new QTag(),
302		new RootTag(),
303		new STag(),
304		new AnonymousTag("samp"),
305		new SkipTag("script"),
306		new SkipTag("select"),
307		new SmallTag(),
308		new IgnoreTag("span"),
309		new AliasTag("strong", new BTag()),
310		new SkipTag("style"),
311		new SubTag(),
312		new SupTag(),
313		new SyncTag(),
314		new TableTag(),
315		new IgnoreTag("tbody"),
316		new TdTag(),
317		new IgnoreTag("tfoot"),
318		new SkipTag("textarea"),
319		new SkipTag("title"),
320		new IgnoreTag("thead"),
321		new ThTag(),
322		new TrTag(),
323		new TtTag(),
324		new UTag(),
325		new UlTag(),
326		new AnonymousTag("var"),
327
328		// ThML adds the following to Voyager
329		// Note: hymn.mod is not here nor are additional head&DC
330		// elements
331		new AnonymousTag("added"),
332		new AnonymousTag("attr"),
333		new AnonymousTag("argument"),
334		new CitationTag(),
335		new AnonymousTag("date"),
336		new AnonymousTag("deleted"),
337		new AnonymousTag("def"),
338		new AliasTag("div1", new DivTag(1)),
339		new AliasTag("div2", new DivTag(2)),
340		new AliasTag("div3", new DivTag(3)),
341		new AliasTag("div4", new DivTag(4)),
342		new AliasTag("div5", new DivTag(5)),
343		new AliasTag("div6", new DivTag(6)),
344		new ForeignTag(),
345		new AnonymousTag("index"),
346		new AnonymousTag("insertIndex"),
347		new AnonymousTag("glossary"),
348		new NoteTag(),
349		new NameTag(),
350		new PbTag(),
351		new AnonymousTag("scripCom"),
352		new AnonymousTag("scripContext"),
353		new ScripRefTag(),
354		new ScriptureTag(),
355		new TermTag(),
356		new AnonymousTag("unclear"),
357		new VerseTag(),
358		};
359	0	for (int i = 0; i < tags.length; i++) {
360	0	Tag t = tags[i];
361	0	String tagName = t.getTagName();
362	0	TAG_MAP.put(tagName, t);
363		}
364	0	}
365
366		}