Coverage Report

 /*
  * HtmlFormatter.java -- HTML document pretty-printer
  * Copyright (C) 1999 Quiotix Corporation.  
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License, version 2, 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 General Public License (http://www.gnu.org/copyleft/gpl.txt)
  * for more details.
  */
 
 package com.quiotix.html.parser;
 
 import java.io.BufferedOutputStream;
 import java.io.FileInputStream;
 import java.io.InputStream;
 import java.io.OutputStream;
 import java.io.PrintWriter;
 import java.util.HashSet;
 import java.util.Iterator;
 import java.util.Set;
 
 /**
  * HtmlFormatter is a Visitor which traverses an HtmlDocument, dumping the
  * contents of the document to a specified output stream.  It assumes that
  * the documents has been preprocessed by HtmlCollector (which matches up
  * beginning and end tags) and by HtmlScrubber (which formats tags in a
  * consistent way).  In particular, HtmlScrubber should be invoked with the
  * TRIM_SPACES option to remove trailing spaces, which can confuse the
  * formatting algorithm.
  * <p>
  * The right margin and indent increment can be specified as properties.
  * </p>
  * 
  * @author Brian Goetz, Quiotix
  * @see com.quiotix.html.parser.HtmlVisitor
  * @see com.quiotix.html.parser.HtmlCollector
  * @see com.quiotix.html.parser.HtmlScrubber
  */
 
 public class HtmlFormatter extends HtmlVisitor {
     protected MarginWriter out;
     protected int rightMargin = 80;
     protected int indentSize = 2;
     protected static Set tagsIndentBlock = new HashSet();
     protected static Set tagsNewlineBefore = new HashSet();
     protected static Set tagsPreformatted = new HashSet();
     protected static Set tagsTryMatch = new HashSet();
     protected static final String[] tagsIndentStrings
             = {"TABLE", "TR", "TD", "TH", "FORM", "HTML", "HEAD", "BODY", "SELECT", "OL", "UL", "LI"};
     protected static final String[] tagsNewlineBeforeStrings
             = {"P", "H1", "H2", "H3", "H4", "H5", "H6", "BR"};
     protected static final String[] tagsPreformattedStrings
             = {"PRE", "SCRIPT", "STYLE"};
     protected static final String[] tagsTryMatchStrings
             = {"A", "TD", "TH", "TR", "I", "B", "EM", "FONT", "TT", "UL", "OL", "LI"};
 
     static {
         for (int i = 0; i < tagsIndentStrings.length; i++)
             tagsIndentBlock.add(tagsIndentStrings[i]);
         for (int i = 0; i < tagsNewlineBeforeStrings.length; i++)
             tagsNewlineBefore.add(tagsNewlineBeforeStrings[i]);
         for (int i = 0; i < tagsPreformattedStrings.length; i++)
             tagsPreformatted.add(tagsPreformattedStrings[i]);
         for (int i = 0; i < tagsTryMatchStrings.length; i++)
             tagsTryMatch.add(tagsTryMatchStrings[i]);
     }
     protected TagBlockRenderer blockRenderer = new TagBlockRenderer();
     protected HtmlDocument.HtmlElement previousElement;
     protected boolean inPreBlock;
 
     /** Constructor. */
     public HtmlFormatter(OutputStream os) throws Exception {
         out = new MarginWriter(new PrintWriter(new BufferedOutputStream(os)));
         out.setRightMargin(rightMargin);
     }
 
     /**
      * @param margin the right margin column to wrap at
      */
     public void setRightMargin(int margin) {
         rightMargin = margin;
         out.setRightMargin(rightMargin);
     }
 
     /**
      * @param indent the number of spaces to indent by
      */
     public void setIndent(int indent) {
         indentSize = indent;
     }
 
     public void visit(HtmlDocument.TagBlock block) {
         boolean indent;
         boolean preformat;
         int wasMargin = 0;
 
         if (tagsTryMatch.contains(block.startTag.tagName.toUpperCase())) {
             blockRenderer.start();
             blockRenderer.setTargetWidth(out.getRightMargin() - out.getLeftMargin());
             blockRenderer.visit(block);
             blockRenderer.finish();
             if (!blockRenderer.hasBlownTarget()) {
                 out.printAutoWrap(blockRenderer.getString());
                 previousElement = block.endTag;
                 return;
             } 
 
         }
 
         // Only will get here if we've failed the try-block test
         indent = tagsIndentBlock.contains(block.startTag.tagName.toUpperCase());
         preformat = tagsPreformatted.contains(block.startTag.tagName.toUpperCase());
         if (preformat) {
             inPreBlock = true;
             visit(block.startTag);
             wasMargin = out.getLeftMargin();
             out.setLeftMargin(0);
             visit(block.body);
             out.setLeftMargin(wasMargin);
             visit(block.endTag);
         } else if (indent) {
             out.printlnSoft();
             visit(block.startTag);
             out.printlnSoft();
             out.setLeftMargin(out.getLeftMargin() + indentSize);
             visit(block.body);
             out.setLeftMargin(out.getLeftMargin() - indentSize);
             out.printlnSoft();
             visit(block.endTag);
             out.printlnSoft();
             inPreBlock = false;
         } else {
             visit(block.startTag);
             visit(block.body);
             visit(block.endTag);
         }
     }
 
     public void visit(HtmlDocument.Tag t) {
         String s = t.toString();
         int hanging;
 
         if (tagsNewlineBefore.contains(t.tagName.toUpperCase())
                 || out.getCurPosition() + s.length() > out.getRightMargin())
             out.printlnSoft();
 
         out.print("<" + t.tagName);
         hanging = t.tagName.length() + 1;
         for (Iterator it = t.attributeList.attributes.iterator(); it.hasNext();) {
             HtmlDocument.Attribute a = (HtmlDocument.Attribute) it.next();
             out.printAutoWrap(" " + a.toString(), hanging);
         }
         if (t.emptyTag) out.print("/");
         out.print(">");
         previousElement = t;
     }
 
     public void visit(HtmlDocument.EndTag t) {
         out.printAutoWrap(t.toString());
         if (tagsNewlineBefore.contains(t.tagName.toUpperCase())) {
             out.printlnSoft();
             out.println();
         }
         previousElement = t;
     }
 
     public void visit(HtmlDocument.Comment c) {
         out.print(c.toString());
         previousElement = c;
     }
 
     public void visit(HtmlDocument.Text t) {
         if (inPreBlock)
             out.print(t.text);
         else {
             int start = 0;
             while (start < t.text.length()) {
                 int index = t.text.indexOf(' ', start) + 1;
                 if (index == 0)
                     index = t.text.length();
                 out.printAutoWrap(t.text.substring(start, index));
                 start = index;
             }
         }
         previousElement = t;
     }
 
     public void visit(HtmlDocument.Newline n) {
         if (inPreBlock)
             out.println();
         else if (previousElement instanceof HtmlDocument.Tag
                 || previousElement instanceof HtmlDocument.EndTag
                 || previousElement instanceof HtmlDocument.Comment
                 || previousElement instanceof HtmlDocument.Newline)
             out.printlnSoft();
         else if (previousElement instanceof HtmlDocument.Text)
             out.print(" ");
         previousElement = n;
     }
 
     public void start() {
         previousElement = null;
         inPreBlock = false;
     }
 
     public void finish() {
         out.flush();
     }
 
     /**
      * Runnable.
      */
     public static void main(String[] args) throws Exception {
         InputStream r = new FileInputStream(args[0]);
         HtmlDocument document;
 
         try {
             document = new HtmlParser(r).HtmlDocument();
             document.accept(new HtmlCollector());
             document.accept(new HtmlScrubber(HtmlScrubber.DEFAULT_OPTIONS
                     | HtmlScrubber.TRIM_SPACES));
             document.accept(new HtmlFormatter(System.out));
         } catch (Exception e) {
             e.printStackTrace();
         } finally {
             r.close();
         }
     }
 }
 
 
 /**
  * Utility class, used by HtmlFormatter, which adds some word-wrapping
  * and hanging indent functionality to a PrintWriter.
  */
 
 class MarginWriter {
     protected int tabStop;
     protected int curPosition;
     protected int leftMargin;
     protected int rightMargin;
     protected java.io.PrintWriter out;
     protected char[] spaces = new char[256];
 
     /** Constructor. */
     MarginWriter(java.io.PrintWriter out) {
         this.out = out;
         for (int i = 0; i < spaces.length; i++)
             spaces[i] = ' ';
     }
 
     void flush() {
         out.flush();
     }
 
     void close() {
         out.close();
     }
 
     void print(String s) {
         if (curPosition == 0 && leftMargin > 0) {
             out.write(spaces, 0, leftMargin);
             curPosition = leftMargin;
         }
         out.print(s);
         curPosition += s.length();
     }
 
     void printAutoWrap(String s) {
         if (curPosition > leftMargin
                 && curPosition + s.length() > rightMargin)
             println();
         print(s);
     }
 
     void printAutoWrap(String s, int hanging) {
         if (curPosition > leftMargin
                 && curPosition + s.length() > rightMargin) {
             println();
             out.write(spaces, 0, hanging + leftMargin);
             curPosition = leftMargin + hanging;
         };
         print(s);
     }
 
     void println() {
         curPosition = 0;
         out.println();
     }
 
     void printlnSoft() {
         if (curPosition > 0)
             println();
     }
 
     void setLeftMargin(int leftMargin) {
         this.leftMargin = leftMargin;
     }
 
     int getLeftMargin() {
         return leftMargin;
     }
 
     void setRightMargin(int rightMargin) {
         this.rightMargin = rightMargin;
     }
 
     int getRightMargin() {
         return rightMargin;
     }
 
     int getCurPosition() {
         return (curPosition == 0 ? leftMargin : curPosition);
     }
 }
 
 /**
  * Utility class, used by HtmlFormatter, which tentatively tries to format
  * the contents of an HtmlDocument.TagBlock to see if the entire block can
  * fit on the rest of the line.  If it cannot, it gives up and indicates
  * failure through the hasBlownTarget method; if it can, the contents can
  * be retrieved through the getString method.
  */
 
 class TagBlockRenderer extends HtmlVisitor {
     protected String s;
     protected boolean multiLine;
     protected boolean blownTarget;
     protected int targetWidth = 80;
 
     public void start() {
         s = "";
         multiLine = false;
         blownTarget = false;
     }
 
     public void finish() {
     }
 
     void setTargetWidth(int w) {
         targetWidth = w;
     }
 
     String getString() {
         return s;
     }
 
     boolean isMultiLine() {
         return multiLine;
     }
 
     boolean hasBlownTarget() {
         return blownTarget;
     }
 
     public void visit(HtmlDocument.Tag t) {
         if (s.length() < targetWidth)
             s += t.toString();
         else
             blownTarget = true;
     }
 
     public void visit(HtmlDocument.EndTag t) {
         if (s.length() < targetWidth)
             s += t.toString();
         else
             blownTarget = true;
     }
 
     public void visit(HtmlDocument.Comment c) {
         if (s.length() < targetWidth)
             s += c.toString();
         else
             blownTarget = true;
     }
 
     public void visit(HtmlDocument.Text t) {
         if (s.length() < targetWidth)
             s += t.toString();
         else
             blownTarget = true;
     }
 
     public void visit(HtmlDocument.Newline n) {
         multiLine = true;
         s += " ";
     }
 }
 
 

1		/*
2		* HtmlFormatter.java -- HTML document pretty-printer
3		* Copyright (C) 1999 Quiotix Corporation.
4		*
5		* This program is free software; you can redistribute it and/or modify
6		* it under the terms of the GNU General Public License, version 2, as
7		* published by the Free Software Foundation.
8		*
9		* This program is distributed in the hope that it will be useful,
10		* but WITHOUT ANY WARRANTY; without even the implied warranty of
11		* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12		* GNU General Public License (http://www.gnu.org/copyleft/gpl.txt)
13		* for more details.
14		*/
15
16		package com.quiotix.html.parser;
17
18		import java.io.BufferedOutputStream;
19		import java.io.FileInputStream;
20		import java.io.InputStream;
21		import java.io.OutputStream;
22		import java.io.PrintWriter;
23		import java.util.HashSet;
24		import java.util.Iterator;
25		import java.util.Set;
26
27		/**
28		* HtmlFormatter is a Visitor which traverses an HtmlDocument, dumping the
29		* contents of the document to a specified output stream. It assumes that
30		* the documents has been preprocessed by HtmlCollector (which matches up
31		* beginning and end tags) and by HtmlScrubber (which formats tags in a
32		* consistent way). In particular, HtmlScrubber should be invoked with the
33		* TRIM_SPACES option to remove trailing spaces, which can confuse the
34		* formatting algorithm.
35		* <p>
36		* The right margin and indent increment can be specified as properties.
37		* </p>
38		*
39		* @author Brian Goetz, Quiotix
40		* @see com.quiotix.html.parser.HtmlVisitor
41		* @see com.quiotix.html.parser.HtmlCollector
42		* @see com.quiotix.html.parser.HtmlScrubber
43		*/
44
45		public class HtmlFormatter extends HtmlVisitor {
46		protected MarginWriter out;
47	4	protected int rightMargin = 80;
48	4	protected int indentSize = 2;
49	2	protected static Set tagsIndentBlock = new HashSet();
50	2	protected static Set tagsNewlineBefore = new HashSet();
51	2	protected static Set tagsPreformatted = new HashSet();
52	2	protected static Set tagsTryMatch = new HashSet();
53	2	protected static final String[] tagsIndentStrings
54		= {"TABLE", "TR", "TD", "TH", "FORM", "HTML", "HEAD", "BODY", "SELECT", "OL", "UL", "LI"};
55	2	protected static final String[] tagsNewlineBeforeStrings
56		= {"P", "H1", "H2", "H3", "H4", "H5", "H6", "BR"};
57	2	protected static final String[] tagsPreformattedStrings
58		= {"PRE", "SCRIPT", "STYLE"};
59	2	protected static final String[] tagsTryMatchStrings
60		= {"A", "TD", "TH", "TR", "I", "B", "EM", "FONT", "TT", "UL", "OL", "LI"};
61
62		static {
63	26	for (int i = 0; i < tagsIndentStrings.length; i++)
64	24	tagsIndentBlock.add(tagsIndentStrings[i]);
65	18	for (int i = 0; i < tagsNewlineBeforeStrings.length; i++)
66	16	tagsNewlineBefore.add(tagsNewlineBeforeStrings[i]);
67	8	for (int i = 0; i < tagsPreformattedStrings.length; i++)
68	6	tagsPreformatted.add(tagsPreformattedStrings[i]);
69	26	for (int i = 0; i < tagsTryMatchStrings.length; i++)
70	24	tagsTryMatch.add(tagsTryMatchStrings[i]);
71	2	}
72	4	protected TagBlockRenderer blockRenderer = new TagBlockRenderer();
73		protected HtmlDocument.HtmlElement previousElement;
74		protected boolean inPreBlock;
75
76		/** Constructor. */
77	4	public HtmlFormatter(OutputStream os) throws Exception {
78	4	out = new MarginWriter(new PrintWriter(new BufferedOutputStream(os)));
79	4	out.setRightMargin(rightMargin);
80	4	}
81
82		/**
83		* @param margin the right margin column to wrap at
84		*/
85		public void setRightMargin(int margin) {
86	0	rightMargin = margin;
87	0	out.setRightMargin(rightMargin);
88	0	}
89
90		/**
91		* @param indent the number of spaces to indent by
92		*/
93		public void setIndent(int indent) {
94	0	indentSize = indent;
95	0	}
96
97		public void visit(HtmlDocument.TagBlock block) {
98		boolean indent;
99		boolean preformat;
100	0	int wasMargin = 0;
101
102	0	if (tagsTryMatch.contains(block.startTag.tagName.toUpperCase())) {
103	0	blockRenderer.start();
104	0	blockRenderer.setTargetWidth(out.getRightMargin() - out.getLeftMargin());
105	0	blockRenderer.visit(block);
106	0	blockRenderer.finish();
107	0	if (!blockRenderer.hasBlownTarget()) {
108	0	out.printAutoWrap(blockRenderer.getString());
109	0	previousElement = block.endTag;
110	0	return;
111		}
112
113		}
114
115		// Only will get here if we've failed the try-block test
116	0	indent = tagsIndentBlock.contains(block.startTag.tagName.toUpperCase());
117	0	preformat = tagsPreformatted.contains(block.startTag.tagName.toUpperCase());
118	0	if (preformat) {
119	0	inPreBlock = true;
120	0	visit(block.startTag);
121	0	wasMargin = out.getLeftMargin();
122	0	out.setLeftMargin(0);
123	0	visit(block.body);
124	0	out.setLeftMargin(wasMargin);
125	0	visit(block.endTag);
126	0	} else if (indent) {
127	0	out.printlnSoft();
128	0	visit(block.startTag);
129	0	out.printlnSoft();
130	0	out.setLeftMargin(out.getLeftMargin() + indentSize);
131	0	visit(block.body);
132	0	out.setLeftMargin(out.getLeftMargin() - indentSize);
133	0	out.printlnSoft();
134	0	visit(block.endTag);
135	0	out.printlnSoft();
136	0	inPreBlock = false;
137		} else {
138	0	visit(block.startTag);
139	0	visit(block.body);
140	0	visit(block.endTag);
141		}
142	0	}
143
144		public void visit(HtmlDocument.Tag t) {
145	20	String s = t.toString();
146		int hanging;
147
148	20	if (tagsNewlineBefore.contains(t.tagName.toUpperCase())
149		\|\| out.getCurPosition() + s.length() > out.getRightMargin())
150	8	out.printlnSoft();
151
152	20	out.print("<" + t.tagName);
153	20	hanging = t.tagName.length() + 1;
154	20	for (Iterator it = t.attributeList.attributes.iterator(); it.hasNext();) {
155	4	HtmlDocument.Attribute a = (HtmlDocument.Attribute) it.next();
156	4	out.printAutoWrap(" " + a.toString(), hanging);
157	4	}
158	20	if (t.emptyTag) out.print("/");
159	20	out.print(">");
160	20	previousElement = t;
161	20	}
162
163		public void visit(HtmlDocument.EndTag t) {
164	0	out.printAutoWrap(t.toString());
165	0	if (tagsNewlineBefore.contains(t.tagName.toUpperCase())) {
166	0	out.printlnSoft();
167	0	out.println();
168		}
169	0	previousElement = t;
170	0	}
171
172		public void visit(HtmlDocument.Comment c) {
173	0	out.print(c.toString());
174	0	previousElement = c;
175	0	}
176
177		public void visit(HtmlDocument.Text t) {
178	8	if (inPreBlock)
179	0	out.print(t.text);
180		else {
181	8	int start = 0;
182	46	while (start < t.text.length()) {
183	38	int index = t.text.indexOf(' ', start) + 1;
184	38	if (index == 0)
185	0	index = t.text.length();
186	38	out.printAutoWrap(t.text.substring(start, index));
187	38	start = index;
188	38	}
189		}
190	8	previousElement = t;
191	8	}
192
193		public void visit(HtmlDocument.Newline n) {
194	4	if (inPreBlock)
195	0	out.println();
196	4	else if (previousElement instanceof HtmlDocument.Tag
197		\|\| previousElement instanceof HtmlDocument.EndTag
198		\|\| previousElement instanceof HtmlDocument.Comment
199		\|\| previousElement instanceof HtmlDocument.Newline)
200	0	out.printlnSoft();
201	4	else if (previousElement instanceof HtmlDocument.Text)
202	4	out.print(" ");
203	4	previousElement = n;
204	4	}
205
206		public void start() {
207	4	previousElement = null;
208	4	inPreBlock = false;
209	4	}
210
211		public void finish() {
212	4	out.flush();
213	4	}
214
215		/**
216		* Runnable.
217		*/
218		public static void main(String[] args) throws Exception {
219	0	InputStream r = new FileInputStream(args[0]);
220		HtmlDocument document;
221
222		try {
223	0	document = new HtmlParser(r).HtmlDocument();
224	0	document.accept(new HtmlCollector());
225	0	document.accept(new HtmlScrubber(HtmlScrubber.DEFAULT_OPTIONS
226		\| HtmlScrubber.TRIM_SPACES));
227	0	document.accept(new HtmlFormatter(System.out));
228	0	} catch (Exception e) {
229	0	e.printStackTrace();
230		} finally {
231	0	r.close();
232	0	}
233	0	}
234		}
235
236
237		/**
238		* Utility class, used by HtmlFormatter, which adds some word-wrapping
239		* and hanging indent functionality to a PrintWriter.
240		*/
241
242		class MarginWriter {
243		protected int tabStop;
244		protected int curPosition;
245		protected int leftMargin;
246		protected int rightMargin;
247		protected java.io.PrintWriter out;
248	4	protected char[] spaces = new char[256];
249
250		/** Constructor. */
251	4	MarginWriter(java.io.PrintWriter out) {
252	4	this.out = out;
253	1028	for (int i = 0; i < spaces.length; i++)
254	1024	spaces[i] = ' ';
255	4	}
256
257		void flush() {
258	4	out.flush();
259	4	}
260
261		void close() {
262	0	out.close();
263	0	}
264
265		void print(String s) {
266	86	if (curPosition == 0 && leftMargin > 0) {
267	0	out.write(spaces, 0, leftMargin);
268	0	curPosition = leftMargin;
269		}
270	86	out.print(s);
271	86	curPosition += s.length();
272	86	}
273
274		void printAutoWrap(String s) {
275	38	if (curPosition > leftMargin
276		&& curPosition + s.length() > rightMargin)
277	0	println();
278	38	print(s);
279	38	}
280
281		void printAutoWrap(String s, int hanging) {
282	4	if (curPosition > leftMargin
283		&& curPosition + s.length() > rightMargin) {
284	0	println();
285	0	out.write(spaces, 0, hanging + leftMargin);
286	0	curPosition = leftMargin + hanging;
287		};
288	4	print(s);
289	4	}
290
291		void println() {
292	8	curPosition = 0;
293	8	out.println();
294	8	}
295
296		void printlnSoft() {
297	8	if (curPosition > 0)
298	8	println();
299	8	}
300
301		void setLeftMargin(int leftMargin) {
302	0	this.leftMargin = leftMargin;
303	0	}
304
305		int getLeftMargin() {
306	0	return leftMargin;
307		}
308
309		void setRightMargin(int rightMargin) {
310	4	this.rightMargin = rightMargin;
311	4	}
312
313		int getRightMargin() {
314	12	return rightMargin;
315		}
316
317		int getCurPosition() {
318	12	return (curPosition == 0 ? leftMargin : curPosition);
319		}
320		}
321
322		/**
323		* Utility class, used by HtmlFormatter, which tentatively tries to format
324		* the contents of an HtmlDocument.TagBlock to see if the entire block can
325		* fit on the rest of the line. If it cannot, it gives up and indicates
326		* failure through the hasBlownTarget method; if it can, the contents can
327		* be retrieved through the getString method.
328		*/
329
330	4	class TagBlockRenderer extends HtmlVisitor {
331		protected String s;
332		protected boolean multiLine;
333		protected boolean blownTarget;
334	4	protected int targetWidth = 80;
335
336		public void start() {
337	0	s = "";
338	0	multiLine = false;
339	0	blownTarget = false;
340	0	}
341
342		public void finish() {
343	0	}
344
345		void setTargetWidth(int w) {
346	0	targetWidth = w;
347	0	}
348
349		String getString() {
350	0	return s;
351		}
352
353		boolean isMultiLine() {
354	0	return multiLine;
355		}
356
357		boolean hasBlownTarget() {
358	0	return blownTarget;
359		}
360
361		public void visit(HtmlDocument.Tag t) {
362	0	if (s.length() < targetWidth)
363	0	s += t.toString();
364		else
365	0	blownTarget = true;
366	0	}
367
368		public void visit(HtmlDocument.EndTag t) {
369	0	if (s.length() < targetWidth)
370	0	s += t.toString();
371		else
372	0	blownTarget = true;
373	0	}
374
375		public void visit(HtmlDocument.Comment c) {
376	0	if (s.length() < targetWidth)
377	0	s += c.toString();
378		else
379	0	blownTarget = true;
380	0	}
381
382		public void visit(HtmlDocument.Text t) {
383	0	if (s.length() < targetWidth)
384	0	s += t.toString();
385		else
386	0	blownTarget = true;
387	0	}
388
389		public void visit(HtmlDocument.Newline n) {
390	0	multiLine = true;
391	0	s += " ";
392	0	}
393		}
394
395