1 // Copyright 2004 Adam Megacz, see the COPYING file for licensing [GPL]
2 package org.ibex.graphics;
8 import org.ibex.util.*;
11 * While entities are limited to a subset of Unicode characters ,
12 * numeric character references can specify any character. Numeric
13 * character references may be given in decimal or hexadecimal, though
14 * browser support is stronger for decimal references. Decimal
15 * references are of the form &#number; while hexadecimal references
16 * take the case-insensitive form &#xnumber;. Examples of numeric
17 * character references include © or © for the copyright
18 * symbol, Α or Α for the Greek capital letter alpha, and
19 * ا or ا for the Arabic letter ALEF.
21 * http://www.htmlhelp.com/reference/html40/entities/special.html
22 * http://www.htmlhelp.com/reference/html40/entities/symbols.html
23 * http://www.htmlhelp.com/reference/html40/entities/latin1.html
27 * This class parses an InputStream containing HTML and returns it
28 * as an XWT DOM tree. Each HTML Element is returned as a struct,
29 * with the following members:
31 * Since HTML may have multiple top level elements (unlike XML),
32 * this class will search all top level elements for one with a tag
33 * name 'html'. If such a node is found, only it is returned. If no
34 * top-level element has the tag name 'html', such a node is
35 * fabricated, and all top level elements become the children of
36 * that node, which is then returned.
40 private final static String[] noEndTag =
41 new String[] { "area", "base", "basefont", "br", "col", "frame", "hr", "img",
42 "input", "isindex", "link", "meta", "param" };
44 /** we keep a char[] around for use by removeRedundantWhitespace() */
45 private static char[] cbuf = null;
47 /** we keep a StringBuffer around for use by removeRedundantWhitespace() */
48 private static StringBuffer sbuf = null;
50 /** true iff we have encountered an LI more recently than the last OL/UL */
51 private static boolean withinLI = false;
53 // FEATURE: This is ugly
54 private static class JS extends org.ibex.js.JS.O {
55 public void put(String key, Object value) throws JSExn {
56 if(value instanceof String) put(JS.S(key),JS.S((String)value));
57 else if(value instanceof Number) put(JS.S(key), JS.N((Number)value));
58 else if(value == null) put(JS.S(key),null);
59 else throw new Error("FIXME");
61 public Object _get(String key) throws JSExn {
62 org.ibex.js.JS js = get(JS.S(key));
63 if(JS.isInt(js)) return new Integer(JS.toInt(js));
64 return JS.toString(js);
68 public static synchronized JS parseReader(Reader r) throws IOException, JSExn {
69 CharStream cs = new CharStream(r);
73 h.put("$name", "html");
76 while (true) parseBody(cs, h, null);
77 } catch (EOFException e) {
78 // continue until we get an EOFException
82 Object[] ids = h.keys();
83 for(int i=0; i<ids.length; i++) {
84 Object el = h.get((String)ids[i]);
85 if (el instanceof JS && "html".equals(((JS)el).get("$name")))
93 * Parses a single element and stores it in <tt>h</tt>. The
94 * CharStream should be positioned immediately <i>after</i> the
97 * If a close tag not matching this open tag is found, the
98 * tagname on the close tag will be returned in order to
99 * facilitate correcting broken HTML. Otherwise, this returns
102 private static String parseElement(CharStream cs, JS h) throws IOException, JSExn {
104 while(Character.isSpace(cs.peek())) cs.get();
105 String elementName = parseElementName(cs);
107 boolean saveWithinLI = withinLI;
108 if (elementName.equals("li")) {
110 cs.unread(new char[] { '<', 'l', 'i', ' ' });
115 } else if (elementName.equals("ol") || elementName.equals("ul")) {
119 h.put("$name", elementName);
120 if (elementName.equals("!--")) {
121 h.put("0", parseComment(cs));
122 h.put("$numchildren", new Integer(0));
127 while (cs.peek() != '>') {
128 String name = parseAttributeName(cs);
129 if (name.equals("")) break;
130 String value = expandEntities(parseAttributeValue(cs));
134 // eat the close-angle bracket
137 // bodyless tags return here
138 for(int i=0; i<noEndTag.length; i++)
139 if (noEndTag[i].equals(elementName))
143 String ret = parseBody(cs, h, elementName);
144 withinLI = saveWithinLI;
149 * Parses the body of an element. The CharStream should be
150 * positioned at the character immediately after the right
151 * bracket closing the start-tag
153 private static String parseBody(CharStream cs, JS h, String elementName) throws IOException, JSExn {
155 int length = h._get("$numchildren") == null ? 0 : Integer.parseInt(h._get("$numchildren").toString());
157 String closetag = null;
161 if (c != '<') { cdata += c; continue; }
162 String expanded = removeRedundantWhitespace(expandEntities(cdata));
163 if (expanded.length() > 0) {
164 h.put(String.valueOf(length), expanded);
165 h.put("$numchildren", new Integer(++length));
169 } catch (EOFException e) {
170 String expanded = removeRedundantWhitespace(expandEntities(cdata));
171 if (expanded.length() > 0) {
172 h.put(String.valueOf(length), expanded);
173 h.put("$numchildren", new Integer(++length));
180 if (cs.peek() != '/') {
182 closetag = parseElement(cs, kid);
183 h.put(String.valueOf(length), kid);
184 h.put("$numchildren", new Integer(++length));
188 cs.get(); // drop the slash
189 closetag = parseElementName(cs);
190 while(cs.get() != '>');
192 } catch (EOFException e) {
197 if (closetag != null)
198 return closetag.equals(elementName) ? null : closetag;
202 /** Parses an element name and returns it. The CharStream should
203 * be positioned at the first character of the name.
205 private static String parseElementName(CharStream cs) throws IOException, JSExn {
207 while (cs.peek() != '>' && !Character.isSpace(cs.peek())) ret += cs.get();
208 return ret.toLowerCase();
211 /** Parses an attribute name and returns it. The CharStream should
212 * be positioned at the first character of the name, possibly
213 * with intervening whitespace.
215 private static String parseAttributeName(CharStream cs) throws IOException, JSExn {
216 while(Character.isSpace(cs.peek())) cs.get();
218 while(!Character.isSpace(cs.peek()) && cs.peek() != '=' && cs.peek() != '>') ret += cs.get();
219 return ret.toLowerCase();
222 /** Parses an attribute value and returns it. The CharStream
223 * should be positioned at the equals sign, possibly with
224 * intervening whitespace.
226 private static String parseAttributeValue(CharStream cs) throws IOException, JSExn {
228 // eat whitespace and equals sign
229 while(Character.isSpace(cs.peek())) cs.get();
230 if (cs.peek() != '=') return "";
232 while(Character.isSpace(cs.peek())) cs.get();
234 boolean doublequoted = false;
235 boolean singlequoted = false;
238 if (cs.peek() == '\"') { doublequoted = true; cs.get(); }
239 else if (cs.peek() == '\'') { singlequoted = true; cs.get(); }
243 if (!doublequoted && !singlequoted && (Character.isSpace(c) || c == '>')) break;
244 if (singlequoted && c == '\'') { cs.get(); break; }
245 if (doublequoted && c == '\"') { cs.get(); break; }
251 /** Parses a comment and returns its body. The CharStream should
252 * be positioned immediately after the <!--
254 private static String parseComment(CharStream cs) throws IOException, JSExn {
259 if (c == '>' && dashes == 2) return ret.substring(0, ret.length() - 2);
260 if (c == '-') dashes++;
266 /** Expands all SGML entities in string <tt>s</tt> */
267 public static String expandEntities(String s) throws IOException, JSExn {
268 if (s.indexOf('&') == -1) return s;
269 StringBuffer sb = new StringBuffer();
272 while(nextamp != -1) {
273 nextamp = s.indexOf('&', i);
274 sb.append(nextamp == -1 ? s.substring(i) : s.substring(i, nextamp));
275 if (nextamp == -1) break;
276 if (s.regionMatches(nextamp, "&", 0, 5)) {
279 } else if (s.regionMatches(nextamp, ">", 0, 4)) {
282 } else if (s.regionMatches(nextamp, "<", 0, 4)) {
285 } else if (s.regionMatches(nextamp, """, 0, 6)) {
288 } else if (s.regionMatches(nextamp, " ", 0, 6)) {
289 // FEATURE: perhaps we should distinguish this somehow
297 return sb.toString();
300 /** removes all redundant whitespace */
301 private static String removeRedundantWhitespace(String s) throws JSExn {
303 if (s.indexOf(' ') == -1 && s.indexOf('\n') == -1 && s.indexOf('\t') == -1 && s.indexOf('\r') == -1) return s;
305 int len = s.length();
306 if (cbuf == null || cbuf.length < len) {
307 cbuf = new char[len * 2];
308 sbuf = new StringBuffer(len * 2);
311 s.getChars(0, len, cbuf, 0);
314 boolean lastWasWhitespace = false;
315 for(int i=0; i<len; i++) {
316 boolean lastlast = lastWasWhitespace;
318 case '\n': case '\r': case '\t':
321 lastWasWhitespace = true;
324 lastWasWhitespace = false;
327 if (lastWasWhitespace && lastlast) {
328 if (last != i) sbuf.append(cbuf, last, i - last);
333 if (last != len) sbuf.append(cbuf, last, len - last);
334 return sbuf.toString().trim();
337 // CharStream /////////////////////////////////////////////////////////////////////
339 private static class CharStream extends PushbackReader {
340 public CharStream(Reader r) { super(r, 1024); }
342 public char peek() throws IOException {
348 public char get() throws IOException {
350 if (i == -1) throw new EOFException();