|
1
47
48 package com.lowagie.text.pdf.hyphenation;
49
50 import java.io.FileInputStream  ;
51 import java.io.IOException ;
52 import java.io.InputStream ;
53 import java.util.ArrayList ;
54 import java.util.StringTokenizer ;
55
56 import com.lowagie.text.ExceptionConverter;
57 import com.lowagie.text.xml.simpleparser.SimpleXMLDocHandler;
58 import com.lowagie.text.xml.simpleparser.SimpleXMLParser;
59
60
64 public class SimplePatternParser implements SimpleXMLDocHandler, PatternConsumer {
65 int currElement;
66 PatternConsumer consumer;
67 StringBuffer token;
68 ArrayList exception;
69 char hyphenChar;
70 SimpleXMLParser parser;
71
72 static final int ELEM_CLASSES = 1;
73 static final int ELEM_EXCEPTIONS = 2;
74 static final int ELEM_PATTERNS = 3;
75 static final int ELEM_HYPHEN = 4;
76
77
78 public SimplePatternParser() {
79 token = new StringBuffer ();
80 hyphenChar = '-'; }
82
83 public void parse(InputStream stream, PatternConsumer consumer) {
84 this.consumer = consumer;
85 try {
86 SimpleXMLParser.parse(this, stream);
87 }
88 catch (IOException e) {
89 throw new ExceptionConverter(e);
90 }
91 finally {
92 try{stream.close();}catch(Exception e){}
93 }
94 }
95
96 protected static String getPattern(String word) {
97 StringBuffer pat = new StringBuffer ();
98 int len = word.length();
99 for (int i = 0; i < len; i++) {
100 if (!Character.isDigit(word.charAt(i))) {
101 pat.append(word.charAt(i));
102 }
103 }
104 return pat.toString();
105 }
106
107 protected ArrayList normalizeException(ArrayList ex) {
108 ArrayList res = new ArrayList ();
109 for (int i = 0; i < ex.size(); i++) {
110 Object item = ex.get(i);
111 if (item instanceof String ) {
112 String str = (String )item;
113 StringBuffer buf = new StringBuffer ();
114 for (int j = 0; j < str.length(); j++) {
115 char c = str.charAt(j);
116 if (c != hyphenChar) {
117 buf.append(c);
118 } else {
119 res.add(buf.toString());
120 buf.setLength(0);
121 char[] h = new char[1];
122 h[0] = hyphenChar;
123 res.add(new Hyphen(new String (h), null, null));
126 }
127 }
128 if (buf.length() > 0) {
129 res.add(buf.toString());
130 }
131 } else {
132 res.add(item);
133 }
134 }
135 return res;
136 }
137
138 protected String getExceptionWord(ArrayList ex) {
139 StringBuffer res = new StringBuffer ();
140 for (int i = 0; i < ex.size(); i++) {
141 Object item = ex.get(i);
142 if (item instanceof String ) {
143 res.append((String )item);
144 } else {
145 if (((Hyphen)item).noBreak != null) {
146 res.append(((Hyphen)item).noBreak);
147 }
148 }
149 }
150 return res.toString();
151 }
152
153 protected static String getInterletterValues(String pat) {
154 StringBuffer il = new StringBuffer ();
155 String word = pat + "a"; int len = word.length();
157 for (int i = 0; i < len; i++) {
158 char c = word.charAt(i);
159 if (Character.isDigit(c)) {
160 il.append(c);
161 i++;
162 } else {
163 il.append('0');
164 }
165 }
166 return il.toString();
167 }
168
169 public void endDocument() {
170 }
171
172 public void endElement(String tag) {
173 if (token.length() > 0) {
174 String word = token.toString();
175 switch (currElement) {
176 case ELEM_CLASSES:
177 consumer.addClass(word);
178 break;
179 case ELEM_EXCEPTIONS:
180 exception.add(word);
181 exception = normalizeException(exception);
182 consumer.addException(getExceptionWord(exception),
183 (ArrayList )exception.clone());
184 break;
185 case ELEM_PATTERNS:
186 consumer.addPattern(getPattern(word),
187 getInterletterValues(word));
188 break;
189 case ELEM_HYPHEN:
190 break;
192 }
193 if (currElement != ELEM_HYPHEN) {
194 token.setLength(0);
195 }
196 }
197 if (currElement == ELEM_HYPHEN) {
198 currElement = ELEM_EXCEPTIONS;
199 } else {
200 currElement = 0;
201 }
202 }
203
204 public void startDocument() {
205 }
206
207 public void startElement(String tag, java.util.HashMap h) {
208 if (tag.equals("hyphen-char")) {
209 String hh = (String )h.get("value");
210 if (hh != null && hh.length() == 1) {
211 hyphenChar = hh.charAt(0);
212 }
213 } else if (tag.equals("classes")) {
214 currElement = ELEM_CLASSES;
215 } else if (tag.equals("patterns")) {
216 currElement = ELEM_PATTERNS;
217 } else if (tag.equals("exceptions")) {
218 currElement = ELEM_EXCEPTIONS;
219 exception = new ArrayList ();
220 } else if (tag.equals("hyphen")) {
221 if (token.length() > 0) {
222 exception.add(token.toString());
223 }
224 exception.add(new Hyphen((String )h.get("pre"),
225 (String )h.get("no"),
226 (String )h.get("post")));
227 currElement = ELEM_HYPHEN;
228 }
229 token.setLength(0);
230 }
231
232 public void text(String str) {
233 StringTokenizer tk = new StringTokenizer (str);
234 while (tk.hasMoreTokens()) {
235 String word = tk.nextToken();
236 switch (currElement) {
238 case ELEM_CLASSES:
239 consumer.addClass(word);
240 break;
241 case ELEM_EXCEPTIONS:
242 exception.add(word);
243 exception = normalizeException(exception);
244 consumer.addException(getExceptionWord(exception),
245 (ArrayList )exception.clone());
246 exception.clear();
247 break;
248 case ELEM_PATTERNS:
249 consumer.addPattern(getPattern(word),
250 getInterletterValues(word));
251 break;
252 }
253 }
254 }
255
256 public void addClass(String c) {
258 System.out.println("class: " + c);
259 }
260
261 public void addException(String w, ArrayList e) {
262 System.out.println("exception: " + w + " : " + e.toString());
263 }
264
265 public void addPattern(String p, String v) {
266 System.out.println("pattern: " + p + " : " + v);
267 }
268
269 public static void main(String [] args) throws Exception {
270 try {
271 if (args.length > 0) {
272 SimplePatternParser pp = new SimplePatternParser();
273 pp.parse(new FileInputStream (args[0]), pp);
274 }
275 }
276 catch (Exception e) {
277 e.printStackTrace();
278 }
279 }
280 }
281 |
Popular Tags |
Java API By Example, From Geeks To Geeks.