|
1
47
48 package com.lowagie.text.html.simpleparser;
49
50 import java.io.File  ;
51 import java.io.IOException ;
52 import java.io.Reader ;
53 import java.util.ArrayList ;
54 import java.util.HashMap ;
55 import java.util.Stack ;
56 import java.util.StringTokenizer ;
57
58 import com.lowagie.text.Chunk;
59 import com.lowagie.text.DocListener;
60 import com.lowagie.text.DocumentException;
61 import com.lowagie.text.Element;
62 import com.lowagie.text.ExceptionConverter;
63 import com.lowagie.text.FontFactoryImp;
64 import com.lowagie.text.HeaderFooter;
65 import com.lowagie.text.Image;
66 import com.lowagie.text.ListItem;
67 import com.lowagie.text.Paragraph;
68 import com.lowagie.text.Phrase;
69 import com.lowagie.text.Rectangle;
70 import com.lowagie.text.TextElementArray;
71 import com.lowagie.text.pdf.PdfPTable;
72 import com.lowagie.text.xml.simpleparser.SimpleXMLDocHandler;
73 import com.lowagie.text.xml.simpleparser.SimpleXMLParser;
74
75 public class HTMLWorker implements SimpleXMLDocHandler, DocListener {
76
77 protected ArrayList objectList;
78 protected DocListener document;
79 private Paragraph currentParagraph;
80 private ChainedProperties cprops = new ChainedProperties();
81 private Stack stack = new Stack ();
82 private boolean pendingTR = false;
83 private boolean pendingTD = false;
84 private boolean pendingLI = false;
85 private StyleSheet style = new StyleSheet();
86 private boolean isPRE = false;
87 private Stack tableState = new Stack ();
88 private boolean skipText = false;
89 private HashMap interfaceProps;
90 private FactoryProperties factoryProperties = new FactoryProperties();
91
92
93 public HTMLWorker(DocListener document) {
94 this.document = document;
95 }
96
97 public void setStyleSheet(StyleSheet style) {
98 this.style = style;
99 }
100
101 public StyleSheet getStyleSheet() {
102 return style;
103 }
104
105 public void setInterfaceProps(HashMap interfaceProps) {
106 this.interfaceProps = interfaceProps;
107 FontFactoryImp ff = null;
108 if (interfaceProps != null)
109 ff = (FontFactoryImp)interfaceProps.get("font_factory");
110 if (ff != null)
111 factoryProperties.setFontImp(ff);
112 }
113
114 public HashMap getInterfaceProps() {
115 return interfaceProps;
116 }
117
118 public void parse(Reader reader) throws IOException {
119 SimpleXMLParser.parse(this, null, reader, true);
120 }
121
122 public static ArrayList parseToList(Reader reader, StyleSheet style) throws IOException {
123 return parseToList(reader, style, null);
124 }
125
126 public static ArrayList parseToList(Reader reader, StyleSheet style, HashMap interfaceProps) throws IOException {
127 HTMLWorker worker = new HTMLWorker(null);
128 if (style != null)
129 worker.style = style;
130 worker.document = worker;
131 worker.setInterfaceProps(interfaceProps);
132 worker.objectList = new ArrayList ();
133 worker.parse(reader);
134 return worker.objectList;
135 }
136
137 public void endDocument() {
138 try {
139 for (int k = 0; k < stack.size(); ++k)
140 document.add((Element)stack.elementAt(k));
141 if (currentParagraph != null)
142 document.add(currentParagraph);
143 currentParagraph = null;
144 }
145 catch (Exception e) {
146 throw new ExceptionConverter(e);
147 }
148 }
149
150 public void startDocument() {
151 HashMap h = new HashMap ();
152 style.applyStyle("body", h);
153 cprops.addToChain("body", h);
154 }
155
156 public void startElement(String tag, HashMap h) {
157 if (!tagsSupported.containsKey(tag))
158 return;
159 try {
160 style.applyStyle(tag, h);
161 String follow = (String )FactoryProperties.followTags.get(tag);
162 if (follow != null) {
163 HashMap prop = new HashMap ();
164 prop.put(follow, null);
165 cprops.addToChain(follow, prop);
166 return;
167 }
168 FactoryProperties.insertStyle(h);
169 if (tag.equals("a")) {
170 cprops.addToChain(tag, h);
171 if (currentParagraph == null)
172 currentParagraph = new Paragraph();
173 stack.push(currentParagraph);
174 currentParagraph = new Paragraph();
175 return;
176 }
177 if (tag.equals("br")) {
178 if (currentParagraph == null)
179 currentParagraph = new Paragraph();
180 currentParagraph.add(factoryProperties.createChunk("\n", cprops));
181 return;
182 }
183 if (tag.equals("font") || tag.equals("span")) {
184 cprops.addToChain(tag, h);
185 return;
186 }
187 if (tag.equals("img")) {
188 String src = (String )h.get("src");
189 if (src == null)
190 return;
191 cprops.addToChain(tag, h);
192 Image img = null;
193 if (interfaceProps != null) {
194 ImageProvider ip = (ImageProvider)interfaceProps.get("img_provider");
195 if (ip != null)
196 img = ip.getImage(src, h, cprops, document);
197 if (img == null) {
198 HashMap images = (HashMap )interfaceProps.get("img_static");
199 if (images != null) {
200 Image tim = (Image)images.get(src);
201 if (tim != null)
202 img = Image.getInstance(tim);
203 } else {
204 if (!src.startsWith("http")) { String baseurl = (String )interfaceProps.get("img_baseurl");
206 if (baseurl != null) {
207 src = baseurl+src;
208 img = Image.getInstance(src);
209 }
210 }
211 }
212 }
213 }
214 if (img == null) {
215 if (!src.startsWith("http")) {
216 String path = cprops.getProperty("image_path");
217 if (path == null)
218 path = "";
219 src = new File (path, src).getPath();
220 }
221 img = Image.getInstance(src);
222 }
223 String align = (String )h.get("align");
224 String width = (String )h.get("width");
225 String height = (String )h.get("height");
226 String before = cprops.getProperty("before");
227 String after = cprops.getProperty("after");
228 if (before != null)
229 img.setSpacingBefore(Float.parseFloat(before));
230 if (after != null)
231 img.setSpacingAfter(Float.parseFloat(after));
232 float wp = lengthParse(width, (int)img.getWidth());
233 float lp = lengthParse(height, (int)img.getHeight());
234 if (wp > 0 && lp > 0)
235 img.scalePercent(wp > lp ? lp : wp);
236 else if (wp > 0)
237 img.scalePercent(wp);
238 else if (lp > 0)
239 img.scalePercent(lp);
240 img.setWidthPercentage(0);
241 if (align != null) {
242 endElement("p");
243 int ralign = Image.MIDDLE;
244 if (align.equalsIgnoreCase("left"))
245 ralign = Image.LEFT;
246 else if (align.equalsIgnoreCase("right"))
247 ralign = Image.RIGHT;
248 img.setAlignment(ralign);
249 Img i = null;
250 boolean skip = false;
251 if (interfaceProps != null) {
252 i = (Img)interfaceProps.get("img_interface");
253 if (i != null)
254 skip = i.process(img, h, cprops, document);
255 }
256 if (!skip)
257 document.add(img);
258 cprops.removeChain(tag);
259 }
260 else {
261 cprops.removeChain(tag);
262 if (currentParagraph == null)
263 currentParagraph = FactoryProperties.createParagraph(cprops);
264 currentParagraph.add(new Chunk(img, 0, 0));
265 }
266 return;
267 }
268 endElement("p");
269 if (tag.equals("h1") || tag.equals("h2") || tag.equals("h3") || tag.equals("h4") || tag.equals("h5") || tag.equals("h6")) {
270 if (!h.containsKey("size")) {
271 int v = 7 - Integer.parseInt(tag.substring(1));
272 h.put("size", Integer.toString(v));
273 }
274 cprops.addToChain(tag, h);
275 return;
276 }
277 if (tag.equals("ul")) {
278 if (pendingLI)
279 endElement("li");
280 skipText = true;
281 cprops.addToChain(tag, h);
282 com.lowagie.text.List list = new com.lowagie.text.List(false, 10);
283 list.setListSymbol("\u2022");
284 stack.push(list);
285 return;
286 }
287 if (tag.equals("ol")) {
288 if (pendingLI)
289 endElement("li");
290 skipText = true;
291 cprops.addToChain(tag, h);
292 com.lowagie.text.List list = new com.lowagie.text.List(true, 10);
293 stack.push(list);
294 return;
295 }
296 if (tag.equals("li")) {
297 if (pendingLI)
298 endElement("li");
299 skipText = false;
300 pendingLI = true;
301 cprops.addToChain(tag, h);
302 stack.push(FactoryProperties.createListItem(cprops));
303 return;
304 }
305 if (tag.equals("div") || tag.equals("body")) {
306 cprops.addToChain(tag, h);
307 return;
308 }
309 if (tag.equals("pre")) {
310 if (!h.containsKey("face")) {
311 h.put("face", "Courier");
312 }
313 cprops.addToChain(tag, h);
314 isPRE = true;
315 return;
316 }
317 if (tag.equals("p")) {
318 cprops.addToChain(tag, h);
319 currentParagraph = FactoryProperties.createParagraph(h);
320 return;
321 }
322 if (tag.equals("tr")) {
323 if (pendingTR)
324 endElement("tr");
325 skipText = true;
326 pendingTR = true;
327 cprops.addToChain("tr", h);
328 return;
329 }
330 if (tag.equals("td") || tag.equals("th")) {
331 if (pendingTD)
332 endElement(tag);
333 skipText = false;
334 pendingTD = true;
335 cprops.addToChain("td", h);
336 stack.push(new IncCell(tag, cprops));
337 return;
338 }
339 if (tag.equals("table")) {
340 cprops.addToChain("table", h);
341 IncTable table = new IncTable(h);
342 stack.push(table);
343 tableState.push(new boolean[]{pendingTR, pendingTD});
344 pendingTR = pendingTD = false;
345 skipText = true;
346 return;
347 }
348 }
349 catch (Exception e) {
350 throw new ExceptionConverter(e);
351 }
352 }
353
354 public void endElement(String tag) {
355 if (!tagsSupported.containsKey(tag))
356 return;
357 try {
358 String follow = (String )FactoryProperties.followTags.get(tag);
359 if (follow != null) {
360 cprops.removeChain(follow);
361 return;
362 }
363 if (tag.equals("font") || tag.equals("span")) {
364 cprops.removeChain(tag);
365 return;
366 }
367 if (tag.equals("a")) {
368 if (currentParagraph == null)
369 currentParagraph = new Paragraph();
370 ALink i = null;
371 boolean skip = false;
372 if (interfaceProps != null) {
373 i = (ALink)interfaceProps.get("alink_interface");
374 if (i != null)
375 skip = i.process(currentParagraph, cprops);
376 }
377 if (!skip) {
378 String href = cprops.getProperty("href");
379 if (href != null) {
380 ArrayList chunks = currentParagraph.getChunks();
381 for (int k = 0; k < chunks.size(); ++k) {
382 Chunk ck = (Chunk)chunks.get(k);
383 ck.setAnchor(href);
384 }
385 }
386 }
387 Paragraph tmp = (Paragraph)stack.pop();
388 Phrase tmp2 = new Phrase();
389 tmp2.add(currentParagraph);
390 tmp.add(tmp2);
391 currentParagraph = tmp;
392 cprops.removeChain("a");
393 return;
394 }
395 if (tag.equals("br")) {
396 return;
397 }
398 if (currentParagraph != null) {
399 if (stack.empty())
400 document.add(currentParagraph);
401 else {
402 Object obj = stack.pop();
403 if (obj instanceof TextElementArray) {
404 TextElementArray current = (TextElementArray)obj;
405 current.add(currentParagraph);
406 }
407 stack.push(obj);
408 }
409 }
410 currentParagraph = null;
411 if (tag.equals("ul") || tag.equals("ol")) {
412 if (pendingLI)
413 endElement("li");
414 skipText = false;
415 cprops.removeChain(tag);
416 if (stack.empty())
417 return;
418 Object obj = stack.pop();
419 if (!(obj instanceof com.lowagie.text.List)) {
420 stack.push(obj);
421 return;
422 }
423 if (stack.empty())
424 document.add((Element)obj);
425 else
426 ((TextElementArray)stack.peek()).add(obj);
427 return;
428 }
429 if (tag.equals("li")) {
430 pendingLI = false;
431 skipText = true;
432 cprops.removeChain(tag);
433 if (stack.empty())
434 return;
435 Object obj = stack.pop();
436 if (!(obj instanceof ListItem)) {
437 stack.push(obj);
438 return;
439 }
440 if (stack.empty()) {
441 document.add((Element)obj);
442 return;
443 }
444 Object list = stack.pop();
445 if (!(list instanceof com.lowagie.text.List)) {
446 stack.push(list);
447 return;
448 }
449 ListItem item = (ListItem)obj;
450 ((com.lowagie.text.List)list).add(item);
451 ArrayList cks = item.getChunks();
452 if (!cks.isEmpty())
453 item.getListSymbol().setFont(((Chunk)cks.get(0)).getFont());
454 stack.push(list);
455 return;
456 }
457 if (tag.equals("div") || tag.equals("body")) {
458 cprops.removeChain(tag);
459 return;
460 }
461 if (tag.equals("pre")) {
462 cprops.removeChain(tag);
463 isPRE = false;
464 return;
465 }
466 if (tag.equals("p")) {
467 cprops.removeChain(tag);
468 return;
469 }
470 if (tag.equals("h1") || tag.equals("h2") || tag.equals("h3") || tag.equals("h4") || tag.equals("h5") || tag.equals("h6")) {
471 cprops.removeChain(tag);
472 return;
473 }
474 if (tag.equals("table")) {
475 if (pendingTR)
476 endElement("tr");
477 cprops.removeChain("table");
478 IncTable table = (IncTable) stack.pop();
479 PdfPTable tb = table.buildTable();
480 tb.setSplitRows(true);
481 if (stack.empty())
482 document.add(tb);
483 else
484 ((TextElementArray)stack.peek()).add(tb);
485 boolean state[] = (boolean[])tableState.pop();
486 pendingTR = state[0];
487 pendingTD = state[1];
488 skipText = false;
489 return;
490 }
491 if (tag.equals("tr")) {
492 if (pendingTD)
493 endElement("td");
494 pendingTR = false;
495 cprops.removeChain("tr");
496 ArrayList cells = new ArrayList ();
497 IncTable table = null;
498 while (true) {
499 Object obj = stack.pop();
500 if (obj instanceof IncCell) {
501 cells.add(((IncCell)obj).getCell());
502 }
503 if (obj instanceof IncTable) {
504 table = (IncTable)obj;
505 break;
506 }
507 }
508 table.addCols(cells);
509 table.endRow();
510 stack.push(table);
511 skipText = true;
512 return;
513 }
514 if (tag.equals("td") || tag.equals("th")) {
515 pendingTD = false;
516 cprops.removeChain("td");
517 skipText = true;
518 return;
519 }
520 }
521 catch (Exception e) {
522 throw new ExceptionConverter(e);
523 }
524 }
525
526 public void text(String str) {
527 if (skipText)
528 return;
529 String content = str;
530 if (isPRE) {
531 if (currentParagraph == null)
532 currentParagraph = new Paragraph();
533 currentParagraph.add(factoryProperties.createChunk(content, cprops));
534 return;
535 }
536 if (content.trim().length() == 0 && content.indexOf(' ') < 0) {
537 return;
538 }
539
540 StringBuffer buf = new StringBuffer ();
541 int len = content.length();
542 char character;
543 boolean newline = false;
544 for (int i = 0; i < len; i++) {
545 switch(character = content.charAt(i)) {
546 case ' ':
547 if (!newline) {
548 buf.append(character);
549 }
550 break;
551 case '\n':
552 if (i > 0) {
553 newline = true;
554 buf.append(' ');
555 }
556 break;
557 case '\r':
558 break;
559 case '\t':
560 break;
561 default:
562 newline = false;
563 buf.append(character);
564 }
565 }
566 if (currentParagraph == null)
567 currentParagraph = FactoryProperties.createParagraph(cprops);
568 currentParagraph.add(factoryProperties.createChunk(buf.toString(), cprops));
569 }
570
571 public boolean add(Element element) throws DocumentException {
572 objectList.add(element);
573 return true;
574 }
575
576 public void clearTextWrap() throws DocumentException {
577 }
578
579 public void close() {
580 }
581
582 public boolean newPage() {
583 return true;
584 }
585
586 public void open() {
587 }
588
589 public void resetFooter() {
590 }
591
592 public void resetHeader() {
593 }
594
595 public void resetPageCount() {
596 }
597
598 public void setFooter(HeaderFooter footer) {
599 }
600
601 public void setHeader(HeaderFooter header) {
602 }
603
604 public boolean setMarginMirroring(boolean marginMirroring) {
605 return true;
606 }
607
608 public boolean setMargins(float marginLeft, float marginRight, float marginTop, float marginBottom) {
609 return true;
610 }
611
612 public void setPageCount(int pageN) {
613 }
614
615 public boolean setPageSize(Rectangle pageSize) {
616 return true;
617 }
618
619 public static final String tagsSupportedString = "ol ul li a pre font span br p div body table td th tr i b u sub sup em strong"
620 + " h1 h2 h3 h4 h5 h6 img";
621
622 public static final HashMap tagsSupported = new HashMap ();
623
624 static {
625 StringTokenizer tok = new StringTokenizer (tagsSupportedString);
626 while (tok.hasMoreTokens())
627 tagsSupported.put(tok.nextToken(), null);
628 }
629
630 private static float lengthParse(String txt, int c) {
631 if (txt == null)
632 return -1;
633 if (txt.endsWith("%")) {
634 float vf = Float.parseFloat(txt.substring(0, txt.length() - 1));
635 return vf;
636 }
637 int v = Integer.parseInt(txt);
638 return (float)v / c * 100f;
639 }
640 }
641 |
Popular Tags |
Java API By Example, From Geeks To Geeks.