1 | package org.expeditee.io;
|
---|
2 |
|
---|
3 | import java.awt.Color;
|
---|
4 | import java.awt.Font;
|
---|
5 | import java.awt.Graphics;
|
---|
6 | import java.awt.image.BufferedImage;
|
---|
7 | import java.io.File;
|
---|
8 | import java.io.IOException;
|
---|
9 | import java.lang.reflect.InvocationTargetException;
|
---|
10 | import java.net.HttpURLConnection;
|
---|
11 | import java.net.MalformedURLException;
|
---|
12 | import java.net.URL;
|
---|
13 | import java.text.SimpleDateFormat;
|
---|
14 | import java.util.Arrays;
|
---|
15 |
|
---|
16 | /*
|
---|
17 | * JavaFX is not on the default java classpath until Java 8 (but is still included with Java 7), so your IDE will probably complain that the imports below can't be resolved.
|
---|
18 | * In Eclipse hitting'Proceed' when told 'Errors exist in project' should allow you to run Expeditee without any issues (although the JFX Browser widget will not display),
|
---|
19 | * or you can just exclude JfxBrowser, WebParser and JfxbrowserActions from the build path.
|
---|
20 | *
|
---|
21 | * If you are using Ant to build/run, 'ant build' will try to build with JavaFX jar added to the classpath.
|
---|
22 | * If this fails, 'ant build-nojfx' will build with the JfxBrowser, WebParser and JfxbrowserActions excluded from the build path.
|
---|
23 | */
|
---|
24 |
|
---|
25 |
|
---|
26 | import java.util.Date;
|
---|
27 |
|
---|
28 | import javafx.animation.AnimationTimer;
|
---|
29 | import javafx.application.Platform;
|
---|
30 | import javafx.beans.value.ChangeListener;
|
---|
31 | import javafx.beans.value.ObservableValue;
|
---|
32 | import javafx.concurrent.Worker.State;
|
---|
33 | import javafx.scene.transform.Rotate;
|
---|
34 | import javafx.scene.web.WebEngine;
|
---|
35 |
|
---|
36 | import javax.imageio.ImageIO;
|
---|
37 | import javax.swing.JComponent;
|
---|
38 |
|
---|
39 | import netscape.javascript.JSObject;
|
---|
40 |
|
---|
41 | import org.expeditee.gui.DisplayIO;
|
---|
42 | import org.expeditee.gui.Frame;
|
---|
43 | import org.expeditee.gui.FrameCreator;
|
---|
44 | import org.expeditee.gui.FrameGraphics;
|
---|
45 | import org.expeditee.gui.FrameIO;
|
---|
46 | import org.expeditee.gui.FrameUtils;
|
---|
47 | import org.expeditee.gui.MessageBay;
|
---|
48 | import org.expeditee.gui.MessageBay.Progress;
|
---|
49 | import org.expeditee.importer.FrameDNDTransferHandler;
|
---|
50 | import org.expeditee.items.ItemUtils;
|
---|
51 | import org.expeditee.items.Justification;
|
---|
52 | import org.expeditee.items.Picture;
|
---|
53 | import org.expeditee.items.Text;
|
---|
54 | import org.w3c.dom.Node;
|
---|
55 | import org.w3c.dom.html.HTMLBodyElement;
|
---|
56 |
|
---|
57 | import com.sun.org.apache.bcel.internal.generic.NEW;
|
---|
58 |
|
---|
59 | /**
|
---|
60 | * Methods to convert webpages to Expeditee frames
|
---|
61 | *
|
---|
62 | * @author ngw8
|
---|
63 | * @author jts21
|
---|
64 | */
|
---|
65 | public class WebParser {
|
---|
66 |
|
---|
67 |
|
---|
68 | /**
|
---|
69 | * Loads a webpage and renders it as Expeditee frame(s)
|
---|
70 | *
|
---|
71 | * @param URL
|
---|
72 | * Page to load
|
---|
73 | * @param frame
|
---|
74 | * The Expeditee frame to output the converted page to
|
---|
75 | */
|
---|
76 | public static void parseURL(final String URL, final Frame frame) {
|
---|
77 | try {
|
---|
78 | Platform.runLater(new Runnable() {
|
---|
79 | @Override
|
---|
80 | public void run() {
|
---|
81 | try {
|
---|
82 | WebEngine webEngine = new WebEngine(URL);
|
---|
83 | loadPage(webEngine, frame);
|
---|
84 | } catch (Exception e) {
|
---|
85 | e.printStackTrace();
|
---|
86 | }
|
---|
87 | }
|
---|
88 | });
|
---|
89 | } catch (Exception e) {
|
---|
90 | e.printStackTrace();
|
---|
91 | }
|
---|
92 | }
|
---|
93 |
|
---|
94 | protected static void loadPage(final WebEngine webEngine, final Frame frame) throws Exception {
|
---|
95 | webEngine.getLoadWorker().stateProperty().addListener(new ChangeListener<State>() {
|
---|
96 |
|
---|
97 | @Override
|
---|
98 | public void changed(ObservableValue<? extends State> ov, State oldState, State newState) {
|
---|
99 |
|
---|
100 | switch (newState) {
|
---|
101 | case READY: // READY
|
---|
102 | // MessageBay.displayMessage("WebEngine ready");
|
---|
103 | break;
|
---|
104 | case SCHEDULED: // SCHEDULED
|
---|
105 | // MessageBay.displayMessage("Scheduled page load");
|
---|
106 | break;
|
---|
107 | case RUNNING: // RUNNING
|
---|
108 | System.out.println("Loading page!");
|
---|
109 | // MessageBay.displayMessage("WebEngine running");
|
---|
110 | break;
|
---|
111 | case SUCCEEDED: // SUCCEEDED
|
---|
112 | // MessageBay.displayMessage("Finished loading page");
|
---|
113 | System.out.println("Parsing page!");
|
---|
114 | webEngine.executeScript("window.resizeTo(800, 800);"
|
---|
115 | + "document.body.style.width = '1000px'");
|
---|
116 | parsePage(webEngine, frame);
|
---|
117 | System.out.println("Parsed page!");
|
---|
118 | break;
|
---|
119 | case CANCELLED: // CANCELLED
|
---|
120 | MessageBay.displayMessage("Cancelled loading page");
|
---|
121 | break;
|
---|
122 | case FAILED: // FAILED
|
---|
123 | MessageBay.displayMessage("Failed to load page");
|
---|
124 | break;
|
---|
125 | }
|
---|
126 | }
|
---|
127 | });
|
---|
128 | }
|
---|
129 |
|
---|
130 | /**
|
---|
131 | * Converts a loaded page to Expeditee frame(s)
|
---|
132 | *
|
---|
133 | * @param webEngine
|
---|
134 | * The JavaFX WebEngine in which the page to be converted is loaded
|
---|
135 | * @param frame
|
---|
136 | * The Expeditee frame to output the converted page to
|
---|
137 | */
|
---|
138 | public static void parsePage(final WebEngine webEngine, final Frame frame) {
|
---|
139 | try {
|
---|
140 | Platform.runLater(new Runnable() {
|
---|
141 | @Override
|
---|
142 | public void run() {
|
---|
143 | try {
|
---|
144 | Progress progressBar = MessageBay.displayProgress("Converting web page");
|
---|
145 |
|
---|
146 | Node doc = (Node) webEngine.executeScript("document.body");
|
---|
147 |
|
---|
148 | JSObject window = (JSObject) webEngine.executeScript("window");
|
---|
149 |
|
---|
150 | frame.setBackgroundColor(rgbStringToColor((String) ((JSObject) (window.call("getComputedStyle", new Object[] { doc }))).call("getPropertyValue",
|
---|
151 | new Object[] { "background-color" })));
|
---|
152 |
|
---|
153 | // Functions to be used later in JavaScript
|
---|
154 | webEngine.executeScript(""
|
---|
155 | + "function addToSpan(text) {"
|
---|
156 | + " span = document.createElement('wordSpan');"
|
---|
157 | + " span.textContent = text;"
|
---|
158 | + " par.insertBefore(span, refNode);"
|
---|
159 | // Checking if the current word is on a new line (i.e. lower than the previous word)
|
---|
160 | + " if (prevSpan !== null && span.getBoundingClientRect().top > prevSpan.getBoundingClientRect().top) {"
|
---|
161 | // If it is, prepend a new line character to it. The new line characters doesn't affect the rendered HTML
|
---|
162 | + " span.textContent = '\\n' + span.textContent;"
|
---|
163 |
|
---|
164 | // Checking if the previous word is horizontally aligned with the one before it.
|
---|
165 | // If it is, merge the text of the two spans
|
---|
166 | + " if ( prevPrevSpan !== null && prevPrevSpan.getBoundingClientRect().left == prevSpan.getBoundingClientRect().left) {"
|
---|
167 | + " prevPrevSpan.textContent = prevPrevSpan.textContent + prevSpan.textContent;"
|
---|
168 | + " par.removeChild(prevSpan);"
|
---|
169 | + " } else {"
|
---|
170 | + " prevPrevSpan = prevSpan;"
|
---|
171 | + " }"
|
---|
172 | + " prevSpan = span;"
|
---|
173 | + " } else if ( prevSpan !== null) {"
|
---|
174 | // Word is on the same line as the previous one, so merge the second into the span of the first
|
---|
175 | + " prevSpan.textContent = prevSpan.textContent + span.textContent;"
|
---|
176 | + " par.removeChild(span);"
|
---|
177 | + " } else {"
|
---|
178 | + " prevSpan = span;"
|
---|
179 | + " }"
|
---|
180 | + "}"
|
---|
181 |
|
---|
182 | + "function splitIntoWords(toSplit) {"
|
---|
183 | + " var words = [];"
|
---|
184 | + " var pattern = /\\s+/g;"
|
---|
185 | + " var words = toSplit.split(pattern);"
|
---|
186 | + ""
|
---|
187 | + " for (var i = 0; i < words.length - 1; i++) {"
|
---|
188 | + " words[i] = words[i] + ' ';"
|
---|
189 | + " }"
|
---|
190 | + " return words;"
|
---|
191 | + "}"
|
---|
192 | );
|
---|
193 |
|
---|
194 | // Using Javascript to get an array of all the text nodes in the document so they can be wrapped in spans. Have to
|
---|
195 | // loop through twice (once to build the array and once actually going through the array, otherwise when the
|
---|
196 | // textnode is removed from the document items end up being skipped)
|
---|
197 | JSObject textNodes = (JSObject) webEngine.executeScript(""
|
---|
198 | + "function getTextNodes(rootNode){"
|
---|
199 | + "var node;"
|
---|
200 | + "var textNodes=[];"
|
---|
201 | + "var walk = document.createTreeWalker(rootNode, NodeFilter.SHOW_TEXT);"
|
---|
202 | + "while(node=walk.nextNode()) {"
|
---|
203 | + "if((node.textContent.trim().length > 0)) { "
|
---|
204 | + "textNodes.push(node);"
|
---|
205 | + "}"
|
---|
206 | + "}"
|
---|
207 | + "return textNodes;"
|
---|
208 | + "}; "
|
---|
209 | + "getTextNodes(document.body)"
|
---|
210 | );
|
---|
211 |
|
---|
212 | int nodesLength = (Integer) textNodes.getMember("length");
|
---|
213 |
|
---|
214 | // Looping through all the text nodes in the document
|
---|
215 | for (int j = 0; j < nodesLength; j++) {
|
---|
216 | Node currentNode = (Node) textNodes.getSlot(j);
|
---|
217 |
|
---|
218 | // Making the current node accessible in JavaScript
|
---|
219 | window.setMember("currentNode", currentNode);
|
---|
220 |
|
---|
221 | webEngine.executeScript(""
|
---|
222 | + "var span = null, prevSpan = null, prevPrevSpan = null;"
|
---|
223 |
|
---|
224 | // Removing repeated whitespace from the text node's content then splitting it into individual words
|
---|
225 | + "var textContent = currentNode.textContent.replace(/\\n|\\r/g, '').replace(/\\s+/g, ' ');"
|
---|
226 | + "var words = splitIntoWords(textContent);"
|
---|
227 |
|
---|
228 | + "var refNode = currentNode.nextSibling;"
|
---|
229 | + "var par = currentNode.parentElement;"
|
---|
230 | + "currentNode.parentElement.removeChild(currentNode);"
|
---|
231 |
|
---|
232 | + "for (var i = 0; i < words.length; i++) {"
|
---|
233 | + " addToSpan(words[i]);"
|
---|
234 | + "}"
|
---|
235 |
|
---|
236 | + "if (prevPrevSpan !== null && prevPrevSpan.getBoundingClientRect().left == prevSpan.getBoundingClientRect().left) {"
|
---|
237 | + " prevPrevSpan.textContent = prevPrevSpan.textContent + prevSpan.textContent;"
|
---|
238 | + " par.removeChild(prevSpan);"
|
---|
239 | + "}"
|
---|
240 | );
|
---|
241 |
|
---|
242 | // Will never reach 100% here, as the processing is not quite finished - progress is set to 100% at the end of
|
---|
243 | // the addPageToFrame loop below
|
---|
244 | progressBar.set((100 * (j)) / nodesLength);
|
---|
245 | }
|
---|
246 |
|
---|
247 | // Finding all links within the page, then setting the href attribute of all their descendants to be the same
|
---|
248 | // link/URL.
|
---|
249 | // This is needed because there is no apparent and efficient way to check if an element is a child of a link when
|
---|
250 | // running through the document when added each element to Expeditee
|
---|
251 | webEngine.executeScript(""
|
---|
252 | + "var anchors = document.getElementsByTagName('a');"
|
---|
253 | + ""
|
---|
254 | + "for (var i = 0; i < anchors.length; i++) {"
|
---|
255 | + "var currentAnchor = anchors.item(i);"
|
---|
256 | + "var anchorDescendants = currentAnchor.querySelectorAll('*');"
|
---|
257 | + "for (var j = 0; j < anchorDescendants.length; j++) {"
|
---|
258 | + "anchorDescendants.item(j).href = currentAnchor.href;"
|
---|
259 | + "}"
|
---|
260 | + "}"
|
---|
261 | );
|
---|
262 |
|
---|
263 | WebParser.addPageToFrame(doc, window, webEngine, frame);
|
---|
264 |
|
---|
265 | progressBar.set(100);
|
---|
266 |
|
---|
267 | } catch (Exception e) {
|
---|
268 | e.printStackTrace();
|
---|
269 | }
|
---|
270 | System.out.println("Parsed frame");
|
---|
271 | FrameUtils.Parse(frame);
|
---|
272 | frame.setChanged(true);
|
---|
273 | FrameIO.SaveFrame(frame);
|
---|
274 | }
|
---|
275 | });
|
---|
276 | } catch (Exception e) {
|
---|
277 | e.printStackTrace();
|
---|
278 | }
|
---|
279 | }
|
---|
280 |
|
---|
281 | /**
|
---|
282 | * Converts a loaded page to Expeditee frame(s)
|
---|
283 | *
|
---|
284 | * @param webEngine
|
---|
285 | * The JavaFX WebEngine in which the page to be converted is loaded
|
---|
286 | * @param frame
|
---|
287 | * The Expeditee frame to output the converted page to
|
---|
288 | */
|
---|
289 | public static void parsePageSimple(final WebEngine webEngine, final Object webView, final JComponent jfxPanel, final Frame frame) {
|
---|
290 | try {
|
---|
291 | final Object notifier = new Object();
|
---|
292 | final MutableBool bottomReached = new MutableBool(false);
|
---|
293 |
|
---|
294 | String pageTitle = webEngine.getTitle();
|
---|
295 |
|
---|
296 | String framesetName = FrameIO.ConvertToValidFramesetName((new SimpleDateFormat("yy-MM-dd-HH-mm-ss").format(new Date())) + pageTitle);
|
---|
297 |
|
---|
298 | final Progress progressBar = MessageBay.displayProgress("Converting web page");
|
---|
299 |
|
---|
300 | final Frame frameset = FrameIO.CreateNewFrameset(framesetName);
|
---|
301 |
|
---|
302 | frameset.setTitle(pageTitle);
|
---|
303 | frameset.getTitleItem().setSize(14);
|
---|
304 |
|
---|
305 | Text link = new Text(DisplayIO.getCurrentFrame().getNextItemID(), webEngine.getTitle());
|
---|
306 |
|
---|
307 | link.setPosition(100, 100);
|
---|
308 |
|
---|
309 | DisplayIO.getCurrentFrame().addItem(link);
|
---|
310 |
|
---|
311 | link.setLink(framesetName + "1");
|
---|
312 |
|
---|
313 | // Timer that fires every time JFX is redrawn. After a few redraws, the handle method of this takes a screenshot of the page,
|
---|
314 | // adds it to the frame, then adds the text on top
|
---|
315 | AnimationTimer timer = new AnimationTimer() {
|
---|
316 |
|
---|
317 | int frameCount = 0;
|
---|
318 | Frame frameToAddTo = frameset;
|
---|
319 |
|
---|
320 | @Override
|
---|
321 | public void handle(long arg0) {
|
---|
322 | System.out.println(frameCount);
|
---|
323 | // Must wait 2 frames before taking a snapshot of the webview, otherwise JavaFX won't have redrawn
|
---|
324 | if (frameCount++ > 3) {
|
---|
325 | frameCount = 0;
|
---|
326 | this.stop();
|
---|
327 |
|
---|
328 | BufferedImage image = new BufferedImage(jfxPanel.getWidth(), jfxPanel.getHeight(), BufferedImage.TYPE_INT_ARGB);
|
---|
329 |
|
---|
330 | Graphics graphics = image.createGraphics();
|
---|
331 |
|
---|
332 | // Drawing the JfxPanel (containing the webview) to the image
|
---|
333 | jfxPanel.paint(graphics);
|
---|
334 |
|
---|
335 | try {
|
---|
336 | frameToAddTo = FrameIO.CreateFrame(frameToAddTo.getFramesetName(), null, null);
|
---|
337 |
|
---|
338 | int hashcode = Arrays.hashCode(image.getData().getPixels(0, 0, image.getWidth(), image.getHeight(), (int[]) null));
|
---|
339 |
|
---|
340 | File out = new File(FrameIO.IMAGES_PATH + "webpage-" + Integer.toHexString(hashcode) + ".png");
|
---|
341 | out.mkdirs();
|
---|
342 | ImageIO.write(image, "png", out);
|
---|
343 |
|
---|
344 | // Adding the image
|
---|
345 | frameToAddTo.addText(0, 0, "@i: " + out.getName(), null);
|
---|
346 |
|
---|
347 | Text thumb = frameset.addText(100, 100, "@i: " + out.getName() + " " + 200, null);
|
---|
348 | thumb.setLink(frameToAddTo.getName());
|
---|
349 |
|
---|
350 | // Button to go to the next page
|
---|
351 | Text nextButton = (Text) FrameCreator.createButton("Next", null, null, 10F, 10F);
|
---|
352 | nextButton.setID(frameToAddTo.getNextItemID());
|
---|
353 | nextButton.addAction("next");
|
---|
354 | frameToAddTo.addItem(nextButton);
|
---|
355 |
|
---|
356 | FrameIO.SaveFrame(frameToAddTo);
|
---|
357 | FrameIO.SaveFrame(frameset);
|
---|
358 |
|
---|
359 | System.out.println("Screenshot taken and added " + frameToAddTo.getName());
|
---|
360 |
|
---|
361 | } catch (IOException e) {
|
---|
362 | e.printStackTrace();
|
---|
363 | }
|
---|
364 |
|
---|
365 | graphics.dispose();
|
---|
366 | image.flush();
|
---|
367 |
|
---|
368 | try {
|
---|
369 | Platform.runLater(new Runnable() {
|
---|
370 | @Override
|
---|
371 | public void run() {
|
---|
372 | try {
|
---|
373 | HTMLBodyElement doc = (HTMLBodyElement) webEngine.executeScript("document.body");
|
---|
374 |
|
---|
375 | JSObject window = (JSObject) webEngine.executeScript("window");
|
---|
376 |
|
---|
377 | int visibleWidth = (int) webEngine.executeScript("window.innerWidth");
|
---|
378 | int visibleHeight = (int) webEngine.executeScript("window.innerHeight");
|
---|
379 |
|
---|
380 | System.out.println("Adding text " + frameToAddTo.getName());
|
---|
381 | WebParser.addTextToFrame(doc, visibleWidth, visibleHeight, window, webEngine, frameToAddTo);
|
---|
382 | FrameIO.SaveFrame(frameToAddTo);
|
---|
383 | System.out.println("Added text " + frameToAddTo.getName());
|
---|
384 | } catch (Exception ex) {
|
---|
385 | ex.printStackTrace();
|
---|
386 | }
|
---|
387 |
|
---|
388 | synchronized (notifier) {
|
---|
389 | notifier.notify();
|
---|
390 | }
|
---|
391 | }
|
---|
392 | });
|
---|
393 | } catch (Exception ex) {
|
---|
394 | ex.printStackTrace();
|
---|
395 | }
|
---|
396 | }
|
---|
397 | }
|
---|
398 | };
|
---|
399 |
|
---|
400 | Platform.runLater(new Runnable() {
|
---|
401 | @Override
|
---|
402 | public void run() {
|
---|
403 | try {
|
---|
404 | webEngine.executeScript(""
|
---|
405 | // Initializing the counter used when scrolling the page
|
---|
406 | + "var scrollCounter = 0;"
|
---|
407 |
|
---|
408 | // Setting all text to be hidden
|
---|
409 | + "var css = document.createElement('style');"
|
---|
410 | + "css.type = 'text/css';"
|
---|
411 | + "var style = 'WordSpan { visibility: hidden }';"
|
---|
412 | + "css.appendChild(document.createTextNode(style));"
|
---|
413 | + "document.getElementsByTagName('head')[0].appendChild(css);");
|
---|
414 |
|
---|
415 | HTMLBodyElement doc = (HTMLBodyElement) webEngine.executeScript("document.body");
|
---|
416 |
|
---|
417 | JSObject window = (JSObject) webEngine.executeScript("window");
|
---|
418 |
|
---|
419 | frame.setBackgroundColor(rgbStringToColor((String) ((JSObject) (window.call("getComputedStyle", new Object[] { doc }))).call("getPropertyValue",
|
---|
420 | new Object[] { "background-color" })));
|
---|
421 |
|
---|
422 | // Functions to be used later in JavaScript
|
---|
423 | webEngine.executeScript(""
|
---|
424 | + "function addToSpan(text) {"
|
---|
425 | + " span = document.createElement('wordSpan');"
|
---|
426 | + " span.textContent = text;"
|
---|
427 | + " par.insertBefore(span, refNode);"
|
---|
428 | + " if (prevSpan !== null && span.getBoundingClientRect().top > prevSpan.getBoundingClientRect().top) {"
|
---|
429 | + " span.textContent = '\\n' + span.textContent;"
|
---|
430 | + " if ( prevPrevSpan !== null && prevPrevSpan.getBoundingClientRect().left == prevSpan.getBoundingClientRect().left) {"
|
---|
431 | + " prevPrevSpan.textContent = prevPrevSpan.textContent + prevSpan.textContent;"
|
---|
432 | + " par.removeChild(prevSpan);"
|
---|
433 | + " } else {"
|
---|
434 | + " prevPrevSpan = prevSpan;"
|
---|
435 | + " }"
|
---|
436 | + " prevSpan = span;"
|
---|
437 | + " } else if ( prevSpan !== null) {"
|
---|
438 | + " prevSpan.textContent = prevSpan.textContent + span.textContent;"
|
---|
439 | + " par.removeChild(span);"
|
---|
440 | + " } else {"
|
---|
441 | + " prevSpan = span;"
|
---|
442 | + " }"
|
---|
443 | + "}"
|
---|
444 |
|
---|
445 | + "function splitIntoWords(toSplit) {"
|
---|
446 | + " var words = [];"
|
---|
447 | + " var pattern = /\\s+/g;"
|
---|
448 | + " var words = toSplit.split(pattern);"
|
---|
449 | + ""
|
---|
450 | + " for (var i = 0; i < words.length - 1; i++) {"
|
---|
451 | + " words[i] = words[i] + ' ';"
|
---|
452 | + " }"
|
---|
453 | + " return words;"
|
---|
454 | + "}"
|
---|
455 | );
|
---|
456 |
|
---|
457 | // Using Javascript to get an array of all the text nodes in the document so they can be wrapped in spans. Have to
|
---|
458 | // loop through twice (once to build the array and once actually going through the array, otherwise when the
|
---|
459 | // textnode is removed from the document items end up being skipped)
|
---|
460 | JSObject textNodes = (JSObject) webEngine.executeScript(""
|
---|
461 | + "function getTextNodes(rootNode){"
|
---|
462 | + "var node;"
|
---|
463 | + "var textNodes=[];"
|
---|
464 | + "var walk = document.createTreeWalker(rootNode, NodeFilter.SHOW_TEXT);"
|
---|
465 | + "while(node=walk.nextNode()) {"
|
---|
466 | + "if((node.textContent.trim().length > 0)) { "
|
---|
467 | + "textNodes.push(node);"
|
---|
468 | + "}"
|
---|
469 | + "}"
|
---|
470 | + "return textNodes;"
|
---|
471 | + "}; "
|
---|
472 | + "getTextNodes(document.body)"
|
---|
473 | );
|
---|
474 |
|
---|
475 | int nodesLength = (Integer) textNodes.getMember("length");
|
---|
476 |
|
---|
477 | // Looping through all the text nodes in the document
|
---|
478 | for (int j = 0; j < nodesLength; j++) {
|
---|
479 | Node currentNode = (Node) textNodes.getSlot(j);
|
---|
480 |
|
---|
481 | // Making the current node accessible in JavaScript
|
---|
482 | window.setMember("currentNode", currentNode);
|
---|
483 |
|
---|
484 | webEngine.executeScript(""
|
---|
485 | + "var span = null, prevSpan = null, prevPrevSpan = null;"
|
---|
486 |
|
---|
487 | // Removing repeated whitespace from the text node's content then splitting it into individual words
|
---|
488 | + "var textContent = currentNode.textContent.replace(/\\n|\\r/g, '').replace(/\\s+/g, ' ');"
|
---|
489 | + "var words = splitIntoWords(textContent);"
|
---|
490 |
|
---|
491 | + "var refNode = currentNode.nextSibling;"
|
---|
492 | + "var par = currentNode.parentElement;"
|
---|
493 | + "currentNode.parentElement.removeChild(currentNode);"
|
---|
494 |
|
---|
495 | + "for (var i = 0; i < words.length; i++) {"
|
---|
496 | + " addToSpan(words[i]);"
|
---|
497 | + "}"
|
---|
498 |
|
---|
499 | + "if (prevPrevSpan !== null && prevPrevSpan.getBoundingClientRect().left == prevSpan.getBoundingClientRect().left) {"
|
---|
500 | + " prevPrevSpan.textContent = prevPrevSpan.textContent + prevSpan.textContent;"
|
---|
501 | + " par.removeChild(prevSpan);"
|
---|
502 | + "}"
|
---|
503 | );
|
---|
504 |
|
---|
505 | // Will never reach 100% here, as the processing is not quite finished - progress is set to 100% at the end of
|
---|
506 | // the addPageToFrame loop below
|
---|
507 | progressBar.set((100 * (j)) / nodesLength);
|
---|
508 | }
|
---|
509 |
|
---|
510 | // Finding all links within the page, then setting the href attribute of all their descendants to be the same
|
---|
511 | // link/URL.
|
---|
512 | // This is needed because there is no apparent and efficient way to check if an element is a child of a link when
|
---|
513 | // running through the document when added each element to Expeditee
|
---|
514 | webEngine.executeScript(""
|
---|
515 | + "var anchors = document.getElementsByTagName('a');"
|
---|
516 | + ""
|
---|
517 | + "for (var i = 0; i < anchors.length; i++) {"
|
---|
518 | + "var currentAnchor = anchors.item(i);"
|
---|
519 | + "var anchorDescendants = currentAnchor.querySelectorAll('*');"
|
---|
520 | + "for (var j = 0; j < anchorDescendants.length; j++) {"
|
---|
521 | + "anchorDescendants.item(j).href = currentAnchor.href;"
|
---|
522 | + "}"
|
---|
523 | + "}"
|
---|
524 | );
|
---|
525 |
|
---|
526 | } catch (Exception ex) {
|
---|
527 | ex.printStackTrace();
|
---|
528 | }
|
---|
529 |
|
---|
530 | synchronized (notifier) {
|
---|
531 | notifier.notify();
|
---|
532 | }
|
---|
533 | }
|
---|
534 | });
|
---|
535 |
|
---|
536 | synchronized (notifier) {
|
---|
537 | try {
|
---|
538 | // Waiting for the JavaFX thread to finish
|
---|
539 | notifier.wait();
|
---|
540 | } catch (InterruptedException e) {
|
---|
541 | // TODO Auto-generated catch block
|
---|
542 | e.printStackTrace();
|
---|
543 | }
|
---|
544 | }
|
---|
545 |
|
---|
546 | while (!bottomReached.getValue()) {
|
---|
547 | Platform.runLater(new Runnable() {
|
---|
548 | @Override
|
---|
549 | public void run() {
|
---|
550 | try {
|
---|
551 | // Scrolling down the page
|
---|
552 | webEngine.executeScript(""
|
---|
553 | + "window.scrollTo(0, scrollCounter * 0.9 * window.innerHeight);"
|
---|
554 | + "scrollCounter = scrollCounter+1;");
|
---|
555 |
|
---|
556 | System.out.println("Scrolling");
|
---|
557 |
|
---|
558 | bottomReached.setValue((Boolean) webEngine.executeScript("(window.pageYOffset + window.innerHeight >= document.documentElement.scrollHeight)"));
|
---|
559 |
|
---|
560 | synchronized (notifier) {
|
---|
561 | notifier.notify();
|
---|
562 | }
|
---|
563 |
|
---|
564 | } catch (Exception e) {
|
---|
565 | e.printStackTrace();
|
---|
566 | }
|
---|
567 | }
|
---|
568 | });
|
---|
569 |
|
---|
570 | synchronized (notifier) {
|
---|
571 | try {
|
---|
572 | // Waiting for the JavaFX thread to finish
|
---|
573 | notifier.wait();
|
---|
574 | } catch (InterruptedException e) {
|
---|
575 | // TODO Auto-generated catch block
|
---|
576 | e.printStackTrace();
|
---|
577 | }
|
---|
578 | }
|
---|
579 |
|
---|
580 | timer.start();
|
---|
581 |
|
---|
582 | synchronized (notifier) {
|
---|
583 | try {
|
---|
584 | // Waiting for the timer thread to finish before looping again
|
---|
585 | notifier.wait();
|
---|
586 | } catch (InterruptedException e) {
|
---|
587 | // TODO Auto-generated catch block
|
---|
588 | e.printStackTrace();
|
---|
589 | }
|
---|
590 | }
|
---|
591 |
|
---|
592 | }
|
---|
593 |
|
---|
594 | } catch (Exception ex) {
|
---|
595 | ex.printStackTrace();
|
---|
596 | }
|
---|
597 |
|
---|
598 |
|
---|
599 |
|
---|
600 | }
|
---|
601 |
|
---|
602 | /**
|
---|
603 | * @param rgbString
|
---|
604 | * string in the format <i>rgb(x,x,x)</i> or <i>rgba(x,x,x,x)</i>
|
---|
605 | * @return A Color object that should match the rgb string passed int. Returns null if alpha is 0
|
---|
606 | */
|
---|
607 | private static Color rgbStringToColor(String rgbString) {
|
---|
608 |
|
---|
609 | if (rgbString == null) {
|
---|
610 | return null;
|
---|
611 | }
|
---|
612 |
|
---|
613 | // Splitting the string into 'rgb' and 'x, x, x'
|
---|
614 | String[] tmpStrings = rgbString.split("\\(|\\)");
|
---|
615 |
|
---|
616 | // Splitting up the RGB(A) components into an array
|
---|
617 | tmpStrings = tmpStrings[1].split(",");
|
---|
618 |
|
---|
619 | int[] components = new int[4];
|
---|
620 | Arrays.fill(components, 255);
|
---|
621 |
|
---|
622 | for (int i = 0; i < tmpStrings.length; i++) {
|
---|
623 | Float d = Float.parseFloat(tmpStrings[i].trim());
|
---|
624 |
|
---|
625 | components[i] = Math.round(d);
|
---|
626 | }
|
---|
627 |
|
---|
628 | if (components[3] > 0) {
|
---|
629 | return new Color(components[0], components[1], components[2], components[3]);
|
---|
630 | } else {
|
---|
631 | return null;
|
---|
632 | }
|
---|
633 | }
|
---|
634 |
|
---|
635 | /**
|
---|
636 | * @param rootElement
|
---|
637 | * Element that will be converted (including all sub-elements)
|
---|
638 | * @param backgroundColor
|
---|
639 | * String to be used as the background color of this element when added. In the format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
---|
640 | * @param window
|
---|
641 | * 'window' from Javascript
|
---|
642 | * @param webEngine
|
---|
643 | * Web engine that the page is loaded in
|
---|
644 | * @param frame
|
---|
645 | * Expeditee frame to add the converted page to
|
---|
646 | * @throws IllegalArgumentException
|
---|
647 | * @throws IllegalAccessException
|
---|
648 | */
|
---|
649 | private static void addPageToFrame(Node rootElement, JSObject window, WebEngine webEngine, Frame frame) throws InvocationTargetException, IllegalAccessException,
|
---|
650 | IllegalArgumentException {
|
---|
651 |
|
---|
652 | Node currentNode = rootElement;
|
---|
653 |
|
---|
654 | if (currentNode.getNodeType() == Node.TEXT_NODE || currentNode.getNodeType() == Node.ELEMENT_NODE) {
|
---|
655 |
|
---|
656 | JSObject style;
|
---|
657 | JSObject bounds;
|
---|
658 |
|
---|
659 | if (currentNode.getNodeType() == Node.TEXT_NODE) {
|
---|
660 | // CSS style for the element
|
---|
661 | style = (JSObject) window.call("getComputedStyle", new Object[] { currentNode.getParentNode() });
|
---|
662 |
|
---|
663 | // Getting a rectangle that represents the area and position of the element
|
---|
664 | bounds = (JSObject) ((JSObject) currentNode.getParentNode()).call("getBoundingClientRect", new Object[] {});
|
---|
665 | } else {
|
---|
666 | style = (JSObject) window.call("getComputedStyle", new Object[] { currentNode });
|
---|
667 |
|
---|
668 | bounds = (JSObject) ((JSObject) currentNode).call("getBoundingClientRect", new Object[] {});
|
---|
669 | }
|
---|
670 |
|
---|
671 | // Bounding rectangle position is relative to the current view, so scroll position must be added to x/y
|
---|
672 | // TODO: This doesn't check if an element or any of its parent elements have position:fixed set - the only
|
---|
673 | // way to check seems to be to walking through the element's parents until the document root is reached
|
---|
674 | float x = Float.valueOf(bounds.getMember("left").toString()) + Float.valueOf(webEngine.executeScript("window.pageXOffset").toString());
|
---|
675 | float y = Float.valueOf(bounds.getMember("top").toString()) + Float.valueOf(webEngine.executeScript("window.pageYOffset").toString());
|
---|
676 |
|
---|
677 | float width = Float.valueOf(bounds.getMember("width").toString());
|
---|
678 | float height = Float.valueOf(bounds.getMember("height").toString());
|
---|
679 |
|
---|
680 | // Checking if the element is actually visible on the page
|
---|
681 | if (WebParser.elementVisible(x, y, width, height, style)) {
|
---|
682 |
|
---|
683 | // Filtering the node type, starting with text nodes
|
---|
684 | if (currentNode.getNodeType() == Node.TEXT_NODE) {
|
---|
685 |
|
---|
686 | String fontSize = ((String) style.call("getPropertyValue", new Object[] { "font-size" }));
|
---|
687 |
|
---|
688 | // Trimming off the units (always px) from the font size
|
---|
689 | fontSize = fontSize.substring(0, fontSize.length() - 2);
|
---|
690 |
|
---|
691 | // Always returns in format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
---|
692 | String color = (String) style.call("getPropertyValue", new Object[] { "color" });
|
---|
693 |
|
---|
694 | // Always returns in format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
---|
695 | String bgColorString = (String) style.call("getPropertyValue", new Object[] { "background-color" });
|
---|
696 |
|
---|
697 | String align = (String) style.call("getPropertyValue", new Object[] { "text-align" });
|
---|
698 |
|
---|
699 | // Returns comma-separated list of typefaces
|
---|
700 | String typeface = (String) style.call("getPropertyValue", new Object[] { "font-family" });
|
---|
701 |
|
---|
702 | String[] typefaces = typeface.split(", |,");
|
---|
703 |
|
---|
704 | String weight = (String) style.call("getPropertyValue", new Object[] { "font-weight" });
|
---|
705 |
|
---|
706 | String fontStyle = (String) style.call("getPropertyValue", new Object[] { "font-style" });
|
---|
707 |
|
---|
708 | // Returns "normal" or a value in pixels (e.g. "10px")
|
---|
709 | String letterSpacing = (String) style.call("getPropertyValue", new Object[] { "letter-spacing" });
|
---|
710 |
|
---|
711 | // Returns a value in pixels (e.g. "10px")
|
---|
712 | String lineHeight = (String) style.call("getPropertyValue", new Object[] { "line-height" });
|
---|
713 |
|
---|
714 | String textTransform = (String) style.call("getPropertyValue", new Object[] { "text-transform" });
|
---|
715 |
|
---|
716 | String linkUrl = (String) ((JSObject) currentNode.getParentNode()).getMember("href");
|
---|
717 |
|
---|
718 | Boolean fontFound = false;
|
---|
719 | Font font = new Font(null);
|
---|
720 |
|
---|
721 | // Looping through all font-families listed in the element's CSS until one that is installed is
|
---|
722 | // found, or the end of the list is reached, in which case the default font is used
|
---|
723 | for (int j = 0; j < typefaces.length && !fontFound; j++) {
|
---|
724 | if (typefaces[j].toLowerCase().equals("sans-serif")) {
|
---|
725 | typefaces[j] = "Arial Unicode MS";
|
---|
726 | } else if (typefaces[j].toLowerCase().equals("serif")) {
|
---|
727 | typefaces[j] = "Times New Roman";
|
---|
728 | } else if ((typefaces[j].toLowerCase().equals("arial"))) {
|
---|
729 | // Have to use Arial Unicode, otherwise unicode characters display incorrectly
|
---|
730 | typefaces[j] = "Arial Unicode MS";
|
---|
731 | }
|
---|
732 |
|
---|
733 | // Regex will remove any inverted commas surrounding multi-word typeface names
|
---|
734 | font = new Font(typefaces[j].replaceAll("^'|'$", ""), Font.PLAIN, 12);
|
---|
735 |
|
---|
736 | // If the font isn't found, Java just uses Font.DIALOG, so this check checks whether the font was found
|
---|
737 | if (!(font.getFamily().toLowerCase().equals(Font.DIALOG.toLowerCase()))) {
|
---|
738 | fontFound = true;
|
---|
739 | }
|
---|
740 | }
|
---|
741 |
|
---|
742 | if (font.getFamily().toLowerCase().equals(Font.DIALOG.toLowerCase())) {
|
---|
743 | font = new Font("Times New Roman", Font.PLAIN, 12);
|
---|
744 | }
|
---|
745 |
|
---|
746 | String fontStyleComplete = "";
|
---|
747 |
|
---|
748 | int weightInt = 0;
|
---|
749 |
|
---|
750 | try {
|
---|
751 | weightInt = Integer.parseInt(weight);
|
---|
752 | } catch (NumberFormatException nfe) {
|
---|
753 | // Use default value as set above
|
---|
754 | }
|
---|
755 |
|
---|
756 | // checking if font is bold - i.e. 'bold', 'bolder' or weight over 500
|
---|
757 | if (weight.toLowerCase().startsWith("bold") || weightInt > 500) {
|
---|
758 | fontStyleComplete = fontStyleComplete.concat("bold");
|
---|
759 | }
|
---|
760 |
|
---|
761 | if (fontStyle.toLowerCase().equals("italic") || fontStyle.toLowerCase().equals("oblique")) {
|
---|
762 | fontStyleComplete = fontStyleComplete.concat("italic");
|
---|
763 | }
|
---|
764 |
|
---|
765 | float fontSizeFloat = 12;
|
---|
766 |
|
---|
767 | try {
|
---|
768 | fontSizeFloat = Float.valueOf(fontSize);
|
---|
769 | } catch (NumberFormatException nfe) {
|
---|
770 | // Use default value as set above
|
---|
771 | }
|
---|
772 |
|
---|
773 | float letterSpacingFloat = -0.008f;
|
---|
774 |
|
---|
775 | try {
|
---|
776 | letterSpacingFloat = (Integer.parseInt(letterSpacing.substring(0, letterSpacing.length() - 2)) / (fontSizeFloat));
|
---|
777 | } catch (NumberFormatException nfe) {
|
---|
778 | // Use default value as set above
|
---|
779 | }
|
---|
780 |
|
---|
781 | float lineHeightInt = -1;
|
---|
782 |
|
---|
783 | try {
|
---|
784 | lineHeightInt = (Float.parseFloat(lineHeight.substring(0, lineHeight.length() - 2)));
|
---|
785 | } catch (NumberFormatException nfe) {
|
---|
786 | // Use default value as set above
|
---|
787 | }
|
---|
788 |
|
---|
789 | Text t;
|
---|
790 |
|
---|
791 | String textContent = currentNode.getTextContent().replaceAll("[^\\S\\n]+", " ");
|
---|
792 | textContent = textContent.replaceAll("^(\\s)(\\n|\\r)", "");
|
---|
793 |
|
---|
794 | if (textTransform.equals("uppercase")) {
|
---|
795 | textContent = textContent.toUpperCase();
|
---|
796 | } else if (textTransform.equals("lowercase")) {
|
---|
797 | textContent = textContent.toUpperCase();
|
---|
798 | }
|
---|
799 |
|
---|
800 | // Adding the text to the frame. Expeditee text seems to be positioned relative to the baseline of the first line, so
|
---|
801 | // the font size has to be added to the y-position
|
---|
802 | t = frame.addText(Math.round(x), Math.round(y + fontSizeFloat), textContent, null);
|
---|
803 |
|
---|
804 | t.setColor(rgbStringToColor(color));
|
---|
805 | t.setBackgroundColor(rgbStringToColor(bgColorString));
|
---|
806 | t.setFont(font);
|
---|
807 | t.setSize(fontSizeFloat);
|
---|
808 | t.setFontStyle(fontStyleComplete);
|
---|
809 | t.setLetterSpacing(letterSpacingFloat);
|
---|
810 |
|
---|
811 | // Removing any spacing between lines allowing t.getLineHeight() to be used to get the actual height
|
---|
812 | // of just the characters (i.e. distance from ascenders to descenders)
|
---|
813 | t.setSpacing(0);
|
---|
814 |
|
---|
815 | t.setSpacing(lineHeightInt - t.getLineHeight());
|
---|
816 |
|
---|
817 | if (align.equals("left")) {
|
---|
818 | t.setJustification(Justification.left);
|
---|
819 | } else if (align.equals("right")) {
|
---|
820 | t.setJustification(Justification.right);
|
---|
821 | } else if (align.equals("center")) {
|
---|
822 | t.setJustification(Justification.center);
|
---|
823 | } else if (align.equals("justify")) {
|
---|
824 | t.setJustification(Justification.full);
|
---|
825 | }
|
---|
826 |
|
---|
827 | // Font size is added to the item width to give a little breathing room
|
---|
828 | t.setWidth(Math.round(width + (t.getSize())));
|
---|
829 |
|
---|
830 | if (!linkUrl.equals("undefined")) {
|
---|
831 | t.setAction("gotourl " + linkUrl);
|
---|
832 | t.setActionMark(false);
|
---|
833 | }
|
---|
834 |
|
---|
835 | } else if (currentNode.getNodeType() == Node.ELEMENT_NODE) {
|
---|
836 |
|
---|
837 | // Always returns in format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
---|
838 | String bgColorString = (String) style.call("getPropertyValue", new Object[] { "background-color" });
|
---|
839 |
|
---|
840 | Color bgColor = rgbStringToColor(bgColorString);
|
---|
841 |
|
---|
842 | // If the element has a background color then add it (to Expeditee) as a rectangle with that background color
|
---|
843 | if (bgColor != null) {
|
---|
844 | System.out.println("bg");
|
---|
845 | frame.addRectangle(Math.round(x), Math.round(y), Math.round(width), Math.round(height), 0, null, bgColor);
|
---|
846 | }
|
---|
847 |
|
---|
848 | // background image, returns in format "url(protocol://absolute/path/to/img.extension)" for images,
|
---|
849 | // may also return gradients, data, etc. (not handled yet). Only need to add bg image on
|
---|
850 | // 'ELEMENT_NODE' (and not 'TEXT_NODE' otherwise there would be double-ups
|
---|
851 | String bgImage = (String) style.call("getPropertyValue", new Object[] { "background-image" });
|
---|
852 |
|
---|
853 | String linkUrl = (String) ((JSObject) currentNode).getMember("href");
|
---|
854 |
|
---|
855 | if (bgImage.startsWith("url(")) {
|
---|
856 | bgImage = bgImage.substring(4, bgImage.length() - 1);
|
---|
857 |
|
---|
858 | String bgSize = ((String) style.call("getPropertyValue", new Object[] { "background-size" })).toLowerCase();
|
---|
859 | String bgRepeat = ((String) style.call("getPropertyValue", new Object[] { "background-repeat" })).toLowerCase();
|
---|
860 |
|
---|
861 | // Returns "[x]px [y]px", "[x]% [y]%", "[x]px [y]%" or "[x]% [y]px"
|
---|
862 | String bgPosition = ((String) style.call("getPropertyValue", new Object[] { "background-position" })).toLowerCase();
|
---|
863 |
|
---|
864 | String[] bgOffsetCoords = bgPosition.split(" ");
|
---|
865 |
|
---|
866 | int bgOffsetX = 0, bgOffsetY = 0;
|
---|
867 |
|
---|
868 | float originXPercent = 0, originYPercent = 0;
|
---|
869 |
|
---|
870 | int cropStartX, cropStartY, cropEndX, cropEndY;
|
---|
871 |
|
---|
872 | // Converting the x and y offset values to integers (and from % to px if needed)
|
---|
873 | if (bgOffsetCoords[0].endsWith("%")) {
|
---|
874 | bgOffsetX = (int) ((Integer.valueOf(bgOffsetCoords[0].substring(0, bgOffsetCoords[0].length() - 1)) / 100.0) * width);
|
---|
875 | originXPercent = (Integer.valueOf(bgOffsetCoords[0].substring(0, bgOffsetCoords[0].length() - 1))) / 100f;
|
---|
876 | } else if (bgOffsetCoords[0].endsWith("px")) {
|
---|
877 | bgOffsetX = (int) (Integer.valueOf(bgOffsetCoords[0].substring(0, bgOffsetCoords[0].length() - 2)));
|
---|
878 | }
|
---|
879 |
|
---|
880 | if (bgOffsetCoords[1].endsWith("%")) {
|
---|
881 | bgOffsetY = (int) ((Integer.valueOf(bgOffsetCoords[1].substring(0, bgOffsetCoords[1].length() - 1)) / 100.0) * height);
|
---|
882 | originYPercent = (Integer.valueOf(bgOffsetCoords[1].substring(0, bgOffsetCoords[1].length() - 1))) / 100f;
|
---|
883 | } else if (bgOffsetCoords[1].endsWith("px")) {
|
---|
884 | bgOffsetY = (int) (Integer.valueOf(bgOffsetCoords[1].substring(0, bgOffsetCoords[1].length() - 2)));
|
---|
885 | }
|
---|
886 |
|
---|
887 | // Converting from an offset to crop coords
|
---|
888 | cropStartX = -1 * bgOffsetX;
|
---|
889 | cropEndX = (int) (cropStartX + width);
|
---|
890 |
|
---|
891 | cropStartY = -1 * bgOffsetY;
|
---|
892 | cropEndY = (int) (cropStartY + height);
|
---|
893 |
|
---|
894 | int bgWidth = -1;
|
---|
895 |
|
---|
896 | if (bgSize.equals("cover")) {
|
---|
897 | bgWidth = (int) width;
|
---|
898 | } else if (bgSize.equals("contain")) {
|
---|
899 | // TODO: actually compute the appropriate width
|
---|
900 | bgWidth = (int) width;
|
---|
901 | } else if (bgSize.equals("auto")) {
|
---|
902 | bgWidth = -1;
|
---|
903 | } else {
|
---|
904 | bgSize = bgSize.split(" ")[0];
|
---|
905 |
|
---|
906 | if (bgSize.endsWith("%")) {
|
---|
907 | bgWidth = (int) ((Integer.parseInt(bgSize.replaceAll("\\D", "")) / 100.0) * width);
|
---|
908 | } else if (bgSize.endsWith("px")) {
|
---|
909 | bgWidth = Integer.parseInt(bgSize.replaceAll("\\D", ""));
|
---|
910 | }
|
---|
911 | }
|
---|
912 |
|
---|
913 | try {
|
---|
914 | WebParser.addImageFromUrl(bgImage, linkUrl, frame, x, y, bgWidth, cropStartX, cropStartY, cropEndX, cropEndY, bgRepeat, originXPercent, originYPercent);
|
---|
915 | } catch (MalformedURLException mue) {
|
---|
916 | // probably a 'data:' url, not supported yet
|
---|
917 | mue.printStackTrace();
|
---|
918 | } catch (IOException e) {
|
---|
919 | // TODO Auto-generated catch block
|
---|
920 | e.printStackTrace();
|
---|
921 | }
|
---|
922 | }
|
---|
923 |
|
---|
924 | String imgSrc;
|
---|
925 |
|
---|
926 | if (currentNode.getNodeName().toLowerCase().equals("img") && (imgSrc = ((JSObject) currentNode).getMember("src").toString()) != null) {
|
---|
927 | try {
|
---|
928 | WebParser.addImageFromUrl(imgSrc, linkUrl, frame, x, y, (int) width, null, null, null, null, null, 0, 0);
|
---|
929 | } catch (MalformedURLException mue) {
|
---|
930 | // probably a 'data:' url, not supported yet
|
---|
931 | mue.printStackTrace();
|
---|
932 | } catch (IOException e) {
|
---|
933 | // TODO Auto-generated catch block
|
---|
934 | e.printStackTrace();
|
---|
935 | }
|
---|
936 | }
|
---|
937 | }
|
---|
938 | }
|
---|
939 |
|
---|
940 | Node childNode = currentNode.getFirstChild();
|
---|
941 |
|
---|
942 | while (childNode != null) {
|
---|
943 | addPageToFrame(childNode, window, webEngine, frame);
|
---|
944 | childNode = childNode.getNextSibling();
|
---|
945 | }
|
---|
946 | }
|
---|
947 | }
|
---|
948 |
|
---|
949 | private static boolean elementVisible(float x, float y, float width, float height, JSObject style) {
|
---|
950 | if (width <= 0 || height <= 0 || x + width <= 0 || y + height <= 0 || ((String) style.call("getPropertyValue", new Object[] { "visibility" })).equals("hidden")
|
---|
951 | || ((String) style.call("getPropertyValue", new Object[] { "display" })).equals("none")) {
|
---|
952 | return false;
|
---|
953 | } else {
|
---|
954 | return true;
|
---|
955 | }
|
---|
956 | }
|
---|
957 |
|
---|
958 | /**
|
---|
959 | * @param imgSrc
|
---|
960 | * URL of the image to add
|
---|
961 | * @param linkUrl
|
---|
962 | * Absolute URL that the image should link to when clicked
|
---|
963 | * @param frame
|
---|
964 | * Frame to add the image to
|
---|
965 | * @param x
|
---|
966 | * X-coordinate at which the image should be placed on the frame
|
---|
967 | * @param y
|
---|
968 | * Y-coordinate at which the image should be placed on the frame
|
---|
969 | * @param width
|
---|
970 | * Width of the image once added to the frame. Negative 1 (-1) will cause the actual width of the image file to be used
|
---|
971 | *
|
---|
972 | * @param cropStartX
|
---|
973 | * X-coordinate at which to start crop, or null for no crop
|
---|
974 | * @param cropStartY
|
---|
975 | * Y-coordinate at which to start crop, or null for no crop
|
---|
976 | * @param cropEndX
|
---|
977 | * X-coordinate at which to end the crop, or null for no crop
|
---|
978 | * @param cropEndY
|
---|
979 | * Y-coordinate at which to end the crop, or null for no crop
|
---|
980 | *
|
---|
981 | * @param repeat
|
---|
982 | * String determining how the image should be tiled/repeated. Valid strings are: <i>no-repeat</i>, <i>repeat-x</i>, or
|
---|
983 | * <i>repeat-y</i>. All other values (including null) will cause the image to repeat in both directions
|
---|
984 | *
|
---|
985 | * @param originXPercent
|
---|
986 | * Percentage into the image to use as the x coordinate of the image's origin point
|
---|
987 | * @param originYPercent
|
---|
988 | * Percentage into the image to use as the y coordinate of the image's origin point
|
---|
989 | *
|
---|
990 | * @throws MalformedURLException
|
---|
991 | * @throws IOException
|
---|
992 | */
|
---|
993 | private static void addImageFromUrl(String imgSrc, String linkUrl, final Frame frame, float x, float y, int width, Integer cropStartX, Integer cropStartY, Integer cropEndX, Integer cropEndY, String repeat,
|
---|
994 | float originXPercent, float originYPercent)
|
---|
995 | throws MalformedURLException,
|
---|
996 | IOException {
|
---|
997 |
|
---|
998 | URL imgUrl = new URL(imgSrc);
|
---|
999 |
|
---|
1000 | HttpURLConnection connection = (HttpURLConnection) (imgUrl.openConnection());
|
---|
1001 |
|
---|
1002 | // Spoofing a widely accepted User Agent, since some sites refuse to serve non-webbrowser clients
|
---|
1003 | connection.setRequestProperty("User-Agent", "Mozilla/5.0");
|
---|
1004 |
|
---|
1005 | BufferedImage img = ImageIO.read(connection.getInputStream());
|
---|
1006 |
|
---|
1007 | int hashcode = Arrays.hashCode(img.getData().getPixels(0, 0, img.getWidth(), img.getHeight(), (int[]) null));
|
---|
1008 | File out = new File(FrameIO.IMAGES_PATH + Integer.toHexString(hashcode) + ".png");
|
---|
1009 | out.mkdirs();
|
---|
1010 | ImageIO.write(img, "png", out);
|
---|
1011 |
|
---|
1012 | if (cropEndX == null || cropStartX == null || cropEndY == null || cropStartY == null) {
|
---|
1013 | cropStartX = 0;
|
---|
1014 | cropStartY = 0;
|
---|
1015 | cropEndX = img.getWidth();
|
---|
1016 | cropEndY = img.getHeight();
|
---|
1017 | } else if (cropStartX < 0) {
|
---|
1018 | cropEndX = cropEndX - cropStartX;
|
---|
1019 | x = x + Math.abs(cropStartX);
|
---|
1020 | cropStartX = 0;
|
---|
1021 | }
|
---|
1022 |
|
---|
1023 | if (cropStartY < 0) {
|
---|
1024 | cropEndY = cropEndY - cropStartY;
|
---|
1025 | y = y + Math.abs(cropStartY);
|
---|
1026 | cropStartY = 0;
|
---|
1027 | }
|
---|
1028 |
|
---|
1029 | if (width < 0) {
|
---|
1030 | width = img.getWidth();
|
---|
1031 | }
|
---|
1032 |
|
---|
1033 | if (repeat != null) {
|
---|
1034 | if (repeat.equals("no-repeat")) {
|
---|
1035 | int tmpCropEndY = (int) (cropStartY + ((float) width / img.getWidth()) * img.getHeight());
|
---|
1036 | int tmpCropEndX = cropStartX + width;
|
---|
1037 |
|
---|
1038 | cropEndX = (cropEndX < tmpCropEndX) ? cropEndX : tmpCropEndX;
|
---|
1039 | cropEndY = (cropEndY < tmpCropEndY) ? cropEndY : tmpCropEndY;
|
---|
1040 | } else if (repeat.equals("repeat-x")) {
|
---|
1041 | int tmpCropEndY = (int) (cropStartY + ((float) width / img.getWidth()) * img.getHeight());
|
---|
1042 | cropEndY = (cropEndY < tmpCropEndY) ? cropEndY : tmpCropEndY;
|
---|
1043 | } else if (repeat.equals("repeat-y")) {
|
---|
1044 | int tmpCropEndX = cropStartX + width;
|
---|
1045 | cropEndX = (cropEndX < tmpCropEndX) ? cropEndX : tmpCropEndX;
|
---|
1046 | }
|
---|
1047 | }
|
---|
1048 |
|
---|
1049 | if (originXPercent > 0) {
|
---|
1050 | int actualWidth = cropEndX - cropStartX;
|
---|
1051 |
|
---|
1052 | int originXPixels = Math.round(originXPercent * actualWidth);
|
---|
1053 |
|
---|
1054 | x = x - originXPixels;
|
---|
1055 |
|
---|
1056 | cropStartX = (int) (cropStartX + (width - actualWidth) * originXPercent);
|
---|
1057 | cropEndX = (int) (cropEndX + (width - actualWidth) * originXPercent);
|
---|
1058 | }
|
---|
1059 |
|
---|
1060 | if (originYPercent > 0) {
|
---|
1061 | int height = (int) ((img.getHeight() / (float) img.getWidth()) * width);
|
---|
1062 | int actualHeight = (cropEndY - cropStartY);
|
---|
1063 | int originYPixels = Math.round(originYPercent * actualHeight);
|
---|
1064 |
|
---|
1065 | y = y - originYPixels;
|
---|
1066 |
|
---|
1067 | cropStartY = (int) (cropStartY + (height - actualHeight) * originYPercent);
|
---|
1068 | cropEndY = (int) (cropEndY + (height - actualHeight) * originYPercent);
|
---|
1069 | }
|
---|
1070 |
|
---|
1071 | Text text = new Text("@i: " + out.getName() + " " + width);
|
---|
1072 | text.setPosition(x, y);
|
---|
1073 |
|
---|
1074 | Picture pic = ItemUtils.CreatePicture(text, frame);
|
---|
1075 |
|
---|
1076 | float invScale = 1 / pic.getScale();
|
---|
1077 |
|
---|
1078 | pic.setCrop((int)(cropStartX * invScale), (int)(cropStartY * invScale), (int)(cropEndX * invScale), (int)(cropEndY * invScale));
|
---|
1079 |
|
---|
1080 | if (linkUrl != null && !linkUrl.equals("undefined")) {
|
---|
1081 | pic.setAction("goto " + linkUrl);
|
---|
1082 | pic.setActionMark(false);
|
---|
1083 | }
|
---|
1084 |
|
---|
1085 | frame.addItem(pic);
|
---|
1086 | pic.anchor();
|
---|
1087 | pic.getSource().anchor();
|
---|
1088 | }
|
---|
1089 |
|
---|
1090 | /**
|
---|
1091 | * @param rootElement
|
---|
1092 | * Element that will be converted (including all sub-elements)
|
---|
1093 | * @param backgroundColor
|
---|
1094 | * String to be used as the background color of this element when added. In the format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
---|
1095 | * @param window
|
---|
1096 | * 'window' from Javascript
|
---|
1097 | * @param webEngine
|
---|
1098 | * Web engine that the page is loaded in
|
---|
1099 | * @param frame
|
---|
1100 | * Expeditee frame to add the converted page to
|
---|
1101 | * @throws IllegalArgumentException
|
---|
1102 | * @throws IllegalAccessException
|
---|
1103 | */
|
---|
1104 | private static void addTextToFrame(Node rootElement, int visibleWidth, int visibleHeight, JSObject window, WebEngine webEngine, Frame frame) throws InvocationTargetException,
|
---|
1105 | IllegalAccessException, IllegalArgumentException {
|
---|
1106 |
|
---|
1107 | Node currentNode = rootElement;
|
---|
1108 |
|
---|
1109 | if (currentNode.getNodeType() == Node.TEXT_NODE) {
|
---|
1110 |
|
---|
1111 | JSObject style;
|
---|
1112 | JSObject bounds;
|
---|
1113 |
|
---|
1114 |
|
---|
1115 | // CSS style for the element
|
---|
1116 | style = (JSObject) window.call("getComputedStyle", new Object[] { currentNode.getParentNode() });
|
---|
1117 |
|
---|
1118 | // Getting a rectangle that represents the area and position of the element
|
---|
1119 | bounds = (JSObject) ((JSObject) currentNode.getParentNode()).call("getBoundingClientRect", new Object[] {});
|
---|
1120 |
|
---|
1121 |
|
---|
1122 | // Bounding rectangle position is relative to the current view, so scroll position must be added to x/y
|
---|
1123 | // TODO: This doesn't check if an element or any of its parent elements have position:fixed set - the only
|
---|
1124 | // way to check seems to be to walking through the element's parents until the document root is reached
|
---|
1125 | float x = Float.valueOf(bounds.getMember("left").toString());
|
---|
1126 | float y = Float.valueOf(bounds.getMember("top").toString());
|
---|
1127 |
|
---|
1128 | float width = Float.valueOf(bounds.getMember("width").toString());
|
---|
1129 | float height = Float.valueOf(bounds.getMember("height").toString());
|
---|
1130 |
|
---|
1131 | // Checking if the element is actually visible on the page
|
---|
1132 | if (width > 0 && height > 0 && x + width > 0 && y + height > 0 && x <= visibleWidth && y <= visibleHeight
|
---|
1133 | && !(((String) style.call("getPropertyValue", new Object[] { "display" })).equals("none"))) {
|
---|
1134 |
|
---|
1135 | String fontSize = ((String) style.call("getPropertyValue", new Object[] { "font-size" }));
|
---|
1136 |
|
---|
1137 | // Trimming off the units (always px) from the font size
|
---|
1138 | fontSize = fontSize.substring(0, fontSize.length() - 2);
|
---|
1139 |
|
---|
1140 | // Always returns in format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
---|
1141 | String color = (String) style.call("getPropertyValue", new Object[] { "color" });
|
---|
1142 |
|
---|
1143 | // Always returns in format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
---|
1144 | String bgColorString = (String) style.call("getPropertyValue", new Object[] { "background-color" });
|
---|
1145 |
|
---|
1146 | String align = (String) style.call("getPropertyValue", new Object[] { "text-align" });
|
---|
1147 |
|
---|
1148 | // Returns comma-separated list of typefaces
|
---|
1149 | String typeface = (String) style.call("getPropertyValue", new Object[] { "font-family" });
|
---|
1150 |
|
---|
1151 | String[] typefaces = typeface.split(", |,");
|
---|
1152 |
|
---|
1153 | String weight = (String) style.call("getPropertyValue", new Object[] { "font-weight" });
|
---|
1154 |
|
---|
1155 | String fontStyle = (String) style.call("getPropertyValue", new Object[] { "font-style" });
|
---|
1156 |
|
---|
1157 | // Returns "normal" or a value in pixels (e.g. "10px")
|
---|
1158 | String letterSpacing = (String) style.call("getPropertyValue", new Object[] { "letter-spacing" });
|
---|
1159 |
|
---|
1160 | // Returns a value in pixels (e.g. "10px")
|
---|
1161 | String lineHeight = (String) style.call("getPropertyValue", new Object[] { "line-height" });
|
---|
1162 |
|
---|
1163 | String textTransform = (String) style.call("getPropertyValue", new Object[] { "text-transform" });
|
---|
1164 |
|
---|
1165 | String linkUrl = (String) ((JSObject) currentNode.getParentNode()).getMember("href");
|
---|
1166 |
|
---|
1167 | Boolean fontFound = false;
|
---|
1168 | Font font = new Font(null);
|
---|
1169 |
|
---|
1170 | // Looping through all font-families listed in the element's CSS until one that is installed is
|
---|
1171 | // found, or the end of the list is reached, in which case the default font is used
|
---|
1172 | for (int j = 0; j < typefaces.length && !fontFound; j++) {
|
---|
1173 | if (typefaces[j].toLowerCase().equals("sans-serif")) {
|
---|
1174 | typefaces[j] = "Arial Unicode MS";
|
---|
1175 | } else if (typefaces[j].toLowerCase().equals("serif")) {
|
---|
1176 | typefaces[j] = "Times New Roman";
|
---|
1177 | } else if ((typefaces[j].toLowerCase().equals("arial"))) {
|
---|
1178 | // Have to use Arial Unicode, otherwise unicode characters display incorrectly
|
---|
1179 | typefaces[j] = "Arial Unicode MS";
|
---|
1180 | }
|
---|
1181 |
|
---|
1182 | // Regex will remove any inverted commas surrounding multi-word typeface names
|
---|
1183 | font = new Font(typefaces[j].replaceAll("^'|'$", ""), Font.PLAIN, 12);
|
---|
1184 |
|
---|
1185 | // If the font isn't found, Java just uses Font.DIALOG, so this check checks whether the font was found
|
---|
1186 | if (!(font.getFamily().toLowerCase().equals(Font.DIALOG.toLowerCase()))) {
|
---|
1187 | fontFound = true;
|
---|
1188 | }
|
---|
1189 | }
|
---|
1190 |
|
---|
1191 | if (font.getFamily().toLowerCase().equals(Font.DIALOG.toLowerCase())) {
|
---|
1192 | font = new Font("Times New Roman", Font.PLAIN, 12);
|
---|
1193 | }
|
---|
1194 |
|
---|
1195 | String fontStyleComplete = "";
|
---|
1196 |
|
---|
1197 | int weightInt = 0;
|
---|
1198 |
|
---|
1199 | try {
|
---|
1200 | weightInt = Integer.parseInt(weight);
|
---|
1201 | } catch (NumberFormatException nfe) {
|
---|
1202 | // Use default value as set above
|
---|
1203 | }
|
---|
1204 |
|
---|
1205 | // checking if font is bold - i.e. 'bold', 'bolder' or weight over 500
|
---|
1206 | if (weight.toLowerCase().startsWith("bold") || weightInt > 500) {
|
---|
1207 | fontStyleComplete = fontStyleComplete.concat("bold");
|
---|
1208 | }
|
---|
1209 |
|
---|
1210 | if (fontStyle.toLowerCase().equals("italic") || fontStyle.toLowerCase().equals("oblique")) {
|
---|
1211 | fontStyleComplete = fontStyleComplete.concat("italic");
|
---|
1212 | }
|
---|
1213 |
|
---|
1214 | float fontSizeFloat = 12;
|
---|
1215 |
|
---|
1216 | try {
|
---|
1217 | fontSizeFloat = Float.valueOf(fontSize);
|
---|
1218 | } catch (NumberFormatException nfe) {
|
---|
1219 | // Use default value as set above
|
---|
1220 | }
|
---|
1221 |
|
---|
1222 | float letterSpacingFloat = -0.008f;
|
---|
1223 |
|
---|
1224 | try {
|
---|
1225 | letterSpacingFloat = (Integer.parseInt(letterSpacing.substring(0, letterSpacing.length() - 2)) / (fontSizeFloat));
|
---|
1226 | } catch (NumberFormatException nfe) {
|
---|
1227 | // Use default value as set above
|
---|
1228 | }
|
---|
1229 |
|
---|
1230 | float lineHeightInt = -1;
|
---|
1231 |
|
---|
1232 | try {
|
---|
1233 | lineHeightInt = (Float.parseFloat(lineHeight.substring(0, lineHeight.length() - 2)));
|
---|
1234 | } catch (NumberFormatException nfe) {
|
---|
1235 | // Use default value as set above
|
---|
1236 | }
|
---|
1237 |
|
---|
1238 | Text t;
|
---|
1239 |
|
---|
1240 | String textContent = currentNode.getTextContent().replaceAll("[^\\S\\n]+", " ");
|
---|
1241 | textContent = textContent.replaceAll("^(\\s)(\\n|\\r)", "");
|
---|
1242 |
|
---|
1243 | if (textTransform.equals("uppercase")) {
|
---|
1244 | textContent = textContent.toUpperCase();
|
---|
1245 | } else if (textTransform.equals("lowercase")) {
|
---|
1246 | textContent = textContent.toUpperCase();
|
---|
1247 | }
|
---|
1248 |
|
---|
1249 | // Adding the text to the frame. Expeditee text seems to be positioned relative to the baseline of the first line, so
|
---|
1250 | // the font size has to be added to the y-position
|
---|
1251 | t = frame.addText(Math.round(x), Math.round(y + fontSizeFloat), textContent, null);
|
---|
1252 |
|
---|
1253 | t.setColor(rgbStringToColor(color));
|
---|
1254 | t.setBackgroundColor(rgbStringToColor(bgColorString));
|
---|
1255 | t.setFont(font);
|
---|
1256 | t.setSize(fontSizeFloat);
|
---|
1257 | t.setFontStyle(fontStyleComplete);
|
---|
1258 | t.setLetterSpacing(letterSpacingFloat);
|
---|
1259 |
|
---|
1260 | // Removing any spacing between lines allowing t.getLineHeight() to be used to get the actual height
|
---|
1261 | // of just the characters (i.e. distance from ascenders to descenders)
|
---|
1262 | t.setSpacing(0);
|
---|
1263 |
|
---|
1264 | t.setSpacing(lineHeightInt - t.getLineHeight());
|
---|
1265 |
|
---|
1266 | if (align.equals("left")) {
|
---|
1267 | t.setJustification(Justification.left);
|
---|
1268 | } else if (align.equals("right")) {
|
---|
1269 | t.setJustification(Justification.right);
|
---|
1270 | } else if (align.equals("center")) {
|
---|
1271 | t.setJustification(Justification.center);
|
---|
1272 | } else if (align.equals("justify")) {
|
---|
1273 | t.setJustification(Justification.full);
|
---|
1274 | }
|
---|
1275 |
|
---|
1276 | // Font size is added to the item width to give a little breathing room
|
---|
1277 | t.setWidth(Math.round(width + (t.getSize())));
|
---|
1278 |
|
---|
1279 | if (!linkUrl.equals("undefined")) {
|
---|
1280 | t.setAction("gotourl " + linkUrl);
|
---|
1281 | t.setActionMark(false);
|
---|
1282 | }
|
---|
1283 | }
|
---|
1284 |
|
---|
1285 | } else if (currentNode.getNodeType() == Node.ELEMENT_NODE) {
|
---|
1286 | Node childNode = currentNode.getFirstChild();
|
---|
1287 |
|
---|
1288 | while (childNode != null) {
|
---|
1289 | addTextToFrame(childNode, visibleWidth, visibleHeight, window, webEngine, frame);
|
---|
1290 | childNode = childNode.getNextSibling();
|
---|
1291 | }
|
---|
1292 | }
|
---|
1293 | }
|
---|
1294 |
|
---|
1295 | private static class MutableBool {
|
---|
1296 | private boolean value;
|
---|
1297 |
|
---|
1298 | public MutableBool(boolean value) {
|
---|
1299 | this.value = value;
|
---|
1300 | }
|
---|
1301 |
|
---|
1302 | public boolean getValue() {
|
---|
1303 | return value;
|
---|
1304 | }
|
---|
1305 |
|
---|
1306 | public void setValue(boolean value) {
|
---|
1307 | this.value = value;
|
---|
1308 | }
|
---|
1309 | }
|
---|
1310 | }
|
---|