| 1 | /**
|
|---|
| 2 | * WebParser.java
|
|---|
| 3 | * Copyright (C) 2010 New Zealand Digital Library, http://expeditee.org
|
|---|
| 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 as published by
|
|---|
| 7 | * the Free Software Foundation, either version 3 of the License, or
|
|---|
| 8 | * (at your option) any later version.
|
|---|
| 9 | *
|
|---|
| 10 | * This program is distributed in the hope that it will be useful,
|
|---|
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|---|
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|---|
| 13 | * GNU General Public License for more details.
|
|---|
| 14 | *
|
|---|
| 15 | * You should have received a copy of the GNU General Public License
|
|---|
| 16 | * along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|---|
| 17 | */
|
|---|
| 18 |
|
|---|
| 19 | package org.expeditee.io;
|
|---|
| 20 |
|
|---|
| 21 | import java.io.File;
|
|---|
| 22 | import java.io.IOException;
|
|---|
| 23 | import java.lang.reflect.InvocationTargetException;
|
|---|
| 24 | import java.net.HttpURLConnection;
|
|---|
| 25 | import java.net.MalformedURLException;
|
|---|
| 26 | import java.net.URL;
|
|---|
| 27 | import java.text.SimpleDateFormat;
|
|---|
| 28 | /*
|
|---|
| 29 | * 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.
|
|---|
| 30 | * 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),
|
|---|
| 31 | * or you can just exclude JfxBrowser, WebParser and JfxbrowserActions from the build path.
|
|---|
| 32 | *
|
|---|
| 33 | * If you are using Ant to build/run, 'ant build' will try to build with JavaFX jar added to the classpath.
|
|---|
| 34 | * If this fails, 'ant build-nojfx' will build with the JfxBrowser, WebParser and JfxbrowserActions excluded from the build path.
|
|---|
| 35 | */
|
|---|
| 36 | import java.util.Arrays;
|
|---|
| 37 | import java.util.Date;
|
|---|
| 38 |
|
|---|
| 39 | import javafx.animation.AnimationTimer;
|
|---|
| 40 | import javafx.application.Platform;
|
|---|
| 41 | import javafx.beans.value.ChangeListener;
|
|---|
| 42 | import javafx.beans.value.ObservableValue;
|
|---|
| 43 | import javafx.concurrent.Worker.State;
|
|---|
| 44 | import javafx.scene.SnapshotParameters;
|
|---|
| 45 | import javafx.scene.image.WritableImage;
|
|---|
| 46 | import javafx.scene.web.WebEngine;
|
|---|
| 47 | import javafx.scene.web.WebView;
|
|---|
| 48 |
|
|---|
| 49 | import netscape.javascript.JSObject;
|
|---|
| 50 |
|
|---|
| 51 | import org.expeditee.core.Colour;
|
|---|
| 52 | import org.expeditee.core.Font;
|
|---|
| 53 | import org.expeditee.core.Image;
|
|---|
| 54 | import org.expeditee.core.InOutReference;
|
|---|
| 55 | import org.expeditee.gio.EcosystemManager;
|
|---|
| 56 | import org.expeditee.gio.gesture.StandardGestureActions;
|
|---|
| 57 | import org.expeditee.gio.swing.SwingMiscManager;
|
|---|
| 58 | import org.expeditee.gui.DisplayController;
|
|---|
| 59 | import org.expeditee.gui.Frame;
|
|---|
| 60 | import org.expeditee.gui.FrameIO;
|
|---|
| 61 | import org.expeditee.gui.FrameUtils;
|
|---|
| 62 | import org.expeditee.gui.MessageBay;
|
|---|
| 63 | import org.expeditee.gui.MessageBay.Progress;
|
|---|
| 64 | import org.expeditee.items.ItemUtils;
|
|---|
| 65 | import org.expeditee.items.Justification;
|
|---|
| 66 | import org.expeditee.items.Picture;
|
|---|
| 67 | import org.expeditee.items.Text;
|
|---|
| 68 | import org.expeditee.items.widgets.JfxBrowser;
|
|---|
| 69 | import org.w3c.dom.Node;
|
|---|
| 70 |
|
|---|
| 71 | /**
|
|---|
| 72 | * Methods to convert webpages to Expeditee frames
|
|---|
| 73 | *
|
|---|
| 74 | * @author ngw8
|
|---|
| 75 | * @author jts21
|
|---|
| 76 | */
|
|---|
| 77 | public class WebParser {
|
|---|
| 78 |
|
|---|
| 79 |
|
|---|
| 80 | /**
|
|---|
| 81 | * Loads a webpage and renders it as Expeditee frame(s)
|
|---|
| 82 | *
|
|---|
| 83 | * @param URL
|
|---|
| 84 | * Page to load
|
|---|
| 85 | * @param frame
|
|---|
| 86 | * The Expeditee frame to output the converted page to
|
|---|
| 87 | */
|
|---|
| 88 | public static void parseURL(final String URL, final Frame frame) {
|
|---|
| 89 | try {
|
|---|
| 90 | Platform.runLater(new Runnable() {
|
|---|
| 91 | @Override
|
|---|
| 92 | public void run() {
|
|---|
| 93 | try {
|
|---|
| 94 | WebEngine webEngine = new WebEngine(URL);
|
|---|
| 95 | loadPage(webEngine, frame);
|
|---|
| 96 | } catch (Exception e) {
|
|---|
| 97 | e.printStackTrace();
|
|---|
| 98 | }
|
|---|
| 99 | }
|
|---|
| 100 | });
|
|---|
| 101 | } catch (Exception e) {
|
|---|
| 102 | e.printStackTrace();
|
|---|
| 103 | }
|
|---|
| 104 | }
|
|---|
| 105 |
|
|---|
| 106 | protected static void loadPage(final WebEngine webEngine, final Frame frame) throws Exception {
|
|---|
| 107 | webEngine.getLoadWorker().stateProperty().addListener(new ChangeListener<State>() {
|
|---|
| 108 |
|
|---|
| 109 | @Override
|
|---|
| 110 | public void changed(ObservableValue<? extends State> ov, State oldState, State newState) {
|
|---|
| 111 |
|
|---|
| 112 | switch (newState) {
|
|---|
| 113 | case READY: // READY
|
|---|
| 114 | // MessageBay.displayMessage("WebEngine ready");
|
|---|
| 115 | break;
|
|---|
| 116 | case SCHEDULED: // SCHEDULED
|
|---|
| 117 | // MessageBay.displayMessage("Scheduled page load");
|
|---|
| 118 | break;
|
|---|
| 119 | case RUNNING: // RUNNING
|
|---|
| 120 | System.out.println("Loading page!");
|
|---|
| 121 | // MessageBay.displayMessage("WebEngine running");
|
|---|
| 122 | break;
|
|---|
| 123 | case SUCCEEDED: // SUCCEEDED
|
|---|
| 124 | // MessageBay.displayMessage("Finished loading page");
|
|---|
| 125 | System.out.println("Parsing page!");
|
|---|
| 126 | webEngine.executeScript("window.resizeTo(800, 800);"
|
|---|
| 127 | + "document.body.style.width = '1000px'");
|
|---|
| 128 | parsePage(webEngine, frame);
|
|---|
| 129 | System.out.println("Parsed page!");
|
|---|
| 130 | break;
|
|---|
| 131 | case CANCELLED: // CANCELLED
|
|---|
| 132 | MessageBay.displayMessage("Cancelled loading page");
|
|---|
| 133 | break;
|
|---|
| 134 | case FAILED: // FAILED
|
|---|
| 135 | MessageBay.displayMessage("Failed to load page");
|
|---|
| 136 | break;
|
|---|
| 137 | }
|
|---|
| 138 | }
|
|---|
| 139 | });
|
|---|
| 140 | }
|
|---|
| 141 |
|
|---|
| 142 | /**
|
|---|
| 143 | * Converts a loaded page to Expeditee frame(s)
|
|---|
| 144 | *
|
|---|
| 145 | * @param webEngine
|
|---|
| 146 | * The JavaFX WebEngine in which the page to be converted is loaded
|
|---|
| 147 | * @param frame
|
|---|
| 148 | * The Expeditee frame to output the converted page to
|
|---|
| 149 | */
|
|---|
| 150 | public static void parsePage(final WebEngine webEngine, final Frame frame) {
|
|---|
| 151 | try {
|
|---|
| 152 | Platform.runLater(new Runnable() {
|
|---|
| 153 | @Override
|
|---|
| 154 | public void run() {
|
|---|
| 155 | try {
|
|---|
| 156 | Progress progressBar = MessageBay.displayProgress("Converting web page");
|
|---|
| 157 |
|
|---|
| 158 | Node doc = (Node) webEngine.executeScript("document.body");
|
|---|
| 159 |
|
|---|
| 160 | JSObject window = (JSObject) webEngine.executeScript("window");
|
|---|
| 161 |
|
|---|
| 162 | frame.setBackgroundColor(rgbStringToColor((String) ((JSObject) (window.call("getComputedStyle", new Object[] { doc }))).call("getPropertyValue",
|
|---|
| 163 | new Object[] { "background-color" })));
|
|---|
| 164 |
|
|---|
| 165 | // Functions to be used later in JavaScript
|
|---|
| 166 | webEngine.executeScript(""
|
|---|
| 167 | + "function addToSpan(text) {"
|
|---|
| 168 | + " span = document.createElement('wordSpan');"
|
|---|
| 169 | + " span.textContent = text;"
|
|---|
| 170 | + " par.insertBefore(span, refNode);"
|
|---|
| 171 | // Checking if the current word is on a new line (i.e. lower than the previous word)
|
|---|
| 172 | + " if (prevSpan !== null && span.getBoundingClientRect().top > prevSpan.getBoundingClientRect().top) {"
|
|---|
| 173 | // If it is, prepend a new line character to it. The new line characters doesn't affect the rendered HTML
|
|---|
| 174 | + " span.textContent = '\\n' + span.textContent;"
|
|---|
| 175 |
|
|---|
| 176 | // Checking if the previous word is horizontally aligned with the one before it.
|
|---|
| 177 | // If it is, merge the text of the two spans
|
|---|
| 178 | + " if ( prevPrevSpan !== null && prevPrevSpan.getBoundingClientRect().left == prevSpan.getBoundingClientRect().left) {"
|
|---|
| 179 | + " prevPrevSpan.textContent = prevPrevSpan.textContent + prevSpan.textContent;"
|
|---|
| 180 | + " par.removeChild(prevSpan);"
|
|---|
| 181 | + " } else {"
|
|---|
| 182 | + " prevPrevSpan = prevSpan;"
|
|---|
| 183 | + " }"
|
|---|
| 184 | + " prevSpan = span;"
|
|---|
| 185 | + " } else if ( prevSpan !== null) {"
|
|---|
| 186 | // Word is on the same line as the previous one, so merge the second into the span of the first
|
|---|
| 187 | + " prevSpan.textContent = prevSpan.textContent + span.textContent;"
|
|---|
| 188 | + " par.removeChild(span);"
|
|---|
| 189 | + " } else {"
|
|---|
| 190 | + " prevSpan = span;"
|
|---|
| 191 | + " }"
|
|---|
| 192 | + "}"
|
|---|
| 193 |
|
|---|
| 194 | + "function splitIntoWords(toSplit) {"
|
|---|
| 195 | + " var words = [];"
|
|---|
| 196 | + " var pattern = /\\s+/g;"
|
|---|
| 197 | + " var words = toSplit.split(pattern);"
|
|---|
| 198 | + ""
|
|---|
| 199 | + " for (var i = 0; i < words.length - 1; i++) {"
|
|---|
| 200 | + " words[i] = words[i] + ' ';"
|
|---|
| 201 | + " }"
|
|---|
| 202 | + " return words;"
|
|---|
| 203 | + "}"
|
|---|
| 204 | );
|
|---|
| 205 |
|
|---|
| 206 | // Using Javascript to get an array of all the text nodes in the document so they can be wrapped in spans. Have to
|
|---|
| 207 | // loop through twice (once to build the array and once actually going through the array, otherwise when the
|
|---|
| 208 | // textnode is removed from the document items end up being skipped)
|
|---|
| 209 | JSObject textNodes = (JSObject) webEngine.executeScript(""
|
|---|
| 210 | + "function getTextNodes(rootNode){"
|
|---|
| 211 | + "var node;"
|
|---|
| 212 | + "var textNodes=[];"
|
|---|
| 213 | + "var walk = document.createTreeWalker(rootNode, NodeFilter.SHOW_TEXT);"
|
|---|
| 214 | + "while(node=walk.nextNode()) {"
|
|---|
| 215 | + "if((node.textContent.trim().length > 0)) { "
|
|---|
| 216 | + "textNodes.push(node);"
|
|---|
| 217 | + "}"
|
|---|
| 218 | + "}"
|
|---|
| 219 | + "return textNodes;"
|
|---|
| 220 | + "}; "
|
|---|
| 221 | + "getTextNodes(document.body)"
|
|---|
| 222 | );
|
|---|
| 223 |
|
|---|
| 224 | int nodesLength = (Integer) textNodes.getMember("length");
|
|---|
| 225 |
|
|---|
| 226 | // Looping through all the text nodes in the document
|
|---|
| 227 | for (int j = 0; j < nodesLength; j++) {
|
|---|
| 228 | Node currentNode = (Node) textNodes.getSlot(j);
|
|---|
| 229 |
|
|---|
| 230 | // Making the current node accessible in JavaScript
|
|---|
| 231 | window.setMember("currentNode", currentNode);
|
|---|
| 232 |
|
|---|
| 233 | webEngine.executeScript(""
|
|---|
| 234 | + "var span = null, prevSpan = null, prevPrevSpan = null;"
|
|---|
| 235 |
|
|---|
| 236 | // Removing repeated whitespace from the text node's content then splitting it into individual words
|
|---|
| 237 | + "var textContent = currentNode.textContent.replace(/\\n|\\r/g, '').replace(/\\s+/g, ' ');"
|
|---|
| 238 | + "var words = splitIntoWords(textContent);"
|
|---|
| 239 |
|
|---|
| 240 | + "var refNode = currentNode.nextSibling;"
|
|---|
| 241 | + "var par = currentNode.parentElement;"
|
|---|
| 242 | + "currentNode.parentElement.removeChild(currentNode);"
|
|---|
| 243 |
|
|---|
| 244 | + "for (var i = 0; i < words.length; i++) {"
|
|---|
| 245 | + " addToSpan(words[i]);"
|
|---|
| 246 | + "}"
|
|---|
| 247 |
|
|---|
| 248 | + "if (prevPrevSpan !== null && prevPrevSpan.getBoundingClientRect().left == prevSpan.getBoundingClientRect().left) {"
|
|---|
| 249 | + " prevPrevSpan.textContent = prevPrevSpan.textContent + prevSpan.textContent;"
|
|---|
| 250 | + " par.removeChild(prevSpan);"
|
|---|
| 251 | + "}"
|
|---|
| 252 | );
|
|---|
| 253 |
|
|---|
| 254 | // Will never reach 100% here, as the processing is not quite finished - progress is set to 100% at the end of
|
|---|
| 255 | // the addPageToFrame loop below
|
|---|
| 256 | progressBar.set((100 * (j)) / nodesLength);
|
|---|
| 257 | }
|
|---|
| 258 |
|
|---|
| 259 | // Finding all links within the page, then setting the href attribute of all their descendants to be the same
|
|---|
| 260 | // link/URL.
|
|---|
| 261 | // This is needed because there is no apparent and efficient way to check if an element is a child of a link when
|
|---|
| 262 | // running through the document when added each element to Expeditee
|
|---|
| 263 | webEngine.executeScript(""
|
|---|
| 264 | + "var anchors = document.getElementsByTagName('a');"
|
|---|
| 265 | + ""
|
|---|
| 266 | + "for (var i = 0; i < anchors.length; i++) {"
|
|---|
| 267 | + "var currentAnchor = anchors.item(i);"
|
|---|
| 268 | + "var anchorDescendants = currentAnchor.querySelectorAll('*');"
|
|---|
| 269 | + "for (var j = 0; j < anchorDescendants.length; j++) {"
|
|---|
| 270 | + "anchorDescendants.item(j).href = currentAnchor.href;"
|
|---|
| 271 | + "}"
|
|---|
| 272 | + "}"
|
|---|
| 273 | );
|
|---|
| 274 |
|
|---|
| 275 | WebParser.addPageToFrame(doc, window, webEngine, frame);
|
|---|
| 276 |
|
|---|
| 277 | progressBar.set(100);
|
|---|
| 278 |
|
|---|
| 279 | } catch (Exception e) {
|
|---|
| 280 | e.printStackTrace();
|
|---|
| 281 | }
|
|---|
| 282 | System.out.println("Parsed frame");
|
|---|
| 283 | FrameUtils.Parse(frame);
|
|---|
| 284 | frame.setChanged(true);
|
|---|
| 285 | FrameIO.SaveFrame(frame);
|
|---|
| 286 | }
|
|---|
| 287 | });
|
|---|
| 288 | } catch (Exception e) {
|
|---|
| 289 | e.printStackTrace();
|
|---|
| 290 | }
|
|---|
| 291 | }
|
|---|
| 292 |
|
|---|
| 293 | /**
|
|---|
| 294 | * Converts a loaded page to Expeditee frame(s)
|
|---|
| 295 | *
|
|---|
| 296 | * @param webEngine
|
|---|
| 297 | * The JavaFX WebEngine in which the page to be converted is loaded
|
|---|
| 298 | * @param frame
|
|---|
| 299 | * The Expeditee frame to output the converted page to
|
|---|
| 300 | */
|
|---|
| 301 | public static void parsePageSimple(final JfxBrowser browserWidget, final WebEngine webEngine, final WebView webView, final Frame frame) {
|
|---|
| 302 | try {
|
|---|
| 303 |
|
|---|
| 304 | final int verticalScrollPerPage = (int) (DisplayController.getFramePaintAreaHeight() * 0.85);
|
|---|
| 305 | final int horizontalScrollPerPage = (int) (DisplayController.getFramePaintAreaWidth() * 0.85);
|
|---|
| 306 |
|
|---|
| 307 | Platform.runLater(new Runnable() {
|
|---|
| 308 |
|
|---|
| 309 | @Override
|
|---|
| 310 | public void run() {
|
|---|
| 311 | browserWidget.setOverlayVisible(true);
|
|---|
| 312 |
|
|---|
| 313 | // Webview area is set to slightly larger than the size of a converted page, to give some overlap between each page
|
|---|
| 314 | browserWidget.setWebViewSize(horizontalScrollPerPage * 1.1, verticalScrollPerPage * 1.1);
|
|---|
| 315 | browserWidget.setScrollbarsVisible(false);
|
|---|
| 316 | }
|
|---|
| 317 | });
|
|---|
| 318 |
|
|---|
| 319 | final Object notifier = new Object();
|
|---|
| 320 |
|
|---|
| 321 | final InOutReference<Integer> verticalCount = new InOutReference<Integer>(0);
|
|---|
| 322 | final InOutReference<Integer> horizontalCount = new InOutReference<Integer>(0);
|
|---|
| 323 |
|
|---|
| 324 | final InOutReference<Integer> pagesVertical = new InOutReference<Integer>(1);
|
|---|
| 325 | final InOutReference<Integer> pagesHorizontal = new InOutReference<Integer>(1);
|
|---|
| 326 |
|
|---|
| 327 | final String pageTitle;
|
|---|
| 328 |
|
|---|
| 329 | if (webEngine.getTitle() != null) {
|
|---|
| 330 | pageTitle = webEngine.getTitle();
|
|---|
| 331 | } else {
|
|---|
| 332 | pageTitle = "Untitled Page";
|
|---|
| 333 | }
|
|---|
| 334 |
|
|---|
| 335 | final Progress progressBar = MessageBay.displayProgress("Converting web page");
|
|---|
| 336 |
|
|---|
| 337 | final Frame frameset = FrameIO.CreateNewFrameset(FrameIO.ConvertToValidFramesetName((new SimpleDateFormat("yy-MM-dd-HH-mm-ss").format(new Date())) + pageTitle));
|
|---|
| 338 |
|
|---|
| 339 | frameset.setTitle(pageTitle);
|
|---|
| 340 | frameset.getTitleItem().setSize(14);
|
|---|
| 341 |
|
|---|
| 342 | WebParser.addButton("Return to original frame", frame.getName(), null, 200, frameset, null, 0, 10, null);
|
|---|
| 343 |
|
|---|
| 344 | Text link = DisplayController.getCurrentFrame().addText(DisplayController.getMouseX(), DisplayController.getMouseY(), pageTitle, null);
|
|---|
| 345 | link.setLink(frameset.getName());
|
|---|
| 346 |
|
|---|
| 347 | StandardGestureActions.pickup(link);
|
|---|
| 348 |
|
|---|
| 349 | // Timer that fires every time JFX is redrawn. After a few redraws, the handle method of this takes a screenshot of the page,
|
|---|
| 350 | // adds it to the frame, then adds the text on top
|
|---|
| 351 | AnimationTimer timer = new AnimationTimer() {
|
|---|
| 352 |
|
|---|
| 353 | int frameCount = 0;
|
|---|
| 354 |
|
|---|
| 355 | Frame frameToAddTo = frameset;
|
|---|
| 356 | int thumbWidth = 100;
|
|---|
| 357 |
|
|---|
| 358 | @Override
|
|---|
| 359 | public void handle(long arg0) {
|
|---|
| 360 | // Must wait 2 frames before taking a snapshot of the webview, otherwise JavaFX won't have redrawn
|
|---|
| 361 | if (frameCount++ > 1) {
|
|---|
| 362 | frameCount = 0;
|
|---|
| 363 | this.stop();
|
|---|
| 364 |
|
|---|
| 365 | verticalCount.set(verticalCount.get() + 1);
|
|---|
| 366 |
|
|---|
| 367 | frameToAddTo = FrameIO.CreateFrame(frameToAddTo.getFramesetName(), pageTitle, null);
|
|---|
| 368 | frameToAddTo.removeAllItems(frameToAddTo.getSortedItems());
|
|---|
| 369 |
|
|---|
| 370 | try {
|
|---|
| 371 | // removing the CSS that hides the text (otherwise the text would not pass the visibility check that is run on
|
|---|
| 372 | // it before adding it to the frame)
|
|---|
| 373 | webEngine.executeScript("cssHide.innerHTML = '';");
|
|---|
| 374 |
|
|---|
| 375 | JSObject window = (JSObject) webEngine.executeScript("window");
|
|---|
| 376 |
|
|---|
| 377 | int visibleWidth = (Integer) webEngine.executeScript("window.innerWidth");
|
|---|
| 378 | int visibleHeight = (Integer) webEngine.executeScript("window.innerHeight");
|
|---|
| 379 |
|
|---|
| 380 | WebParser.addTextToFrame(visibleWidth, visibleHeight, window, webEngine, frameToAddTo);
|
|---|
| 381 |
|
|---|
| 382 | FrameIO.SaveFrame(frameToAddTo);
|
|---|
| 383 | } catch (Exception ex) {
|
|---|
| 384 | ex.printStackTrace();
|
|---|
| 385 | }
|
|---|
| 386 |
|
|---|
| 387 | webEngine.executeScript(""
|
|---|
| 388 | // Setting all text to be hidden before taking the screenshot
|
|---|
| 389 | + "cssHide.appendChild(document.createTextNode(wordSpanHiddenStyle));");
|
|---|
| 390 |
|
|---|
| 391 | WritableImage img = new WritableImage((int)webView.getWidth(), (int)webView.getHeight());
|
|---|
| 392 |
|
|---|
| 393 | webView.snapshot(new SnapshotParameters(), img);
|
|---|
| 394 |
|
|---|
| 395 | // Getting a BufferedImage from the JavaFX image
|
|---|
| 396 | //BufferedImage image = SwingFXUtils.fromFXImage(img, null);
|
|---|
| 397 | Image image = SwingMiscManager.getImageForJavaFXImage(img);
|
|---|
| 398 |
|
|---|
| 399 | try {
|
|---|
| 400 | // TODO: tidy. cts16
|
|---|
| 401 | //int hashcode = Arrays.hashCode(image.getData().getPixels(0, 0, image.getWidth(), image.getHeight(), (int[]) null));
|
|---|
| 402 | int hashcode = image.hashCode();
|
|---|
| 403 |
|
|---|
| 404 | File out = new File(FrameIO.IMAGES_PATH + "webpage-" + Integer.toHexString(hashcode) + ".png");
|
|---|
| 405 | out.mkdirs();
|
|---|
| 406 | image.writeToDisk("png", out);
|
|---|
| 407 |
|
|---|
| 408 | // Adding the image to the frame
|
|---|
| 409 | frameToAddTo.addText(0, 0, "@i: " + out.getName(), null);
|
|---|
| 410 |
|
|---|
| 411 | // Adding thumbnail to the overview page
|
|---|
| 412 | Text thumb = frameset.addText((int) (thumbWidth * 1.1 * horizontalCount.get()) + 10,
|
|---|
| 413 | (int) ((((float) thumbWidth / image.getWidth()) * image.getHeight()) * 1.1 * verticalCount.get()),
|
|---|
| 414 | "@i: " + out.getName() + " " + thumbWidth,
|
|---|
| 415 | null);
|
|---|
| 416 |
|
|---|
| 417 | thumb.setLink(frameToAddTo.getName());
|
|---|
| 418 | thumb.setBorderColor(Colour.LIGHT_GREY);
|
|---|
| 419 | thumb.setThickness(1);
|
|---|
| 420 |
|
|---|
| 421 | // Button to go to the next frame/page
|
|---|
| 422 | WebParser.addButton("Next", null, "next", 70, frameToAddTo, null, 0, 10, null);
|
|---|
| 423 |
|
|---|
| 424 | // Button to go to the previous frame/page
|
|---|
| 425 | if (verticalCount.get() > 1 || horizontalCount.get() > 0) {
|
|---|
| 426 | WebParser.addButton("Previous", null, "previous", 70, frameToAddTo, null, 85, 10, null);
|
|---|
| 427 | }
|
|---|
| 428 |
|
|---|
| 429 | // Button to return to the index/overview page
|
|---|
| 430 | WebParser.addButton("Index", frameset.getName(), null, 70, frameToAddTo, null, null, 10, 5);
|
|---|
| 431 |
|
|---|
| 432 | FrameIO.SaveFrame(frameToAddTo);
|
|---|
| 433 | FrameIO.SaveFrame(frameset);
|
|---|
| 434 |
|
|---|
| 435 | } catch (IOException e) {
|
|---|
| 436 | e.printStackTrace();
|
|---|
| 437 | }
|
|---|
| 438 |
|
|---|
| 439 | image.releaseImage();
|
|---|
| 440 |
|
|---|
| 441 | synchronized (notifier) {
|
|---|
| 442 | // Notifying that the timer has finished
|
|---|
| 443 | notifier.notify();
|
|---|
| 444 | }
|
|---|
| 445 | }
|
|---|
| 446 | }
|
|---|
| 447 | };
|
|---|
| 448 |
|
|---|
| 449 | Platform.runLater(new Runnable() {
|
|---|
| 450 | @Override
|
|---|
| 451 | public void run() {
|
|---|
| 452 | try {
|
|---|
| 453 | JSObject window = (JSObject) webEngine.executeScript("window");
|
|---|
| 454 |
|
|---|
| 455 | webEngine.executeScript(""
|
|---|
| 456 | // Initializing the counter used when scrolling the page
|
|---|
| 457 | + "var scrollCounter = 0;"
|
|---|
| 458 | + "var scrollCounterHorizontal = 0;"
|
|---|
| 459 |
|
|---|
| 460 | // Storing the current scroll position
|
|---|
| 461 | + "var originalScrollX = window.pageXOffset;"
|
|---|
| 462 | + "var originalScrollY = window.pageYOffset;");
|
|---|
| 463 |
|
|---|
| 464 | window.setMember("horizontalScrollPerPage", horizontalScrollPerPage);
|
|---|
| 465 | window.setMember("verticalScrollPerPage", verticalScrollPerPage);
|
|---|
| 466 |
|
|---|
| 467 |
|
|---|
| 468 |
|
|---|
| 469 | // The scrollPerPage will always be less than the page's height, due to the overlap being added/allowed for between pages,
|
|---|
| 470 | // but if the webpage fits in a single converted page, there's no need for any overlap, so just use 1 as the number of pages
|
|---|
| 471 | if((Boolean) webEngine.executeScript("document.documentElement.scrollHeight > window.innerHeight")) {
|
|---|
| 472 | pagesVertical.set((int) Math.ceil((Integer) webEngine.executeScript("document.documentElement.scrollHeight") / (float) verticalScrollPerPage));
|
|---|
| 473 | }
|
|---|
| 474 |
|
|---|
| 475 | if((Boolean) webEngine.executeScript("document.documentElement.scrollWidth > window.innerWidth")) {
|
|---|
| 476 | pagesHorizontal.set((int) Math.ceil((Integer) webEngine.executeScript("document.documentElement.scrollWidth") / (float) horizontalScrollPerPage));
|
|---|
| 477 | }
|
|---|
| 478 |
|
|---|
| 479 | System.out.println(webEngine.executeScript("document.documentElement.scrollWidth") + "/" + horizontalScrollPerPage);
|
|---|
| 480 | System.out.println(pagesVertical.get() + "x" + pagesHorizontal.get());
|
|---|
| 481 |
|
|---|
| 482 | // Setting up the element that contains the CSS to hide all text. Also hiding readability mode buttons.
|
|---|
| 483 | // This is wiped before the text is converted, then re-added before taking the screenshot
|
|---|
| 484 | webEngine.executeScript(""
|
|---|
| 485 | + "var cssHide = document.createElement('style');"
|
|---|
| 486 | + "cssHide.type = 'text/css';"
|
|---|
| 487 | + "var wordSpanHiddenStyle = 'WordSpan, #readOverlay #readTools { visibility: hidden !important;}';"
|
|---|
| 488 | + "cssHide.appendChild(document.createTextNode(wordSpanHiddenStyle));"
|
|---|
| 489 | + "document.getElementsByTagName('head')[0].appendChild(cssHide);"
|
|---|
| 490 | );
|
|---|
| 491 |
|
|---|
| 492 | // Replacing line breaks in all <pre> tags with <br> tags, otherwise they are lost during the conversion
|
|---|
| 493 | webEngine.executeScript(""
|
|---|
| 494 | + "var pres = document.getElementsByTagName ('pre');"
|
|---|
| 495 | + "for(var i = 0; i < pres.length; i++){"
|
|---|
| 496 | + " pres[i].innerHTML = pres[i].innerHTML.replace(/\\n|\\r/g, '<br />');"
|
|---|
| 497 | + "}");
|
|---|
| 498 |
|
|---|
| 499 | // Functions to be used later in JavaScript
|
|---|
| 500 | webEngine.executeScript(""
|
|---|
| 501 | + "function addToSpan(text) {"
|
|---|
| 502 | + " span = document.createElement('wordSpan');"
|
|---|
| 503 | + " span.textContent = text;"
|
|---|
| 504 | + " par.insertBefore(span, refNode);"
|
|---|
| 505 | + " if (prevSpan !== null && span.getBoundingClientRect().top > prevSpan.getBoundingClientRect().top) {"
|
|---|
| 506 | + " span.textContent = '\\n' + span.textContent;"
|
|---|
| 507 | + " if ( prevPrevSpan !== null && prevPrevSpan.getBoundingClientRect().left == prevSpan.getBoundingClientRect().left) {"
|
|---|
| 508 | + " prevPrevSpan.textContent = prevPrevSpan.textContent + prevSpan.textContent;"
|
|---|
| 509 | + " par.removeChild(prevSpan);"
|
|---|
| 510 | + " } else {"
|
|---|
| 511 | + " prevPrevSpan = prevSpan;"
|
|---|
| 512 | + " }"
|
|---|
| 513 | + " prevSpan = span;"
|
|---|
| 514 | + " } else if ( prevSpan !== null) {"
|
|---|
| 515 | + " prevSpan.textContent = prevSpan.textContent + span.textContent;"
|
|---|
| 516 | + " par.removeChild(span);"
|
|---|
| 517 | + " } else {"
|
|---|
| 518 | + " prevSpan = span;"
|
|---|
| 519 | + " }"
|
|---|
| 520 | + "}"
|
|---|
| 521 |
|
|---|
| 522 | + "function splitIntoWords(toSplit) {"
|
|---|
| 523 | + " var words = [];"
|
|---|
| 524 | + " var pattern = /\\s+/g;"
|
|---|
| 525 | + " var words = toSplit.split(pattern);"
|
|---|
| 526 | + ""
|
|---|
| 527 | + " for (var i = 0; i < words.length - 1; i++) {"
|
|---|
| 528 | + " words[i] = words[i] + ' ';"
|
|---|
| 529 | + " }"
|
|---|
| 530 | + " return words;"
|
|---|
| 531 | + "}"
|
|---|
| 532 | );
|
|---|
| 533 |
|
|---|
| 534 | // Using Javascript to get an array of all the text nodes in the document so they can be wrapped in spans. Have to
|
|---|
| 535 | // loop through twice (once here to build the array and once later actually going through the array, otherwise when the
|
|---|
| 536 | // textnode is removed from the document items end up being skipped)
|
|---|
| 537 | webEngine.executeScript(""
|
|---|
| 538 | + "var node;"
|
|---|
| 539 | + "var textNodes=[];"
|
|---|
| 540 | + "var walk = document.createTreeWalker(document.body, NodeFilter.SHOW_TEXT);"
|
|---|
| 541 | );
|
|---|
| 542 |
|
|---|
| 543 | while(webEngine.executeScript("node=walk.nextNode()") != null && browserWidget.isParserRunning()) {
|
|---|
| 544 |
|
|---|
| 545 | webEngine.executeScript(""
|
|---|
| 546 | + "if((node.textContent.trim().length > 0)) { "
|
|---|
| 547 | + "textNodes.push(node);"
|
|---|
| 548 | + "}"
|
|---|
| 549 | );
|
|---|
| 550 | }
|
|---|
| 551 |
|
|---|
| 552 | JSObject textNodes = (JSObject) webEngine.executeScript("textNodes");
|
|---|
| 553 |
|
|---|
| 554 | int nodesLength = (Integer) textNodes.getMember("length");
|
|---|
| 555 |
|
|---|
| 556 | // Looping through all the text nodes in the document
|
|---|
| 557 | for (int j = 0; j < nodesLength && browserWidget.isParserRunning(); j++) {
|
|---|
| 558 | Node currentNode = (Node) textNodes.getSlot(j);
|
|---|
| 559 |
|
|---|
| 560 | // Making the current node accessible in JavaScript
|
|---|
| 561 | window.setMember("currentNode", currentNode);
|
|---|
| 562 |
|
|---|
| 563 | webEngine.executeScript(""
|
|---|
| 564 | + "var span = null, prevSpan = null, prevPrevSpan = null;"
|
|---|
| 565 |
|
|---|
| 566 | // Removing repeated whitespace from the text node's content then splitting it into individual words
|
|---|
| 567 | + "var textContent = currentNode.textContent.replace(/\\n|\\r/g, '').replace(/\\s+/g, ' ');"
|
|---|
| 568 | + "var words = splitIntoWords(textContent);"
|
|---|
| 569 |
|
|---|
| 570 | + "var refNode = currentNode.nextSibling;"
|
|---|
| 571 | + "var par = currentNode.parentElement;"
|
|---|
| 572 | + "currentNode.parentElement.removeChild(currentNode);"
|
|---|
| 573 |
|
|---|
| 574 | + "for (var i = 0; i < words.length; i++) {"
|
|---|
| 575 | + " addToSpan(words[i]);"
|
|---|
| 576 | + "}"
|
|---|
| 577 |
|
|---|
| 578 | + "if (prevPrevSpan !== null && prevPrevSpan.getBoundingClientRect().left == prevSpan.getBoundingClientRect().left) {"
|
|---|
| 579 | + " prevPrevSpan.textContent = prevPrevSpan.textContent + prevSpan.textContent;"
|
|---|
| 580 | + " par.removeChild(prevSpan);"
|
|---|
| 581 | + "}"
|
|---|
| 582 | );
|
|---|
| 583 |
|
|---|
| 584 | // Will never reach 100% here, as the processing is not quite finished - progress is set to 100% at the end of
|
|---|
| 585 | // the addPageToFrame loop below
|
|---|
| 586 | try {
|
|---|
| 587 | progressBar.set((50 * (j + 1)) / nodesLength);
|
|---|
| 588 | } catch (Exception e) {
|
|---|
| 589 | // Seems to be a bug somewhere along the line when updating the progressbar, so am catching any exception
|
|---|
| 590 | // thrown here to avoid it stuffing up the rest of the parsing
|
|---|
| 591 | e.printStackTrace();
|
|---|
| 592 | }
|
|---|
| 593 | }
|
|---|
| 594 |
|
|---|
| 595 | // Finding all links within the page, then setting the href attribute of all their descendants to be the same
|
|---|
| 596 | // link/URL.
|
|---|
| 597 | // This is needed because there is no apparent and efficient way to check if an element is a child of a link when
|
|---|
| 598 | // running through the document when added each element to Expeditee
|
|---|
| 599 | webEngine.executeScript(""
|
|---|
| 600 | + "var anchors = document.getElementsByTagName('a');"
|
|---|
| 601 | + ""
|
|---|
| 602 | + "for (var i = 0; i < anchors.length; i++) {"
|
|---|
| 603 | + "var currentAnchor = anchors.item(i);"
|
|---|
| 604 | + "var anchorDescendants = currentAnchor.querySelectorAll('*');"
|
|---|
| 605 | + "for (var j = 0; j < anchorDescendants.length; j++) {"
|
|---|
| 606 | + "anchorDescendants.item(j).href = currentAnchor.href;"
|
|---|
| 607 | + "}"
|
|---|
| 608 | + "}"
|
|---|
| 609 | );
|
|---|
| 610 |
|
|---|
| 611 | } catch (Exception ex) {
|
|---|
| 612 | ex.printStackTrace();
|
|---|
| 613 | }
|
|---|
| 614 |
|
|---|
| 615 | synchronized (notifier) {
|
|---|
| 616 | notifier.notify();
|
|---|
| 617 | }
|
|---|
| 618 | }
|
|---|
| 619 | });
|
|---|
| 620 |
|
|---|
| 621 | synchronized (notifier) {
|
|---|
| 622 | try {
|
|---|
| 623 | // Waiting for the page setup (splitting into spans) to finish
|
|---|
| 624 | notifier.wait();
|
|---|
| 625 | } catch (InterruptedException e) {
|
|---|
| 626 | // TODO Auto-generated catch block
|
|---|
| 627 | e.printStackTrace();
|
|---|
| 628 | }
|
|---|
| 629 | }
|
|---|
| 630 |
|
|---|
| 631 | // Loop that scrolls the page horizontally
|
|---|
| 632 | for(int i = 0; i < pagesHorizontal.get() && browserWidget.isParserRunning(); i++) {
|
|---|
| 633 |
|
|---|
| 634 | Platform.runLater(new Runnable() {
|
|---|
| 635 | @Override
|
|---|
| 636 | public void run() {
|
|---|
| 637 | try {
|
|---|
| 638 | // Scrolling down the page
|
|---|
| 639 | webEngine.executeScript(""
|
|---|
| 640 | + "scrollCounter = 0;"
|
|---|
| 641 | + "window.scrollTo(scrollCounterHorizontal * horizontalScrollPerPage, 0);"
|
|---|
| 642 | + "scrollCounterHorizontal = scrollCounterHorizontal+1;");
|
|---|
| 643 |
|
|---|
| 644 | } catch (Exception e) {
|
|---|
| 645 | e.printStackTrace();
|
|---|
| 646 | }
|
|---|
| 647 | }
|
|---|
| 648 | });
|
|---|
| 649 |
|
|---|
| 650 | // Loop that scrolls the page vertically (for each horizontal scroll position)
|
|---|
| 651 | for(int j = 0; j < pagesVertical.get() && browserWidget.isParserRunning(); j++) {
|
|---|
| 652 |
|
|---|
| 653 | try {
|
|---|
| 654 | progressBar.set((int) (50 + ((float)(j+1)/(pagesVertical.get() * pagesHorizontal.get()) + ((float)(i) / pagesHorizontal.get())) * 50));
|
|---|
| 655 | } catch (Exception e) {
|
|---|
| 656 | e.printStackTrace();
|
|---|
| 657 | }
|
|---|
| 658 |
|
|---|
| 659 | Platform.runLater(new Runnable() {
|
|---|
| 660 | @Override
|
|---|
| 661 | public void run() {
|
|---|
| 662 | try {
|
|---|
| 663 | // Scrolling down the page
|
|---|
| 664 | webEngine.executeScript(""
|
|---|
| 665 | + "window.scrollTo(window.pageXOffset, scrollCounter * verticalScrollPerPage);"
|
|---|
| 666 | + "scrollCounter = scrollCounter+1;");
|
|---|
| 667 |
|
|---|
| 668 | synchronized (notifier) {
|
|---|
| 669 | notifier.notify();
|
|---|
| 670 | }
|
|---|
| 671 |
|
|---|
| 672 | } catch (Exception e) {
|
|---|
| 673 | e.printStackTrace();
|
|---|
| 674 | }
|
|---|
| 675 | }
|
|---|
| 676 | });
|
|---|
| 677 |
|
|---|
| 678 | synchronized (notifier) {
|
|---|
| 679 | try {
|
|---|
| 680 | // Waiting for the page to be scrolled
|
|---|
| 681 | notifier.wait();
|
|---|
| 682 | } catch (InterruptedException e) {
|
|---|
| 683 | // TODO Auto-generated catch block
|
|---|
| 684 | e.printStackTrace();
|
|---|
| 685 | }
|
|---|
| 686 | }
|
|---|
| 687 |
|
|---|
| 688 | timer.start();
|
|---|
| 689 |
|
|---|
| 690 | synchronized (notifier) {
|
|---|
| 691 | try {
|
|---|
| 692 | // Waiting for the timer thread to finish before looping again
|
|---|
| 693 | notifier.wait();
|
|---|
| 694 | } catch (InterruptedException e) {
|
|---|
| 695 | // TODO Auto-generated catch block
|
|---|
| 696 | e.printStackTrace();
|
|---|
| 697 | }
|
|---|
| 698 | }
|
|---|
| 699 |
|
|---|
| 700 | }
|
|---|
| 701 |
|
|---|
| 702 | horizontalCount.set(horizontalCount.get() + 1);
|
|---|
| 703 | verticalCount.set(0);
|
|---|
| 704 | }
|
|---|
| 705 |
|
|---|
| 706 | if(browserWidget.isParserRunning()) {
|
|---|
| 707 | progressBar.set(100);
|
|---|
| 708 | } else {
|
|---|
| 709 | MessageBay.displayMessage("Web page conversion cancelled");
|
|---|
| 710 | }
|
|---|
| 711 |
|
|---|
| 712 | browserWidget.parserFinished();
|
|---|
| 713 |
|
|---|
| 714 | Platform.runLater(new Runnable() {
|
|---|
| 715 | @Override
|
|---|
| 716 | public void run() {
|
|---|
| 717 | // Scrolling to the original position on the page
|
|---|
| 718 | webEngine.executeScript("window.scrollTo(originalScrollX, originalScrollY)");
|
|---|
| 719 | // Reloading the page once the parsing is done - only realistic way to reset (i.e. remove all the added WordSpan tags)
|
|---|
| 720 | // the page
|
|---|
| 721 | webEngine.reload();
|
|---|
| 722 | }
|
|---|
| 723 | });
|
|---|
| 724 |
|
|---|
| 725 | } catch (Exception ex) {
|
|---|
| 726 | ex.printStackTrace();
|
|---|
| 727 | }
|
|---|
| 728 |
|
|---|
| 729 | Platform.runLater(new Runnable() {
|
|---|
| 730 |
|
|---|
| 731 | @Override
|
|---|
| 732 | public void run() {
|
|---|
| 733 | browserWidget.setOverlayVisible(false);
|
|---|
| 734 | browserWidget.rebindWebViewSize();
|
|---|
| 735 | browserWidget.setScrollbarsVisible(true);
|
|---|
| 736 | }
|
|---|
| 737 | });
|
|---|
| 738 |
|
|---|
| 739 | }
|
|---|
| 740 |
|
|---|
| 741 | /**
|
|---|
| 742 | * @param rgbString
|
|---|
| 743 | * string in the format <i>rgb(x,x,x)</i> or <i>rgba(x,x,x,x)</i>
|
|---|
| 744 | * @return A Color object that should match the rgb string passed int. Returns null if alpha is 0
|
|---|
| 745 | */
|
|---|
| 746 | private static Colour rgbStringToColor(String rgbString) {
|
|---|
| 747 |
|
|---|
| 748 | if (rgbString == null) {
|
|---|
| 749 | return null;
|
|---|
| 750 | }
|
|---|
| 751 |
|
|---|
| 752 | // Splitting the string into 'rgb' and 'x, x, x'
|
|---|
| 753 | String[] tmpStrings = rgbString.split("\\(|\\)");
|
|---|
| 754 |
|
|---|
| 755 | // Splitting up the RGB(A) components into an array
|
|---|
| 756 | tmpStrings = tmpStrings[1].split(",");
|
|---|
| 757 |
|
|---|
| 758 | int[] components = new int[4];
|
|---|
| 759 | Arrays.fill(components, 255);
|
|---|
| 760 |
|
|---|
| 761 | for (int i = 0; i < tmpStrings.length; i++) {
|
|---|
| 762 | Float d = Float.parseFloat(tmpStrings[i].trim());
|
|---|
| 763 |
|
|---|
| 764 | components[i] = Math.round(d);
|
|---|
| 765 | }
|
|---|
| 766 |
|
|---|
| 767 | if (components[3] > 0) {
|
|---|
| 768 | return Colour.FromRGBA255(components[0], components[1], components[2], components[3]);
|
|---|
| 769 | } else {
|
|---|
| 770 | return null;
|
|---|
| 771 | }
|
|---|
| 772 | }
|
|---|
| 773 |
|
|---|
| 774 | /**
|
|---|
| 775 | * @param rootElement
|
|---|
| 776 | * Element that will be converted (including all sub-elements)
|
|---|
| 777 | * @param backgroundColor
|
|---|
| 778 | * 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)"
|
|---|
| 779 | * @param window
|
|---|
| 780 | * 'window' from Javascript
|
|---|
| 781 | * @param webEngine
|
|---|
| 782 | * Web engine that the page is loaded in
|
|---|
| 783 | * @param frame
|
|---|
| 784 | * Expeditee frame to add the converted page to
|
|---|
| 785 | * @throws IllegalArgumentException
|
|---|
| 786 | * @throws IllegalAccessException
|
|---|
| 787 | */
|
|---|
| 788 | private static void addPageToFrame(Node rootElement, JSObject window, WebEngine webEngine, Frame frame) throws InvocationTargetException, IllegalAccessException,
|
|---|
| 789 | IllegalArgumentException {
|
|---|
| 790 |
|
|---|
| 791 | Node currentNode = rootElement;
|
|---|
| 792 |
|
|---|
| 793 | if (currentNode.getNodeType() == Node.TEXT_NODE || currentNode.getNodeType() == Node.ELEMENT_NODE) {
|
|---|
| 794 |
|
|---|
| 795 | JSObject style;
|
|---|
| 796 | JSObject bounds;
|
|---|
| 797 |
|
|---|
| 798 | if (currentNode.getNodeType() == Node.TEXT_NODE) {
|
|---|
| 799 | // CSS style for the element
|
|---|
| 800 | style = (JSObject) window.call("getComputedStyle", new Object[] { currentNode.getParentNode() });
|
|---|
| 801 |
|
|---|
| 802 | // Getting a rectangle that represents the area and position of the element
|
|---|
| 803 | bounds = (JSObject) ((JSObject) currentNode.getParentNode()).call("getBoundingClientRect", new Object[] {});
|
|---|
| 804 | } else {
|
|---|
| 805 | style = (JSObject) window.call("getComputedStyle", new Object[] { currentNode });
|
|---|
| 806 |
|
|---|
| 807 | bounds = (JSObject) ((JSObject) currentNode).call("getBoundingClientRect", new Object[] {});
|
|---|
| 808 | }
|
|---|
| 809 |
|
|---|
| 810 | // Bounding rectangle position is relative to the current view, so scroll position must be added to x/y
|
|---|
| 811 | // TODO: This doesn't check if an element or any of its parent elements have position:fixed set - the only
|
|---|
| 812 | // way to check seems to be to walking through the element's parents until the document root is reached
|
|---|
| 813 | float x = Float.valueOf(bounds.getMember("left").toString()) + Float.valueOf(webEngine.executeScript("window.pageXOffset").toString());
|
|---|
| 814 | float y = Float.valueOf(bounds.getMember("top").toString()) + Float.valueOf(webEngine.executeScript("window.pageYOffset").toString());
|
|---|
| 815 |
|
|---|
| 816 | float width = Float.valueOf(bounds.getMember("width").toString());
|
|---|
| 817 | float height = Float.valueOf(bounds.getMember("height").toString());
|
|---|
| 818 |
|
|---|
| 819 | // Checking if the element is actually visible on the page
|
|---|
| 820 | if (WebParser.elementVisible(x, y, width, height, style)) {
|
|---|
| 821 |
|
|---|
| 822 | // Filtering the node type, starting with text nodes
|
|---|
| 823 | if (currentNode.getNodeType() == Node.TEXT_NODE) {
|
|---|
| 824 |
|
|---|
| 825 | String fontSize = ((String) style.call("getPropertyValue", new Object[] { "font-size" }));
|
|---|
| 826 |
|
|---|
| 827 | // Trimming off the units (always px) from the font size
|
|---|
| 828 | fontSize = fontSize.substring(0, fontSize.length() - 2);
|
|---|
| 829 |
|
|---|
| 830 | // Always returns in format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
|---|
| 831 | String color = (String) style.call("getPropertyValue", new Object[] { "color" });
|
|---|
| 832 |
|
|---|
| 833 | // Always returns in format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
|---|
| 834 | String bgColorString = (String) style.call("getPropertyValue", new Object[] { "background-color" });
|
|---|
| 835 |
|
|---|
| 836 | String align = (String) style.call("getPropertyValue", new Object[] { "text-align" });
|
|---|
| 837 |
|
|---|
| 838 | // Returns comma-separated list of typefaces
|
|---|
| 839 | String typeface = (String) style.call("getPropertyValue", new Object[] { "font-family" });
|
|---|
| 840 |
|
|---|
| 841 | String[] typefaces = typeface.split(", |,");
|
|---|
| 842 |
|
|---|
| 843 | String weight = (String) style.call("getPropertyValue", new Object[] { "font-weight" });
|
|---|
| 844 |
|
|---|
| 845 | String fontStyle = (String) style.call("getPropertyValue", new Object[] { "font-style" });
|
|---|
| 846 |
|
|---|
| 847 | // Returns "normal" or a value in pixels (e.g. "10px")
|
|---|
| 848 | String letterSpacing = (String) style.call("getPropertyValue", new Object[] { "letter-spacing" });
|
|---|
| 849 |
|
|---|
| 850 | // Returns a value in pixels (e.g. "10px")
|
|---|
| 851 | String lineHeight = (String) style.call("getPropertyValue", new Object[] { "line-height" });
|
|---|
| 852 |
|
|---|
| 853 | String textTransform = (String) style.call("getPropertyValue", new Object[] { "text-transform" });
|
|---|
| 854 |
|
|---|
| 855 | String linkUrl = (String) ((JSObject) currentNode.getParentNode()).getMember("href");
|
|---|
| 856 |
|
|---|
| 857 | Boolean fontFound = false;
|
|---|
| 858 | Font font = null;
|
|---|
| 859 |
|
|---|
| 860 | // Looping through all font-families listed in the element's CSS until one that is installed is
|
|---|
| 861 | // found, or the end of the list is reached, in which case the default font is used
|
|---|
| 862 | for (int j = 0; j < typefaces.length && !fontFound; j++) {
|
|---|
| 863 | if (typefaces[j].toLowerCase().equals("sans-serif")) {
|
|---|
| 864 | typefaces[j] = "Arial Unicode MS";
|
|---|
| 865 | } else if (typefaces[j].toLowerCase().equals("serif")) {
|
|---|
| 866 | typefaces[j] = "Times New Roman";
|
|---|
| 867 | } else if ((typefaces[j].toLowerCase().equals("arial"))) {
|
|---|
| 868 | // Have to use Arial Unicode, otherwise unicode characters display incorrectly
|
|---|
| 869 | typefaces[j] = "Arial Unicode MS";
|
|---|
| 870 | }
|
|---|
| 871 |
|
|---|
| 872 | // Regex will remove any inverted commas surrounding multi-word typeface names
|
|---|
| 873 | String familyName = typefaces[j].replaceAll("^'|'$", "");
|
|---|
| 874 | font = new Font(familyName);
|
|---|
| 875 | font.setStyle(Font.Style.PLAIN);
|
|---|
| 876 | font.setSize(12);
|
|---|
| 877 |
|
|---|
| 878 | // Check whether the font was found
|
|---|
| 879 | if (EcosystemManager.getFontManager().getActualFont(font).getFamilyName().toLowerCase().equals(familyName.toLowerCase())) {
|
|---|
| 880 | fontFound = true;
|
|---|
| 881 | }
|
|---|
| 882 | }
|
|---|
| 883 |
|
|---|
| 884 | if (!fontFound) {
|
|---|
| 885 | font = new Font("Times New Roman");
|
|---|
| 886 | font.setStyle(Font.Style.PLAIN);
|
|---|
| 887 | font.setSize(12);
|
|---|
| 888 | }
|
|---|
| 889 |
|
|---|
| 890 | String fontStyleComplete = "";
|
|---|
| 891 |
|
|---|
| 892 | int weightInt = 0;
|
|---|
| 893 |
|
|---|
| 894 | try {
|
|---|
| 895 | weightInt = Integer.parseInt(weight);
|
|---|
| 896 | } catch (NumberFormatException nfe) {
|
|---|
| 897 | // Use default value as set above
|
|---|
| 898 | }
|
|---|
| 899 |
|
|---|
| 900 | // checking if font is bold - i.e. 'bold', 'bolder' or weight over 500
|
|---|
| 901 | if (weight.toLowerCase().startsWith("bold") || weightInt > 500) {
|
|---|
| 902 | fontStyleComplete = fontStyleComplete.concat("bold");
|
|---|
| 903 | }
|
|---|
| 904 |
|
|---|
| 905 | if (fontStyle.toLowerCase().equals("italic") || fontStyle.toLowerCase().equals("oblique")) {
|
|---|
| 906 | fontStyleComplete = fontStyleComplete.concat("italic");
|
|---|
| 907 | }
|
|---|
| 908 |
|
|---|
| 909 | float fontSizeFloat = 12;
|
|---|
| 910 |
|
|---|
| 911 | try {
|
|---|
| 912 | fontSizeFloat = Float.valueOf(fontSize);
|
|---|
| 913 | } catch (NumberFormatException nfe) {
|
|---|
| 914 | // Use default value as set above
|
|---|
| 915 | }
|
|---|
| 916 |
|
|---|
| 917 | float letterSpacingFloat = -0.008f;
|
|---|
| 918 |
|
|---|
| 919 | try {
|
|---|
| 920 | letterSpacingFloat = (Integer.parseInt(letterSpacing.substring(0, letterSpacing.length() - 2)) / (fontSizeFloat));
|
|---|
| 921 | } catch (NumberFormatException nfe) {
|
|---|
| 922 | // Use default value as set above
|
|---|
| 923 | }
|
|---|
| 924 |
|
|---|
| 925 | float lineHeightInt = -1;
|
|---|
| 926 |
|
|---|
| 927 | try {
|
|---|
| 928 | lineHeightInt = (Float.parseFloat(lineHeight.substring(0, lineHeight.length() - 2)));
|
|---|
| 929 | } catch (NumberFormatException nfe) {
|
|---|
| 930 | // Use default value as set above
|
|---|
| 931 | }
|
|---|
| 932 |
|
|---|
| 933 | Text t;
|
|---|
| 934 |
|
|---|
| 935 | String textContent = currentNode.getTextContent().replaceAll("[^\\S\\n]+", " ");
|
|---|
| 936 | textContent = textContent.replaceAll("^(\\s)(\\n|\\r)", "");
|
|---|
| 937 |
|
|---|
| 938 | if (textTransform.equals("uppercase")) {
|
|---|
| 939 | textContent = textContent.toUpperCase();
|
|---|
| 940 | } else if (textTransform.equals("lowercase")) {
|
|---|
| 941 | textContent = textContent.toUpperCase();
|
|---|
| 942 | }
|
|---|
| 943 |
|
|---|
| 944 | // Adding the text to the frame. Expeditee text seems to be positioned relative to the baseline of the first line, so
|
|---|
| 945 | // the font size has to be added to the y-position
|
|---|
| 946 | t = frame.addText(Math.round(x), Math.round(y + fontSizeFloat), textContent, null);
|
|---|
| 947 |
|
|---|
| 948 | t.setColor(rgbStringToColor(color));
|
|---|
| 949 | t.setBackgroundColor(rgbStringToColor(bgColorString));
|
|---|
| 950 | t.setFont(font.clone());
|
|---|
| 951 | t.setSize(fontSizeFloat);
|
|---|
| 952 | t.setFontStyle(fontStyleComplete);
|
|---|
| 953 | t.setLetterSpacing(letterSpacingFloat);
|
|---|
| 954 |
|
|---|
| 955 | // Removing any spacing between lines allowing t.getLineHeight() to be used to get the actual height
|
|---|
| 956 | // of just the characters (i.e. distance from ascenders to descenders)
|
|---|
| 957 | t.setSpacing(0);
|
|---|
| 958 |
|
|---|
| 959 | t.setSpacing(lineHeightInt - t.getLineHeight());
|
|---|
| 960 |
|
|---|
| 961 | if (align.equals("left")) {
|
|---|
| 962 | t.setJustification(Justification.left);
|
|---|
| 963 | } else if (align.equals("right")) {
|
|---|
| 964 | t.setJustification(Justification.right);
|
|---|
| 965 | } else if (align.equals("center")) {
|
|---|
| 966 | t.setJustification(Justification.center);
|
|---|
| 967 | } else if (align.equals("justify")) {
|
|---|
| 968 | t.setJustification(Justification.full);
|
|---|
| 969 | }
|
|---|
| 970 |
|
|---|
| 971 | // Font size is added to the item width to give a little breathing room
|
|---|
| 972 | t.setWidth(Math.round(width + (t.getSize())));
|
|---|
| 973 |
|
|---|
| 974 | if (!linkUrl.equals("undefined")) {
|
|---|
| 975 | t.setAction("gotourl " + linkUrl);
|
|---|
| 976 | t.setActionMark(false);
|
|---|
| 977 | }
|
|---|
| 978 |
|
|---|
| 979 | } else if (currentNode.getNodeType() == Node.ELEMENT_NODE) {
|
|---|
| 980 |
|
|---|
| 981 | // Always returns in format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
|---|
| 982 | String bgColorString = (String) style.call("getPropertyValue", new Object[] { "background-color" });
|
|---|
| 983 |
|
|---|
| 984 | Colour bgColor = rgbStringToColor(bgColorString);
|
|---|
| 985 |
|
|---|
| 986 | // If the element has a background color then add it (to Expeditee) as a rectangle with that background color
|
|---|
| 987 | if (bgColor != null) {
|
|---|
| 988 | frame.addRectangle(Math.round(x), Math.round(y), Math.round(width), Math.round(height), 0, null, bgColor);
|
|---|
| 989 | }
|
|---|
| 990 |
|
|---|
| 991 | String linkUrl = (String) ((JSObject) currentNode).getMember("href");
|
|---|
| 992 |
|
|---|
| 993 | // background image, returns in format "url(protocol://absolute/path/to/img.extension)" for images,
|
|---|
| 994 | // may also return gradients, data, etc. (not handled yet). Only need to add bg image on
|
|---|
| 995 | // 'ELEMENT_NODE' (and not 'TEXT_NODE' otherwise there would be double-ups
|
|---|
| 996 | if (((String) style.call("getPropertyValue", new Object[] { "background-image" })).startsWith("url(")) {
|
|---|
| 997 |
|
|---|
| 998 | try {
|
|---|
| 999 | WebParser.addBackgroundImageFromNode(currentNode, style, frame, linkUrl, x, y, width, height);
|
|---|
| 1000 | } catch (MalformedURLException mue) {
|
|---|
| 1001 | // probably a 'data:' url, not supported yet
|
|---|
| 1002 | mue.printStackTrace();
|
|---|
| 1003 | } catch (IOException e) {
|
|---|
| 1004 | // TODO Auto-generated catch block
|
|---|
| 1005 | e.printStackTrace();
|
|---|
| 1006 | }
|
|---|
| 1007 | }
|
|---|
| 1008 |
|
|---|
| 1009 | String imgSrc;
|
|---|
| 1010 |
|
|---|
| 1011 | if (currentNode.getNodeName().toLowerCase().equals("img") && (imgSrc = ((JSObject) currentNode).getMember("src").toString()) != null) {
|
|---|
| 1012 | try {
|
|---|
| 1013 | WebParser.addImageFromUrl(imgSrc, linkUrl, frame, x, y, (int) width, null, null, null, null, null, 0, 0);
|
|---|
| 1014 | } catch (MalformedURLException mue) {
|
|---|
| 1015 | // probably a 'data:' url, not supported yet
|
|---|
| 1016 | mue.printStackTrace();
|
|---|
| 1017 | } catch (IOException e) {
|
|---|
| 1018 | // TODO Auto-generated catch block
|
|---|
| 1019 | e.printStackTrace();
|
|---|
| 1020 | }
|
|---|
| 1021 | }
|
|---|
| 1022 | }
|
|---|
| 1023 | }
|
|---|
| 1024 |
|
|---|
| 1025 | Node childNode = currentNode.getFirstChild();
|
|---|
| 1026 |
|
|---|
| 1027 | while (childNode != null) {
|
|---|
| 1028 | addPageToFrame(childNode, window, webEngine, frame);
|
|---|
| 1029 | childNode = childNode.getNextSibling();
|
|---|
| 1030 | }
|
|---|
| 1031 | }
|
|---|
| 1032 | }
|
|---|
| 1033 |
|
|---|
| 1034 | private static boolean elementVisible(float x, float y, float width, float height, JSObject style) {
|
|---|
| 1035 | if (width <= 0 || height <= 0 || x + width <= 0 || y + height <= 0 || ((String) style.call("getPropertyValue", new Object[] { "visibility" })).equals("hidden")
|
|---|
| 1036 | || ((String) style.call("getPropertyValue", new Object[] { "display" })).equals("none")) {
|
|---|
| 1037 | return false;
|
|---|
| 1038 | } else {
|
|---|
| 1039 | return true;
|
|---|
| 1040 | }
|
|---|
| 1041 | }
|
|---|
| 1042 |
|
|---|
| 1043 | /**
|
|---|
| 1044 | * @param imgSrc
|
|---|
| 1045 | * URL of the image to add
|
|---|
| 1046 | * @param linkUrl
|
|---|
| 1047 | * Absolute URL that the image should link to when clicked
|
|---|
| 1048 | * @param frame
|
|---|
| 1049 | * Frame to add the image to
|
|---|
| 1050 | * @param x
|
|---|
| 1051 | * X-coordinate at which the image should be placed on the frame
|
|---|
| 1052 | * @param y
|
|---|
| 1053 | * Y-coordinate at which the image should be placed on the frame
|
|---|
| 1054 | * @param width
|
|---|
| 1055 | * Width of the image once added to the frame. Negative 1 (-1) will cause the actual width of the image file to be used
|
|---|
| 1056 | *
|
|---|
| 1057 | * @param cropStartX
|
|---|
| 1058 | * X-coordinate at which to start crop, or null for no crop
|
|---|
| 1059 | * @param cropStartY
|
|---|
| 1060 | * Y-coordinate at which to start crop, or null for no crop
|
|---|
| 1061 | * @param cropEndX
|
|---|
| 1062 | * X-coordinate at which to end the crop, or null for no crop
|
|---|
| 1063 | * @param cropEndY
|
|---|
| 1064 | * Y-coordinate at which to end the crop, or null for no crop
|
|---|
| 1065 | *
|
|---|
| 1066 | * @param repeat
|
|---|
| 1067 | * String determining how the image should be tiled/repeated. Valid strings are: <i>no-repeat</i>, <i>repeat-x</i>, or
|
|---|
| 1068 | * <i>repeat-y</i>. All other values (including null) will cause the image to repeat in both directions
|
|---|
| 1069 | *
|
|---|
| 1070 | * @param originXPercent
|
|---|
| 1071 | * Percentage into the image to use as the x coordinate of the image's origin point
|
|---|
| 1072 | * @param originYPercent
|
|---|
| 1073 | * Percentage into the image to use as the y coordinate of the image's origin point
|
|---|
| 1074 | *
|
|---|
| 1075 | * @throws MalformedURLException
|
|---|
| 1076 | * @throws IOException
|
|---|
| 1077 | */
|
|---|
| 1078 | public static Picture getImageFromUrl(String imgSrc, String linkUrl, final Frame frame, float x, float y, int width,
|
|---|
| 1079 | Integer cropStartX, Integer cropStartY, Integer cropEndX, Integer cropEndY, String repeat, float originXPercent, float originYPercent)
|
|---|
| 1080 | throws IOException {
|
|---|
| 1081 |
|
|---|
| 1082 | URL imgUrl = new URL(imgSrc);
|
|---|
| 1083 |
|
|---|
| 1084 | HttpURLConnection connection = (HttpURLConnection) (imgUrl.openConnection());
|
|---|
| 1085 |
|
|---|
| 1086 | Image img = Image.getImage(connection);
|
|---|
| 1087 |
|
|---|
| 1088 | int hashcode = img.hashCode();
|
|---|
| 1089 |
|
|---|
| 1090 | File out = new File(FrameIO.IMAGES_PATH + Integer.toHexString(hashcode) + ".png");
|
|---|
| 1091 | out.mkdirs();
|
|---|
| 1092 | img.writeToDisk("png", out);
|
|---|
| 1093 |
|
|---|
| 1094 | if (repeat == null && cropEndX == null && cropStartX == null && cropEndY == null && cropStartY == null) {
|
|---|
| 1095 | repeat = "no-repeat";
|
|---|
| 1096 | }
|
|---|
| 1097 |
|
|---|
| 1098 | if (cropEndX == null || cropStartX == null || cropEndY == null || cropStartY == null) {
|
|---|
| 1099 | cropStartX = 0;
|
|---|
| 1100 | cropStartY = 0;
|
|---|
| 1101 | cropEndX = img.getWidth();
|
|---|
| 1102 | cropEndY = img.getHeight();
|
|---|
| 1103 | } else if (cropStartX < 0) {
|
|---|
| 1104 | cropEndX = cropEndX - cropStartX;
|
|---|
| 1105 | x = x + Math.abs(cropStartX);
|
|---|
| 1106 | cropStartX = 0;
|
|---|
| 1107 | }
|
|---|
| 1108 |
|
|---|
| 1109 | if (cropStartY < 0) {
|
|---|
| 1110 | cropEndY = cropEndY - cropStartY;
|
|---|
| 1111 | y = y + Math.abs(cropStartY);
|
|---|
| 1112 | cropStartY = 0;
|
|---|
| 1113 | }
|
|---|
| 1114 |
|
|---|
| 1115 | if (width < 0) {
|
|---|
| 1116 | width = img.getWidth();
|
|---|
| 1117 | }
|
|---|
| 1118 |
|
|---|
| 1119 | if (repeat != null) {
|
|---|
| 1120 | if (repeat.equals("no-repeat")) {
|
|---|
| 1121 | int tmpCropEndY = (int) (cropStartY + ((float) width / img.getWidth()) * img.getHeight());
|
|---|
| 1122 | int tmpCropEndX = cropStartX + width;
|
|---|
| 1123 |
|
|---|
| 1124 | cropEndX = (cropEndX < tmpCropEndX) ? cropEndX : tmpCropEndX;
|
|---|
| 1125 | cropEndY = (cropEndY < tmpCropEndY) ? cropEndY : tmpCropEndY;
|
|---|
| 1126 | } else if (repeat.equals("repeat-x")) {
|
|---|
| 1127 | int tmpCropEndY = (int) (cropStartY + ((float) width / img.getWidth()) * img.getHeight());
|
|---|
| 1128 | cropEndY = (cropEndY < tmpCropEndY) ? cropEndY : tmpCropEndY;
|
|---|
| 1129 | } else if (repeat.equals("repeat-y")) {
|
|---|
| 1130 | int tmpCropEndX = cropStartX + width;
|
|---|
| 1131 | cropEndX = (cropEndX < tmpCropEndX) ? cropEndX : tmpCropEndX;
|
|---|
| 1132 | }
|
|---|
| 1133 | }
|
|---|
| 1134 |
|
|---|
| 1135 | if (originXPercent > 0) {
|
|---|
| 1136 | int actualWidth = cropEndX - cropStartX;
|
|---|
| 1137 |
|
|---|
| 1138 | int originXPixels = Math.round(originXPercent * actualWidth);
|
|---|
| 1139 |
|
|---|
| 1140 | x = x - originXPixels;
|
|---|
| 1141 |
|
|---|
| 1142 | cropStartX = (int) (cropStartX + (width - actualWidth) * originXPercent);
|
|---|
| 1143 | cropEndX = (int) (cropEndX + (width - actualWidth) * originXPercent);
|
|---|
| 1144 | }
|
|---|
| 1145 |
|
|---|
| 1146 | if (originYPercent > 0) {
|
|---|
| 1147 | int height = (int) ((img.getHeight() / (float) img.getWidth()) * width);
|
|---|
| 1148 | int actualHeight = (cropEndY - cropStartY);
|
|---|
| 1149 | int originYPixels = Math.round(originYPercent * actualHeight);
|
|---|
| 1150 |
|
|---|
| 1151 | y = y - originYPixels;
|
|---|
| 1152 |
|
|---|
| 1153 | cropStartY = (int) (cropStartY + (height - actualHeight) * originYPercent);
|
|---|
| 1154 | cropEndY = (int) (cropEndY + (height - actualHeight) * originYPercent);
|
|---|
| 1155 | }
|
|---|
| 1156 |
|
|---|
| 1157 | Text text = new Text("@i: " + out.getName() + " " + width);
|
|---|
| 1158 | text.setPosition(x, y);
|
|---|
| 1159 |
|
|---|
| 1160 | Picture pic = ItemUtils.CreatePicture(text);
|
|---|
| 1161 |
|
|---|
| 1162 | float invScale = 1 / pic.getScale();
|
|---|
| 1163 |
|
|---|
| 1164 | pic.setCrop((int)(cropStartX * invScale), (int)(cropStartY * invScale), (int)(cropEndX * invScale), (int)(cropEndY * invScale));
|
|---|
| 1165 |
|
|---|
| 1166 | if (linkUrl != null && !linkUrl.equals("undefined")) {
|
|---|
| 1167 | pic.setAction("goto " + linkUrl);
|
|---|
| 1168 | pic.setActionMark(false);
|
|---|
| 1169 | }
|
|---|
| 1170 |
|
|---|
| 1171 | return pic;
|
|---|
| 1172 | }
|
|---|
| 1173 |
|
|---|
| 1174 | 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,
|
|---|
| 1175 | float originXPercent, float originYPercent)
|
|---|
| 1176 | throws IOException {
|
|---|
| 1177 | Picture pic = getImageFromUrl(imgSrc, linkUrl, frame, x, y, width, cropStartX, cropStartY, cropEndX, cropEndY, repeat, originXPercent, originYPercent);
|
|---|
| 1178 | frame.addItem(pic);
|
|---|
| 1179 | pic.anchor();
|
|---|
| 1180 | pic.getSource().anchor();
|
|---|
| 1181 | }
|
|---|
| 1182 |
|
|---|
| 1183 | public static Picture getBackgroundImageFromNode(Node node, JSObject style, final Frame frame, String linkUrl, float x, float y, float width, float height) throws IOException {
|
|---|
| 1184 |
|
|---|
| 1185 |
|
|---|
| 1186 | String bgImage = (String) style.call("getPropertyValue", new Object[] { "background-image" });
|
|---|
| 1187 | bgImage = bgImage.substring(4, bgImage.length() - 1);
|
|---|
| 1188 |
|
|---|
| 1189 | String bgSize = ((String) style.call("getPropertyValue", new Object[] { "background-size" })).toLowerCase();
|
|---|
| 1190 | String bgRepeat = ((String) style.call("getPropertyValue", new Object[] { "background-repeat" })).toLowerCase();
|
|---|
| 1191 |
|
|---|
| 1192 | // Returns "[x]px [y]px", "[x]% [y]%", "[x]px [y]%" or "[x]% [y]px"
|
|---|
| 1193 | String bgPosition = ((String) style.call("getPropertyValue", new Object[] { "background-position" })).toLowerCase();
|
|---|
| 1194 |
|
|---|
| 1195 | String[] bgOffsetCoords = bgPosition.split(" ");
|
|---|
| 1196 |
|
|---|
| 1197 | int bgOffsetX = 0, bgOffsetY = 0;
|
|---|
| 1198 |
|
|---|
| 1199 | float originXPercent = 0, originYPercent = 0;
|
|---|
| 1200 |
|
|---|
| 1201 | int cropStartX, cropStartY, cropEndX, cropEndY;
|
|---|
| 1202 |
|
|---|
| 1203 | // Converting the x and y offset values to integers (and from % to px if needed)
|
|---|
| 1204 | if (bgOffsetCoords[0].endsWith("%")) {
|
|---|
| 1205 | bgOffsetX = (int) ((Integer.valueOf(bgOffsetCoords[0].substring(0, bgOffsetCoords[0].length() - 1)) / 100.0) * width);
|
|---|
| 1206 | originXPercent = (Integer.valueOf(bgOffsetCoords[0].substring(0, bgOffsetCoords[0].length() - 1))) / 100f;
|
|---|
| 1207 | } else if (bgOffsetCoords[0].endsWith("px")) {
|
|---|
| 1208 | bgOffsetX = (int) (Integer.valueOf(bgOffsetCoords[0].substring(0, bgOffsetCoords[0].length() - 2)));
|
|---|
| 1209 | }
|
|---|
| 1210 |
|
|---|
| 1211 | if (bgOffsetCoords[1].endsWith("%")) {
|
|---|
| 1212 | bgOffsetY = (int) ((Integer.valueOf(bgOffsetCoords[1].substring(0, bgOffsetCoords[1].length() - 1)) / 100.0) * height);
|
|---|
| 1213 | originYPercent = (Integer.valueOf(bgOffsetCoords[1].substring(0, bgOffsetCoords[1].length() - 1))) / 100f;
|
|---|
| 1214 | } else if (bgOffsetCoords[1].endsWith("px")) {
|
|---|
| 1215 | bgOffsetY = (int) (Integer.valueOf(bgOffsetCoords[1].substring(0, bgOffsetCoords[1].length() - 2)));
|
|---|
| 1216 | }
|
|---|
| 1217 |
|
|---|
| 1218 | // Converting from an offset to crop coords
|
|---|
| 1219 | cropStartX = -1 * bgOffsetX;
|
|---|
| 1220 | cropEndX = (int) (cropStartX + width);
|
|---|
| 1221 |
|
|---|
| 1222 | cropStartY = -1 * bgOffsetY;
|
|---|
| 1223 | cropEndY = (int) (cropStartY + height);
|
|---|
| 1224 |
|
|---|
| 1225 | int bgWidth = -1;
|
|---|
| 1226 |
|
|---|
| 1227 | if (bgSize.equals("cover")) {
|
|---|
| 1228 | bgWidth = (int) width;
|
|---|
| 1229 | } else if (bgSize.equals("contain")) {
|
|---|
| 1230 | // TODO: actually compute the appropriate width
|
|---|
| 1231 | bgWidth = (int) width;
|
|---|
| 1232 | } else if (bgSize.equals("auto")) {
|
|---|
| 1233 | bgWidth = -1;
|
|---|
| 1234 | } else {
|
|---|
| 1235 | bgSize = bgSize.split(" ")[0];
|
|---|
| 1236 |
|
|---|
| 1237 | if (bgSize.endsWith("%")) {
|
|---|
| 1238 | bgWidth = (int) ((Integer.parseInt(bgSize.replaceAll("\\D", "")) / 100.0) * width);
|
|---|
| 1239 | } else if (bgSize.endsWith("px")) {
|
|---|
| 1240 | bgWidth = Integer.parseInt(bgSize.replaceAll("\\D", ""));
|
|---|
| 1241 | }
|
|---|
| 1242 | }
|
|---|
| 1243 |
|
|---|
| 1244 | return getImageFromUrl(bgImage, linkUrl, frame, x, y, bgWidth, cropStartX, cropStartY, cropEndX, cropEndY, bgRepeat, originXPercent, originYPercent);
|
|---|
| 1245 | }
|
|---|
| 1246 |
|
|---|
| 1247 | private static void addBackgroundImageFromNode(Node node, JSObject style, final Frame frame, String linkUrl, float x, float y, float width, float height) throws IOException {
|
|---|
| 1248 | Picture pic = getBackgroundImageFromNode(node, style, frame, linkUrl, x, y, width, height);
|
|---|
| 1249 | frame.addItem(pic);
|
|---|
| 1250 | pic.anchor();
|
|---|
| 1251 | pic.getSource().anchor();
|
|---|
| 1252 | }
|
|---|
| 1253 |
|
|---|
| 1254 | /**
|
|---|
| 1255 | * @param rootElement
|
|---|
| 1256 | * Element that will be converted (including all sub-elements)
|
|---|
| 1257 | * @param backgroundColor
|
|---|
| 1258 | * 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)"
|
|---|
| 1259 | * @param window
|
|---|
| 1260 | * 'window' from Javascript
|
|---|
| 1261 | * @param webEngine
|
|---|
| 1262 | * Web engine that the page is loaded in
|
|---|
| 1263 | * @param frame
|
|---|
| 1264 | * Expeditee frame to add the converted page to
|
|---|
| 1265 | * @throws IllegalArgumentException
|
|---|
| 1266 | * @throws IllegalAccessException
|
|---|
| 1267 | */
|
|---|
| 1268 | private static void addTextToFrame(int visibleWidth, int visibleHeight, JSObject window, WebEngine webEngine, Frame frame) throws InvocationTargetException,
|
|---|
| 1269 | IllegalAccessException, IllegalArgumentException {
|
|---|
| 1270 |
|
|---|
| 1271 | webEngine.executeScript("var walker = document.createTreeWalker(document.body, NodeFilter.SHOW_TEXT, null, false);");
|
|---|
| 1272 |
|
|---|
| 1273 | Node currentNode;
|
|---|
| 1274 |
|
|---|
| 1275 | while ((currentNode = (Node) webEngine.executeScript("walker.nextNode()")) != null) {
|
|---|
| 1276 | JSObject style;
|
|---|
| 1277 | JSObject bounds;
|
|---|
| 1278 |
|
|---|
| 1279 | // CSS style for the element
|
|---|
| 1280 | style = (JSObject) window.call("getComputedStyle", new Object[] { currentNode.getParentNode() });
|
|---|
| 1281 |
|
|---|
| 1282 | // Getting a rectangle that represents the area and position of the element
|
|---|
| 1283 | bounds = (JSObject) ((JSObject) currentNode.getParentNode()).call("getBoundingClientRect", new Object[] {});
|
|---|
| 1284 |
|
|---|
| 1285 | // TODO: This doesn't check if an element or any of its parent elements have position:fixed set - the only way to check seems to
|
|---|
| 1286 | // be to walking through the element's parents until the document root is reached (or a recursive function)
|
|---|
| 1287 | float x = Float.valueOf(bounds.getMember("left").toString());
|
|---|
| 1288 | float y = Float.valueOf(bounds.getMember("top").toString());
|
|---|
| 1289 |
|
|---|
| 1290 | float width = Float.valueOf(bounds.getMember("width").toString());
|
|---|
| 1291 | float height = Float.valueOf(bounds.getMember("height").toString());
|
|---|
| 1292 |
|
|---|
| 1293 | // Checking if the element is actually visible on the page
|
|---|
| 1294 | if (width > 0 && height > 0 && x + width > 0 && y + height > 0 && x <= visibleWidth && y <= visibleHeight
|
|---|
| 1295 | && !(((String) style.call("getPropertyValue", new Object[] { "display" })).equals("none"))
|
|---|
| 1296 | && !(((String) style.call("getPropertyValue", new Object[] { "visibility" })).equals("hidden"))) {
|
|---|
| 1297 |
|
|---|
| 1298 | String fontSize = ((String) style.call("getPropertyValue", new Object[] { "font-size" }));
|
|---|
| 1299 |
|
|---|
| 1300 | // Trimming off the units (always px) from the font size
|
|---|
| 1301 | fontSize = fontSize.substring(0, fontSize.length() - 2);
|
|---|
| 1302 |
|
|---|
| 1303 | // Always returns in format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
|---|
| 1304 | String color = (String) style.call("getPropertyValue", new Object[] { "color" });
|
|---|
| 1305 |
|
|---|
| 1306 | // Always returns in format "rgb(x,x,x)" or "rgba(x,x,x,x)"
|
|---|
| 1307 | String bgColorString = (String) style.call("getPropertyValue", new Object[] { "background-color" });
|
|---|
| 1308 |
|
|---|
| 1309 | String align = (String) style.call("getPropertyValue", new Object[] { "text-align" });
|
|---|
| 1310 |
|
|---|
| 1311 | // Returns comma-separated list of typefaces
|
|---|
| 1312 | String typeface = (String) style.call("getPropertyValue", new Object[] { "font-family" });
|
|---|
| 1313 |
|
|---|
| 1314 | String[] typefaces = typeface.split(", |,");
|
|---|
| 1315 |
|
|---|
| 1316 | String weight = (String) style.call("getPropertyValue", new Object[] { "font-weight" });
|
|---|
| 1317 |
|
|---|
| 1318 | String fontStyle = (String) style.call("getPropertyValue", new Object[] { "font-style" });
|
|---|
| 1319 |
|
|---|
| 1320 | // Returns "normal" or a value in pixels (e.g. "10px")
|
|---|
| 1321 | String letterSpacing = (String) style.call("getPropertyValue", new Object[] { "letter-spacing" });
|
|---|
| 1322 |
|
|---|
| 1323 | // Returns a value in pixels (e.g. "10px")
|
|---|
| 1324 | String lineHeight = (String) style.call("getPropertyValue", new Object[] { "line-height" });
|
|---|
| 1325 |
|
|---|
| 1326 | String textTransform = (String) style.call("getPropertyValue", new Object[] { "text-transform" });
|
|---|
| 1327 |
|
|---|
| 1328 | String linkUrl = (String) ((JSObject) currentNode.getParentNode()).getMember("href");
|
|---|
| 1329 |
|
|---|
| 1330 | Boolean fontFound = false;
|
|---|
| 1331 | Font font = null;
|
|---|
| 1332 |
|
|---|
| 1333 | // Looping through all font-families listed in the element's CSS until one that is installed is
|
|---|
| 1334 | // found, or the end of the list is reached, in which case the default font is used
|
|---|
| 1335 | for (int j = 0; j < typefaces.length && !fontFound; j++) {
|
|---|
| 1336 | if (typefaces[j].toLowerCase().equals("sans-serif")) {
|
|---|
| 1337 | typefaces[j] = "SansSerif";
|
|---|
| 1338 | } else if ((typefaces[j].toLowerCase().equals("arial"))) {
|
|---|
| 1339 | // Have to use Arial Unicode, otherwise unicode characters display incorrectly
|
|---|
| 1340 | // It seems that not all systems have this font (including some Windows machines),
|
|---|
| 1341 | // but as long as the website has a general font type specified (e.g. "font-family: Arial, Sans-Serif"),
|
|---|
| 1342 | // there should be no noticeable difference.
|
|---|
| 1343 | typefaces[j] = "Arial Unicode MS";
|
|---|
| 1344 | } else if ((typefaces[j].toLowerCase().equals("monospace"))) {
|
|---|
| 1345 | typefaces[j] = "monospaced";
|
|---|
| 1346 | }
|
|---|
| 1347 |
|
|---|
| 1348 | // Regex will remove any inverted commas surrounding multi-word typeface names
|
|---|
| 1349 | String familyName = typefaces[j].replaceAll("^'|'$", "");
|
|---|
| 1350 | font = new Font(familyName);
|
|---|
| 1351 | font.setStyle(Font.Style.PLAIN);
|
|---|
| 1352 | font.setSize(12);
|
|---|
| 1353 |
|
|---|
| 1354 | // If the font isn't found, Java just uses Font.DIALOG, so this check checks whether the font was found
|
|---|
| 1355 | if (EcosystemManager.getFontManager().getActualFont(font).getFamilyName().toLowerCase().equals(familyName.toLowerCase())) {
|
|---|
| 1356 | fontFound = true;
|
|---|
| 1357 | }
|
|---|
| 1358 | }
|
|---|
| 1359 |
|
|---|
| 1360 | if (!fontFound) {
|
|---|
| 1361 | font = new Font("Times New Roman");
|
|---|
| 1362 | font.setStyle(Font.Style.PLAIN);
|
|---|
| 1363 | font.setSize(12);
|
|---|
| 1364 | }
|
|---|
| 1365 |
|
|---|
| 1366 | String fontStyleComplete = "";
|
|---|
| 1367 |
|
|---|
| 1368 | int weightInt = 0;
|
|---|
| 1369 |
|
|---|
| 1370 | try {
|
|---|
| 1371 | weightInt = Integer.parseInt(weight);
|
|---|
| 1372 | } catch (NumberFormatException nfe) {
|
|---|
| 1373 | // Use default value as set above
|
|---|
| 1374 | }
|
|---|
| 1375 |
|
|---|
| 1376 | // checking if font is bold - i.e. 'bold', 'bolder' or weight over 500
|
|---|
| 1377 | if (weight.toLowerCase().startsWith("bold") || weightInt > 500) {
|
|---|
| 1378 | fontStyleComplete = fontStyleComplete.concat("bold");
|
|---|
| 1379 | }
|
|---|
| 1380 |
|
|---|
| 1381 | if (fontStyle.toLowerCase().equals("italic") || fontStyle.toLowerCase().equals("oblique")) {
|
|---|
| 1382 | fontStyleComplete = fontStyleComplete.concat("italic");
|
|---|
| 1383 | }
|
|---|
| 1384 |
|
|---|
| 1385 | float fontSizeFloat = 12;
|
|---|
| 1386 |
|
|---|
| 1387 | try {
|
|---|
| 1388 | fontSizeFloat = Float.valueOf(fontSize);
|
|---|
| 1389 | } catch (NumberFormatException nfe) {
|
|---|
| 1390 | // Use default value as set above
|
|---|
| 1391 | }
|
|---|
| 1392 |
|
|---|
| 1393 | float letterSpacingFloat = -0.008f;
|
|---|
| 1394 |
|
|---|
| 1395 | try {
|
|---|
| 1396 | letterSpacingFloat = (Integer.parseInt(letterSpacing.substring(0, letterSpacing.length() - 2)) / (fontSizeFloat));
|
|---|
| 1397 | } catch (NumberFormatException nfe) {
|
|---|
| 1398 | // Use default value as set above
|
|---|
| 1399 | }
|
|---|
| 1400 |
|
|---|
| 1401 | float lineHeightInt = -1;
|
|---|
| 1402 |
|
|---|
| 1403 | try {
|
|---|
| 1404 | lineHeightInt = (Float.parseFloat(lineHeight.substring(0, lineHeight.length() - 2)));
|
|---|
| 1405 | } catch (NumberFormatException nfe) {
|
|---|
| 1406 | // Use default value as set above
|
|---|
| 1407 | }
|
|---|
| 1408 |
|
|---|
| 1409 | Text t;
|
|---|
| 1410 |
|
|---|
| 1411 | String textContent = currentNode.getTextContent().replaceAll("[^\\S\\n]+", " ");
|
|---|
| 1412 | textContent = textContent.replaceAll("^(\\s)(\\n|\\r)", "");
|
|---|
| 1413 |
|
|---|
| 1414 | if (textTransform.equals("uppercase")) {
|
|---|
| 1415 | textContent = textContent.toUpperCase();
|
|---|
| 1416 | } else if (textTransform.equals("lowercase")) {
|
|---|
| 1417 | textContent = textContent.toUpperCase();
|
|---|
| 1418 | }
|
|---|
| 1419 |
|
|---|
| 1420 | // Adding the text to the frame. Expeditee text seems to be positioned relative to the baseline of the first line, so
|
|---|
| 1421 | // the font size has to be added to the y-position
|
|---|
| 1422 | t = frame.addText(Math.round(x), Math.round(y + fontSizeFloat), textContent, null);
|
|---|
| 1423 |
|
|---|
| 1424 | t.setColor(rgbStringToColor(color));
|
|---|
| 1425 | t.setBackgroundColor(rgbStringToColor(bgColorString));
|
|---|
| 1426 | t.setFont(font.clone());
|
|---|
| 1427 | t.setSize(fontSizeFloat);
|
|---|
| 1428 | t.setFontStyle(fontStyleComplete);
|
|---|
| 1429 | t.setLetterSpacing(letterSpacingFloat);
|
|---|
| 1430 |
|
|---|
| 1431 | // Removing any spacing between lines allowing t.getLineHeight() to be used to get the actual height
|
|---|
| 1432 | // of just the characters (i.e. distance from ascenders to descenders)
|
|---|
| 1433 | t.setSpacing(0);
|
|---|
| 1434 |
|
|---|
| 1435 | t.setSpacing(lineHeightInt - t.getLineHeight());
|
|---|
| 1436 |
|
|---|
| 1437 | if (align.equals("left")) {
|
|---|
| 1438 | t.setJustification(Justification.left);
|
|---|
| 1439 | } else if (align.equals("right")) {
|
|---|
| 1440 | t.setJustification(Justification.right);
|
|---|
| 1441 | } else if (align.equals("center")) {
|
|---|
| 1442 | t.setJustification(Justification.center);
|
|---|
| 1443 | } else if (align.equals("justify")) {
|
|---|
| 1444 | t.setJustification(Justification.full);
|
|---|
| 1445 | }
|
|---|
| 1446 |
|
|---|
| 1447 | // Font size is added to the item width to give a little breathing room
|
|---|
| 1448 | t.setWidth(Math.round(width + (t.getSize())));
|
|---|
| 1449 |
|
|---|
| 1450 | if (!linkUrl.equals("undefined")) {
|
|---|
| 1451 | t.setAction("createFrameWithBrowser " + linkUrl);
|
|---|
| 1452 | t.setActionMark(false);
|
|---|
| 1453 | }
|
|---|
| 1454 | }
|
|---|
| 1455 | }
|
|---|
| 1456 | }
|
|---|
| 1457 |
|
|---|
| 1458 | /**
|
|---|
| 1459 | * Used by the web parser to add Next, Previous, etc. buttons to the converted pages
|
|---|
| 1460 | *
|
|---|
| 1461 | * @param text
|
|---|
| 1462 | * text to display on the button
|
|---|
| 1463 | * @param link
|
|---|
| 1464 | * Frame that the button will link to
|
|---|
| 1465 | * @param action
|
|---|
| 1466 | * Action to run when button is clicked
|
|---|
| 1467 | * @param width
|
|---|
| 1468 | * Width of the button
|
|---|
| 1469 | * @param toAddTo
|
|---|
| 1470 | * Frame to add the button to
|
|---|
| 1471 | * @param anchorTop
|
|---|
| 1472 | * @param anchorRight
|
|---|
| 1473 | * @param anchorBottom
|
|---|
| 1474 | * @param anchorLeft
|
|---|
| 1475 | */
|
|---|
| 1476 | private static void addButton(String text, String link, String action, int width, Frame toAddTo, Integer anchorTop, Integer anchorRight, Integer anchorBottom, Integer anchorLeft) {
|
|---|
| 1477 | // Button to go to the next frame/page
|
|---|
| 1478 | Text button = new Text(text);
|
|---|
| 1479 |
|
|---|
| 1480 | button.setLink(link);
|
|---|
| 1481 | button.addAction(action);
|
|---|
| 1482 | button.setBorderColor(new Colour(0.7f, 0.7f, 0.7f));
|
|---|
| 1483 | button.setBackgroundColor(new Colour(0.9f, 0.9f, 0.9f));
|
|---|
| 1484 | button.setThickness(1);
|
|---|
| 1485 | button.setLinkMark(false);
|
|---|
| 1486 | button.setActionMark(false);
|
|---|
| 1487 | button.setFamily("Roboto Condensed Light");
|
|---|
| 1488 | button.setJustification(Justification.center);
|
|---|
| 1489 | button.setWidth(width);
|
|---|
| 1490 |
|
|---|
| 1491 | if (anchorTop != null) {
|
|---|
| 1492 | button.setAnchorTop(anchorTop);
|
|---|
| 1493 | }
|
|---|
| 1494 |
|
|---|
| 1495 | if (anchorRight != null) {
|
|---|
| 1496 | button.setAnchorRight(anchorRight);
|
|---|
| 1497 | }
|
|---|
| 1498 |
|
|---|
| 1499 | if (anchorBottom != null) {
|
|---|
| 1500 | button.setAnchorBottom(anchorBottom);
|
|---|
| 1501 | }
|
|---|
| 1502 |
|
|---|
| 1503 | if (anchorLeft != null) {
|
|---|
| 1504 | button.setAnchorLeft(anchorLeft);
|
|---|
| 1505 | }
|
|---|
| 1506 |
|
|---|
| 1507 | button.setID(toAddTo.getNextItemID());
|
|---|
| 1508 | toAddTo.addItem(button);
|
|---|
| 1509 |
|
|---|
| 1510 | }
|
|---|
| 1511 | }
|
|---|
