Context Navigation

source: trunk/src/org/expeditee/io/WebParser.java@ 753

Last change on this file since 753 was 753, checked in by jts21, 10 years ago
Can't cast from Object to boolean, changed to Boolean
File size: 38.9 KB

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

Note: See TracBrowser for help on using the repository browser.

Download in other formats: