More flexible HTML parsing for mobile BKU

Way more lenient towards slight page changes now

3 files changed, 136 insertions, 40 deletions

diff --git a/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/ATrustHandler.java b/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/ATrustHandler.java
index b7e457c3..a493d109 100644
--- a/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/ATrustHandler.java
+++ b/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/ATrustHandler.java
@@ -79,23 +79,23 @@ public class ATrustHandler extends MobileBKUHandler {
 		ATrustStatus status = getStatus();
 
 		if (responseData.contains("<sl:ErrorResponse")) { //$NON-NLS-1$
-			String errorCode = MobileBKUHelper.extractTag(responseData,
+			String errorCode = MobileBKUHelper.extractSubstring(responseData,
 					"<sl:ErrorCode>", "</sl:ErrorCode>"); //$NON-NLS-1$ //$NON-NLS-2$
-			String errorMsg = MobileBKUHelper.extractTag(responseData,
+			String errorMsg = MobileBKUHelper.extractSubstring(responseData,
 					"<sl:Info>", "</sl:Info>"); //$NON-NLS-1$ //$NON-NLS-2$
 			throw new Exception("Error from mobile BKU: " + //$NON-NLS-1$
 					errorCode + " - " + errorMsg); //$NON-NLS-1$
 		}
 
 		// Extract infos:
-		String sessionID = MobileBKUHelper.extractTag(responseData,
+		String sessionID = MobileBKUHelper.extractSubstring(responseData,
 				"identification.aspx?sid=", "\""); //$NON-NLS-1$ //$NON-NLS-2$
 
-		String viewState = MobileBKUHelper.extractTag(responseData,
-				"id=\"__VIEWSTATE\" value=\"", "\""); //$NON-NLS-1$  //$NON-NLS-2$
+		String viewState = MobileBKUHelper.extractValueFromTagWithParam(
+				responseData, "", "id", "__VIEWSTATE", "value"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
 
-		String eventValidation = MobileBKUHelper.extractTag(responseData,
-				"id=\"__EVENTVALIDATION\" value=\"", "\""); //$NON-NLS-1$  //$NON-NLS-2$
+		String eventValidation = MobileBKUHelper.extractValueFromTagWithParam(
+				responseData, "", "id", "__EVENTVALIDATION", "value"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
 
 		log.info("sessionID: " + sessionID); //$NON-NLS-1$
 		log.info("viewState: " + viewState); //$NON-NLS-1$
@@ -117,7 +117,7 @@ public class ATrustHandler extends MobileBKUHandler {
 
 		MobileBKUHelper.registerTrustedSocketFactory();
 		HttpClient client = BKUHelper.getHttpClient();
-	
+
 		PostMethod post = new PostMethod(status.getBaseURL() + "/identification.aspx?sid=" + status.getSessionID()); //$NON-NLS-1$
 		post.getParams().setContentCharset("utf-8"); //$NON-NLS-1$
 		post.addParameter("__VIEWSTATE", status.getViewstate()); //$NON-NLS-1$
@@ -125,7 +125,7 @@ public class ATrustHandler extends MobileBKUHandler {
 		post.addParameter("handynummer", status.getPhoneNumber()); //$NON-NLS-1$
 		post.addParameter("signaturpasswort", status.getMobilePassword()); //$NON-NLS-1$
 		post.addParameter("Button_Identification", "Identifizieren"); //$NON-NLS-1$ //$NON-NLS-2$
-	
+
 		return executePost(client, post);
 	}
 
@@ -147,7 +147,7 @@ public class ATrustHandler extends MobileBKUHandler {
 		if (responseData.contains("ExpiresInfo.aspx?sid=")) { //$NON-NLS-1$
 			// Certification expiration interstitial - skip
 			String notice = Messages.getString("mobileBKU.notice") + " " + //$NON-NLS-1$ //$NON-NLS-2$
-					StringEscapeUtils.unescapeHtml4(MobileBKUHelper.extractTag(responseData, "<span id=\"Label2\">", "</span>")) //$NON-NLS-1$ //$NON-NLS-2$
+					StringEscapeUtils.unescapeHtml4(MobileBKUHelper.extractContentFromTagWithParam(responseData, "span", "id", "Label2")) //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
 					.replaceAll("\\<.*?\\>", ""); //$NON-NLS-1$ //$NON-NLS-2$
 			log.info(notice);
 
@@ -174,9 +174,9 @@ public class ATrustHandler extends MobileBKUHandler {
 				expiryNoticeDisplayed = true;
 			}
 
-			String t_sessionID = MobileBKUHelper.extractTag(responseData, "ExpiresInfo.aspx?sid=", "\""); //$NON-NLS-1$ //$NON-NLS-2$
-			String t_viewState = MobileBKUHelper.extractTag(responseData, "id=\"__VIEWSTATE\" value=\"", "\""); //$NON-NLS-1$  //$NON-NLS-2$
-			String t_eventValidation = MobileBKUHelper.extractTag(responseData, "id=\"__EVENTVALIDATION\" value=\"", "\""); //$NON-NLS-1$  //$NON-NLS-2$
+			String t_sessionID = MobileBKUHelper.extractSubstring(responseData, "ExpiresInfo.aspx?sid=", "\""); //$NON-NLS-1$ //$NON-NLS-2$
+			String t_viewState = MobileBKUHelper.extractValueFromTagWithParam(responseData, "", "id", "__VIEWSTATE", "value"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
+			String t_eventValidation = MobileBKUHelper.extractValueFromTagWithParam(responseData, "", "id", "__EVENTVALIDATION", "value"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
 
 			// Post again to skip
 			MobileBKUHelper.registerTrustedSocketFactory();
@@ -195,14 +195,14 @@ public class ATrustHandler extends MobileBKUHandler {
 		if (responseData.contains("signature.aspx?sid=")) { //$NON-NLS-1$
 			// credentials ok! TAN entry
 			log.debug("Credentials accepted - TAN required"); //$NON-NLS-1$
-			sessionID = MobileBKUHelper.extractTag(responseData, "signature.aspx?sid=", "\""); //$NON-NLS-1$ //$NON-NLS-2$
-			viewState = MobileBKUHelper.extractTag(responseData, "id=\"__VIEWSTATE\" value=\"", "\""); //$NON-NLS-1$ //$NON-NLS-2$
-			eventValidation = MobileBKUHelper.extractTag(responseData, "id=\"__EVENTVALIDATION\" value=\"", "\""); //$NON-NLS-1$ //$NON-NLS-2$
-			refVal = MobileBKUHelper.extractTag(responseData, "id='vergleichswert'><b>Vergleichswert:</b>", "</div>"); //$NON-NLS-1$ //$NON-NLS-2$
+			sessionID = MobileBKUHelper.extractSubstring(responseData, "signature.aspx?sid=", "\""); //$NON-NLS-1$ //$NON-NLS-2$
+			viewState = MobileBKUHelper.extractValueFromTagWithParam(responseData, "", "id", "__VIEWSTATE", "value"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
+			eventValidation = MobileBKUHelper.extractValueFromTagWithParam(responseData, "", "id", "__EVENTVALIDATION", "value"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
+			refVal = MobileBKUHelper.extractSubstring(responseData, "id='vergleichswert'><b>Vergleichswert:</b>", "</div>"); //$NON-NLS-1$ //$NON-NLS-2$
 			signatureDataURL = status.getBaseURL() + "/ShowSigobj.aspx" + //$NON-NLS-1$
-					MobileBKUHelper.extractTag(responseData, "ShowSigobj.aspx", "'"); //$NON-NLS-1$ //$NON-NLS-2$
+					MobileBKUHelper.extractSubstring(responseData, "ShowSigobj.aspx", "'"); //$NON-NLS-1$ //$NON-NLS-2$
 			try {
-				qrCode = MobileBKUHelper.extractTag(responseData, "<img class='qrcode' src='", "'"); //$NON-NLS-1$ //$NON-NLS-2$
+				qrCode = MobileBKUHelper.extractValueFromTagWithParam(responseData, "img", "class", "qrcode", "src"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
 				log.debug("QR Code found: " + qrCode); //$NON-NLS-1$
 				status.setQRCode(qrCode);
 			} catch (Exception e) {
@@ -218,11 +218,11 @@ public class ATrustHandler extends MobileBKUHandler {
 			// error page
 			// extract error text!
 			try {
-				String errorMessage = MobileBKUHelper.extractTag(responseData, "<span id=\"Label1\" class=\"ErrorClass\">", "</span>"); //$NON-NLS-1$ //$NON-NLS-2$
+				String errorMessage = MobileBKUHelper.extractContentFromTagWithParam(responseData, "span", "class", "ErrorClass"); //$NON-NLS-1$ //$NON-NLS-2$  //$NON-NLS-3$
 				status.setErrorMessage(errorMessage);
 			} catch (Exception e) {
-				throw new SignatureException(MobileBKUHelper.extractTag(responseData, "<sl:ErrorCode>", "</sl:ErrorCode>") + ": " + //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
-						MobileBKUHelper.extractTag(responseData, "<sl:Info>", "</sl:Info>")); //$NON-NLS-1$ //$NON-NLS-2$
+				throw new SignatureException(MobileBKUHelper.extractSubstring(responseData, "<sl:ErrorCode>", "</sl:ErrorCode>") + ": " + //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
+						MobileBKUHelper.extractSubstring(responseData, "<sl:Info>", "</sl:Info>")); //$NON-NLS-1$ //$NON-NLS-2$
 			}
 
 			// force UI again!
@@ -279,7 +279,7 @@ public class ATrustHandler extends MobileBKUHandler {
 					new SLResponse(responseData, getStatus().getServer(), null, null));
 		} else {
 			try {
-				String tries = MobileBKUHelper.extractTag(
+				String tries = MobileBKUHelper.extractSubstring(
 						responseData, "Sie haben noch", "Versuch"); //$NON-NLS-1$ //$NON-NLS-2$
 				getStatus().setTanTries(Integer.parseInt(tries.trim()));
 				getStatus().setErrorMessage("mobileBKU.wrong_tan"); //$NON-NLS-1$
diff --git a/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/IAIKHandler.java b/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/IAIKHandler.java
index 0904f045..b282811f 100644
--- a/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/IAIKHandler.java
+++ b/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/IAIKHandler.java
@@ -68,7 +68,7 @@ public class IAIKHandler extends MobileBKUHandler {
 		IAIKStatus status = getStatus();
 
 		// Extract infos:
-		String credentialURL = MobileBKUHelper.extractTag(responseData,
+		String credentialURL = MobileBKUHelper.extractSubstring(responseData,
 				"name=\"userCredLogon\" method=\"post\" action=\"", "\""); //$NON-NLS-1$ //$NON-NLS-2$
 		URL baseURL = new URL(status.getBaseURL());
 		int p = baseURL.getPort();
@@ -76,7 +76,7 @@ public class IAIKHandler extends MobileBKUHandler {
 		credentialURL = baseURL.getProtocol() + "://" + baseURL.getHost() + port + //$NON-NLS-1$
 		(credentialURL.startsWith("/") ? "" : "/") + credentialURL; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
 
-		String viewState = MobileBKUHelper.extractTag(responseData,
+		String viewState = MobileBKUHelper.extractSubstring(responseData,
 				"id=\"javax.faces.ViewState\" value=\"", "\""); //$NON-NLS-1$ //$NON-NLS-2$
 
 		String sessionID = credentialURL.substring(credentialURL.indexOf("jsessionid=") + 11); //$NON-NLS-1$
@@ -133,7 +133,7 @@ public class IAIKHandler extends MobileBKUHandler {
 
 			String errorMessage;
 			try {
-				errorMessage = MobileBKUHelper.extractTag(responseData, ":errorMessage\">", "</span>"); //$NON-NLS-1$ //$NON-NLS-2$
+				errorMessage = MobileBKUHelper.extractSubstring(responseData, ":errorMessage\">", "</span>"); //$NON-NLS-1$ //$NON-NLS-2$
 			} catch (Exception e) {
 				errorMessage = Messages.getString("error.Unexpected"); //$NON-NLS-1$
 			}
@@ -149,16 +149,16 @@ public class IAIKHandler extends MobileBKUHandler {
 		String redirectURL = status.getBaseURL().substring(0,
 				status.getBaseURL().lastIndexOf('/',
 						status.getBaseURL().lastIndexOf('/') - 1) + 1); //Cut off last directory
-		redirectURL += MobileBKUHelper.extractTag(responseData,
+		redirectURL += MobileBKUHelper.extractSubstring(responseData,
 				"redirection_url\":\"", "\""); //$NON-NLS-1$ //$NON-NLS-2$
 		redirectURL = status.ensureSessionID(redirectURL);
 
 		responseData = getRedirect(client, redirectURL);
 
-		refVal = MobileBKUHelper.extractTag(responseData,
+		refVal = MobileBKUHelper.extractSubstring(responseData,
 				"id=\"j_idt6:refValue\" class=\"strong\">", "</"); //$NON-NLS-1$ //$NON-NLS-2$
 
-		String viewState = MobileBKUHelper.extractTag(responseData,
+		String viewState = MobileBKUHelper.extractSubstring(responseData,
 				"id=\"javax.faces.ViewState\" value=\"", "\""); //$NON-NLS-1$ //$NON-NLS-2$
 		status.setViewState(viewState);
 
@@ -186,7 +186,7 @@ public class IAIKHandler extends MobileBKUHandler {
 				return;
 			}
 
-			redirectURL = MobileBKUHelper.extractTag(responseData,
+			redirectURL = MobileBKUHelper.extractSubstring(responseData,
 					"redirect url=\"", "\""); //$NON-NLS-1$ //$NON-NLS-2$
 			URL baseURL = new URL(status.getBaseURL());
 			int p = baseURL.getPort();
@@ -196,7 +196,7 @@ public class IAIKHandler extends MobileBKUHandler {
 
 			responseData = getRedirect(client, redirectURL);
 
-			viewState = MobileBKUHelper.extractTag(responseData,
+			viewState = MobileBKUHelper.extractSubstring(responseData,
 					"id=\"javax.faces.ViewState\" value=\"", "\""); //$NON-NLS-1$ //$NON-NLS-2$
 			status.setViewState(viewState);
 		}
@@ -204,11 +204,11 @@ public class IAIKHandler extends MobileBKUHandler {
 		signatureDataURL = status.getBaseURL();
 		signatureDataURL = signatureDataURL.substring(0, signatureDataURL.lastIndexOf('/') + 1);
 		signatureDataURL += "viewer.jsf" + //$NON-NLS-1$
-				MobileBKUHelper.extractTag(responseData, "viewer.jsf", "\""); //$NON-NLS-1$ //$NON-NLS-2$
+				MobileBKUHelper.extractSubstring(responseData, "viewer.jsf", "\""); //$NON-NLS-1$ //$NON-NLS-2$
 		signatureDataURL += (signatureDataURL.contains("?") ? "&" : "?") + //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
 				"pdfoversessionid=" + status.getSessionID(); //$NON-NLS-1$
 
-		String tanURL = MobileBKUHelper.extractTag(responseData,
+		String tanURL = MobileBKUHelper.extractSubstring(responseData,
 				"name=\"j_idt6\" method=\"post\" action=\"", "\""); //$NON-NLS-1$ //$NON-NLS-2$
 		URL baseURL = new URL(status.getBaseURL());
 		int p = baseURL.getPort();
@@ -259,7 +259,7 @@ public class IAIKHandler extends MobileBKUHandler {
 					new SLResponse(responseData, getStatus().getServer(), null, null));
 		} else {
 			try {
-				String errorMessage = MobileBKUHelper.extractTag(responseData,
+				String errorMessage = MobileBKUHelper.extractSubstring(responseData,
 						":errorMessage\">", "</span>"); //$NON-NLS-1$ //$NON-NLS-2$
 				getStatus().setErrorMessage(errorMessage);
 			} catch (Exception e) {
diff --git a/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/MobileBKUHelper.java b/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/MobileBKUHelper.java
index 05746c7c..b9975167 100644
--- a/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/MobileBKUHelper.java
+++ b/pdf-over-gui/src/main/java/at/asit/pdfover/gui/bku/mobile/MobileBKUHelper.java
@@ -56,10 +56,11 @@ public class MobileBKUHelper {
 	 *            the start marker
 	 * @param end
 	 *            the end marker
-	 * @return the substring
+	 * @return    the substring
 	 * @throws Exception
+	 *            not found
 	 */
-	public static String extractTag(String data, String start, String end)
+	public static String extractSubstring(String data, String start, String end)
 			throws Exception {
 		int startidx = data.indexOf(start);
 		if (startidx > 0) {
@@ -68,11 +69,106 @@ public class MobileBKUHelper {
 			if (endidx > startidx) {
 				return data.substring(startidx, endidx);
 			}
-			log.error("extracting Tag: end tag not valid!: " + start + " ... " + end); //$NON-NLS-1$//$NON-NLS-2$
-			throw new Exception("end tag not available! Mobile BKU site changed?"); //$NON-NLS-1$
+			log.error("extracting substring: end not valid!: " + start + " ... " + end); //$NON-NLS-1$//$NON-NLS-2$
+			throw new Exception("End string not available! Mobile BKU site changed?"); //$NON-NLS-1$
 		}
-		log.error("extracting Tag: start tag not valid!: " + start + " ... " + end); //$NON-NLS-1$//$NON-NLS-2$
-		throw new Exception("start tag not available! Mobile BKU site changed?"); //$NON-NLS-1$
+		log.error("extracting substring: start not valid!: " + start + " ... " + end); //$NON-NLS-1$//$NON-NLS-2$
+		throw new Exception("Start string not available! Mobile BKU site changed?"); //$NON-NLS-1$
+	}
+
+	/**
+	 * Extracts an XML tag from data with the given param="value"
+	 * 
+	 * @param data
+	 *            the whole data string
+	 * @param tag
+	 *            the tag name (empty string to match all tags)
+	 * @param param
+	 *            the parameter to look for
+	 * @param value
+	 *            the parameter value to look for
+	 * @return    the found tag
+	 * @throws Exception
+	 *            not found
+	 */
+	public static String extractTagWithParam(String data, String tag,
+			String param, String value) throws Exception {
+		String start = '<' + tag;
+		int startidx, endidx = 0;
+		while ((startidx = data.indexOf(start, endidx)) != -1) {
+			endidx = data.indexOf('>', startidx);
+			if (endidx == -1) {
+				log.error("extracting tag: unterminated tag! " + tag + " (" + param + "=" + value + ")"); //$NON-NLS-1$//$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
+				throw new Exception("Tag not found! Mobile BKU site changed?"); //$NON-NLS-1$
+			}
+			String found = data.substring(startidx, endidx + 1);
+			if (found.contains(param + "='" + value + "'") || //$NON-NLS-1$ //$NON-NLS-2$
+			    found.contains(param + "=\"" + value + "\"")) //$NON-NLS-1$ //$NON-NLS-2$
+				return found;
+		}
+		log.error("extracting tag: not found!: " + tag + " (" + param + "='" + value + "')"); //$NON-NLS-1$//$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
+		throw new Exception("Tag not found! Mobile BKU site changed?"); //$NON-NLS-1$
+	}
+
+	/**
+	 * Extracts a parameter value from an XML tag from data with the given param="value"
+	 * 
+	 * @param data
+	 *            the whole data string
+	 * @param tag
+	 *            the tag name (empty string to match all tags)
+	 * @param param
+	 *            the parameter to look for
+	 * @param value
+	 *            the parameter value to look for
+	 * @param returnparam
+	 *            the parameter whose value to return
+	 * @return    the found tag
+	 * @throws Exception
+	 *            not found
+	 */
+	public static String extractValueFromTagWithParam(String data, String tag,
+			String param, String value, String returnparam) throws Exception {
+		String found = extractTagWithParam(data, tag, param, value);
+		int startidx = found.indexOf(returnparam + '=');
+		if (startidx == -1) {
+			log.error("extracting tag: param not found! " + tag + " (" + param + "=" + value + ") - " + returnparam); //$NON-NLS-1$//$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
+			throw new Exception("Tag not found! Mobile BKU site changed?"); //$NON-NLS-1$
+		}
+		startidx += returnparam.length() + 1;
+		int endidx = found.indexOf(found.charAt(startidx), startidx + 1);
+		if (endidx == -1) {
+			log.error("extracting tag: unterminated param value! " + tag + " (" + param + "=" + value + ") - " + returnparam); //$NON-NLS-1$//$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
+			throw new Exception("Tag not found! Mobile BKU site changed?"); //$NON-NLS-1$
+		}
+		return found.substring(startidx + 1, endidx);
+	}
+
+	/**
+	 * Extracts the content from an XML tag from data with the given param="value"
+	 * 
+	 * @param data
+	 *            the whole data string
+	 * @param tag
+	 *            the tag name
+	 * @param param
+	 *            the parameter to look for
+	 * @param value
+	 *            the parameter value to look for
+	 * @return    the found tag's content
+	 * @throws Exception
+	 *            not found
+	 */
+	public static String extractContentFromTagWithParam(String data, String tag,
+			String param, String value) throws Exception {
+		String found = extractTagWithParam(data, tag, param, value);
+		int startidx = data.indexOf(found) + found.length();
+		int endidx = data.indexOf("</" + tag + ">"); //$NON-NLS-1$ //$NON-NLS-2$
+		if (endidx == -1) {
+			log.error("extracting tag: closing tag not found! " + tag + " (" + param + "=" + value + ")"); //$NON-NLS-1$//$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
+			throw new Exception("Tag not found! Mobile BKU site changed?"); //$NON-NLS-1$
+		}
+		return data.substring(startidx, endidx);
 	}
 
 	/**