// appends text to the string builder with a simple word wrap method
private void append2(String text) {
// if (text.contains('\n'))
width =
0; // reset counter if starts with a newline. only from formats above, not in natural text
if (text.equals(" ")
&& (accum.length() == 0 || StringUtil.in(accum.substring(accum.length() - 1), " ")))
return; // don't accumulate long runs of empty spaces
if (text.length() + width > maxWidth) { // won't fit, needs to wrap
String words[] = text.split("\\s+");
for (int i = 0; i < words.length; i++) {
String word = words[i];
boolean last = i == words.length - 1;
if (!last) // insert a space if not the last word
word = word + " ";
if (word.length() + width > maxWidth) { // wrap and reset counter
accum.append('\n').append(word);
width = word.length();
} else {
accum.append(word);
width += word.length();
}
}
} else { // fits as is, without need to wrap text
accum.append(text);
width += text.length();
}
}
private String[][] processItems(List<Item> items) {
if (items.size() == 0) {
return null;
}
DSpaceMapping mapping = new DSpaceMapping();
Set<String> headings = new HashSet<String>();
List<Map<String, List<String>>> records = new ArrayList<Map<String, List<String>>>();
for (Item item : items) {
try {
Map<String, List<String>> values = mapping.getMap(item);
Set<String> itemHeadings = values.keySet();
headings.addAll(itemHeadings);
values.put("itemId", Arrays.asList(item.getItemId().toString()));
List<String> handles = new ArrayList<String>();
for (ItemHandle handle : item.getHandles()) {
String handleVal = "http://hdl.handle.net/" + handle.getHandle();
handles.add(handleVal);
}
if (handles.size() > 0) {
values.put("handle", handles);
}
records.add(values);
} catch (IOException e) {
}
}
List<String> excludeFields = getExcludeFieldTypes();
headings.removeAll(excludeFields);
LOGGER.info("Number of new records: {}, Number of Fields: {}", records.size(), headings.size());
String[][] itemsMatrix = new String[records.size() + 1][headings.size() + 2];
itemsMatrix[0][0] = "itemId";
itemsMatrix[0][1] = "handle";
Iterator<String> it = headings.iterator();
int i = 2;
while (it.hasNext()) {
String heading = it.next();
itemsMatrix[0][i] = heading;
i++;
}
for (int j = 0; j < records.size(); j++) {
Map<String, List<String>> record = records.get(j);
for (int k = 0; k < itemsMatrix[0].length; k++) {
String header = itemsMatrix[0][k];
List<String> values = record.get(header);
if (values != null && values.size() > 0) {
String val = StringUtil.join(values, ";");
itemsMatrix[j + 1][k] = val;
}
}
}
return itemsMatrix;
}
// hit when all of the node's children (if any) have been visited
public void tail(Node node, int depth) {
String name = node.nodeName();
if (StringUtil.in(name, "p", "h1", "h2", "h3", "h4", "h5", "a", "span")) append("<br>");
if (name.equals("ul")) {
append("</ul>");
}
if (name.equals("li")) {
append("</li>");
}
}
/**
* @param url
* @param parameter
* @return
*/
public static String getParameter(URI uri, String parameter) {
String queryString = uri.getQuery();
if (StringUtil.isBlank(queryString)) {
return null;
}
String[] kvs = queryString.split("&");
String[] kv = null;
for (String s : kvs) {
if (StringUtil.isBlank(s)) {
continue;
} else {
kv = s.split("=");
if (kv.length == 2) {
if (kv[0].equals(parameter)) {
return kv[1];
}
}
}
}
return null;
}
// start setting of list in right rail
public void rightRailList(
Node listNode, Element rightListEle, Map<String, String> urlMap, String locale) {
try {
Element title;
Element description;
Elements headElements = rightListEle.getElementsByTag("h2");
if (headElements.size() > 1) {
title = rightListEle.getElementsByTag("h2").last();
description = rightListEle.getElementsByTag("p").last();
sb.append("<li>Mismatch in count of list panel component in right rail.</li>");
} else {
title = rightListEle.getElementsByTag("h2").first();
description = rightListEle.getElementsByTag("p").first();
}
listNode.setProperty("title", title.text());
javax.jcr.Node introNode = listNode.getNode("intro");
introNode.setProperty("paragraph_rte", description.text());
javax.jcr.Node eleListNode = listNode.getNode("element_list_0");
Elements ulList = rightListEle.getElementsByTag("ul");
for (Element element : ulList) {
java.util.List<String> list = new ArrayList<String>();
Elements menuLiList = element.getElementsByTag("li");
for (Element li : menuLiList) {
JSONObject jsonObjrr = new JSONObject();
Element listItemAnchor = li.getElementsByTag("a").first();
String anchorText = listItemAnchor != null ? listItemAnchor.text() : "";
String anchorHref = listItemAnchor.absUrl("href");
if (StringUtil.isBlank(anchorHref)) {
anchorHref = listItemAnchor.attr("href");
}
// Start extracting valid href
log.debug("Before right list LinkUrl" + anchorHref + "\n");
anchorHref = FrameworkUtils.getLocaleReference(anchorHref, urlMap, locale, sb);
log.debug("after right list LinkUrl" + anchorHref + "\n");
// End extracting valid href
jsonObjrr.put("linktext", anchorText);
jsonObjrr.put("linkurl", anchorHref);
jsonObjrr.put("icon", "none");
jsonObjrr.put("size", "");
jsonObjrr.put("description", "");
jsonObjrr.put("openInNewWindow", "false");
list.add(jsonObjrr.toString());
}
eleListNode.setProperty("listitems", list.toArray(new String[list.size()]));
}
log.debug("Updated title, descriptoin and linktext at " + listNode.getPath());
} catch (Exception e) {
e.printStackTrace();
}
}
private boolean checkIfStringIsSimilar(
String firstPartOfCurrentCompanyName, String firstPartOfHistoricalCompanyName) {
if (!StringUtil.isBlank(firstPartOfCurrentCompanyName)
&& !StringUtil.isBlank(firstPartOfHistoricalCompanyName)) {
firstPartOfCurrentCompanyName = firstPartOfCurrentCompanyName.toLowerCase().trim();
firstPartOfHistoricalCompanyName = firstPartOfHistoricalCompanyName.toLowerCase().trim();
int currentCmpNameLen = firstPartOfCurrentCompanyName.length();
int histCmpNameLen = firstPartOfHistoricalCompanyName.length();
String endOfFirstPartCurrent =
firstPartOfCurrentCompanyName.substring(
currentCmpNameLen >= 4 ? currentCmpNameLen - 4 : 0, currentCmpNameLen - 1);
String endOfLastParHist =
firstPartOfHistoricalCompanyName.substring(
histCmpNameLen >= 4 ? histCmpNameLen - 4 : 0, histCmpNameLen - 1);
if (endOfFirstPartCurrent.equalsIgnoreCase(endOfLastParHist)) {
return true;
}
}
return false;
}
public boolean scrapInternal(String url, String marker) throws IOException {
Document document;
try {
document = Jsoup.connect(url).get();
} catch (IOException e) {
// try again
document = Jsoup.connect(url).get();
}
Elements hrefElements = document.select("a");
Iterator<Element> eltIterator = hrefElements.iterator();
boolean postFound = false;
while (eltIterator.hasNext()) {
String href = eltIterator.next().attr("href");
if (StringUtil.isBlank(href)) {
continue;
} else if (parser.isPostPage(href)) {
try {
BlogPost blogPost = parser.getRepository().findOneByUrl(href);
if (blogPost == null) {
scrapBlogPost(href, null, false, marker);
postFound = true;
} else if (!marker.equals(blogPost.getMarker())) {
scrapBlogPost(href, blogPost, true, marker);
}
} catch (Exception exp) {
exp.printStackTrace();
}
} else if (parser.isListPage(href) && !summaryQueue.contains(href)) {
summaryQueue.add(href);
}
}
return postFound;
}
protected void indent(StringBuilder accum, int depth, Document.OutputSettings out) {
accum.append("\n").append(StringUtil.padding(depth * out.indentAmount()));
}
// start setting of spotlight
public void spotLightTranslate(
Node slNode, Element spElement, String locale, Map<String, String> urlMap) {
try {
Element spotLightTitle = spElement.getElementsByTag("h2").first();
Element spotLightDescription = spElement.getElementsByTag("p").first();
Elements spotLightAnchorElements = spElement.getElementsByTag("a");
Element spotLightAnchor = spotLightAnchorElements.first();
// start image
String spotLightImage = FrameworkUtils.extractImagePath(spElement, sb);
log.debug("spotLightImage " + spotLightImage + "\n");
if (slNode != null) {
if (slNode.hasNode("image")) {
Node spotLightImageNode = slNode.getNode("image");
String fileReference =
spotLightImageNode.hasProperty("fileReference")
? spotLightImageNode.getProperty("fileReference").getString()
: "";
spotLightImage =
FrameworkUtils.migrateDAMContent(spotLightImage, fileReference, locale, sb);
log.debug("spotLightImage " + spotLightImage + "\n");
if (StringUtils.isNotBlank(spotLightImage)) {
spotLightImageNode.setProperty("fileReference", spotLightImage);
} else {
sb.append("<li> Image not found for spot light component on locale page. </li>");
}
} else {
sb.append("<li>spotlight image node doesn't exist</li>");
}
}
// end image
if (spotLightDescription.getElementsByTag("a") != null
&& !spotLightDescription.getElementsByTag("a").isEmpty()) {
slNode.setProperty("description", spotLightDescription.html());
// start
if (spotLightAnchorElements.size() > 1) {
spotLightAnchor = spotLightAnchorElements.get(1);
} else {
spotLightAnchor = null;
sb.append(
"<li>Link is not found on locale page for the spotlight component. This needs to be deleted manually.</li>");
}
// end
} else {
slNode.setProperty("description", spotLightDescription.text());
}
if (spotLightAnchor != null) {
String linkText = spotLightAnchor.text();
String linkUrl = spotLightAnchor.absUrl("href");
if (StringUtil.isBlank(linkUrl)) {
linkUrl = spotLightAnchor.attr("href");
}
// Start extracting valid href
log.debug("Before spotlight LinkUrl" + linkUrl + "\n");
linkUrl = FrameworkUtils.getLocaleReference(linkUrl, urlMap, locale, sb);
log.debug("after spotlight LinkUrl" + linkUrl + "\n");
// End extracting valid href
slNode.setProperty("linktext", linkText);
javax.jcr.Node ctaNode = slNode.getNode("cta");
if (ctaNode != null) {
if (linkUrl != null) {
// ctaNode.setProperty("linktype", "Url");
ctaNode.setProperty("url", linkUrl);
}
}
}
if (spotLightTitle != null) {
Elements spotLightLink = spotLightTitle.getElementsByTag("a");
if (spotLightLink.isEmpty()) {
slNode.setProperty("title", spotLightTitle.text());
} else {
Element spotLightLinkEle = spotLightLink.first();
String slLinkText = spotLightLinkEle.text();
String slLinkUrl = spotLightLinkEle.absUrl("href");
if (StringUtil.isBlank(slLinkUrl)) {
slLinkUrl = spotLightLinkEle.attr("href");
}
// Start extracting valid href
log.debug("Before spotlight LinkUrl" + slLinkUrl + "\n");
slLinkUrl = FrameworkUtils.getLocaleReference(slLinkUrl, urlMap, locale, sb);
log.debug("after spotlight LinkUrl" + slLinkUrl + "\n");
// End extracting valid href
slNode.setProperty("title", slLinkText);
javax.jcr.Node titleLinkNode = slNode.getNode("titlelink");
if (titleLinkNode != null) {
if (slLinkUrl != null) {
titleLinkNode.setProperty("linktype", "Url");
titleLinkNode.setProperty("url", slLinkUrl);
}
}
}
}
log.debug("Updated title, descriptoin and linktext at " + slNode.getPath());
} catch (Exception e) {
e.printStackTrace();
}
}
// start setting of selectorbar
public void selectorBarTranslate(
Node selectorBarPanelNode, Element ele, Map<String, String> urlMap, String locale) {
try {
String title = (ele != null ? ele.getElementsByTag("a").first().text() : "");
String titleUrl = ele.getElementsByTag("a").first().absUrl("href");
if (StringUtil.isBlank(titleUrl)) {
titleUrl = ele.getElementsByTag("a").first().attr("href");
}
// Start extracting valid href
log.debug("Before selector bar title LinkUrl" + titleUrl + "\n");
titleUrl = FrameworkUtils.getLocaleReference(titleUrl, urlMap, locale, sb);
log.debug("after selector bar title LinkUrl" + titleUrl + "\n");
// End extracting valid href
log.debug("selector component titleUrl: " + titleUrl);
selectorBarPanelNode.setProperty("title", title);
selectorBarPanelNode.setProperty("titleurl", titleUrl);
if (ele.childNodeSize() >= 2) {
log.debug("Child node size is greater than 1.");
if (ele.select("div.menu").isEmpty()) {
log.debug("Menu is not available.");
sb.append(
"<li>Selector bar drop down menu elements does not exist on the locale page.</li>");
} else {
log.debug("Menu is available.");
Element menuEle = ele.child(1);
if (menuEle != null) {
log.debug("selector component menuEle: " + menuEle.toString());
Element anchor = menuEle.getElementsByTag("a").last();
String allLinkText = anchor != null ? anchor.text() : "";
String allLinkUrl = anchor != null ? anchor.absUrl("href") : "";
if (StringUtil.isBlank(allLinkUrl)) {
allLinkUrl = anchor.attr("href");
}
// Start extracting valid href
log.debug("Before selector bar menu LinkUrl" + allLinkUrl + "\n");
allLinkUrl = FrameworkUtils.getLocaleReference(allLinkUrl, urlMap, locale, sb);
log.debug("after selector bar menu LinkUrl" + allLinkUrl + "\n");
// End extracting valid href
selectorBarPanelNode.setProperty("alllinktext", allLinkText);
selectorBarPanelNode.setProperty("alllinkurl", allLinkUrl);
Elements menuUlList = menuEle.getElementsByTag("ul");
for (Element element : menuUlList) {
java.util.List<String> list = new ArrayList<String>();
Elements menuLiList = element.getElementsByTag("li");
System.out.println(menuLiList.size());
for (Element li : menuLiList) {
JSONObject jsonObj = new JSONObject();
Element listItemAnchor = li.getElementsByTag("a").first();
String anchorText = listItemAnchor != null ? listItemAnchor.text() : "";
String anchorHref = listItemAnchor.absUrl("href");
if (StringUtil.isBlank(anchorHref)) {
anchorHref = listItemAnchor.attr("href");
}
// Start extracting valid href
log.debug("Before selectorbarLinkUrl" + anchorHref + "\n");
anchorHref = FrameworkUtils.getLocaleReference(anchorHref, urlMap, locale, sb);
log.debug("after selectorbarLinkUrl" + anchorHref + "\n");
// End extracting valid href
jsonObj.put("linktext", anchorText);
jsonObj.put("linkurl", anchorHref);
jsonObj.put("size", "");
list.add(jsonObj.toString());
}
selectorBarPanelNode.setProperty("panelitems", list.toArray(new String[list.size()]));
}
} else {
sb.append(
"<li>Selector bar drop down menu elements does not exist on the locale page.</li>");
}
}
} else {
sb.append(
"<li>Selector bar drop down menu elements does not exist on the locale page.</li>");
}
} catch (Exception e) {
e.printStackTrace();
}
}
// start setting of heropanel
public void heroPanelTranslate(
Node heroPanelNode,
Element ele,
String locale,
Map<String, String> urlMap,
int imageSrcEmptyCount) {
try {
String title = ele.getElementsByTag("h2") != null ? ele.getElementsByTag("h2").text() : "";
String desc =
ele.getElementsByTag("p") != null ? ele.getElementsByTag("p").first().text() : "";
Element anchor = ele.getElementsByTag("a").first();
String anchorText = anchor != null ? anchor.text() : "";
String anchorHref = anchor.absUrl("href");
if (StringUtil.isBlank(anchorHref)) {
anchorHref = anchor.attr("href");
}
// Start extracting valid href
log.debug("Before heroPanelLinkUrl" + anchorHref + "\n");
anchorHref = FrameworkUtils.getLocaleReference(anchorHref, urlMap, locale, sb);
log.debug("after heroPanelLinkUrl" + anchorHref + "\n");
// End extracting valid href
// start image
String heroImage = FrameworkUtils.extractImagePath(ele, sb);
log.debug("heroImage before migration : " + heroImage + "\n");
if (heroPanelNode != null) {
Node heroPanelPopUpNode = null;
Element lightBoxElement = null;
Elements lightBoxElements = ele.select("div.c50-image").select("a.c26v4-lightbox");
heroPanelPopUpNode = FrameworkUtils.getHeroPopUpNode(heroPanelNode);
lightBoxElements =
!lightBoxElements.isEmpty()
? lightBoxElements
: ele.select("div.c50-text").select("a.c26v4-lightbox");
if (lightBoxElements != null && !lightBoxElements.isEmpty()) {
lightBoxElement = lightBoxElements.first();
}
if (heroPanelPopUpNode == null && lightBoxElement != null) {
sb.append(
"<li>video pop up is present in WEB page but it is not present in WEM page.</li>");
}
if (heroPanelPopUpNode != null && lightBoxElement == null) {
sb.append(
"<li>video pop up is present in WEM page but it is not present in WEB page.</li>");
}
if (heroPanelPopUpNode != null
&& lightBoxElement != null
&& StringUtils.isNotBlank(title)) {
heroPanelPopUpNode.setProperty("popupHeader", title);
}
if (heroPanelNode.hasNode("image")) {
Node imageNode = heroPanelNode.getNode("image");
String fileReference =
imageNode.hasProperty("fileReference")
? imageNode.getProperty("fileReference").getString()
: "";
heroImage = FrameworkUtils.migrateDAMContent(heroImage, fileReference, locale, sb);
log.debug("heroImage after migration : " + heroImage + "\n");
if (StringUtils.isNotBlank(heroImage)) {
imageNode.setProperty("fileReference", heroImage);
} else {
imageSrcEmptyCount++;
}
} else {
sb.append("<li>hero image node doesn't exist</li>");
}
heroPanelNode.setProperty("title", title);
heroPanelNode.setProperty("description", desc);
heroPanelNode.setProperty("linktext", anchorText);
heroPanelNode.setProperty("linkurl", anchorHref);
}
// end image
if (imageSrcEmptyCount > 0) {
sb.append(
"<li> "
+ imageSrcEmptyCount
+ " image(s) are not found on hero elements on locale page.</li>");
}
} catch (Exception e) {
e.printStackTrace();
}
}
public String translate(
String host,
String loc,
String prod,
String type,
String catType,
String locale,
Session session,
Map<String, String> urlMap)
throws IOException, ValueFormatException, VersionException, LockException,
ConstraintViolationException, RepositoryException {
log.debug("In the translate method");
log.debug("In the translate method, catType is :" + catType);
String pagePropertiesPath = "/content/<locale>/" + catType + "/<prod>/index/jcr:content";
String indexLeft =
"/content/<locale>/"
+ catType
+ "/<prod>/index/jcr:content/content_parsys/overview/layout-overview/gd12v2/gd12v2-left";
String indexRight =
"/content/<locale>/"
+ catType
+ "/<prod>/index/jcr:content/content_parsys/overview/layout-overview/gd12v2/gd12v2-right";
String pageUrl = host + "/content/<locale>/" + catType + "/<prod>/index.html";
pageUrl = pageUrl.replace("<locale>", locale).replace("<prod>", prod);
pagePropertiesPath = pagePropertiesPath.replace("<locale>", locale).replace("<prod>", prod);
sb.append("<td>" + "<a href=" + pageUrl + ">" + pageUrl + "</a>" + "</td>");
sb.append("<td>" + "<a href=" + loc + ">" + loc + "</a>" + "</td>");
sb.append("<td><ul>");
indexLeft = indexLeft.replace("<locale>", locale).replace("<prod>", prod);
indexRight = indexRight.replace("<locale>", locale).replace("<prod>", prod);
javax.jcr.Node indexLeftNode = null;
javax.jcr.Node indexRightNode = null;
javax.jcr.Node pageJcrNode = null;
try {
indexLeftNode = session.getNode(indexLeft);
indexRightNode = session.getNode(indexRight);
pageJcrNode = session.getNode(pagePropertiesPath);
try {
// doc = Jsoup.connect(loc).get();
doc = getConnection(loc);
if (doc != null) {
log.debug("Connected to the provided URL");
// ------------------------------------------------------------------------------------------------------------------------------------------
// start set page properties.
FrameworkUtils.setPageProperties(pageJcrNode, doc, session, sb);
// end set page properties.
// ------------------------------------------------------------------------------------------------------------------------------------------
// start set hero large component properties.
try {
log.debug("Start of Hero component");
Elements heroElements = doc.select("div.frame");
Node heroNode =
indexLeftNode.hasNode("hero_large") ? indexLeftNode.getNode("hero_large") : null;
if (heroNode != null) {
log.debug("hero node found: " + heroNode.getPath());
if (heroElements.isEmpty()) {
log.debug("No hero element found with div class name frame.");
sb.append(
"<li>Hero component with class name 'frame' does not exist on locale page.</li>");
} else {
int eleSize = heroElements.size();
log.debug("hero node element size: " + eleSize);
NodeIterator heroPanelNodeIterator = heroNode.getNodes("heropanel*");
int nodeSize = (int) heroPanelNodeIterator.getSize();
log.debug("hero node nodeSize : " + nodeSize);
if (eleSize == nodeSize) {
setForHero(heroElements, heroNode, locale, urlMap);
} else if (nodeSize < eleSize) {
setForHero(heroElements, heroNode, locale, urlMap);
sb.append(
"<li>Mismatch in the count of hero panels. Additional panel(s) found on locale page. Locale page has "
+ eleSize
+ " panels and there are "
+ nodeSize
+ " nodes.</li>");
} else if (nodeSize > eleSize) {
setForHero(heroElements, heroNode, locale, urlMap);
sb.append(
"<li>Mismatch in the count of hero panels. Additional node(s) found. Locale page has "
+ eleSize
+ " panels and there are "
+ nodeSize
+ " nodes.</li>");
}
}
} else {
log.debug("No hero node found at " + indexLeftNode);
sb.append("<li>Node for hero large component does not exist.</li>");
}
log.debug("End of Hero component");
} catch (Exception e) {
sb.append("<li>Unable to update hero large component.</li>");
}
// end set Hero Large component properties.
// start set selectorbar large component properties.
try {
Elements selectorBarLargeElements = doc.select("div.selectorbarpanel");
if (selectorBarLargeElements.size() == 0) {
selectorBarLargeElements =
doc.select("div.c58-pilot")
.select("div.left,div.mid,div.right"); // ("div.selectorbarpanel");
}
Node selectorBarNode =
indexLeftNode.hasNode("selectorbarlarge_0")
? indexLeftNode.getNode("selectorbarlarge_0")
: null;
if (selectorBarNode != null) {
if (selectorBarLargeElements.isEmpty()) {
log.debug("No selector bar element found with div class name selectorbarpanel.");
sb.append(
"<li>Selector bar component with class name 'selectorbarpanel' does not exist on locale page.</li>");
} else {
log.debug("selector bar node component found at : " + indexLeftNode.getPath());
int eleSize = selectorBarLargeElements.size();
log.debug("selector component element size: " + eleSize);
NodeIterator selectorBarPanel = selectorBarNode.getNodes("selectorbarpanel*");
int nodeSize = (int) selectorBarPanel.getSize();
log.debug("selector component nodeSize : " + nodeSize);
if (eleSize == nodeSize) {
for (Element ele : selectorBarLargeElements) {
Node selectorBarPanelNode;
if (selectorBarPanel.hasNext()) {
selectorBarPanelNode = (Node) selectorBarPanel.next();
selectorBarTranslate(selectorBarPanelNode, ele, urlMap, locale);
} else {
log.debug("Next node not found");
}
}
} else if (eleSize > nodeSize) {
for (Element ele : selectorBarLargeElements) {
Node selectorBarPanelNode;
if (selectorBarPanel.hasNext()) {
selectorBarPanelNode = (Node) selectorBarPanel.next();
selectorBarTranslate(selectorBarPanelNode, ele, urlMap, locale);
} else {
log.debug("Next node not found");
sb.append(
"<li>Mismatch in the count of selector bar panel. Additional panel(s) found on locale page. Locale page has "
+ eleSize
+ " panels and there are "
+ nodeSize
+ " nodes.</li>");
break;
}
}
} else if (eleSize < nodeSize) {
for (Element ele : selectorBarLargeElements) {
Node selectorBarPanelNode;
if (selectorBarPanel.hasNext()) {
selectorBarPanelNode = (Node) selectorBarPanel.next();
selectorBarTranslate(selectorBarPanelNode, ele, urlMap, locale);
}
}
sb.append(
"<li>Mismatch in the count of selector bar panels. Additional node(s) found. Locale page has "
+ eleSize
+ " panels and there are "
+ nodeSize
+ " nodes.</li>");
}
}
} else {
log.debug("No selector bar node found at " + indexLeftNode);
sb.append("<li>Node for selector bar component does not exist.</li>");
}
} catch (Exception e) {
sb.append("<li>Unable to update Selecotr bar large component." + e + "</li>");
}
// end set Selector bar.
// Start of button
Node textNode2 = null;
try {
textNode2 =
indexLeftNode.hasNode("gd22v2")
? indexLeftNode.getNode("gd22v2").getNode("gd22v2-right")
: null;
Element a00v1CqElement =
!doc.select("div.cl-grids").isEmpty()
? !doc.select("div.cl-grids").select("div.a00v1-cq").isEmpty()
? doc.select("div.cl-grids").select("div.a00v1-cq").first()
: null
: null;
if (a00v1CqElement == null) {
a00v1CqElement =
!doc.select("div.cl-grids").isEmpty()
? !doc.select("div.cl-grids").select("a.a00v1").isEmpty()
? doc.select("div.cl-grids").select("a.a00v1").first()
: null
: null;
}
if (textNode2 != null) {
Node buttonNode = textNode2.getNode("a00v1_cq");
if (buttonNode != null) {
if (a00v1CqElement != null) {
Element cqAnchor = a00v1CqElement.getElementsByTag("a").first();
String anchorText = cqAnchor != null ? cqAnchor.text() : "";
String anchorHref = cqAnchor.absUrl("href");
if (StringUtil.isBlank(anchorHref)) {
anchorHref = cqAnchor.attr("href");
}
// Start extracting valid href
log.debug("Before ButtonLinkUrl" + anchorHref + "\n");
anchorHref = FrameworkUtils.getLocaleReference(anchorHref, urlMap, locale, sb);
log.debug("after ButtonLinkUrl" + anchorHref + "\n");
// End extracting valid href
buttonNode.setProperty("linkText", anchorText);
buttonNode.setProperty("linkUrl", anchorHref);
a00v1CqElement.remove();
//
// doc.select("div.gd22v2-right").first().getElementsByTag("ul").first().remove();
} else {
sb.append("<li>Button is not available on the locale page.</li>");
log.debug("This button does not exist.");
}
} else {
if (a00v1CqElement != null) {
sb.append("<li>Additional button is available on the locale page.</li>");
}
}
} else {
if (a00v1CqElement != null) {
sb.append("<li>Additional button is available on the locale page.</li>");
}
}
} catch (Exception e) {
sb.append("<li>Unable to update button component." + e + "</li>");
}
// End of button
// start of text component
try {
Elements textElements = doc.select("div.gd-left").select("div.c00-pilot");
Node textNode1 =
indexLeftNode.hasNode("gd22v2")
? indexLeftNode.getNode("gd22v2").getNode("gd22v2-left").getNode("text")
: indexLeftNode.getNode("text");
if (textNode1 != null) {
if (textElements.isEmpty()) {
sb.append("<li>The first text element is not available on the locale page.</li>");
} else {
if (textElements.first().html() != null) {
textNode1.setProperty("text", textElements.first().html());
} else {
sb.append("<li>The first text element is not available on the locale page.</li>");
}
}
} else {
if (textElements.first().html() != null) {
sb.append("<li>The first text node is not available on the locale page.</li>");
}
}
if (textNode2 != null) {
Node textChildNode = textNode2.hasNode("text") ? textNode2.getNode("text") : null;
if (textElements.size() >= 2) {
if (textElements.get(1).html() != null) {
textChildNode.setProperty("text", textElements.get(1).html());
} else {
sb.append(
"<li>The second text element is not available on the locale page.</li>");
}
} else {
sb.append("<li>The second text element is not available on the locale page.</li>");
}
} else {
if (textElements.size() >= 2) {
if (textElements.get(1).html() != null) {
sb.append("<li>The second text node is not available on the locale page.</li>");
}
}
}
} catch (Exception e) {
log.error("Exception " + e);
sb.append("<li>Unable to update text components.</li>");
}
// end of text component
// start set spotlight component.
try {
Elements spotLightElements = doc.select("div.c11-pilot");
if (indexLeftNode != null) {
log.debug("Spotlight node found: " + indexLeftNode.getPath());
if (spotLightElements.isEmpty()) {
log.debug("No spot light element found with div class name c11-pilot.");
sb.append(
"<li>Spot light component with class name 'c11-pilot' does not exist on locale page.</li>");
} else {
int eleSize = spotLightElements.size();
log.debug("Spotlight node element size: " + eleSize);
NodeIterator slNodeIterator = indexLeftNode.getNodes("spotlight_large_v2*");
int nodeSize = (int) slNodeIterator.getSize();
log.debug("Spotlight node nodeSize : " + nodeSize);
if (eleSize == nodeSize) {
for (Element spElement : spotLightElements) {
Node slNode;
if (slNodeIterator.hasNext()) {
slNode = (Node) slNodeIterator.next();
spotLightTranslate(slNode, spElement, locale, urlMap);
} else {
log.debug("Next node not found");
}
}
} else if (nodeSize < eleSize) {
for (Element spElement : spotLightElements) {
Node slNode;
if (slNodeIterator.hasNext()) {
slNode = (Node) slNodeIterator.next();
spotLightTranslate(slNode, spElement, locale, urlMap);
} else {
log.debug("Next node not found");
sb.append(
"<li>Mismatch in the count of spot light panels. Additional panel(s) found on locale page. Locale page has "
+ eleSize
+ " panels and there are "
+ nodeSize
+ " nodes.</li>");
break;
}
}
} else if (nodeSize > eleSize) {
for (Element spElement : spotLightElements) {
Node slNode;
if (slNodeIterator.hasNext()) {
slNode = (Node) slNodeIterator.next();
spotLightTranslate(slNode, spElement, locale, urlMap);
} else {
log.debug("Next node not found");
}
}
sb.append(
"<li>Mismatch in the count of spot light panels. Additional node(s) found. Locale page has "
+ eleSize
+ " panels and there are "
+ nodeSize
+ " nodes.</li>");
}
}
} else {
log.debug("No spot light node found at " + indexLeftNode);
sb.append("<li>Node for spot light component does not exist.</li>");
}
} catch (Exception e) {
sb.append("<li>Unable to update Spot light component.</li>");
log.error("Exception : " + e);
}
// end set spotlight properties.
// start right rail properties
try {
boolean migrate = true;
Elements rightRailList =
doc.select("div.gd-right").select("div.mlb-pilot").select("div.c00-pilot");
if (rightRailList.isEmpty()) {
rightRailList = doc.select("div.gd-right").select("div.n13-pilot");
if (rightRailList != null) {
int eleSize = rightRailList.size();
if (eleSize == 1) {
Element rightListElem = rightRailList.first();
if (rightListElem != null) {
Elements ulElements = rightListElem.getElementsByTag("ul");
if (ulElements.size() > 1) {
sb.append(
"<li>The HTML structure for list component in right rail on the locale page is different and hence migration needs to be done manually.</li>");
migrate = false;
}
}
}
}
}
if (migrate) {
if (rightRailList.isEmpty()) {
log.debug(
"No right rail elements found with div class name c00-pilot or n13-pilot.");
sb.append(
"<li>Right rail component with class name 'c00-pilot' or 'n13-pilot' does not exist on locale page.</li>");
} else {
int eleSize = rightRailList.size();
NodeIterator listIterator = indexRightNode.getNodes("list*");
int nodeSize = (int) listIterator.getSize();
if (eleSize == nodeSize) {
for (Element rightListEle : rightRailList) {
Node listNode;
if (listIterator.hasNext()) {
listNode = (Node) listIterator.next();
rightRailList(listNode, rightListEle, urlMap, locale);
}
}
} else if (eleSize > nodeSize) {
for (Element rightListEle : rightRailList) {
Node listNode;
if (listIterator.hasNext()) {
listNode = (Node) listIterator.next();
rightRailList(listNode, rightListEle, urlMap, locale);
}
}
sb.append(
"<li>Mismatch in the count of list panels. Additional panel(s) found on locale page. Locale page has "
+ eleSize
+ " panels and there are "
+ nodeSize
+ " nodes.</li>");
} else if (eleSize < nodeSize) {
for (Element rightListEle : rightRailList) {
Node listNode;
if (listIterator.hasNext()) {
listNode = (Node) listIterator.next();
rightRailList(listNode, rightListEle, urlMap, locale);
}
}
sb.append(
"<li>Mismatch in the count of list panels. Additional node(s) found. Locale page has "
+ eleSize
+ " panels and there are "
+ nodeSize
+ " nodes.</li>");
}
}
}
} catch (Exception e) {
sb.append("<li>Unable to update benefits tile_bordered component.</li>");
log.error("Exception : ", e);
}
// end set index list.
// start of follow us component
try {
String h2Content = "";
boolean followUsEle = true;
boolean followUsNode = true;
List<String> list = new ArrayList<String>();
Elements rightRailPilotElements = doc.select("div.gd-right").select("div.s14-pilot");
if (rightRailPilotElements != null) {
Element rightRailPilotElement = rightRailPilotElements.first();
if (rightRailPilotElement != null) {
Elements h2Elements = rightRailPilotElement.getElementsByTag("h2");
if (h2Elements != null) {
Element h2Element = h2Elements.first();
h2Content = h2Element.text();
} else {
sb.append("<li>h2 of right rail with class 'div.s14-pilot' is blank.</li>");
}
Elements liElements = rightRailPilotElement.getElementsByTag("li");
for (Element ele : liElements) {
JSONObject obj = new JSONObject();
String icon = ele.attr("class");
obj.put("icon", icon);
Elements aElements = ele.getElementsByTag("a");
if (aElements != null) {
Element aElement = aElements.first();
String title = aElement.attr("title");
String href = aElement.absUrl("href");
if (StringUtil.isBlank(href)) {
href = aElement.attr("href");
}
// Start extracting valid href
log.debug("Before pilotLinkUrl" + href + "\n");
href = FrameworkUtils.getLocaleReference(href, urlMap, locale, sb);
log.debug("after pilotLinkUrl" + href + "\n");
// End extracting valid href
obj.put("linktext", title);
obj.put("linkurl", href);
} else {
sb.append("<li>No anchor tag found in the right rail social links</li>");
}
list.add(obj.toString());
}
} else {
followUsEle = false;
}
} else {
followUsEle = false;
}
if (indexRightNode.hasNode("followus")) {
Node followus = indexRightNode.getNode("followus");
if (StringUtils.isNotBlank(h2Content)) {
followus.setProperty("title", h2Content);
}
if (list.size() > 1) {
followus.setProperty("links", list.toArray(new String[list.size()]));
}
} else {
followUsNode = false;
}
if (followUsEle) {
if (followUsNode) {
log.debug("Follow us element and node are available.");
} else {
sb.append(
"<li>Additional follow us element found in right rail on the locale page.</li>");
}
} else {
if (followUsNode) {
sb.append("<li>No follow us element found in right rail on the locale page.</li>");
} else {
log.debug("Follow us element and node are not available.");
}
}
} catch (Exception e) {
sb.append("<li>Unable to update follow us component.</li>");
log.error("Exception : ", e);
}
// end of follow us component
// start of image
Elements aEle =
!doc.select("div.gd12-pilot").select("div.gd-right").isEmpty()
? doc.select("div.gd12-pilot").select("div.gd-right").select("a")
: null;
log.debug("Image check:" + aEle);
if (!aEle.isEmpty() && aEle != null) {
for (Element aEl : aEle) {
Element imgEle = aEl.getElementsByTag("img").first();
if (imgEle != null) {
sb.append("<li>Extra Image found in web page.</li>");
}
}
}
/*if(aEle != null && !aEle.isEmpty() && aEle.size() > 1){
sb.append("<li>Extra Image found in web page.</li>");
}*/
// end of image
// start of c42-pilot component
Elements c42Ele = doc.select("div.c42-pilot");
if (!c42Ele.isEmpty()) {
sb.append("<li>Extra component mbwtile found on web page</li>");
}
// end of c42-pilot component
} else {
sb.append(Constants.URL_CONNECTION_EXCEPTION);
}
} catch (Exception e) {
// doc = getConnection(loc);
log.error("Exception : ", e);
}
} catch (Exception e) {
sb.append("<li>Exception " + e + "</li>");
}
sb.append("</ul></td>");
session.save();
return sb.toString();
}
/**
* Set the element's {@code class} attribute to the supplied class names.
*
* @param classNames set of classes
* @return this element, for chaining
*/
public Element classNames(Set<String> classNames) {
Validate.notNull(classNames);
attributes.put("class", StringUtil.join(classNames, " "));
return this;
}
// hit when all of the node's children (if any) have been visited
public void tail(Node node, int depth) {
String name = node.nodeName();
if (name.equals("br")) append("\n");
else if (StringUtil.in(name, "p", "h1", "h2", "h3", "h4", "h5")) append("\n\n");
else if (name.equals("a")) append(String.format(" <%s>", node.absUrl("href")));
}
private void migrateHeroLarge(
Document doc, Node trainingAndEventsLeftNode, String locale, Map<String, String> urlMap) {
// TODO Auto-generated method stub
try {
String h2Text = "";
String pText = "";
String aText = "";
String aHref = "";
int pCount = 0;
Node heroPanelNode = null;
Elements heroElements = doc.select("div.c50-pilot");
if (heroElements.size() > 0) {
if (!heroElements.select("div.frame").isEmpty()) {
heroElements = heroElements.select("div.frame");
}
Node heroNode =
trainingAndEventsLeftNode.hasNode("hero_large")
? trainingAndEventsLeftNode.getNode("hero_large")
: null;
if (heroNode != null) {
NodeIterator heroPanelNodeIterator =
heroNode.hasNode("heropanel_0") ? heroNode.getNodes("heropanel*") : null;
if (heroPanelNodeIterator != null) {
if (heroElements != null) {
int eleSize = heroElements.size();
int nodeSize = (int) heroPanelNodeIterator.getSize();
if (eleSize != nodeSize) {
log.debug("Hero component node count mismatch!");
sb.append(
"<li>Hero Component count mis match. Elements on page are: "
+ eleSize
+ " Node Count is: "
+ nodeSize
+ "</li>");
}
for (Element ele : heroElements) {
if (heroPanelNodeIterator.hasNext()) {
heroPanelNode = (Node) heroPanelNodeIterator.next();
Elements h2TagText = ele.getElementsByTag("h2");
if (h2TagText != null) {
h2Text = h2TagText.text();
heroPanelNode.setProperty("title", h2Text);
} else {
sb.append(Constants.HERO_CONTENT_HEADING_ELEMENT_DOESNOT_EXISTS);
}
Elements descriptionText = ele.getElementsByTag("p");
if (descriptionText != null) {
if (descriptionText.size() > 2) {
for (Element desEle : descriptionText) {
pCount++;
pText = pText + desEle.text();
if (pCount == 2) {
break;
}
}
} else {
pText = descriptionText.first().text();
}
heroPanelNode.setProperty("description", pText);
} else {
sb.append(Constants.HERO_CONTENT_DESCRIPTION_ELEMENT_DOESNOT_EXISTS);
}
Element anchorText = ele.getElementsByTag("a").first();
if (anchorText != null) {
aText = anchorText.text();
aHref = anchorText.absUrl("href");
if (StringUtil.isBlank(aHref)) {
aHref = anchorText.attr("href");
}
// Start extracting valid href
log.debug("Before heroPanelLinkUrl" + aHref + "\n");
aHref = FrameworkUtils.getLocaleReference(aHref, urlMap, locale, sb);
log.debug("after heroPanelLinkUrl" + aHref + "\n");
// End extracting valid href
heroPanelNode.setProperty("linktext", aText);
heroPanelNode.setProperty("linkurl", aHref);
} else {
sb.append(Constants.HERO_CONTENT_ANCHOR_TEXT_IS_BLANK);
}
// start image
String heroImage = FrameworkUtils.extractImagePath(ele, sb);
log.debug("heroImage before migration : " + heroImage + "\n");
if (heroPanelNode.hasNode("image")) {
Node imageNode = heroPanelNode.getNode("image");
String fileReference =
imageNode.hasProperty("fileReference")
? imageNode.getProperty("fileReference").getString()
: "";
heroImage =
FrameworkUtils.migrateDAMContent(heroImage, fileReference, locale, sb);
log.debug("heroImage after migration : " + heroImage + "\n");
if (StringUtils.isNotBlank(heroImage)) {
imageNode.setProperty("fileReference", heroImage);
}
} else {
sb.append("<li>hero image node doesn't exist</li>");
}
// end image
}
}
} else {
sb.append(Constants.HERO_CONTENT_PANEL_ELEMENT_NOT_FOUND);
}
} else {
log.debug("hero panel node is not found");
}
} else {
if (heroElements.isEmpty()) {
log.debug("Hero Large node and elements are not found");
} else {
sb.append(Constants.HERO_NODE_NOT_AVAILABLE);
}
}
} else {
sb.append("<li>Hero component not found in web url</li>");
}
} catch (Exception e) {
sb.append(Constants.EXCEPTOIN_IN_UPDATING_HERO_CONTENT);
log.error("hero Error" + e);
}
}
private static void appendNormalisedText(StringBuilder accum, TextNode textNode) {
String text = textNode.getWholeText();
if (preserveWhitespace(textNode.parentNode())) accum.append(text);
else StringUtil.appendNormalisedWhitespace(accum, text, lastCharIsWhitespace(accum));
}
/**
* 对用户输入内容进行过滤
*
* @param html
* @return
*/
public static String filter(String html) {
return StringUtil.isBlank(html) ? "" : Jsoup.clean(html, content_filter);
}
private void setListElements(
Element ele, Node rightListNode, Session session, String locale, Map<String, String> urlMap) {
try {
String ownPdfText = "";
String pdfIcon = "";
String pdfSize = "";
Elements h2Ele = ele.getElementsByTag("h2");
Elements h3Ele = ele.getElementsByTag("h3");
Elements ulEle = ele.getElementsByTag("ul");
String h2Text = null;
String h3Text = null;
// start of handling title of list component
if (!h2Ele.isEmpty()) {
h2Text = h2Ele.first().text();
rightListNode.setProperty("title", h2Text);
if (h2Ele.size() > 1) {
sb.append(Constants.MISMATCH_IN_RIGHT_LIST_COUNT);
}
} else {
sb.append(Constants.LIST_HEADING_COMPONENT_NOT_FOUND);
}
// end of handling title of list component
// start of handling title of list component
NodeIterator h3Iterator =
rightListNode.hasNode("element_subtitle_0")
? rightListNode.getNodes("element_subtitle*")
: null;
if (h3Iterator != null) {
if (!h3Ele.isEmpty()) {
int eleSize = h3Ele.size();
int nodeSize = (int) h3Iterator.getSize();
Node h3nodeList;
if (eleSize == nodeSize) {
for (Element h3Itr : h3Ele) {
h3nodeList = (Node) h3Iterator.next();
h3Text = h3Itr.text();
h3nodeList.setProperty("subtitle", h3Text);
}
}
if (nodeSize < eleSize) {
for (Element h3Itr : h3Ele) {
if (h3Iterator.hasNext()) {
h3nodeList = (Node) h3Iterator.next();
h3Text = h3Itr.text();
h3nodeList.setProperty("subtitle", h3Text);
}
}
sb.append(
Constants.MISMATCH_IN_LIST_ELEMENT
+ nodeSize
+ Constants.SPOTLIGHT_ELEMENT_COUNT
+ eleSize);
}
if (nodeSize > eleSize) {
for (Element h3Itr : h3Ele) {
h3nodeList = (Node) h3Iterator.next();
h3Text = h3Itr.text();
h3nodeList.setProperty("subtitle", h3Text);
}
sb.append(
Constants.LIST_ELEMENTS_COUNT_MISMATCH
+ nodeSize
+ Constants.SPOTLIGHT_ELEMENT_COUNT
+ eleSize);
}
} else {
sb.append(Constants.LIST_HEADING_COMPONENT_NOT_FOUND);
log.debug("h3 text is not avalable");
}
} else {
if (!h3Ele.isEmpty()) {
log.debug("subtitle node doesnot exist but ele exist");
sb.append(Constants.LIST_COMPONENT_NOT_FOUND);
}
}
// end of handling title of list component
// Element List
NodeIterator ulNodeIterator =
rightListNode.hasNode("element_list_0") ? rightListNode.getNodes("element_list*") : null;
if (ulNodeIterator != null) {
Node ulnodeList;
for (Element ulItr : ulEle) {
if (ulNodeIterator.hasNext()) {
ulnodeList = (Node) ulNodeIterator.next();
Elements list = ulItr.getElementsByTag("li");
List<String> listAdd = new ArrayList<String>();
for (Element li : list) {
pdfIcon = "";
pdfSize = "";
boolean openNewWindow = false;
// pdf content
try {
ownPdfText = li.ownText();
if (StringUtils.isNotEmpty(ownPdfText)) {
log.debug("OWn text is:" + ownPdfText);
if (ownPdfText.toLowerCase().contains("pdf")
|| ownPdfText.toLowerCase().contains("video")) {
pdfIcon = "pdf";
if (ownPdfText.toLowerCase().contains("video")) {
pdfIcon = "video";
}
int i = 0;
for (; i < ownPdfText.length(); i++) {
char character = ownPdfText.charAt(i);
boolean isDigit = Character.isDigit(character);
if (isDigit) {
break;
}
}
pdfSize = ownPdfText.substring(i, ownPdfText.length() - 1);
pdfSize = pdfSize.replace(")", "");
pdfSize = pdfSize.trim();
}
}
} catch (Exception e) {
sb.append(Constants.Exception_BY_SPECIAL_CHARACTER);
log.error("Exception : ", e);
}
// fix for new win icon
Elements newwinCheck = li.select("span.newwin");
if (!newwinCheck.isEmpty()) {
log.debug("extra new win icon found");
sb.append(Constants.EXTRA_ICON_FOUND_IN_LIST);
}
// check for the lock icon
Elements imgInList = li.getElementsByTag("img");
if (!imgInList.isEmpty()) {
String altImg = imgInList.attr("alt");
if (altImg.equals("lock_icon")) {
log.debug("lock icon found in the list");
sb.append(Constants.EXTRA_LOCK_IMG_FOUND_IN_LIST);
}
}
if (!li.getElementsByTag("a").isEmpty()) {
Element a = li.getElementsByTag("a").first();
String aHref = a.absUrl("href");
if (StringUtil.isBlank(aHref)) {
aHref = a.attr("href");
}
// Start extracting valid href
log.debug("Before anchorHref" + a.absUrl("href") + "\n");
String anchorHref = FrameworkUtils.getLocaleReference(aHref, urlMap, locale, sb);
log.debug("after anchorHref" + anchorHref + "\n");
// End extracting valid href
JSONObject obj = new JSONObject();
obj.put("linktext", a.text());
obj.put("linkurl", anchorHref);
obj.put("icon", pdfIcon);
obj.put("size", pdfSize);
obj.put("description", "");
obj.put("openInNewWindow", openNewWindow);
listAdd.add(obj.toString());
}
}
ulnodeList.setProperty("listitems", listAdd.toArray(new String[listAdd.size()]));
}
}
if (ulNodeIterator.hasNext()) {
sb.append(Constants.MISMATCH_IN_RIGHT_LIST_COUNT);
}
} else {
sb.append(Constants.NO_LIST_NODES_FOUND);
}
// End of Element List
} catch (Exception e) {
sb.append(Constants.UNABLE_TO_MIGRATE_LIST_COMPONENT);
log.error("Exception : ", e);
}
}
