protected void checkHRef(XMLElement e, String attrNS, String attr) {
String href = e.getAttributeNS(attrNS, attr);
if (href == null) {
return;
}
href = href.trim();
if (href.isEmpty()) {
// if href="" then selfreference which is valid,
// but as per issue 225, issue a hint
report.message(
MessageId.HTM_045,
EPUBLocation.create(path, parser.getLineNumber(), parser.getColumnNumber(), href));
return;
} else if (href.contains("#epubcfi")) {
return; // temp until cfi implemented
} else if (".".equals(href)) {
// selfreference, no need to check
return;
}
URI uri = checkURI(href);
if (uri == null) return;
if ("http".equals(uri.getScheme())) {
report.info(path, FeatureEnum.REFERENCE, href);
}
/*
* mgy 20120417 adding check for base to initial if clause as part of
* solution to issue 155
*/
if (URISchemes.contains(uri.getScheme())
|| (null != base && URISchemes.contains(base.getScheme()))) {
return;
}
// This if statement is needed to make sure XML Fragment identifiers
// are not reported as non-registered URI scheme types
else if (uri.getScheme() != null) {
report.message(
MessageId.HTM_025,
EPUBLocation.create(path, parser.getLineNumber(), parser.getColumnNumber(), href));
return;
}
try {
href = PathUtil.resolveRelativeReference(path, href, base == null ? null : base.toString());
} catch (IllegalArgumentException err) {
report.message(
MessageId.OPF_010,
EPUBLocation.create(path, parser.getLineNumber(), parser.getColumnNumber(), href),
err.getMessage());
return;
}
processHyperlink(href);
}
public void runChecks() {
OCFPackage ocf = context.ocf.get();
if (!ocf.hasEntry(path)) {
report.message(MessageId.RSC_001, EPUBLocation.create(ocf.getName()), path);
} else if (!ocf.canDecrypt(path)) {
report.message(MessageId.RSC_004, EPUBLocation.create(ocf.getName()), path);
} else if (!path.endsWith(".xml")) {
report.message(MessageId.OPF_080, EPUBLocation.create(path));
} else {
validate();
}
}
public void endElement() {
openElements--;
XMLElement e = parser.getCurrentElement();
String ns = e.getNamespace();
String name = e.getName();
if (openElements == 0) {
report.info(path, FeatureEnum.CHARS_COUNT, Long.toString(charsCount));
if (!epubTypeInUse) {
if (context.version == EPUBVersion.VERSION_3) {
report.message(MessageId.ACC_007, EPUBLocation.create(path));
}
} else {
epubTypeInUse = false;
}
}
ElementLocation currentLocation = elementLocationStack.pop();
if (EpubConstants.HtmlNamespaceUri.equals(ns)) {
if ("script".equals(name)) {
String attr = e.getAttribute("type");
report.info(path, FeatureEnum.HAS_SCRIPTS, (attr == null) ? "" : attr);
} else if ("style".equals(name)) {
String style = textNode.toString();
if (style.length() > 0) {
CSSCheckerFactory.getInstance()
.newInstance(context, style, currentLocation.getLineNumber(), false)
.runChecks();
}
textNode = null;
} else if ("head".equals(name)) {
checkStylesheetFallback();
} else if ("table".equals(name)) {
if (tableDepth > 0) {
--tableDepth;
EPUBLocation location =
EPUBLocation.create(
path,
currentLocation.getLineNumber(),
currentLocation.getColumnNumber(),
"table");
checkDependentCondition(MessageId.ACC_005, tableDepth == 0, hasTh, location);
checkDependentCondition(MessageId.ACC_006, tableDepth == 0, hasThead, location);
checkDependentCondition(MessageId.ACC_012, tableDepth == 0, hasCaption, location);
hasTh = hasThead = hasCaption = false;
}
}
}
}
protected URI checkURI(String uri) {
try {
return new URI(Preconditions.checkNotNull(uri).trim());
} catch (URISyntaxException e) {
report.message(MessageId.RSC_020, parser.getLocation(), uri);
return null;
}
}
@Override
public boolean validate() {
SearchDictionary validTypes =
new SearchDictionary(SearchDictionary.DictionaryType.SVG_MEDIA_TYPES);
boolean isGlobalFixed = EpubPackage.isGlobalFixed(this.epack);
Hashtable<String, SpineItem> spineItems = new Hashtable<String, SpineItem>();
for (int i = 0; i < epack.getSpine().itemsLength(); ++i) {
SpineItem si = epack.getSpine().getItem(i);
spineItems.put(si.getIdref(), si);
}
for (int i = 0; i < epack.getManifest().itemsLength(); i++) {
ManifestItem itemEntry = epack.getManifest().getItem(i);
if (validTypes.isValidMediaType(itemEntry.getMediaType())) {
String fileToParse = epack.getManifestItemFileName(itemEntry);
ZipEntry entry = epack.getZip().getEntry(fileToParse);
if (entry == null) {
report.message(
MessageId.RSC_001, new MessageLocation(epack.getFileName(), -1, -1), fileToParse);
continue;
}
SpineItem si = spineItems.get(itemEntry.getId());
boolean itemIsFixedFormat = isGlobalFixed;
if (si != null) {
String properties = si.getProperties();
if (properties != null) {
if (properties.length() != 0) {
properties = properties.replaceAll("[\\s]+", " ");
String propertyArray[] = properties.split(" ");
for (String prop : propertyArray) {
if (prop.equals("rendition:layout-pre-paginated")) {
itemIsFixedFormat = true;
} else if (prop.equals("rendition:layout-reflowable")) {
itemIsFixedFormat = false;
}
}
}
}
}
if (itemIsFixedFormat) {
checkSvgDoc(fileToParse);
}
}
}
return true;
}
protected void checkStylesheetFallback() {
// stylesheet is considered as having "built-in" fallback if
// at least one is found with a blessed CMT (i.e. text/css).
// Implem note: xrefChecker is necessarily present if
// nonStandardStylesheetLink is present.
if (nonStandardStylesheetLink.isPresent()
&& !hasCss
&& !xrefChecker
.get()
.hasValidFallback(nonStandardStylesheetLink.get().getContext().get())
.or(false)) {
report.message(MessageId.CSS_010, nonStandardStylesheetLink.get());
}
}
void checkViewBox(String svgDocEntry, Document doc) {
NodeList n = doc.getElementsByTagNameNS(svgNS, "svg");
for (int i = 0; i < n.getLength(); i++) {
Element svgElement = (Element) n.item(i);
String viewport = svgElement.getAttributeNS(svgNS, "viewBox");
if (viewport == null || viewport.length() == 0) {
report.message(
MessageId.HTM_048,
new MessageLocation(
svgDocEntry,
XmlDocParser.getElementLineNumber(svgElement),
XmlDocParser.getElementColumnNumber(svgElement)));
}
}
}
void checkImageXlinkHrefInline(String svgDocEntry, Document doc) {
NodeList n = doc.getElementsByTagNameNS(svgNS, "image");
for (int i = 0; i < n.getLength(); i++) {
Element svgElement = (Element) n.item(i);
String href = svgElement.getAttributeNS(xlinkNS, "href");
if (href != null && href.length() > 0) {
if (!href.startsWith("data:image")) {
report.message(
MessageId.MED_006,
new MessageLocation(
svgDocEntry,
XmlDocParser.getElementLineNumber(svgElement),
XmlDocParser.getElementColumnNumber(svgElement)));
}
}
}
}
private void processRef(String ref, XRefChecker.Type type) {
if (ref != null && context.xrefChecker.isPresent()) {
ref = PathUtil.resolveRelativeReference(path, ref, null);
if (type == XRefChecker.Type.AUDIO) {
String mimeType = context.xrefChecker.get().getMimeType(ref);
if (mimeType != null && !OPFChecker30.isBlessedAudioType(mimeType)) {
report.message(
MessageId.MED_005,
EPUBLocation.create(path, parser.getLineNumber(), parser.getColumnNumber()),
ref,
mimeType);
}
}
context
.xrefChecker
.get()
.registerReference(path, parser.getLineNumber(), parser.getColumnNumber(), ref, type);
}
}
public boolean validate() {
int fatalErrorsSoFar = report.getFatalErrorCount();
int errorsSoFar = report.getErrorCount();
int warningsSoFar = report.getWarningCount();
SearchKeyMapHandler handler;
XMLParser parser = new XMLParser(context);
handler = new SearchKeyMapHandler(context, parser);
parser.addValidator(XMLValidators.SEARCH_KEY_MAP_RNC.get());
parser.addXMLHandler(handler);
parser.process();
return fatalErrorsSoFar == report.getFatalErrorCount()
&& errorsSoFar == report.getErrorCount()
&& warningsSoFar == report.getWarningCount();
}
public void endElement() {
XMLElement e = parser.getCurrentElement();
if (e.getNamespace().equals("http://purl.org/dc/elements/1.1/")) {
String name = e.getName();
if (name.equals("identifier")) {
String idAttr = e.getAttribute("id");
if (idAttr != null && !idAttr.equals("") && idAttr.equals(uniqueIdent)) {
// String idval = (String) e.getPrivateData();
// if (idval != null && ocf != null)
// ocf.setUniqueIdentifier(idval);
}
} else if (name.equals("date")) {
String dateval = (String) e.getPrivateData();
boolean valid = true;
String detail = null;
if (dateval == null || "".equals(dateval)) {
valid = false;
detail = "zero-length string";
} else {
DateParser dateParser = new DateParser();
try {
Date date = dateParser.parse(dateval.trim());
/*
* mg: DateParser does not enforce four-digit years,
* which http://www.w3.org/TR/NOTE-datetime seems to want
*/
String year = new SimpleDateFormat("yyyy").format(date);
if (year.length() > 4) throw new InvalidDateException(year);
} catch (InvalidDateException d) {
valid = false;
detail = d.getMessage();
}
}
if (!valid) {
if (this.version == EPUBVersion.VERSION_3) {
report.warning(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
"date value '"
+ (dateval == null ? "" : dateval)
+ "' does not follow recommended syntax as per http://www.w3.org/TR/NOTE-datetime:"
+ detail);
} else {
report.error(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
"date value '"
+ (dateval == null ? "" : dateval)
+ "' is not valid as per http://www.w3.org/TR/NOTE-datetime:"
+ detail);
}
}
} else if (name.equals("title") || name.equals("language")) {
// issue 138: issue a warning if dc:title and dc:language is empty for 2.0 and 2.0.1
// note that an empty dc:identifier is checked in opf20.rng and will
// therefore be reported as an error, that may or may not be a good idea.
if (version == EPUBVersion.VERSION_2) {
String value = (String) e.getPrivateData();
if (value == null || value.trim().length() < 1) {
report.warning(
path, parser.getLineNumber(), parser.getColumnNumber(), name + " element is empty");
}
}
}
}
}
public void startElement() {
boolean registerEntry = true;
XMLElement e = parser.getCurrentElement();
String ns = e.getNamespace();
if (ns == null
|| ns.equals("")
|| ns.equals("http://openebook.org/namespaces/oeb-package/1.0/")
|| ns.equals("http://www.idpf.org/2007/opf")) {
String name = e.getName();
if (name.equals("package")) {
if (!ns.equals("http://www.idpf.org/2007/opf")) {
report.warning(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
"OPF file is using OEBPS 1.2 syntax allowing backwards compatibility");
opf12PackageFile = true;
}
/*
* This section checks to see the value of the unique-identifier
* attribute and stores it in the String uniqueIdent or reports
* an error if the unique-identifier attribute is missing or
* does not have a value
*/
String uniqueIdentAttr = e.getAttribute("unique-identifier");
if (uniqueIdentAttr != null && !uniqueIdentAttr.equals("")) {
uniqueIdent = uniqueIdentAttr;
} else {
report.error(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
"unique-identifier attribute in package element must be present and have a value");
}
} else if (name.equals("item")) {
String id = e.getAttribute("id");
String href = e.getAttribute("href");
if (href != null && !(version == EPUBVersion.VERSION_3 && href.startsWith("http://"))) {
try {
href = PathUtil.resolveRelativeReference(path, href, null);
} catch (IllegalArgumentException ex) {
report.error(path, parser.getLineNumber(), parser.getColumnNumber(), ex.getMessage());
href = null;
}
}
String mimeType = e.getAttribute("media-type");
String fallback = e.getAttribute("fallback");
String fallbackStyle = e.getAttribute("fallback-style");
String namespace = e.getAttribute("island-type");
String properties = e.getAttribute("properties");
if (properties != null) properties = properties.replaceAll("[\\s]+", " ");
if (version == EPUBVersion.VERSION_3
&& href.startsWith("http://")
&& !OPFChecker30.isBlessedAudioType(mimeType))
if (OPFChecker30.isCoreMediaType(mimeType)) {
report.error(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
"Only audio and video remote resources are permitted");
} else {
// mgy 20120414: this shouldn't even be a warning
// report.warning(
// path,
// parser.getLineNumber(),
// parser.getColumnNumber(),
// "Remote resource not validated");
}
OPFItem item =
new OPFItem(
id,
href,
mimeType,
fallback,
fallbackStyle,
namespace,
properties,
parser.getLineNumber(),
parser.getColumnNumber());
if (id != null) itemMapById.put(id, item);
if (properties != null) {
String propertyArray[] = properties.split(" ");
for (int i = 0; i < propertyArray.length; i++) {
if (propertyArray[i].equals("nav")) item.setNav(true);
if (propertyArray[i].equals("scripted")) item.setScripted(true);
}
}
if (href != null && registerEntry) {
itemMapByPath.put(href, item);
items.add(item);
}
} else if (name.equals("reference")) {
String type = e.getAttribute("type");
String title = e.getAttribute("title");
String href = e.getAttribute("href");
if (href != null && xrefChecker != null) {
try {
href = PathUtil.resolveRelativeReference(path, href, null);
xrefChecker.registerReference(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
href,
XRefChecker.RT_GENERIC);
} catch (IllegalArgumentException ex) {
report.error(path, parser.getLineNumber(), parser.getColumnNumber(), ex.getMessage());
href = null;
}
}
OPFReference ref =
new OPFReference(type, title, href, parser.getLineNumber(), parser.getColumnNumber());
refs.add(ref);
} else if (name.equals("spine")) {
String idref = e.getAttribute("toc");
if (idref != null) {
toc = (OPFItem) itemMapById.get(idref);
if (toc == null)
report.error(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
"item with id '" + idref + "' not found");
else {
toc.setNcx(true);
if (toc.getMimeType() != null && !toc.getMimeType().equals("application/x-dtbncx+xml"))
report.error(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
"toc attribute references resource with non-NCX mime type; \"application/x-dtbncx+xml\" is expected");
}
}
} else if (name.equals("itemref")) {
String idref = e.getAttribute("idref");
if (idref != null) {
OPFItem item = getItemById(idref);
if (item != null) {
spine.add(item);
item.setInSpine(true);
String linear = e.getAttribute("linear");
if (linear != null && "no".equals(linear.trim())) {
item.setSpineLinear(false);
} else {
item.setSpineLinear(true);
}
} else {
report.error(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
"item with id '" + idref + "' not found");
}
}
} else if (name.equals("dc-metadata") || name.equals("x-metadata")) {
if (!opf12PackageFile)
report.error(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
"use of deprecated element '" + name + "'");
}
} else if (ns.equals("http://purl.org/dc/elements/1.1/")) {
// in the DC metadata, when the <identifier> element is parsed, if
// it has a non-null and non-empty id attribute value that is the
// same as the value of the unique-identifier attribute of the
// package element, set uniqueIdentExists = true (to make sure that
// the unique-identifier attribute references an existing
// <identifier> id attribute
String name = e.getName();
if (name.equals("identifier")) {
String idAttr = e.getAttribute("id");
if (idAttr != null && !idAttr.equals("") && idAttr.equals(uniqueIdent))
uniqueIdentExists = true;
} else if (name.equals("creator")) {
String role = e.getAttributeNS("http://www.idpf.org/2007/opf", "role");
if (role != null && !role.equals("")) {
if (!isValidRole(role))
report.error(
path,
parser.getLineNumber(),
parser.getColumnNumber(),
"role value '" + role + "' is not valid");
}
}
}
}
// Report the message id when primary condition1 is true but dependent
// condition2 is false.
protected void checkDependentCondition(
MessageId id, boolean condition1, boolean condition2, EPUBLocation location) {
if (condition1 && !condition2) {
report.message(id, location);
}
}
protected void checkIFrame(XMLElement e) {
report.message(
MessageId.HTM_036,
EPUBLocation.create(path, parser.getLineNumber(), parser.getColumnNumber(), e.getName()));
}