private void buildCases() throws Exception {
List<Element> cases = INDEX_CASES.get().evalAsNativeElementList(doc);
for (Element caseEl : cases) {
WildcardPattern matchNamespace = null;
WildcardPattern matchName = null;
String matchNamespaceAttr = DocumentHelper.getAttribute(caseEl, "matchNamespace", false);
if (matchNamespaceAttr != null) {
// If the matchNamespace attr does not contain a wildcard expression, and its value
// happens to be an existing namespace prefix, than substitute the prefix for the full URI.
if (!WildcardPattern.isWildcardExpression(matchNamespaceAttr)) {
String uri = caseEl.lookupNamespaceURI(matchNamespaceAttr);
if (uri != null) matchNamespaceAttr = uri;
}
matchNamespace = new WildcardPattern(matchNamespaceAttr);
}
String matchNameAttr = DocumentHelper.getAttribute(caseEl, "matchName", false);
if (matchNameAttr != null) {
matchName = new WildcardPattern(matchNameAttr);
}
String vtagsSpec = DocumentHelper.getAttribute(caseEl, "vtags", false);
Map<String, String> varPropsPattern = parseVariantPropertiesPattern(caseEl);
Set<SchemaId> vtags = parseVersionTags(vtagsSpec);
IndexCase indexCase = new IndexCase(matchNamespace, matchName, varPropsPattern, vtags);
conf.addIndexCase(indexCase);
}
}
private Value buildValue(Element fieldEl) throws Exception {
String valueExpr = DocumentHelper.getAttribute(fieldEl, "value", true);
Value value;
boolean extractContent = DocumentHelper.getBooleanAttribute(fieldEl, "extractContent", false);
String formatter = DocumentHelper.getAttribute(fieldEl, "formatter", false);
if (formatter != null && !conf.getFormatters().hasFormatter(formatter)) {
throw new IndexerConfException(
"Formatter does not exist: "
+ formatter
+ " at "
+ LocationAttributes.getLocationString(fieldEl));
}
//
// An index field can basically map to two kinds of values:
// * plain field values
// * dereference expressions (following links to some other record and then taking a field
// value from it)
//
// A dereference expression is specified as "somelink=>somelink=>somefield"
if (valueExpr.contains("=>")) {
//
// A dereference field
//
value = buildDerefValue(fieldEl, valueExpr, extractContent, formatter);
} else {
//
// A plain field
//
value = new FieldValue(getFieldType(valueExpr, fieldEl), extractContent, formatter);
}
if (extractContent
&& !value
.getTargetFieldType()
.getValueType()
.getDeepestValueType()
.getBaseName()
.equals("BLOB")) {
throw new IndexerConfException(
"extractContent is used for a non-blob value at "
+ LocationAttributes.getLocation(fieldEl));
}
return value;
}
private Map<String, String> parseVariantPropertiesPattern(Element caseEl) throws Exception {
String variant = DocumentHelper.getAttribute(caseEl, "matchVariant", false);
Map<String, String> varPropsPattern = new HashMap<String, String>();
if (variant == null) return varPropsPattern;
for (String prop : COMMA_SPLITTER.split(variant)) {
int eqPos = prop.indexOf("=");
if (eqPos != -1) {
String propName = prop.substring(0, eqPos);
String propValue = prop.substring(eqPos + 1);
if (propName.equals("*")) {
throw new IndexerConfException(
String.format(
"Error in matchVariant attribute: the character '*' "
+ "can only be used as wildcard, not as variant dimension name, attribute = %1$s, at: %2$s",
variant, LocationAttributes.getLocation(caseEl)));
}
varPropsPattern.put(propName, propValue);
} else {
varPropsPattern.put(prop, null);
}
}
return varPropsPattern;
}
public static IndexerConf build(InputStream is, Repository repository)
throws IndexerConfException {
Document doc;
try {
doc = DocumentHelper.parse(is);
} catch (Exception e) {
throw new IndexerConfException("Error parsing supplied configuration.", e);
}
return new IndexerConfBuilder().build(doc, repository);
}
private void buildFormatters() throws Exception {
List<Element> formatters = FORMATTERS.get().evalAsNativeElementList(doc);
for (Element formatterEl : formatters) {
String className = DocumentHelper.getAttribute(formatterEl, "class", true);
Formatter formatter = instantiateFormatter(className);
String name = DocumentHelper.getAttribute(formatterEl, "name", true);
if (name != null && conf.getFormatters().hasFormatter(name)) {
throw new IndexerConfException("Duplicate formatter name: " + name);
}
conf.getFormatters().addFormatter(formatter, name);
}
String defaultFormatter = XPathUtils.evalString("/indexer/formatters/@default", doc);
if (defaultFormatter.length() != 0) {
conf.getFormatters().setDefaultFormatter(defaultFormatter);
}
}
private void buildIndexFields() throws Exception {
List<Element> indexFields = INDEX_FIELDS.get().evalAsNativeElementList(doc);
for (Element indexFieldEl : indexFields) {
String name = DocumentHelper.getAttribute(indexFieldEl, "name", true);
validateName(name);
Value value = buildValue(indexFieldEl);
IndexField indexField = new IndexField(name, value);
conf.addIndexField(indexField);
}
}
@Override
public int run(CommandLine cmd) throws Exception {
int result = super.run(cmd);
if (result != 0) {
return result;
}
String collection = cmd.getOptionValue(collectionOption.getOpt());
if (collection == null) {
collection = DEFFAULT_COLLECTION;
}
String branch = cmd.getOptionValue(branchOption.getOpt());
if (branch == null) {
branch = DEFFAULT_BRANCH;
}
RepositoryManager repositoryManager =
new RemoteRepositoryManager(
"http://lilyproject.org:9263", new Credentials("guest", "guest"));
Repository repository = repositoryManager.getRepository(new Credentials("guest", "guest"));
QueryManager queryManager = repository.getQueryManager();
RepositorySchema schema = repository.getRepositorySchema();
String query =
"select id, name where InCollection('" + collection + "') and branch = '" + branch + "'";
VariantKey[] keys = queryManager.performQueryReturnKeys(query, Locale.getDefault());
for (VariantKey key : keys) {
Document doc = repository.getDocument(key, false);
Version version = doc.getLiveVersion();
if (version == null) {
continue;
}
for (Part part : version.getParts().getArray()) {
if (schema.getPartTypeById(part.getTypeId(), false).isDaisyHtml()) {
InputStream is = part.getDataStream();
org.w3c.dom.Document domDoc = DocumentHelper.parse(is);
NodeList links = XPathUtils.evalNodeList("//a/@href", domDoc.getDocumentElement());
for (int j = 0; j < links.getLength(); j++) {
String link = ((Attr) links.item(j)).getValue();
System.out.printf(
"[%1$10.10s][%2$30.30s][%3$10.10s] %4$s\n",
key.getDocumentId(), version.getDocumentName(), part.getTypeName(), link);
}
is.close();
}
}
}
return 0;
}
private void buildDynamicFields() throws Exception {
List<Element> fields = DYNAMIC_INDEX_FIELDS.get().evalAsNativeElementList(doc);
for (Element fieldEl : fields) {
String matchNamespaceAttr = DocumentHelper.getAttribute(fieldEl, "matchNamespace", false);
String matchNameAttr = DocumentHelper.getAttribute(fieldEl, "matchName", false);
String matchTypeAttr = DocumentHelper.getAttribute(fieldEl, "matchType", false);
String matchScopeAttr = DocumentHelper.getAttribute(fieldEl, "matchScope", false);
String nameAttr = DocumentHelper.getAttribute(fieldEl, "name", true);
WildcardPattern matchNamespace = null;
if (matchNamespaceAttr != null) {
// If the matchNamespace attr does not contain a wildcard expression, and its value
// happens to be an existing namespace prefix, than substitute the prefix for the full URI.
if (!WildcardPattern.isWildcardExpression(matchNamespaceAttr)) {
String uri = fieldEl.lookupNamespaceURI(matchNamespaceAttr);
if (uri != null) matchNamespaceAttr = uri;
}
matchNamespace = new WildcardPattern(matchNamespaceAttr);
}
WildcardPattern matchName = null;
if (matchNameAttr != null) {
matchName = new WildcardPattern(matchNameAttr);
}
TypePattern matchTypes = null;
if (matchTypeAttr != null) {
matchTypes = new TypePattern(matchTypeAttr);
}
Set<Scope> matchScopes = null;
if (matchScopeAttr != null) {
matchScopes = EnumSet.noneOf(Scope.class);
for (String scope : COMMA_SPLITTER.split(matchScopeAttr)) {
matchScopes.add(Scope.valueOf(scope));
}
if (matchScopes.isEmpty()) {
matchScopes = null;
}
}
// Be gentle to users of Lily 1.0 and warn them about attributes that are not supported
// anymore
if (DocumentHelper.getAttribute(fieldEl, "matchMultiValue", false) != null) {
log.warn(
"The attribute matchMultiValue on dynamicField is not supported anymore, it will be ignored.");
}
if (DocumentHelper.getAttribute(fieldEl, "matchHierarchical", false) != null) {
log.warn(
"The attribute matchHierarchical on dynamicField is not supported anymore, it will be ignored.");
}
Set<String> variables = new HashSet<String>();
variables.add("namespace");
variables.add("name");
variables.add("type");
variables.add("baseType");
variables.add("nestedType");
variables.add("nestedBaseType");
variables.add("deepestNestedBaseType");
if (matchName != null && matchName.hasWildcard()) variables.add("nameMatch");
if (matchNamespace != null && matchNamespace.hasWildcard()) variables.add("namespaceMatch");
Set<String> booleanVariables = new HashSet<String>();
booleanVariables.add("list");
booleanVariables.add("multiValue");
NameTemplate name = new NameTemplate(nameAttr, variables, booleanVariables);
boolean extractContent = DocumentHelper.getBooleanAttribute(fieldEl, "extractContent", false);
String formatter = DocumentHelper.getAttribute(fieldEl, "formatter", false);
if (formatter != null && !conf.getFormatters().hasFormatter(formatter)) {
throw new IndexerConfException(
"Formatter does not exist: "
+ formatter
+ " at "
+ LocationAttributes.getLocationString(fieldEl));
}
boolean continue_ = DocumentHelper.getBooleanAttribute(fieldEl, "continue", false);
DynamicIndexField field =
new DynamicIndexField(
matchNamespace,
matchName,
matchTypes,
matchScopes,
name,
extractContent,
continue_,
formatter);
conf.addDynamicIndexField(field);
}
}