Search in sources :

Example 1 with Capability

use of nl.knaw.huygens.timbuctoo.remote.rs.xml.Capability in project timbuctoo by HuygensING.

the class RsExplorer method extractCapability.

private Capability extractCapability(Result<RsRoot> result) {
    String xmlString = result.getContent().map(RsRoot::getMetadata).flatMap(RsMd::getCapability).orElse("");
    Capability capa = null;
    try {
        capa = Capability.forString(xmlString);
    } catch (IllegalArgumentException e) {
        result.addError(new RemoteResourceSyncFrameworkException(String.format("invalid value for capability: '%s'", xmlString)));
    }
    return capa;
}
Also used : Capability(nl.knaw.huygens.timbuctoo.remote.rs.xml.Capability) RsRoot(nl.knaw.huygens.timbuctoo.remote.rs.xml.RsRoot)

Example 2 with Capability

use of nl.knaw.huygens.timbuctoo.remote.rs.xml.Capability in project timbuctoo by HuygensING.

the class RsExplorer method explore.

@SuppressWarnings("unchecked")
@Override
public Result<RsRoot> explore(URI uri, ResultIndex index) {
    LOG.debug("Exploring URI " + uri);
    Result<RsRoot> result = execute(uri, getSitemapConverter());
    index.add(result);
    Capability capability = extractCapability(result);
    if (followParentLinks) {
        // rs:ln rel="up" -> points to parent document, a urlset.
        String parentLink = result.getContent().map(rsRoot -> rsRoot.getLinkHref("up")).orElse(null);
        if (parentLink != null && !index.contains(parentLink)) {
            try {
                URI parentUri = new URI(parentLink);
                Result<RsRoot> parentResult = explore(parentUri, index);
                result.addParent(parentResult);
                verifyUpRelation(result, parentResult, capability);
            } catch (URISyntaxException e) {
                index.addInvalidUri(parentLink);
                result.addError(e);
                result.addInvalidUri(parentLink);
            }
        }
    }
    if (followIndexLinks) {
        // rs:ln rel="index" -> points to parent index, a sitemapindex.
        String indexLink = result.getContent().map(rsRoot -> rsRoot.getLinkHref("index")).orElse(null);
        if (indexLink != null && !index.contains(indexLink)) {
            try {
                URI indexUri = new URI(indexLink);
                Result<RsRoot> indexResult = explore(indexUri, index);
                result.addParent(indexResult);
                verifyIndexRelation(result, indexResult, capability);
            } catch (URISyntaxException e) {
                index.addInvalidUri(indexLink);
                result.addError(e);
                result.addInvalidUri(indexLink);
            }
        }
    }
    if (followChildLinks) {
        // elements <url> or <sitemap> have the location of the children of result.
        // children of Urlset with capability resourcelist, resourcedump, changelist, changedump
        // are the resources them selves. do not explore these with this explorer.
        String xmlString = result.getContent().map(RsRoot::getMetadata).flatMap(RsMd::getCapability).orElse("invalid");
        boolean isSitemapindex = result.getContent().map(rsRoot -> rsRoot instanceof Sitemapindex).orElse(false);
        if (Capability.levelfor(xmlString) > Capability.RESOURCELIST.level || isSitemapindex) {
            List<RsItem> itemList = result.getContent().map(RsRoot::getItemList).orElse(Collections.emptyList());
            for (RsItem item : itemList) {
                String childLink = item.getLoc();
                if (childLink != null && !index.contains(childLink)) {
                    try {
                        URI childUri = new URI(childLink);
                        Result<RsRoot> childResult = explore(childUri, index);
                        result.addChild(childResult);
                        verifyChildRelation(result, childResult, capability);
                        Optional<RsLn> maybeDescribedByLink = item.getLink("describedBy");
                        maybeDescribedByLink.ifPresent(rsLn -> loadDescriptionIfApplicable(childResult, rsLn, index));
                    } catch (URISyntaxException e) {
                        index.addInvalidUri(childLink);
                        result.addError(e);
                        result.addInvalidUri(childLink);
                    }
                }
            }
        }
    }
    Optional<RsLn> maybeDescribedByLink = result.getContent().flatMap(rsRoot -> rsRoot.getLink("describedBy"));
    maybeDescribedByLink.ifPresent(rsLn -> loadDescriptionIfApplicable(result, rsLn, index));
    return result;
}
Also used : Capability(nl.knaw.huygens.timbuctoo.remote.rs.xml.Capability) LambdaExceptionUtil(nl.knaw.huygens.timbuctoo.util.LambdaExceptionUtil) CloseableHttpClient(org.apache.http.impl.client.CloseableHttpClient) Arrays(java.util.Arrays) Logger(org.slf4j.Logger) URISyntaxException(java.net.URISyntaxException) RsMd(nl.knaw.huygens.timbuctoo.remote.rs.xml.RsMd) IOUtils(org.apache.commons.io.IOUtils) List(java.util.List) Sitemapindex(nl.knaw.huygens.timbuctoo.remote.rs.xml.Sitemapindex) RsRoot(nl.knaw.huygens.timbuctoo.remote.rs.xml.RsRoot) RsLn(nl.knaw.huygens.timbuctoo.remote.rs.xml.RsLn) Urlset(nl.knaw.huygens.timbuctoo.remote.rs.xml.Urlset) HttpResponse(org.apache.http.HttpResponse) Optional(java.util.Optional) ResourceSyncContext(nl.knaw.huygens.timbuctoo.remote.rs.xml.ResourceSyncContext) RsItem(nl.knaw.huygens.timbuctoo.remote.rs.xml.RsItem) RsBuilder(nl.knaw.huygens.timbuctoo.remote.rs.xml.RsBuilder) URI(java.net.URI) Collections(java.util.Collections) InputStream(java.io.InputStream) Capability(nl.knaw.huygens.timbuctoo.remote.rs.xml.Capability) RsLn(nl.knaw.huygens.timbuctoo.remote.rs.xml.RsLn) RsRoot(nl.knaw.huygens.timbuctoo.remote.rs.xml.RsRoot) URISyntaxException(java.net.URISyntaxException) URI(java.net.URI) RsItem(nl.knaw.huygens.timbuctoo.remote.rs.xml.RsItem) Sitemapindex(nl.knaw.huygens.timbuctoo.remote.rs.xml.Sitemapindex)

Aggregations

Capability (nl.knaw.huygens.timbuctoo.remote.rs.xml.Capability)2 RsRoot (nl.knaw.huygens.timbuctoo.remote.rs.xml.RsRoot)2 InputStream (java.io.InputStream)1 URI (java.net.URI)1 URISyntaxException (java.net.URISyntaxException)1 Arrays (java.util.Arrays)1 Collections (java.util.Collections)1 List (java.util.List)1 Optional (java.util.Optional)1 ResourceSyncContext (nl.knaw.huygens.timbuctoo.remote.rs.xml.ResourceSyncContext)1 RsBuilder (nl.knaw.huygens.timbuctoo.remote.rs.xml.RsBuilder)1 RsItem (nl.knaw.huygens.timbuctoo.remote.rs.xml.RsItem)1 RsLn (nl.knaw.huygens.timbuctoo.remote.rs.xml.RsLn)1 RsMd (nl.knaw.huygens.timbuctoo.remote.rs.xml.RsMd)1 Sitemapindex (nl.knaw.huygens.timbuctoo.remote.rs.xml.Sitemapindex)1 Urlset (nl.knaw.huygens.timbuctoo.remote.rs.xml.Urlset)1 LambdaExceptionUtil (nl.knaw.huygens.timbuctoo.util.LambdaExceptionUtil)1 IOUtils (org.apache.commons.io.IOUtils)1 HttpResponse (org.apache.http.HttpResponse)1 CloseableHttpClient (org.apache.http.impl.client.CloseableHttpClient)1