/**
* Creates a new instance loading any providers found in the provided skin.
*
* @param site The skin to generate a sitemap for
* @param dest The directory to write files to.
* @param url The URI at which this directory is accessible using HTTP.
* @throws IOException If the files can't be written.
*/
public SitemapGenerator(final URI site, final File dest, final URI url) throws IOException {
this(new ArrayList<PageProvider>(), dest, url);
// HACK add site into system properties, a la the global context
System.setProperty("no.sesat.sitemap.SitemapGenerator.site", site.toString());
for (PageProvider provider : ServiceLoader.load(PageProvider.class, getClassLoader(site))) {
LOG.info("Found " + provider.getName());
providers.add(provider);
}
}
/**
* Generate the sitemap files.
*
* @throws IOException if the files could not be created.
* @throws SAXException if a xml error occurs.
*/
public void generate() throws IOException, SAXException {
int totalCount = 0;
AttributesImpl schemaLocation = new AttributesImpl();
transformerHandler.startDocument();
transformerHandler.startPrefixMapping("xsd", XMLConstants.W3C_XML_SCHEMA_NS_URI);
transformerHandler.startPrefixMapping("xsi", XMLConstants.W3C_XML_SCHEMA_INSTANCE_NS_URI);
schemaLocation.addAttribute(
XMLConstants.W3C_XML_SCHEMA_NS_URI,
"schemaLocation",
"xsi:schemaLocation",
"CDATA",
"http://www.sitemaps.org/schemas/sitemap/0.9/siteindex.xsd");
transformerHandler.startElement(NS, "", "sitemapindex", schemaLocation);
for (final PageProvider provider : providers) {
LOG.info("Processing " + provider.getName());
final SiteMap group = new SiteMap(provider.getName());
try {
for (final Page page : provider) {
if (null != page) {
group.addPage(page);
}
}
} finally {
group.finish();
LOG.info(group.getCount() + " entries processed for " + provider.getName());
totalCount += group.getCount();
}
for (final SiteMap.SiteMapFile map : group.getSiteMaps()) {
transformerHandler.startElement("", "", "sitemap", new AttributesImpl());
addElement("loc", uri.resolve(map.getFileName()).toString());
addElement("lastmod", formatDateW3c((new Date())));
transformerHandler.endElement("", "", "sitemap");
}
}
transformerHandler.endElement(NS, "", "sitemapindex");
transformerHandler.endDocument();
writer.close();
LOG.info("All done (" + totalCount + " entries)");
}
