Commit e10e166b authored by Joseph Beard's avatar Joseph Beard Committed by dfabulich
Browse files

Write empty sitemap and index files (#23) 

* Allow SitemapGenerators to write an empty sitemap.

* Allow SitemapIndexGenerators to write an empty sitemap index.
parent 6e9832be

src/main/java/com/redfin/sitemapgenerator/AbstractSitemapGeneratorOptions.java

+12 −0
Original line number Diff line number Diff line
@@ -9,6 +9,7 @@ abstract class AbstractSitemapGeneratorOptions<THIS extends AbstractSitemapGener 
	File baseDir;

	URL baseUrl;

	String fileNamePrefix = "sitemap";

	boolean allowEmptySitemap = false;

	boolean allowMultipleSitemaps = true;

	String suffixStringPattern; // this will store some type of string pattern suitable per needs.

	W3CDateFormat dateFormat;
@@ -38,6 +39,17 @@ abstract class AbstractSitemapGeneratorOptions<THIS extends AbstractSitemapGener 
		return getThis();

	}



	/**

	 * Permit writing a sitemap that contains no URLs.

	 *

	 * @param allowEmpty {@code true} if an empty sitemap is permissible

	 * @return this instance, for chaining

	 */

	public THIS allowEmptySitemap(boolean allowEmpty) {

		this.allowEmptySitemap = allowEmpty;

		return getThis();

	}



	/** When more than the maximum number of URLs are passed in, should we split into multiple sitemaps automatically, or just throw an exception? */

	public THIS allowMultipleSitemaps(boolean allowMultipleSitemaps) {

		this.allowMultipleSitemaps = allowMultipleSitemaps;

src/main/java/com/redfin/sitemapgenerator/SitemapGenerator.java

+4 −2
Original line number Diff line number Diff line
@@ -21,6 +21,7 @@ abstract class SitemapGenerator<U extends ISitemapUrl, THIS extends SitemapGener 
	private final File baseDir;

	private final String fileNamePrefix;

	private final String fileNameSuffix;

	private final boolean allowEmptySitemap;

	private final boolean allowMultipleSitemaps;

	private final ArrayList<U> urls = new ArrayList<U>();

	private final W3CDateFormat dateFormat;
@@ -40,6 +41,7 @@ abstract class SitemapGenerator<U extends ISitemapUrl, THIS extends SitemapGener 
		W3CDateFormat dateFormat = options.dateFormat;

		if (dateFormat == null) dateFormat = new W3CDateFormat();

		this.dateFormat = dateFormat;

		allowEmptySitemap = options.allowEmptySitemap;

		allowMultipleSitemaps = options.allowMultipleSitemaps;

		maxUrls = options.maxUrls;

		autoValidate = options.autoValidate;
@@ -165,7 +167,7 @@ abstract class SitemapGenerator<U extends ISitemapUrl, THIS extends SitemapGener 
	 */

	public List<File> write() {

		if (finished) throw new RuntimeException("Sitemap already printed; you must create a new generator to make more sitemaps");

		if (urls.size() == 0 && mapCount == 0) throw new RuntimeException("No URLs added, sitemap would be empty; you must add some URLs with addUrls");

		if (!allowEmptySitemap && urls.isEmpty() && mapCount == 0) throw new RuntimeException("No URLs added, sitemap would be empty; you must add some URLs with addUrls");

		writeSiteMap();

		finished = true;

		return outFiles;
@@ -221,7 +223,7 @@ abstract class SitemapGenerator<U extends ISitemapUrl, THIS extends SitemapGener 
		if (baseDir == null) {

			throw new NullPointerException("To write to files, baseDir must not be null");

		}

		if (urls.size() == 0) return;

		if (urls.isEmpty() && (mapCount > 0 || !allowEmptySitemap)) return;

		String fileNamePrefix;

		if (mapCount > 0) {

			fileNamePrefix = this.fileNamePrefix + mapCount;

src/main/java/com/redfin/sitemapgenerator/SitemapIndexGenerator.java

+16 −1
Original line number Diff line number Diff line
@@ -19,6 +19,7 @@ import org.xml.sax.SAXException; 
public class SitemapIndexGenerator {

	private final URL baseUrl;	

	private final File outFile;

	private final boolean allowEmptyIndex;

	private final ArrayList<SitemapIndexUrl> urls = new ArrayList<SitemapIndexUrl>();

	private final int maxUrls;

	private final W3CDateFormat dateFormat;
@@ -32,6 +33,7 @@ public class SitemapIndexGenerator { 
		private URL baseUrl;

		private File outFile;

		private W3CDateFormat dateFormat = null;

		private boolean allowEmptyIndex = false;

		private int maxUrls = MAX_SITEMAPS_PER_INDEX;

		private Date defaultLastMod = new Date();

		private boolean autoValidate = false;
@@ -59,6 +61,18 @@ public class SitemapIndexGenerator { 
			this.dateFormat = dateFormat;

			return this;

		}



		/**

		 * Permit writing an index that contains no URLs.

		 *

		 * @param allowEmptyIndex {@code true} if an empty index is permissible

		 * @return this instance, for chaining

		 */

		public Options allowEmptyIndex(boolean allowEmptyIndex) {

			this.allowEmptyIndex = allowEmptyIndex;

			return this;

		}



		/**

		 * The maximum number of sitemaps to allow per sitemap index; the default is the

		 * maximum allowed (1,000), but you can decrease it if you wish (for testing)
@@ -116,6 +130,7 @@ public class SitemapIndexGenerator { 
	private SitemapIndexGenerator(Options options) {

		this.baseUrl = options.baseUrl;		

		this.outFile = options.outFile;

		this.allowEmptyIndex = options.allowEmptyIndex;

		this.maxUrls = options.maxUrls;

		W3CDateFormat dateFormat = options.dateFormat;

		if (dateFormat == null) dateFormat = new W3CDateFormat();
@@ -206,7 +221,7 @@ public class SitemapIndexGenerator { 
	

	/** Writes out the sitemap index */

	public void write() {

		if (urls.size() == 0) throw new RuntimeException("No URLs added, sitemap index would be empty; you must add some URLs with addUrls");

		if (!allowEmptyIndex && urls.isEmpty()) throw new RuntimeException("No URLs added, sitemap index would be empty; you must add some URLs with addUrls");

		try {

			// TODO gzip? is that legal for a sitemap index?

			FileWriter out = new FileWriter(outFile);

src/test/java/com/redfin/sitemapgenerator/SitemapGeneratorTest.java

+18 −0
Original line number Diff line number Diff line
@@ -348,6 +348,24 @@ public class SitemapGeneratorTest extends TestCase { 
		assertEquals("Second string didn't match", SITEMAP_PLUS_ONE, siteMapsAsStrings.get(1));

	}



	public void testWriteEmptySitemap() throws Exception {

		wsg = WebSitemapGenerator.builder("http://www.example.com", dir).allowEmptySitemap(true).build();

		String expected = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" +

				"<urlset xmlns=\"http://www.sitemaps.org/schemas/sitemap/0.9\" >\n" +

				"</urlset>";

		String sitemap = writeSingleSiteMap(wsg);

		assertEquals(expected, sitemap);

	}



	public void testMaxUrlsAllowingEmptyDoesNotWriteExtraSitemap() throws Exception {

		wsg = WebSitemapGenerator.builder("http://www.example.com", dir).allowEmptySitemap(true).maxUrls(10).build();

		for (int i = 0; i < 10; i++) {

			wsg.addUrl("http://www.example.com/"+i);

		}

		String sitemap = writeSingleSiteMap(wsg);

		assertEquals(SITEMAP1, sitemap);

	}

	

	private String writeSingleSiteMap(WebSitemapGenerator wsg) {

		List<File> files = wsg.write();

		assertEquals("Too many files: " + files.toString(), 1, files.size());

src/test/java/com/redfin/sitemapgenerator/SitemapIndexGeneratorTest.java

+10 −0
Original line number Diff line number Diff line
@@ -87,6 +87,16 @@ public class SitemapIndexGeneratorTest extends TestCase { 
		} catch (RuntimeException e) {}

	}



	public void testNoUrlsEmptyIndexAllowed() throws Exception {

		sig = new SitemapIndexGenerator.Options(EXAMPLE, outFile).allowEmptyIndex(true).build();

		sig.write();

		String expected = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" +

				"<sitemapindex xmlns=\"http://www.sitemaps.org/schemas/sitemap/0.9\">\n" +

				"</sitemapindex>";

		String actual = TestUtil.slurpFileAndDelete(outFile);

		assertEquals(expected, actual);

	}

	

	public void testMaxUrls() throws Exception {

		sig = new SitemapIndexGenerator.Options(EXAMPLE, outFile).autoValidate(true)

			.maxUrls(10).defaultLastMod(new Date(0)).dateFormat(ZULU).build();