Skip to content

Commit

Permalink
Fix namespace URI from HTTPS to HTTP
Browse files Browse the repository at this point in the history 
with centralization of constants in one class
  • Loading branch information
@jeromeroucou
jeromeroucou committed Mar 7, 2024
1 parent 1bee6f2 commit d5a1097
Show file tree
Hide file tree
Showing 14 changed files with 189 additions and 89 deletions.
12 changes: 7 additions & 5 deletions src/main/java/com/redfin/sitemapgenerator/AbstractSitemapGeneratorOptions.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@ abstract class AbstractSitemapGeneratorOptions<T extends AbstractSitemapGenerato
boolean allowMultipleSitemaps = true;
String suffixStringPattern; // this will store some type of string pattern suitable per needs.
W3CDateFormat dateFormat;
int maxUrls = SitemapGenerator.MAX_URLS_PER_SITEMAP;
int maxUrls = SitemapConstants.MAX_URLS_PER_SITEMAP;
boolean autoValidate = false;
boolean gzip = false;

Expand Down Expand Up @@ -62,12 +62,14 @@ public T dateFormat(W3CDateFormat dateFormat) {
}
/**
* The maximum number of URLs to allow per sitemap; the default is the
* maximum allowed (50,000), but you can decrease it if you wish (to make
* your auto-generated sitemaps smaller)
* maximum allowed (see {@link SitemapConstants#MAX_URLS_PER_SITEMAP}), but you
* can decrease it if you wish (to make your auto-generated sitemaps smaller)
*/
public T maxUrls(int maxUrls) {
if (maxUrls > SitemapGenerator.MAX_URLS_PER_SITEMAP) {
throw new SitemapException("You can only have 50,000 URLs per sitemap; to use more, allowMultipleSitemaps and generate a sitemap index. You asked for " + maxUrls);
if (maxUrls > SitemapConstants.MAX_URLS_PER_SITEMAP) {
throw new SitemapException(String.format(
"You can only have %d URLs per sitemap; to use more, allowMultipleSitemaps and generate a sitemap index. You asked for %d",
SitemapConstants.MAX_URLS_PER_SITEMAP, maxUrls));
}
this.maxUrls = maxUrls;
return getThis();
Expand Down
18 changes: 8 additions & 10 deletions src/main/java/com/redfin/sitemapgenerator/GoogleImageSitemapGenerator.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -77,27 +77,25 @@ public static SitemapGeneratorBuilder<GoogleImageSitemapGenerator> builder(Strin

private static class Renderer extends AbstractSitemapUrlRenderer<GoogleImageSitemapUrl> implements ISitemapUrlRenderer<GoogleImageSitemapUrl> {

private static final String IMAGE_NS = "image";

public Class<GoogleImageSitemapUrl> getUrlClass() {
return GoogleImageSitemapUrl.class;
}

public String getXmlNamespaces() {
return "xmlns:image=\"https://www.google.com/schemas/sitemap-image/1.1\"";
return String.format("xmlns:%s=\"%s\"", SitemapConstants.GOOGLE_IMAGE_NS, SitemapConstants.GOOGLE_IMAGE_NS_URI);
}

public void render(GoogleImageSitemapUrl url, StringBuilder sb, W3CDateFormat dateFormat) {
StringBuilder tagSb = new StringBuilder();

for(Image image : url.getImages()) {
tagSb.append(" <").append(IMAGE_NS).append(":image>\n");
renderTag(tagSb, IMAGE_NS, "loc", image.getUrl());
renderTag(tagSb, IMAGE_NS, "caption", image.getCaption());
renderTag(tagSb, IMAGE_NS, "title", image.getTitle());
renderTag(tagSb, IMAGE_NS, "geo_location", image.getGeoLocation());
renderTag(tagSb, IMAGE_NS, "license", image.getLicense());
tagSb.append(" </").append(IMAGE_NS).append(":image>\n");
tagSb.append(" <").append(SitemapConstants.GOOGLE_IMAGE_NS).append(":image>\n");
renderTag(tagSb, SitemapConstants.GOOGLE_IMAGE_NS, "loc", image.getUrl());
renderTag(tagSb, SitemapConstants.GOOGLE_IMAGE_NS, "caption", image.getCaption());
renderTag(tagSb, SitemapConstants.GOOGLE_IMAGE_NS, "title", image.getTitle());
renderTag(tagSb, SitemapConstants.GOOGLE_IMAGE_NS, "geo_location", image.getGeoLocation());
renderTag(tagSb, SitemapConstants.GOOGLE_IMAGE_NS, "license", image.getLicense());
tagSb.append(" </").append(SitemapConstants.GOOGLE_IMAGE_NS).append(":image>\n");
}
super.render(url, sb, dateFormat, tagSb.toString());
}
Expand Down
2 changes: 1 addition & 1 deletion src/main/java/com/redfin/sitemapgenerator/GoogleLinkSitemapGenerator.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,7 @@ public Class<GoogleLinkSitemapUrl> getUrlClass() {

public String getXmlNamespaces() {

return "xmlns:xhtml=\"https://www.w3.org/1999/xhtml\"";
return String.format("xmlns:xhtml=\"%s\"", SitemapConstants.GOOGLE_LINK_NS_URI);
}

public void render(final GoogleLinkSitemapUrl url, final StringBuilder sb, final W3CDateFormat dateFormat) {
Expand Down
34 changes: 15 additions & 19 deletions src/main/java/com/redfin/sitemapgenerator/GoogleNewsSitemapGenerator.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10,9 +10,6 @@
* @see <a href="https://developers.google.com/search/docs/crawling-indexing/sitemaps/news-sitemap">Google Developer: News Sitemap</a>
*/
public class GoogleNewsSitemapGenerator extends SitemapGenerator<GoogleNewsSitemapUrl,GoogleNewsSitemapGenerator> {

/** 1000 URLs max in a Google News sitemap. */
public static final int MAX_URLS_PER_SITEMAP = 1000;

/** Configures a builder so you can specify sitemap generator options
*
Expand All @@ -36,14 +33,15 @@ public static SitemapGeneratorBuilder<GoogleNewsSitemapGenerator> builder(URL ba
public static SitemapGeneratorBuilder<GoogleNewsSitemapGenerator> builder(String baseUrl, File baseDir) throws MalformedURLException {
SitemapGeneratorBuilder<GoogleNewsSitemapGenerator> builder =
new SitemapGeneratorBuilder<>(baseUrl, baseDir, GoogleNewsSitemapGenerator.class);
builder.maxUrls = GoogleNewsSitemapGenerator.MAX_URLS_PER_SITEMAP;
builder.maxUrls = SitemapConstants.GOOGLE_NEWS_MAX_URLS_PER_SITEMAP;
return builder;
}

GoogleNewsSitemapGenerator(AbstractSitemapGeneratorOptions<?> options) {
super(options, new Renderer());
if (options.maxUrls > GoogleNewsSitemapGenerator.MAX_URLS_PER_SITEMAP) {
throw new SitemapException("Google News sitemaps can have only 1000 URLs per sitemap: " + options.maxUrls);
if (options.maxUrls > SitemapConstants.GOOGLE_NEWS_MAX_URLS_PER_SITEMAP) {
throw new SitemapException(String.format("Google News sitemaps can have only %d URLs per sitemap: %d",
SitemapConstants.GOOGLE_NEWS_MAX_URLS_PER_SITEMAP, options.maxUrls));
}
}

Expand Down Expand Up @@ -89,28 +87,26 @@ public GoogleNewsSitemapGenerator(URL baseUrl) {

private static class Renderer extends AbstractSitemapUrlRenderer<GoogleNewsSitemapUrl> implements ISitemapUrlRenderer<GoogleNewsSitemapUrl> {

private static final String NEWS_NS = "news";

public Class<GoogleNewsSitemapUrl> getUrlClass() {
return GoogleNewsSitemapUrl.class;
}

public String getXmlNamespaces() {
return "xmlns:news=\"https://www.google.com/schemas/sitemap-news/0.9\"";
return String.format("xmlns:%s=\"%s\"", SitemapConstants.GOOGLE_NEWS_NS, SitemapConstants.GOOGLE_NEWS_NS_URI);
}

public void render(GoogleNewsSitemapUrl url, StringBuilder sb, W3CDateFormat dateFormat) {
StringBuilder tagSb = new StringBuilder();
tagSb.append(" <").append(NEWS_NS).append(":news>\n");
tagSb.append(" <").append(NEWS_NS).append(":publication>\n");
renderSubTag(tagSb, NEWS_NS, "name", url.getPublication().getName());
renderSubTag(tagSb, NEWS_NS, "language", url.getPublication().getLanguage());
tagSb.append(" </").append(NEWS_NS).append(":publication>\n");
renderTag(tagSb, NEWS_NS, "genres", url.getGenres());
renderTag(tagSb, NEWS_NS, "publication_date", dateFormat.format(url.getPublicationDate()));
renderTag(tagSb, NEWS_NS, "title", url.getTitle());
renderTag(tagSb, NEWS_NS, "keywords", url.getKeywords());
tagSb.append(" </").append(NEWS_NS).append(":news>\n");
tagSb.append(" <").append(SitemapConstants.GOOGLE_NEWS_NS).append(":news>\n");
tagSb.append(" <").append(SitemapConstants.GOOGLE_NEWS_NS).append(":publication>\n");
renderSubTag(tagSb, SitemapConstants.GOOGLE_NEWS_NS, "name", url.getPublication().getName());
renderSubTag(tagSb, SitemapConstants.GOOGLE_NEWS_NS, "language", url.getPublication().getLanguage());
tagSb.append(" </").append(SitemapConstants.GOOGLE_NEWS_NS).append(":publication>\n");
renderTag(tagSb, SitemapConstants.GOOGLE_NEWS_NS, "genres", url.getGenres());
renderTag(tagSb, SitemapConstants.GOOGLE_NEWS_NS, "publication_date", dateFormat.format(url.getPublicationDate()));
renderTag(tagSb, SitemapConstants.GOOGLE_NEWS_NS, "title", url.getTitle());
renderTag(tagSb, SitemapConstants.GOOGLE_NEWS_NS, "keywords", url.getKeywords());
tagSb.append(" </").append(SitemapConstants.GOOGLE_NEWS_NS).append(":news>\n");
super.render(url, sb, dateFormat, tagSb.toString());
}

Expand Down
35 changes: 17 additions & 18 deletions src/main/java/com/redfin/sitemapgenerator/GoogleVideoSitemapGenerator.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -77,42 +77,41 @@ public GoogleVideoSitemapGenerator(URL baseUrl) {

private static class Renderer extends AbstractSitemapUrlRenderer<GoogleVideoSitemapUrl> implements ISitemapUrlRenderer<GoogleVideoSitemapUrl> {

private static final String VIDEO_NS = "video";

public Class<GoogleVideoSitemapUrl> getUrlClass() {
return GoogleVideoSitemapUrl.class;
}

public String getXmlNamespaces() {
return "xmlns:video=\"https://www.google.com/schemas/sitemap-video/1.1\"";
return String.format("xmlns:%s=\"%s\"", SitemapConstants.GOOGLE_VIDEO_NS, SitemapConstants.GOOGLE_VIDEO_NS_URI);
}

public void render(GoogleVideoSitemapUrl url, StringBuilder sb, W3CDateFormat dateFormat) {
StringBuilder tagSb = new StringBuilder();
tagSb.append(" <").append(VIDEO_NS).append(":video>\n");
renderTag(tagSb, VIDEO_NS, "content_loc", url.getContentUrl());
tagSb.append(" <").append(SitemapConstants.GOOGLE_VIDEO_NS).append(":video>\n");
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "content_loc", url.getContentUrl());
if (url.getPlayerUrl() != null) {
tagSb.append(" <").append(VIDEO_NS).append(":player_loc allow_embed=\"").append(url.getAllowEmbed()).append("\">");
tagSb.append(" <").append(SitemapConstants.GOOGLE_VIDEO_NS).append(":player_loc allow_embed=\"")
.append(url.getAllowEmbed()).append("\">");
tagSb.append(url.getPlayerUrl());
tagSb.append("</").append(VIDEO_NS).append(":player_loc>\n");
tagSb.append("</").append(SitemapConstants.GOOGLE_VIDEO_NS).append(":player_loc>\n");
}
renderTag(tagSb, VIDEO_NS, "thumbnail_loc", url.getThumbnailUrl());
renderTag(tagSb, VIDEO_NS, "title", url.getTitle());
renderTag(tagSb, VIDEO_NS, "description", url.getDescription());
renderTag(tagSb, VIDEO_NS, "rating", url.getRating());
renderTag(tagSb, VIDEO_NS, "view_count", url.getViewCount());
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "thumbnail_loc", url.getThumbnailUrl());
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "title", url.getTitle());
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "description", url.getDescription());
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "rating", url.getRating());
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "view_count", url.getViewCount());
if (url.getPublicationDate() != null) {
renderTag(tagSb, VIDEO_NS, "publication_date", dateFormat.format(url.getPublicationDate()));
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "publication_date", dateFormat.format(url.getPublicationDate()));
}
if (url.getTags() != null) {
for (String tag : url.getTags()) {
renderTag(tagSb, VIDEO_NS, "tag", tag);
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "tag", tag);
}
}
renderTag(tagSb, VIDEO_NS, "category", url.getCategory());
renderTag(tagSb, VIDEO_NS, "family_friendly", url.getFamilyFriendly());
renderTag(tagSb, VIDEO_NS, "duration", url.getDurationInSeconds());
tagSb.append(" </").append(VIDEO_NS).append(":video>\n");
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "category", url.getCategory());
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "family_friendly", url.getFamilyFriendly());
renderTag(tagSb, SitemapConstants.GOOGLE_VIDEO_NS, "duration", url.getDurationInSeconds());
tagSb.append(" </").append(SitemapConstants.GOOGLE_VIDEO_NS).append(":video>\n");
super.render(url, sb, dateFormat, tagSb.toString());
}

Expand Down
100 changes: 100 additions & 0 deletions src/main/java/com/redfin/sitemapgenerator/SitemapConstants.java
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,100 @@
package com.redfin.sitemapgenerator;

/**
* Utility class to contain basic values as constants.
**/
public final class SitemapConstants {

/**
* Private constructor to prevent instantiation.
*/
private SitemapConstants() {
}

/**
* Google Image sitemap namespace attribute.
*
* <p>See {@link GoogleImageSitemapGenerator} for more information.
*/
public static final String GOOGLE_IMAGE_NS = "image";

/**
* Google Image sitemap namespace URI.
*
* <p>See {@link GoogleImageSitemapGenerator} for more information.
*/
public static final String GOOGLE_IMAGE_NS_URI = "http://www.google.com/schemas/sitemap-image/1.1";

/**
* @see <a href="https://developers.google.com/search/docs/specialty/international/localized-versions#sitemap">
* Google Developers: Sitemap for alternate pages
*/
public static final String GOOGLE_LINK_NS_URI = "http://www.w3.org/1999/xhtml";

/**
* 1000 URLs max in a Google News sitemap file.
*
* <p>See {@link GoogleNewsSitemapGenerator} for more information.
*/
public static final int GOOGLE_NEWS_MAX_URLS_PER_SITEMAP = 1000;

/**
* Google News sitemap namespace attribute.
*
* <p>See {@link GoogleNewsSitemapGenerator} for more information.
*/
public static final String GOOGLE_NEWS_NS = "news";

/**
* Google News sitemap namespace URI.
*
* <p>See {@link GoogleNewsSitemapGenerator} for more information.
*/
public static final String GOOGLE_NEWS_NS_URI = "http://www.google.com/schemas/sitemap-news/0.9";

/**
* Google Video sitemap namespace attribute.
*
* <p>See {@link GoogleVideoSitemapGenerator} for more information.
*/
public static final String GOOGLE_VIDEO_NS = "video";

/**
* Google Video sitemap namespace URI.
*
* <p>See {@link GoogleVideoSitemapGenerator} for more information.
*/
public static final String GOOGLE_VIDEO_NS_URI = "http://www.google.com/schemas/sitemap-video/1.1";

/**
* Maximum 50000 URLs per sitemap file.
*
* @see <a href="https://www.sitemaps.org/protocol.html">
* Sitemaps XML protocol</a>
*/
public static final int MAX_URLS_PER_SITEMAP = 50000;

/** Maximum 50000 sitemaps per index allowed.
*
* @see <a href="https://www.sitemaps.org/protocol.html#index">
* Sitemaps XML protocol</a>
*/
public static final int MAX_SITEMAPS_PER_INDEX = 50000;

/**
* File name of sitemap index files, to group multiple sitemap files.
*
* @see <a href="https://www.sitemaps.org/protocol.html#index">
* Sitemaps XML protocol</a>
*/
public static final String SITEMAP_INDEX_FILE = "sitemap_index.xml";

/**
* Sitemap namespace URI to use with <code>sitemap.xml</code> file.
*
* @see <a href="https://www.sitemaps.org/protocol.html">
* Sitemaps XML protocol</a>
*/
public static final String SITEMAP_NS_URI = "http://www.sitemaps.org/schemas/sitemap/0.9";

}
7 changes: 2 additions & 5 deletions src/main/java/com/redfin/sitemapgenerator/SitemapGenerator.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,9 +13,6 @@
import java.util.zip.GZIPOutputStream;

abstract class SitemapGenerator<U extends ISitemapUrl, T extends SitemapGenerator<U,T>> {
/** 50000 URLs per sitemap maximum */
public static final int MAX_URLS_PER_SITEMAP = 50000;

private final URL baseUrl;
private final File baseDir;
private final String fileNamePrefix;
Expand Down Expand Up @@ -193,7 +190,7 @@ public List<String> writeAsStrings() {

private void writeSiteMapAsString(StringBuilder sb, List<U> urls) {
sb.append("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
sb.append("<urlset xmlns=\"https://www.sitemaps.org/schemas/sitemap/0.9\" ");
sb.append(String.format("<urlset xmlns=\"%s\" ", SitemapConstants.SITEMAP_NS_URI));
if (renderer.getXmlNamespaces() != null) {
sb.append(renderer.getXmlNamespaces());
sb.append(' ');
Expand All @@ -210,7 +207,7 @@ private void writeSiteMapAsString(StringBuilder sb, List<U> urls) {
* The sitemap index is written to {baseDir}/sitemap_index.xml
*/
public File writeSitemapsWithIndex() {
return writeSitemapsWithIndex(new File(baseDir, "sitemap_index.xml"));
return writeSitemapsWithIndex(new File(baseDir, SitemapConstants.SITEMAP_INDEX_FILE));
}

/**
Expand Down
13 changes: 6 additions & 7 deletions src/main/java/com/redfin/sitemapgenerator/SitemapIndexGenerator.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -26,16 +26,14 @@ public class SitemapIndexGenerator {
private final W3CDateFormat dateFormat;
private final Date defaultLastMod;
private final boolean autoValidate;
/** Maximum 50,000 sitemaps per index allowed */
public static final int MAX_SITEMAPS_PER_INDEX = 50000;

/** Options to configure sitemap index generation */
public static class Options {
private final URL baseUrl;
private final File outFile;
private W3CDateFormat dateFormat = null;
private boolean allowEmptyIndex = false;
private int maxUrls = MAX_SITEMAPS_PER_INDEX;
private int maxUrls = SitemapConstants.MAX_SITEMAPS_PER_INDEX;
private Date defaultLastMod = new Date();
private boolean autoValidate = false;
// TODO GZIP? Is that legal for a sitemap index?
Expand Down Expand Up @@ -76,11 +74,12 @@ public Options allowEmptyIndex(boolean allowEmptyIndex) {

/**
* 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)
* maximum allowed (see {@link SitemapConstants#MAX_SITEMAPS_PER_INDEX}), but
* you can decrease it if you wish (for testing)
*/
Options maxUrls(int maxUrls) {
if (maxUrls > MAX_SITEMAPS_PER_INDEX) {
throw new SitemapException("You can't have more than 1000 sitemaps per index");
if (maxUrls > SitemapConstants.MAX_SITEMAPS_PER_INDEX) {
throw new SitemapException(String.format("You can't have more than %d sitemaps per index", SitemapConstants.MAX_SITEMAPS_PER_INDEX));
}
this.maxUrls = maxUrls;
return this;
Expand Down Expand Up @@ -253,7 +252,7 @@ public String writeAsString() {

private void writeAsString(StringBuilder sb) {
sb.append("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
sb.append("<sitemapindex xmlns=\"https://www.sitemaps.org/schemas/sitemap/0.9\">\n");
sb.append(String.format("<sitemapindex xmlns=\"%s\">\n", SitemapConstants.SITEMAP_NS_URI));
for (SitemapIndexUrl url : urls) {
sb.append(" <sitemap>\n");
sb.append(" <loc>");
Expand Down
Loading

0 comments on commit d5a1097

Please sign in to comment.