private void writePages(
Writer writer, String virtualWiki, List<String> topicNames, boolean excludeHistory)
throws DataAccessException, IOException, MigrationException {
// note that effort is being made to re-use temporary objects as this
// code can generate an OOM "GC overhead limit exceeded" with HUGE (500MB) topics
// since the garbage collector ends up being invoked excessively.
TopicVersion topicVersion;
Topic topic;
WikiUser user;
// choose 100,000 as an arbitrary max
Pagination pagination = new Pagination(100000, 0);
List<Integer> topicVersionIds;
Map<String, String> textAttributes = new HashMap<String, String>();
textAttributes.put("xml:space", "preserve");
for (String topicName : topicNames) {
topicVersionIds = new ArrayList<Integer>();
topic = WikiBase.getDataHandler().lookupTopic(virtualWiki, topicName, false);
if (topic == null) {
throw new MigrationException(
"Failure while exporting: topic " + topicName + " does not exist");
}
writer.append("\n<page>");
writer.append('\n');
XMLUtil.buildTag(writer, "title", topic.getName(), true);
writer.append('\n');
XMLUtil.buildTag(writer, "id", topic.getTopicId());
if (excludeHistory) {
// only include the most recent version
topicVersionIds.add(topic.getCurrentVersionId());
} else {
// FIXME - changes sorted newest-to-oldest, should be reverse
List<RecentChange> changes =
WikiBase.getDataHandler().getTopicHistory(topic, pagination, true);
for (int i = (changes.size() - 1); i >= 0; i--) {
topicVersionIds.add(changes.get(i).getTopicVersionId());
}
}
for (int topicVersionId : topicVersionIds) {
topicVersion = WikiBase.getDataHandler().lookupTopicVersion(topicVersionId);
writer.append("\n<revision>");
writer.append('\n');
XMLUtil.buildTag(writer, "id", topicVersion.getTopicVersionId());
writer.append('\n');
XMLUtil.buildTag(
writer, "timestamp", this.parseJAMWikiTimestamp(topicVersion.getEditDate()), true);
writer.append("\n<contributor>");
user =
(topicVersion.getAuthorId() != null)
? WikiBase.getDataHandler().lookupWikiUser(topicVersion.getAuthorId())
: null;
if (user != null) {
writer.append('\n');
XMLUtil.buildTag(writer, "username", user.getUsername(), true);
writer.append('\n');
XMLUtil.buildTag(writer, "id", user.getUserId());
} else if (Utilities.isIpAddress(topicVersion.getAuthorDisplay())) {
writer.append('\n');
XMLUtil.buildTag(writer, "ip", topicVersion.getAuthorDisplay(), true);
} else {
writer.append('\n');
XMLUtil.buildTag(writer, "username", topicVersion.getAuthorDisplay(), true);
}
writer.append("\n</contributor>");
writer.append('\n');
XMLUtil.buildTag(writer, "comment", topicVersion.getEditComment(), true);
writer.append('\n');
XMLUtil.buildTag(writer, "text", topicVersion.getVersionContent(), textAttributes, true);
writer.append("\n</revision>");
// explicitly null out temp variables to improve garbage collection and
// avoid OOM "GC overhead limit exceeded" errors on HUGE (500MB) topics
topicVersion = null;
user = null;
}
writer.append("\n</page>");
}
}
