src/main/java/net/pterodactylus/sone/core/SoneParser.java

   1 package net.pterodactylus.sone.core;
   2
   3 import static java.util.concurrent.TimeUnit.*;
   4 import static java.util.logging.Logger.*;
   5 import static net.pterodactylus.sone.utils.NumberParsers.*;
   6
   7 import java.io.*;
   8 import java.util.*;
   9 import java.util.logging.*;
  10
  11 import javax.annotation.*;
  12 import javax.inject.*;
  13
  14 import net.pterodactylus.sone.data.*;
  15 import net.pterodactylus.sone.data.Profile.*;
  16 import net.pterodactylus.sone.database.*;
  17 import net.pterodactylus.util.xml.*;
  18
  19 import com.codahale.metrics.*;
  20 import com.google.common.base.*;
  21 import org.w3c.dom.*;
  22
  23 /**
  24  * Parses a {@link Sone} from an XML {@link InputStream}.
  25  */
  26 public class SoneParser {
  27
  28         private static final Logger logger = getLogger(SoneParser.class.getName());
  29         private static final int MAX_PROTOCOL_VERSION = 0;
  30         private final Database database;
  31         private final Histogram soneParsingDurationHistogram;
  32
  33         @Inject
  34         public SoneParser(Database database, MetricRegistry metricRegistry) {
  35                 this.database = database;
  36                 this.soneParsingDurationHistogram = metricRegistry.histogram("sone.parse.duration", () -> new Histogram(new ExponentiallyDecayingReservoir(3000, 0)));
  37         }
  38
  39         @Nullable
  40         public Sone parseSone(Sone originalSone, InputStream soneInputStream) throws SoneException {
  41                 /* TODO - impose a size limit? */
  42
  43                 Stopwatch stopwatch = Stopwatch.createStarted();
  44                 Document document;
  45                 /* XML parsing is not thread-safe. */
  46                 synchronized (this) {
  47                         document = XML.transformToDocument(soneInputStream);
  48                 }
  49                 if (document == null) {
  50                         /* TODO - mark Sone as bad. */
  51                         logger.log(Level.WARNING, String.format("Could not parse XML for Sone %s!", originalSone));
  52                         return null;
  53                 }
  54
  55                 SoneBuilder soneBuilder = database.newSoneBuilder().from(originalSone.getIdentity());
  56                 if (originalSone.isLocal()) {
  57                         soneBuilder = soneBuilder.local();
  58                 }
  59                 Sone sone = soneBuilder.build();
  60
  61                 SimpleXML soneXml;
  62                 try {
  63                         soneXml = SimpleXML.fromDocument(document);
  64                 } catch (NullPointerException npe1) {
  65                         /* for some reason, invalid XML can cause NPEs. */
  66                         logger.log(Level.WARNING, String.format("XML for Sone %s can not be parsed!", sone), npe1);
  67                         return null;
  68                 }
  69
  70                 Integer protocolVersion = null;
  71                 String soneProtocolVersion = soneXml.getValue("protocol-version", null);
  72                 if (soneProtocolVersion != null) {
  73                         protocolVersion = parseInt(soneProtocolVersion, null);
  74                 }
  75                 if (protocolVersion == null) {
  76                         logger.log(Level.INFO, "No protocol version found, assuming 0.");
  77                         protocolVersion = 0;
  78                 }
  79
  80                 if (protocolVersion < 0) {
  81                         logger.log(Level.WARNING, String.format("Invalid protocol version: %d! Not parsing Sone.", protocolVersion));
  82                         return null;
  83                 }
  84
  85                 /* check for valid versions. */
  86                 if (protocolVersion > MAX_PROTOCOL_VERSION) {
  87                         logger.log(Level.WARNING, String.format("Unknown protocol version: %d! Not parsing Sone.", protocolVersion));
  88                         return null;
  89                 }
  90
  91                 String soneTime = soneXml.getValue("time", null);
  92                 if (soneTime == null) {
  93                         /* TODO - mark Sone as bad. */
  94                         logger.log(Level.WARNING, String.format("Downloaded time for Sone %s was null!", sone));
  95                         return null;
  96                 }
  97                 try {
  98                         sone.setTime(Long.parseLong(soneTime));
  99                 } catch (NumberFormatException nfe1) {
 100                         /* TODO - mark Sone as bad. */
 101                         logger.log(Level.WARNING, String.format("Downloaded Sone %s with invalid time: %s", sone, soneTime));
 102                         return null;
 103                 }
 104
 105                 SimpleXML clientXml = soneXml.getNode("client");
 106                 if (clientXml != null) {
 107                         String clientName = clientXml.getValue("name", null);
 108                         String clientVersion = clientXml.getValue("version", null);
 109                         if ((clientName == null) || (clientVersion == null)) {
 110                                 logger.log(Level.WARNING, String.format("Download Sone %s with client XML but missing name or version!", sone));
 111                                 return null;
 112                         }
 113                         sone.setClient(new Client(clientName, clientVersion));
 114                 }
 115
 116                 SimpleXML profileXml = soneXml.getNode("profile");
 117                 if (profileXml == null) {
 118                         /* TODO - mark Sone as bad. */
 119                         logger.log(Level.WARNING, String.format("Downloaded Sone %s has no profile!", sone));
 120                         return null;
 121                 }
 122
 123                 /* parse profile. */
 124                 String profileFirstName = profileXml.getValue("first-name", null);
 125                 String profileMiddleName = profileXml.getValue("middle-name", null);
 126                 String profileLastName = profileXml.getValue("last-name", null);
 127                 Integer profileBirthDay = parseInt(profileXml.getValue("birth-day", ""), null);
 128                 Integer profileBirthMonth = parseInt(profileXml.getValue("birth-month", ""), null);
 129                 Integer profileBirthYear = parseInt(profileXml.getValue("birth-year", ""), null);
 130                 Profile profile = new Profile(sone).setFirstName(profileFirstName).setMiddleName(profileMiddleName).setLastName(profileLastName);
 131                 profile.setBirthDay(profileBirthDay).setBirthMonth(profileBirthMonth).setBirthYear(profileBirthYear);
 132                 /* avatar is processed after images are loaded. */
 133                 String avatarId = profileXml.getValue("avatar", null);
 134
 135                 /* parse profile fields. */
 136                 SimpleXML profileFieldsXml = profileXml.getNode("fields");
 137                 if (profileFieldsXml != null) {
 138                         for (SimpleXML fieldXml : profileFieldsXml.getNodes("field")) {
 139                                 String fieldName = fieldXml.getValue("field-name", null);
 140                                 String fieldValue = fieldXml.getValue("field-value", "");
 141                                 if (fieldName == null) {
 142                                         logger.log(Level.WARNING, String.format("Downloaded profile field for Sone %s with missing data! Name: %s, Value: %s", sone, fieldName, fieldValue));
 143                                         return null;
 144                                 }
 145                                 try {
 146                                         profile.addField(fieldName.trim()).setValue(fieldValue);
 147                                 } catch (EmptyFieldName efn1) {
 148                                         logger.log(Level.WARNING, "Empty field name!", efn1);
 149                                         return null;
 150                                 } catch (DuplicateField df1) {
 151                                         logger.log(Level.WARNING, String.format("Duplicate field: %s", fieldName), df1);
 152                                         return null;
 153                                 }
 154                         }
 155                 }
 156
 157                 /* parse posts. */
 158                 SimpleXML postsXml = soneXml.getNode("posts");
 159                 Set<Post> posts = new HashSet<>();
 160                 if (postsXml == null) {
 161                         /* TODO - mark Sone as bad. */
 162                         logger.log(Level.WARNING, String.format("Downloaded Sone %s has no posts!", sone));
 163                 } else {
 164                         for (SimpleXML postXml : postsXml.getNodes("post")) {
 165                                 String postId = postXml.getValue("id", null);
 166                                 String postRecipientId = postXml.getValue("recipient", null);
 167                                 String postTime = postXml.getValue("time", null);
 168                                 String postText = postXml.getValue("text", null);
 169                                 if ((postId == null) || (postTime == null) || (postText == null)) {
 170                                         /* TODO - mark Sone as bad. */
 171                                         logger.log(Level.WARNING, String.format("Downloaded post for Sone %s with missing data! ID: %s, Time: %s, Text: %s", sone, postId, postTime, postText));
 172                                         return null;
 173                                 }
 174                                 try {
 175                                         PostBuilder postBuilder = database.newPostBuilder();
 176                                         /* TODO - parse time correctly. */
 177                                         postBuilder.withId(postId).from(sone.getId()).withTime(Long.parseLong(postTime)).withText(postText);
 178                                         if ((postRecipientId != null) && (postRecipientId.length() == 43)) {
 179                                                 postBuilder.to(postRecipientId);
 180                                         }
 181                                         posts.add(postBuilder.build());
 182                                 } catch (NumberFormatException nfe1) {
 183                                         /* TODO - mark Sone as bad. */
 184                                         logger.log(Level.WARNING, String.format("Downloaded post for Sone %s with invalid time: %s", sone, postTime));
 185                                         return null;
 186                                 }
 187                         }
 188                 }
 189
 190                 /* parse replies. */
 191                 SimpleXML repliesXml = soneXml.getNode("replies");
 192                 Set<PostReply> replies = new HashSet<>();
 193                 if (repliesXml == null) {
 194                         /* TODO - mark Sone as bad. */
 195                         logger.log(Level.WARNING, String.format("Downloaded Sone %s has no replies!", sone));
 196                 } else {
 197                         for (SimpleXML replyXml : repliesXml.getNodes("reply")) {
 198                                 String replyId = replyXml.getValue("id", null);
 199                                 String replyPostId = replyXml.getValue("post-id", null);
 200                                 String replyTime = replyXml.getValue("time", null);
 201                                 String replyText = replyXml.getValue("text", null);
 202                                 if ((replyId == null) || (replyPostId == null) || (replyTime == null) || (replyText == null)) {
 203                                         /* TODO - mark Sone as bad. */
 204                                         logger.log(Level.WARNING, String.format("Downloaded reply for Sone %s with missing data! ID: %s, Post: %s, Time: %s, Text: %s", sone, replyId, replyPostId, replyTime, replyText));
 205                                         return null;
 206                                 }
 207                                 try {
 208                                         PostReplyBuilder postReplyBuilder = database.newPostReplyBuilder();
 209                                         /* TODO - parse time correctly. */
 210                                         postReplyBuilder.withId(replyId).from(sone.getId()).to(replyPostId).withTime(Long.parseLong(replyTime)).withText(replyText);
 211                                         replies.add(postReplyBuilder.build());
 212                                 } catch (NumberFormatException nfe1) {
 213                                         /* TODO - mark Sone as bad. */
 214                                         logger.log(Level.WARNING, String.format("Downloaded reply for Sone %s with invalid time: %s", sone, replyTime));
 215                                         return null;
 216                                 }
 217                         }
 218                 }
 219
 220                 /* parse liked post IDs. */
 221                 SimpleXML likePostIdsXml = soneXml.getNode("post-likes");
 222                 Set<String> likedPostIds = new HashSet<>();
 223                 if (likePostIdsXml == null) {
 224                         /* TODO - mark Sone as bad. */
 225                         logger.log(Level.WARNING, String.format("Downloaded Sone %s has no post likes!", sone));
 226                 } else {
 227                         for (SimpleXML likedPostIdXml : likePostIdsXml.getNodes("post-like")) {
 228                                 String postId = likedPostIdXml.getValue();
 229                                 likedPostIds.add(postId);
 230                         }
 231                 }
 232
 233                 /* parse liked reply IDs. */
 234                 SimpleXML likeReplyIdsXml = soneXml.getNode("reply-likes");
 235                 Set<String> likedReplyIds = new HashSet<>();
 236                 if (likeReplyIdsXml == null) {
 237                         /* TODO - mark Sone as bad. */
 238                         logger.log(Level.WARNING, String.format("Downloaded Sone %s has no reply likes!", sone));
 239                 } else {
 240                         for (SimpleXML likedReplyIdXml : likeReplyIdsXml.getNodes("reply-like")) {
 241                                 String replyId = likedReplyIdXml.getValue();
 242                                 likedReplyIds.add(replyId);
 243                         }
 244                 }
 245
 246                 /* parse albums. */
 247                 SimpleXML albumsXml = soneXml.getNode("albums");
 248                 Map<String, Image> allImages = new HashMap<>();
 249                 List<Album> topLevelAlbums = new ArrayList<>();
 250                 Map<String, Album> allAlbums = new HashMap<>();
 251                 if (albumsXml != null) {
 252                         for (SimpleXML albumXml : albumsXml.getNodes("album")) {
 253                                 String id = albumXml.getValue("id", null);
 254                                 String parentId = albumXml.getValue("parent", null);
 255                                 String title = albumXml.getValue("title", null);
 256                                 String description = albumXml.getValue("description", "");
 257                                 if ((id == null) || (title == null)) {
 258                                         logger.log(Level.WARNING, String.format("Downloaded Sone %s contains invalid album!", sone));
 259                                         return null;
 260                                 }
 261                                 Album parent = null;
 262                                 if (parentId != null) {
 263                                         parent = allAlbums.get(parentId);
 264                                         if (parent == null) {
 265                                                 logger.log(Level.WARNING, String.format("Downloaded Sone %s has album with invalid parent!", sone));
 266                                                 return null;
 267                                         }
 268                                 }
 269                                 Album album = database.newAlbumBuilder()
 270                                                 .withId(id)
 271                                                 .by(sone)
 272                                                 .build()
 273                                                 .modify()
 274                                                 .setTitle(title)
 275                                                 .setDescription(description)
 276                                                 .update();
 277                                 if (parent != null) {
 278                                         parent.addAlbum(album);
 279                                 } else {
 280                                         topLevelAlbums.add(album);
 281                                 }
 282                                 allAlbums.put(album.getId(), album);
 283                                 SimpleXML imagesXml = albumXml.getNode("images");
 284                                 if (imagesXml != null) {
 285                                         for (SimpleXML imageXml : imagesXml.getNodes("image")) {
 286                                                 String imageId = imageXml.getValue("id", null);
 287                                                 String imageCreationTimeString = imageXml.getValue("creation-time", null);
 288                                                 String imageKey = imageXml.getValue("key", null);
 289                                                 String imageTitle = imageXml.getValue("title", null);
 290                                                 String imageDescription = imageXml.getValue("description", "");
 291                                                 String imageWidthString = imageXml.getValue("width", null);
 292                                                 String imageHeightString = imageXml.getValue("height", null);
 293                                                 if ((imageId == null) || (imageCreationTimeString == null) || (imageKey == null) || (imageTitle == null) || (imageWidthString == null) || (imageHeightString == null)) {
 294                                                         logger.log(Level.WARNING, String.format("Downloaded Sone %s contains invalid images!", sone));
 295                                                         return null;
 296                                                 }
 297                                                 long creationTime = parseLong(imageCreationTimeString, 0L);
 298                                                 int imageWidth = parseInt(imageWidthString, 0);
 299                                                 int imageHeight = parseInt(imageHeightString, 0);
 300                                                 if ((imageWidth < 1) || (imageHeight < 1)) {
 301                                                         logger.log(Level.WARNING, String.format("Downloaded Sone %s contains image %s with invalid dimensions (%s, %s)!", sone, imageId, imageWidthString, imageHeightString));
 302                                                         return null;
 303                                                 }
 304                                                 Image image = database.newImageBuilder().withId(imageId).build().modify().setSone(sone).setKey(imageKey).setCreationTime(creationTime).update();
 305                                                 image = image.modify().setTitle(imageTitle).setDescription(imageDescription).update();
 306                                                 image = image.modify().setWidth(imageWidth).setHeight(imageHeight).update();
 307                                                 album.addImage(image);
 308                                                 allImages.put(imageId, image);
 309                                         }
 310                                 }
 311                         }
 312                 }
 313
 314                 /* process avatar. */
 315                 if (avatarId != null) {
 316                         profile.setAvatar(allImages.get(avatarId));
 317                 }
 318
 319                 /* okay, apparently everything was parsed correctly. Now import. */
 320                 sone.setProfile(profile);
 321                 sone.setPosts(posts);
 322                 sone.setReplies(replies);
 323                 sone.setLikePostIds(likedPostIds);
 324                 sone.setLikeReplyIds(likedReplyIds);
 325                 for (Album album : topLevelAlbums) {
 326                         sone.getRootAlbum().addAlbum(album);
 327                 }
 328
 329                 // record the duration
 330                 stopwatch.stop();
 331                 soneParsingDurationHistogram.update(stopwatch.elapsed(MICROSECONDS));
 332                 logger.fine(() -> "Parsed " + originalSone.getIdentity().getId() + "@" + originalSone.getLatestEdition() + " in " + stopwatch.elapsed(MICROSECONDS) + "μs.");
 333
 334                 return sone;
 335
 336         }
 337
 338 }