| 89 | |
| 90 | |
| 91 | /** |
| 92 | * Automatically resolve some tag conflicts. |
| 93 | * The list of automatic resolution is taken from the preferences. |
| 94 | * @param tc the tag collection |
| 95 | */ |
| 96 | public static void applyAutomaticTagConflictResolution(TagCollection tc) { |
| 97 | applyAutomaticTagConflictResolution(tc, getAutomaticTagConflictResolvers()); |
| 98 | } |
| 99 | |
| 100 | |
| 101 | /** |
| 102 | * Default preferences for the list of AutomaticCombine tag conflict resolvers. |
| 103 | */ |
| 104 | public static final Collection<AutomaticCombine> defaultAutomaticTagConflictCombines = Arrays.asList( |
| 105 | new AutomaticCombine("tiger:tlid", "US TIGER tlid", false, ":", "Integer"), |
| 106 | new AutomaticCombine("tiger:(?!tlid$).*", "US TIGER not tlid", true, ":", "String") |
| 107 | ); |
| 108 | |
| 109 | |
| 110 | /** |
| 111 | * Default preferences for the list of AutomaticChoice tag conflict resolvers. |
| 112 | */ |
| 113 | public static final Collection<AutomaticChoice> defaultAutomaticTagConflictChoices = Arrays.asList( |
| 114 | /* "source" "FR:cadastre" |
| 115 | * List of choices for the "source" tag of data exported from the French cadastre, |
| 116 | * which ends by the exported year generating many conflicts. |
| 117 | * The generated score begins with the year number to select the most recent one. |
| 118 | */ |
| 119 | new AutomaticChoice("source", "FR:cadastre", "FR cadastre source, manual value", true, |
| 120 | "cadastre", "0"), |
| 121 | new AutomaticChoice("source", "FR:cadastre", "FR cadastre source, initial format", true, |
| 122 | "extraction vectorielle v1 cadastre-dgi-fr source : Direction G[eé]n[eé]rale des Imp[ôo]ts" |
| 123 | + " - Cadas\\. Mise [àa] jour : ([0-9]{4})", |
| 124 | "$1 1"), |
| 125 | new AutomaticChoice("source", "FR:cadastre", "FR cadastre source, old format", true, |
| 126 | "cadastre-dgi-fr source : Direction G[eé]n[eé]rale des Imp[ôo]ts - Cadastre\\. Mise [àa] jour : ([0-9]{4})", |
| 127 | "$1 2"), |
| 128 | new AutomaticChoice("source", "FR:cadastre", "FR cadastre source, last format", true, |
| 129 | "cadastre-dgi-fr source : Direction G[eé]n[ée]rale des Finances Publiques - Cadastre\\. Mise [aà] jour : ([0-9]{4})", |
| 130 | "$1 3") |
| 131 | ); |
| 132 | |
| 133 | |
| 134 | /** |
| 135 | * Get the AutomaticTagConflictResolvers configured in the Preferences or the default ones. |
| 136 | * @return the configured AutomaticTagConflictResolvers. |
| 137 | */ |
| 138 | public static Collection<AutomaticTagConflictResolver> getAutomaticTagConflictResolvers() { |
| 139 | if (automaticTagConflictResolvers == null) { |
| 140 | Collection<AutomaticCombine> automaticTagConflictCombines = |
| 141 | Main.pref.getListOfStructs( |
| 142 | "automatic-tag-conflict-resolution.combine", |
| 143 | defaultAutomaticTagConflictCombines, AutomaticCombine.class); |
| 144 | Collection<AutomaticChoiceGroup> automaticTagConflictChoiceGroups = |
| 145 | AutomaticChoiceGroup.groupChoices(Main.pref.getListOfStructs( |
| 146 | "automatic-tag-conflict-resolution.choice", |
| 147 | defaultAutomaticTagConflictChoices, AutomaticChoice.class)); |
| 148 | // Use a tmp variable to fully construct the collection before setting |
| 149 | // the volatile variable automaticTagConflictResolvers. |
| 150 | ArrayList<AutomaticTagConflictResolver> tmp = new ArrayList<>(); |
| 151 | tmp.addAll(automaticTagConflictCombines); |
| 152 | tmp.addAll(automaticTagConflictChoiceGroups); |
| 153 | automaticTagConflictResolvers = tmp; |
| 154 | } |
| 155 | return automaticTagConflictResolvers; |
| 156 | } |
| 157 | |
| 158 | private static volatile Collection<AutomaticTagConflictResolver> automaticTagConflictResolvers; |
| 159 | |
| 160 | |
| 161 | /** |
| 162 | * An automatic tag conflict resolver interface. |
| 163 | */ |
| 164 | interface AutomaticTagConflictResolver { |
| 165 | /** |
| 166 | * Check if this resolution apply to the given Tag key. |
| 167 | * @param key The Tag key to match. |
| 168 | * @return true if this automatic resolution apply to the given Tag key. |
| 169 | */ |
| 170 | boolean matchesKey(String key); |
| 171 | |
| 172 | /** |
| 173 | * Try to resolve a conflict between a set of values for a Tag |
| 174 | * @param values the set of conflicting values for the Tag. |
| 175 | * @return the resolved value or null if resolution was not possible. |
| 176 | */ |
| 177 | String resolve(Set<String> values); |
| 178 | } |
| 179 | |
| 180 | |
| 181 | /** |
| 182 | * Automatically resolve some given conflicts using the given resolvers. |
| 183 | * @param tc the tag collection. |
| 184 | * @param resolvers the list of automatic tag conflict resolvers to apply. |
| 185 | */ |
| 186 | public static void applyAutomaticTagConflictResolution(TagCollection tc, |
| 187 | Collection<AutomaticTagConflictResolver> resolvers) { |
| 188 | for (String key: tc.getKeysWithMultipleValues()) { |
| 189 | for (AutomaticTagConflictResolver resolver: resolvers) { |
| 190 | try { |
| 191 | if (resolver.matchesKey(key)) { |
| 192 | String result = resolver.resolve(tc.getValues(key)); |
| 193 | if (result != null) { |
| 194 | tc.setUniqueForKey(key, result); |
| 195 | break; |
| 196 | } |
| 197 | } |
| 198 | } catch (RuntimeException e) { |
| 199 | // Can happen for instance if a particular resolver has an invalid regular |
| 200 | // expression pattern (java.util.regex.PatternSyntaxException) |
| 201 | // but it should not stop the other automatic tag conflict resolution. |
| 202 | Main.error(e); |
| 203 | } |
| 204 | } |
| 205 | } |
| 206 | } |
| 207 | |
| 208 | |
| 209 | /** |
| 210 | * Preference for automatic tag-conflict resolver by combining the tag values |
| 211 | * using a separator. |
| 212 | */ |
| 213 | public static class AutomaticCombine implements AutomaticTagConflictResolver { |
| 214 | |
| 215 | /** The Tag key to match */ |
| 216 | @pref public String key; |
| 217 | |
| 218 | /** A free description */ |
| 219 | @pref public String description = ""; |
| 220 | |
| 221 | /** If regular expression must be used to match the Tag key or the value. */ |
| 222 | @pref public boolean isRegex = false; |
| 223 | |
| 224 | /** The separator to use to combine the values. */ |
| 225 | @pref public String separator = ";"; |
| 226 | |
| 227 | /** If the combined values must be sorted. |
| 228 | * Possible values: |
| 229 | * <ul> |
| 230 | * <li> Integer - Sort using Integer natural order.</li> |
| 231 | * <li> String - Sort using String natural order.</li> |
| 232 | * <li> * - No ordering.</li> |
| 233 | * </ul> |
| 234 | */ |
| 235 | @pref public String sort; |
| 236 | |
| 237 | /** Default constructor needed for instantiation from Preferences */ |
| 238 | public AutomaticCombine() {} |
| 239 | |
| 240 | /** Instantiate an automatic tag-conflict resolver which combining the values using a separator. |
| 241 | * @param key The Tag key to match. |
| 242 | * @param description A free description. |
| 243 | * @param isRegex If regular expression must be used to match the Tag key or the value. |
| 244 | * @param separator The separator to use to combine the values. |
| 245 | * @param sort If the combined values must be sorted. |
| 246 | */ |
| 247 | public AutomaticCombine(String key, String description, boolean isRegex, String separator, String sort) { |
| 248 | this.key = key; |
| 249 | this.description = description; |
| 250 | this.isRegex = isRegex; |
| 251 | this.separator = separator; |
| 252 | this.sort = sort; |
| 253 | } |
| 254 | |
| 255 | @Override |
| 256 | public boolean matchesKey(String k) { |
| 257 | if (isRegex) { |
| 258 | return Pattern.matches(this.key, k); |
| 259 | } else { |
| 260 | return this.key.equals(k); |
| 261 | } |
| 262 | } |
| 263 | |
| 264 | Set<String> instantiateSortedSet() { |
| 265 | if ("String".equals(sort)) { |
| 266 | return new TreeSet<>(); |
| 267 | } else if ("Integer".equals(sort)) { |
| 268 | return new TreeSet<>((String v1, String v2) -> Long.valueOf(v1).compareTo(Long.valueOf(v2))); |
| 269 | } else { |
| 270 | return new LinkedHashSet<>(); |
| 271 | } |
| 272 | } |
| 273 | |
| 274 | @Override |
| 275 | public String resolve(Set<String> values) { |
| 276 | Set<String> results = instantiateSortedSet(); |
| 277 | for (String value: values) { |
| 278 | for (String part: value.split(Pattern.quote(separator))) { |
| 279 | results.add(part); |
| 280 | } |
| 281 | } |
| 282 | return String.join(separator, results); |
| 283 | } |
| 284 | |
| 285 | @Override |
| 286 | public String toString() { |
| 287 | return AutomaticCombine.class.getSimpleName() |
| 288 | + "(key='" + key + "', description='" + description + "', isRegex=" |
| 289 | + isRegex + ", separator='" + separator + "', sort='" + sort + "')"; |
| 290 | } |
| 291 | } |
| 292 | |
| 293 | /** |
| 294 | * Preference for a particular choice from a group for automatic tag conflict resolution. |
| 295 | * AutomaticChoice are grouped into {@link AutomaticChoiceGroup}. |
| 296 | */ |
| 297 | public static class AutomaticChoice { |
| 298 | |
| 299 | /** The Tag key to match. */ |
| 300 | @pref public String key; |
| 301 | |
| 302 | /** The name of the {link AutomaticChoice group} this choice belongs to. */ |
| 303 | @pref public String group; |
| 304 | |
| 305 | /** A free description. */ |
| 306 | @pref public String description = ""; |
| 307 | |
| 308 | /** If regular expression must be used to match the Tag key or the value. */ |
| 309 | @pref public boolean isRegex = false; |
| 310 | |
| 311 | /** The Tag value to match. */ |
| 312 | @pref public String value; |
| 313 | |
| 314 | /** |
| 315 | * The score to give to this choice in order to choose the best value |
| 316 | * Natural String ordering is used to identify the best score. |
| 317 | */ |
| 318 | @pref public String score; |
| 319 | |
| 320 | /** Default constructor needed for instantiation from Preferences */ |
| 321 | public AutomaticChoice() {} |
| 322 | |
| 323 | /** |
| 324 | * Instantiate a particular choice from a group for automatic tag conflict resolution. |
| 325 | * @param key The Tag key to match. |
| 326 | * @param group The name of the {link AutomaticChoice group} this choice belongs to. |
| 327 | * @param description A free description. |
| 328 | * @param isRegex If regular expression must be used to match the Tag key or the value. |
| 329 | * @param value The Tag value to match. |
| 330 | * @param score The score to give to this choice in order to choose the best value. |
| 331 | */ |
| 332 | public AutomaticChoice(String key, String group, String description, boolean isRegex, String value, String score) { |
| 333 | this.key = key; |
| 334 | this.group = group; |
| 335 | this.description = description; |
| 336 | this.isRegex = isRegex; |
| 337 | this.value = value; |
| 338 | this.score = score; |
| 339 | } |
| 340 | |
| 341 | /** |
| 342 | * Check if this choice match the given Tag value. |
| 343 | * @param v the Tag value to match. |
| 344 | * @return true if this choice correspond to the given tag value. |
| 345 | */ |
| 346 | public boolean matchesValue(String v) { |
| 347 | if (isRegex) { |
| 348 | return Pattern.matches(this.value, v); |
| 349 | } else { |
| 350 | return this.value.equals(v); |
| 351 | } |
| 352 | } |
| 353 | |
| 354 | /** |
| 355 | * Return the score associated to this choice for the given Tag value. |
| 356 | * For the result to be valid the given tag value must {@link #matchesValue(String) match} this choice. |
| 357 | * @param v the Tag value of which to get the score. |
| 358 | * @return the score associated to the given Tag value. |
| 359 | */ |
| 360 | public String computeScoreFromValue(String v) { |
| 361 | if (isRegex) { |
| 362 | return v.replaceAll("^" + this.value + "$", this.score); |
| 363 | } else { |
| 364 | return this.score; |
| 365 | } |
| 366 | } |
| 367 | |
| 368 | @Override |
| 369 | public String toString() { |
| 370 | return AutomaticChoice.class.getSimpleName() |
| 371 | + "(key='" + key + "', group='" + group + "', description='" + description |
| 372 | + "', isRegex=" + isRegex + ", value='" + value + "', score='" + score + "')"; |
| 373 | } |
| 374 | } |
| 375 | |
| 376 | |
| 377 | /** |
| 378 | * Preference for an automatic tag conflict resolver which choose from |
| 379 | * a group of possible {@link AutomaticChoice choice} values. |
| 380 | */ |
| 381 | public static class AutomaticChoiceGroup implements AutomaticTagConflictResolver { |
| 382 | |
| 383 | /** The Tag key to match. */ |
| 384 | @pref public String key; |
| 385 | |
| 386 | /** The name of the group. */ |
| 387 | public String group; |
| 388 | |
| 389 | /** If regular expression must be used to match the Tag key. */ |
| 390 | @pref public boolean isRegex = false; |
| 391 | |
| 392 | /** The list of choice to choose from. */ |
| 393 | public List<AutomaticChoice> choices; |
| 394 | |
| 395 | /** Instantiate an automatic tag conflict resolver which choose from |
| 396 | * a given list of {@link AutomaticChoice choice} values. |
| 397 | * |
| 398 | * @param key The Tag key to match. |
| 399 | * @param group The name of the group. |
| 400 | * @param isRegex If regular expression must be used to match the Tag key. |
| 401 | * @param choices The list of choice to choose from. |
| 402 | */ |
| 403 | public AutomaticChoiceGroup(String key, String group, boolean isRegex, List<AutomaticChoice> choices) { |
| 404 | this.key = key; |
| 405 | this.group = group; |
| 406 | this.isRegex = isRegex; |
| 407 | this.choices = choices; |
| 408 | } |
| 409 | |
| 410 | /** |
| 411 | * Group a given list of {@link AutomaticChoice} by the Tag key and the choice group name. |
| 412 | * @param choices the list of {@link AutomaticChoice choices} to group. |
| 413 | * @return the resulting list of group. |
| 414 | */ |
| 415 | public static Collection<AutomaticChoiceGroup> groupChoices(Collection<AutomaticChoice> choices) { |
| 416 | HashMap<Pair<String, String>, AutomaticChoiceGroup> results = new HashMap<>(); |
| 417 | for (AutomaticChoice choice: choices) { |
| 418 | Pair<String, String> id = new Pair<>(choice.key, choice.group); |
| 419 | AutomaticChoiceGroup group = results.get(id); |
| 420 | if (group == null) { |
| 421 | boolean isRegex = choice.isRegex && !Pattern.quote(choice.key).equals(choice.key); |
| 422 | group = new AutomaticChoiceGroup(choice.key, choice.group, isRegex, new ArrayList<>()); |
| 423 | results.put(id, group); |
| 424 | } |
| 425 | group.choices.add(choice); |
| 426 | } |
| 427 | return results.values(); |
| 428 | } |
| 429 | |
| 430 | @Override |
| 431 | public boolean matchesKey(String k) { |
| 432 | if (isRegex) { |
| 433 | return Pattern.matches(this.key, k); |
| 434 | } else { |
| 435 | return this.key.equals(k); |
| 436 | } |
| 437 | } |
| 438 | |
| 439 | @Override |
| 440 | public String resolve(Set<String> values) { |
| 441 | String bestScore = ""; |
| 442 | String bestValue = ""; |
| 443 | for (String value: values) { |
| 444 | String score = null; |
| 445 | for (AutomaticChoice choice: choices) { |
| 446 | if (choice.matchesValue(value)) { |
| 447 | score = choice.computeScoreFromValue(value); |
| 448 | } |
| 449 | } |
| 450 | if (score == null) { |
| 451 | // This value is not matched in this group |
| 452 | // so we can not choose from this group for this key. |
| 453 | return null; |
| 454 | } |
| 455 | if (score.compareTo(bestScore) >= 0) { |
| 456 | bestScore = score; |
| 457 | bestValue = value; |
| 458 | } |
| 459 | } |
| 460 | return bestValue; |
| 461 | } |
| 462 | |
| 463 | @Override |
| 464 | public String toString() { |
| 465 | Collection<String> stringChoices = choices.stream().map(AutomaticChoice::toString).collect(Collectors.toCollection(ArrayList::new)); |
| 466 | return AutomaticChoiceGroup.class.getSimpleName() |
| 467 | + "(key='" + key + "', group='" + group + |
| 468 | "', isRegex=" + isRegex + ", choices=(\n " |
| 469 | + String.join(",\n ", stringChoices) + "))"; |
| 470 | } |
| 471 | |
| 472 | } |
| 473 | |