| 89 | |
| 90 | |
| 91 | /** |
| 92 | * Automatically resolve some configured tag conflicts. |
| 93 | * @param tc the tag collection |
| 94 | */ |
| 95 | public static void applyAutomaticTagConflictResolution(TagCollection tc) { |
| 96 | applyAutomaticTagConflictResolution(tc, getAutomaticTagConflictCombines(), getAutomaticTagConflictChoiceGroups()); |
| 97 | } |
| 98 | |
| 99 | public static Collection<AutomaticCombine> defaultAutomaticTagConflictCombines = Arrays.asList( |
| 100 | new AutomaticCombine("tiger:tlid", "US TIGER tlid", false, ":", "Integer"), |
| 101 | new AutomaticCombine("tiger:(?!tlid$).*", "US TIGER not tlid", true, ":", "String") |
| 102 | ); |
| 103 | |
| 104 | public static Collection<AutomaticChoice> defaultAutomaticTagConflictChoices = Arrays.asList( |
| 105 | // Choice for "source" tag of data exported from the French cadastre, |
| 106 | // which ends by the exported year generating many conflicts |
| 107 | // the generated score begins with the year number to select the most recent one. |
| 108 | new AutomaticChoice("source", "FR cadastre source, manual value", true, "FR:cadastre:source", |
| 109 | "cadastre", "0"), |
| 110 | new AutomaticChoice("source", "FR cadastre source, initial format", true, "FR:cadastre:source", |
| 111 | "extraction vectorielle v1 cadastre-dgi-fr source : Direction G[eé]n[eé]rale des Imp[ôo]ts - Cadas\\. Mise [àa] jour : ([0-9]{4})", |
| 112 | "$1 1"), |
| 113 | new AutomaticChoice("source", "FR cadastre source, old format", true, "FR:cadastre:source", |
| 114 | "cadastre-dgi-fr source : Direction G[eé]n[eé]rale des Imp[ôo]ts - Cadastre\\. Mise [àa] jour : ([0-9]{4})", |
| 115 | "$1 2"), |
| 116 | new AutomaticChoice("source", "FR cadastre source, last format", true, "FR:cadastre:source", |
| 117 | "cadastre-dgi-fr source : Direction G[eé]n[ée]rale des Finances Publiques - Cadastre\\. Mise [aà] jour : ([0-9]{4})", |
| 118 | "$1 3") |
| 119 | ); |
| 120 | |
| 121 | |
| 122 | private static volatile Collection<AutomaticChoiceGroup> automaticTagConflictChoiceGroups; |
| 123 | public static Collection<AutomaticChoiceGroup> getAutomaticTagConflictChoiceGroups() { |
| 124 | if (automaticTagConflictChoiceGroups == null) { |
| 125 | automaticTagConflictChoiceGroups = AutomaticChoiceGroup.groupChoices(Main.pref.getListOfStructs( |
| 126 | "automatic-tag-conflict-resolution.choice", |
| 127 | defaultAutomaticTagConflictChoices, AutomaticChoice.class)); |
| 128 | } |
| 129 | return automaticTagConflictChoiceGroups; |
| 130 | } |
| 131 | |
| 132 | |
| 133 | private static volatile Collection<AutomaticCombine> automaticTagConflictCombines; |
| 134 | public static Collection<AutomaticCombine> getAutomaticTagConflictCombines() { |
| 135 | if (automaticTagConflictCombines == null) { |
| 136 | automaticTagConflictCombines = Main.pref.getListOfStructs( |
| 137 | "automatic-tag-conflict-resolution.combine", |
| 138 | defaultAutomaticTagConflictCombines, AutomaticCombine.class); |
| 139 | } |
| 140 | return automaticTagConflictCombines; |
| 141 | } |
| 142 | |
| 143 | |
| 144 | /** |
| 145 | * Automatically resolve some given conflicts. |
| 146 | * @param tc the tag collection. |
| 147 | * @param combines automatic tag combination to apply. |
| 148 | * @param choiceGroups automatic tag choice to apply. |
| 149 | */ |
| 150 | public static void applyAutomaticTagConflictResolution(TagCollection tc, Collection<AutomaticCombine> combines, Collection<AutomaticChoiceGroup> choiceGroups) { |
| 151 | keyLoop: |
| 152 | for(String key: tc.getKeysWithMultipleValues()) { |
| 153 | for(AutomaticCombine combine: combines) { |
| 154 | if (combine.matchesKey(key)) { |
| 155 | tc.setUniqueForKey(key, combine.resolve(tc.getValues(key))); |
| 156 | continue keyLoop; |
| 157 | } |
| 158 | } |
| 159 | for(AutomaticChoiceGroup choiceGroup: choiceGroups) { |
| 160 | if (choiceGroup.matchesKey(key)) { |
| 161 | String result = choiceGroup.resolve(tc.getValues(key)); |
| 162 | if (result != null) { |
| 163 | tc.setUniqueForKey(key, result); |
| 164 | continue keyLoop; |
| 165 | } |
| 166 | } |
| 167 | } |
| 168 | } |
| 169 | } |
| 170 | |
| 171 | |
| 172 | /** |
| 173 | * Preference for automatic tag-conflict resolution by combining the values |
| 174 | * using a separator. |
| 175 | */ |
| 176 | public static class AutomaticCombine { |
| 177 | |
| 178 | /** The Tag key to match */ |
| 179 | @pref public String key; |
| 180 | |
| 181 | /** A free description */ |
| 182 | @pref public String description; |
| 183 | |
| 184 | /** If regular expression must be used to match the Tag key or the value */ |
| 185 | @pref public boolean isRegex = false; |
| 186 | |
| 187 | /** The separator to use to combine the values */ |
| 188 | @pref public String separator = ";"; |
| 189 | |
| 190 | /** If the combined values must be sorted, possible values are |
| 191 | * <li> |
| 192 | * <ul>String - using String natural ordering |
| 193 | * <ul>Integer- using Integer natural ordering |
| 194 | * <ul> - whatever else value will keep the original order |
| 195 | * </li> |
| 196 | */ |
| 197 | @pref public String sort; |
| 198 | |
| 199 | public AutomaticCombine() {} |
| 200 | |
| 201 | public AutomaticCombine(String key, String description, boolean isRegex, String separator, String sort) { |
| 202 | this.key = key; |
| 203 | this.description = description; |
| 204 | this.isRegex = isRegex; |
| 205 | this.separator = separator; |
| 206 | this.sort = sort; |
| 207 | } |
| 208 | |
| 209 | /** Test if this case apply to the given Tag key */ |
| 210 | public boolean matchesKey(String k) { |
| 211 | if (isRegex) { |
| 212 | return Pattern.matches(this.key, k); |
| 213 | } else { |
| 214 | return this.key.equals(k); |
| 215 | } |
| 216 | } |
| 217 | |
| 218 | Set<String> instantiateSortedSet() { |
| 219 | if ("String".equals(sort)) { |
| 220 | return new TreeSet<String>(); |
| 221 | } else if ("Integer".equals(sort)) { |
| 222 | return new TreeSet<String>( |
| 223 | (String v1, String v2) -> Long.valueOf(v1).compareTo(Long.valueOf(v2))); |
| 224 | } else { |
| 225 | return new LinkedHashSet<String>(); |
| 226 | } |
| 227 | } |
| 228 | |
| 229 | public String resolve(Set<String> values) { |
| 230 | Set<String> results = instantiateSortedSet(); |
| 231 | for (String value: values) { |
| 232 | for(String part: value.split(separator)) { |
| 233 | results.add(part); |
| 234 | } |
| 235 | } |
| 236 | return String.join(separator, results); |
| 237 | } |
| 238 | |
| 239 | @Override |
| 240 | public String toString() { |
| 241 | return AutomaticCombine.class.getSimpleName() |
| 242 | + "(key='" + key + "', description='" + description + "', isRegex=" |
| 243 | + isRegex + ", separator='" + separator + "', sort='" + sort + "')"; |
| 244 | } |
| 245 | } |
| 246 | |
| 247 | /** |
| 248 | * Preference for automatic tag conflict resolution by choosing from |
| 249 | * a group of possible values. |
| 250 | * This class represent a particular choice from a group. |
| 251 | */ |
| 252 | public static class AutomaticChoice { |
| 253 | |
| 254 | /** The Tag key to match */ |
| 255 | @pref public String key; |
| 256 | /** A free description */ |
| 257 | @pref public String description; |
| 258 | |
| 259 | /** If regular expression must be used to match the Tag key or the value */ |
| 260 | @pref public boolean isRegex = false; |
| 261 | |
| 262 | /** The name of the group */ |
| 263 | @pref public String group; |
| 264 | |
| 265 | /** The Tag value to match */ |
| 266 | @pref public String value; |
| 267 | |
| 268 | /** The score to give to this choice in order to choose the best value */ |
| 269 | @pref public String score; |
| 270 | |
| 271 | public AutomaticChoice() {} |
| 272 | |
| 273 | public AutomaticChoice(String key, String description, boolean isRegex, String group, String value, String score) { |
| 274 | this.key = key; |
| 275 | this.description = description; |
| 276 | this.isRegex = isRegex; |
| 277 | this.group = group; |
| 278 | this.value = value; |
| 279 | this.score = score; |
| 280 | } |
| 281 | |
| 282 | public boolean matchValue(String v) { |
| 283 | if (isRegex) { |
| 284 | return Pattern.matches(this.value, v); |
| 285 | } else { |
| 286 | return this.value.equals(v); |
| 287 | } |
| 288 | } |
| 289 | |
| 290 | public String computeScoreFromValue(String v) { |
| 291 | if (isRegex) { |
| 292 | return v.replaceAll("^" + this.value + "$", this.score); |
| 293 | } else { |
| 294 | return this.score; |
| 295 | } |
| 296 | } |
| 297 | |
| 298 | @Override |
| 299 | public String toString() { |
| 300 | return AutomaticChoice.class.getSimpleName() |
| 301 | + "(key='" + key + "', description='" + description + "', isRegex=" |
| 302 | + isRegex + ", group='" + group + "', value='" + value + "', score='" + score + "')"; |
| 303 | } |
| 304 | } |
| 305 | |
| 306 | |
| 307 | /** |
| 308 | * Preference for automatic tag conflict resolution by choosing from |
| 309 | * a group of possible values. |
| 310 | */ |
| 311 | public static class AutomaticChoiceGroup { |
| 312 | |
| 313 | /** The Tag key to match */ |
| 314 | @pref public String key; |
| 315 | |
| 316 | /** If regular expression must be used to match the Tag key */ |
| 317 | @pref public boolean isRegex = false; |
| 318 | |
| 319 | /** The name of the group */ |
| 320 | public String group; |
| 321 | |
| 322 | /** The list of choice to choose from */ |
| 323 | public List<AutomaticChoice> choices; |
| 324 | |
| 325 | public AutomaticChoiceGroup(String key, boolean isRegex, String group, List<AutomaticChoice> choices) { |
| 326 | this.key = key; |
| 327 | this.isRegex = isRegex; |
| 328 | this.group = group; |
| 329 | this.choices = choices; |
| 330 | } |
| 331 | |
| 332 | public static Collection<AutomaticChoiceGroup> groupChoices(Collection<AutomaticChoice> choices) { |
| 333 | HashMap<Pair<String,String>, AutomaticChoiceGroup> results = new HashMap<Pair<String, String>, AutomaticChoiceGroup>(); |
| 334 | for(AutomaticChoice choice: choices) { |
| 335 | Pair<String,String> id = new Pair<String, String>(choice.key, choice.group); |
| 336 | AutomaticChoiceGroup group = results.get(id); |
| 337 | if (group == null) { |
| 338 | boolean isRegex = choice.isRegex && ! Pattern.quote(choice.key).equals(choice.key); |
| 339 | group = new AutomaticChoiceGroup(choice.key, isRegex, choice.group, new ArrayList<>()); |
| 340 | results.put(id, group); |
| 341 | } |
| 342 | group.choices.add(choice); |
| 343 | } |
| 344 | return results.values(); |
| 345 | } |
| 346 | |
| 347 | /** Test if this case apply to the given Tag key */ |
| 348 | public boolean matchesKey(String k) { |
| 349 | if (isRegex) { |
| 350 | return Pattern.matches(this.key, k); |
| 351 | } else { |
| 352 | return this.key.equals(k); |
| 353 | } |
| 354 | } |
| 355 | /** |
| 356 | * Choose the most appropriate value for this group, or return null if not possible. |
| 357 | */ |
| 358 | public String resolve(Set<String> values) { |
| 359 | String bestScore = ""; |
| 360 | String bestValue = ""; |
| 361 | for (String value: values) { |
| 362 | String score = null; |
| 363 | for(AutomaticChoice choice: choices) { |
| 364 | if (choice.matchValue(value)) { |
| 365 | score = choice.computeScoreFromValue(value); |
| 366 | } |
| 367 | } |
| 368 | if (score == null) { |
| 369 | // This value is not matched in this group |
| 370 | // so we can not choose from this group for this key. |
| 371 | return null; |
| 372 | } |
| 373 | if (score.compareTo(bestScore) >= 0) { |
| 374 | bestScore = score; |
| 375 | bestValue = value; |
| 376 | } |
| 377 | } |
| 378 | return bestValue; |
| 379 | } |
| 380 | |
| 381 | @Override |
| 382 | public String toString() { |
| 383 | Collection<String> stringChoices = choices.stream().map(AutomaticChoice::toString).collect(Collectors.toCollection(ArrayList::new)); |
| 384 | return AutomaticChoiceGroup.class.getSimpleName() |
| 385 | + "(key='" + key + "', isRegex=" |
| 386 | + isRegex + ", group='" + group + "', choices=(\n " + String.join("\n ", stringChoices) + "))"; |
| 387 | } |
| 388 | |
| 389 | } |
| 390 | |