| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 1 | package mapper |
| 2 | |
| 3 | import ( |
| 4 | "encoding/json" |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 5 | "testing" |
| 6 | |
| Akron | 2ef703c | 2025-07-03 15:57:42 +0200 | [diff] [blame] | 7 | "github.com/KorAP/Koral-Mapper/ast" |
| 8 | "github.com/KorAP/Koral-Mapper/config" |
| 9 | "github.com/KorAP/Koral-Mapper/matcher" |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 10 | "github.com/stretchr/testify/assert" |
| 11 | "github.com/stretchr/testify/require" |
| 12 | ) |
| 13 | |
| 14 | func TestMapper(t *testing.T) { |
| Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 15 | // Create test mapping list |
| 16 | mappingList := config.MappingList{ |
| 17 | ID: "test-mapper", |
| 18 | FoundryA: "opennlp", |
| 19 | LayerA: "p", |
| 20 | FoundryB: "upos", |
| 21 | LayerB: "p", |
| 22 | Mappings: []config.MappingRule{ |
| 23 | "[PIDAT] <> [opennlp/p=PIDAT & opennlp/p=AdjType:Pdt]", |
| 24 | "[DET] <> [opennlp/p=DET]", |
| 25 | }, |
| 26 | } |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 27 | |
| 28 | // Create a new mapper |
| Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 29 | m, err := NewMapper([]config.MappingList{mappingList}) |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 30 | require.NoError(t, err) |
| 31 | |
| 32 | tests := []struct { |
| 33 | name string |
| 34 | mappingID string |
| 35 | opts MappingOptions |
| 36 | input string |
| 37 | expected string |
| 38 | expectError bool |
| 39 | }{ |
| 40 | { |
| 41 | name: "Simple A to B mapping", |
| 42 | mappingID: "test-mapper", |
| 43 | opts: MappingOptions{ |
| 44 | Direction: AtoB, |
| 45 | }, |
| 46 | input: `{ |
| 47 | "@type": "koral:token", |
| 48 | "wrap": { |
| 49 | "@type": "koral:term", |
| 50 | "foundry": "opennlp", |
| 51 | "key": "PIDAT", |
| 52 | "layer": "p", |
| 53 | "match": "match:eq" |
| 54 | } |
| 55 | }`, |
| 56 | expected: `{ |
| 57 | "@type": "koral:token", |
| 58 | "wrap": { |
| 59 | "@type": "koral:termGroup", |
| 60 | "operands": [ |
| 61 | { |
| 62 | "@type": "koral:term", |
| 63 | "foundry": "opennlp", |
| 64 | "key": "PIDAT", |
| 65 | "layer": "p", |
| 66 | "match": "match:eq" |
| 67 | }, |
| 68 | { |
| 69 | "@type": "koral:term", |
| 70 | "foundry": "opennlp", |
| 71 | "key": "AdjType", |
| 72 | "layer": "p", |
| 73 | "match": "match:eq", |
| 74 | "value": "Pdt" |
| 75 | } |
| 76 | ], |
| 77 | "relation": "relation:and" |
| 78 | } |
| 79 | }`, |
| 80 | }, |
| 81 | { |
| Akron | 0d9117c | 2025-05-27 15:20:21 +0200 | [diff] [blame] | 82 | name: "Simple A to B mapping with rewrites", |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 83 | mappingID: "test-mapper", |
| 84 | opts: MappingOptions{ |
| Akron | 0d9117c | 2025-05-27 15:20:21 +0200 | [diff] [blame] | 85 | Direction: AtoB, |
| 86 | AddRewrites: true, |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 87 | }, |
| 88 | input: `{ |
| 89 | "@type": "koral:token", |
| 90 | "wrap": { |
| Akron | a1a183f | 2025-05-26 17:47:33 +0200 | [diff] [blame] | 91 | "@type": "koral:term", |
| 92 | "foundry": "opennlp", |
| 93 | "key": "PIDAT", |
| 94 | "layer": "p", |
| 95 | "match": "match:eq" |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 96 | } |
| 97 | }`, |
| 98 | expected: `{ |
| 99 | "@type": "koral:token", |
| 100 | "wrap": { |
| Akron | 0d9117c | 2025-05-27 15:20:21 +0200 | [diff] [blame] | 101 | "@type": "koral:termGroup", |
| 102 | "operands": [ |
| 103 | { |
| 104 | "@type": "koral:term", |
| 105 | "foundry": "opennlp", |
| 106 | "key": "PIDAT", |
| 107 | "layer": "p", |
| 108 | "match": "match:eq" |
| 109 | }, |
| 110 | { |
| 111 | "@type": "koral:term", |
| 112 | "foundry": "opennlp", |
| 113 | "key": "AdjType", |
| 114 | "layer": "p", |
| 115 | "match": "match:eq", |
| 116 | "value": "Pdt" |
| 117 | } |
| 118 | ], |
| 119 | "relation": "relation:and", |
| 120 | "rewrites": [ |
| 121 | { |
| 122 | "@type": "koral:rewrite", |
| Akron | 2f93c58 | 2026-02-19 16:49:13 +0100 | [diff] [blame] | 123 | "editor": "Koral-Mapper", |
| Akron | 8a87d9a | 2025-05-27 15:30:48 +0200 | [diff] [blame] | 124 | "original": { |
| Akron | 0d9117c | 2025-05-27 15:20:21 +0200 | [diff] [blame] | 125 | "@type": "koral:term", |
| 126 | "foundry": "opennlp", |
| 127 | "key": "PIDAT", |
| 128 | "layer": "p", |
| 129 | "match": "match:eq" |
| 130 | } |
| 131 | } |
| 132 | ] |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 133 | } |
| 134 | }`, |
| 135 | }, |
| 136 | { |
| Akron | 0d9117c | 2025-05-27 15:20:21 +0200 | [diff] [blame] | 137 | name: "Mapping with foundry override and rewrites", |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 138 | mappingID: "test-mapper", |
| 139 | opts: MappingOptions{ |
| Akron | 0d9117c | 2025-05-27 15:20:21 +0200 | [diff] [blame] | 140 | Direction: AtoB, |
| 141 | FoundryB: "custom", |
| 142 | AddRewrites: true, |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 143 | }, |
| 144 | input: `{ |
| 145 | "@type": "koral:token", |
| 146 | "wrap": { |
| 147 | "@type": "koral:term", |
| 148 | "foundry": "opennlp", |
| 149 | "key": "PIDAT", |
| 150 | "layer": "p", |
| 151 | "match": "match:eq" |
| 152 | } |
| 153 | }`, |
| 154 | expected: `{ |
| 155 | "@type": "koral:token", |
| 156 | "wrap": { |
| 157 | "@type": "koral:termGroup", |
| 158 | "operands": [ |
| 159 | { |
| 160 | "@type": "koral:term", |
| 161 | "foundry": "custom", |
| 162 | "key": "PIDAT", |
| 163 | "layer": "p", |
| 164 | "match": "match:eq" |
| 165 | }, |
| 166 | { |
| 167 | "@type": "koral:term", |
| 168 | "foundry": "custom", |
| 169 | "key": "AdjType", |
| 170 | "layer": "p", |
| 171 | "match": "match:eq", |
| 172 | "value": "Pdt" |
| 173 | } |
| 174 | ], |
| Akron | 0d9117c | 2025-05-27 15:20:21 +0200 | [diff] [blame] | 175 | "relation": "relation:and", |
| 176 | "rewrites": [ |
| 177 | { |
| 178 | "@type": "koral:rewrite", |
| Akron | 2f93c58 | 2026-02-19 16:49:13 +0100 | [diff] [blame] | 179 | "editor": "Koral-Mapper", |
| Akron | 8a87d9a | 2025-05-27 15:30:48 +0200 | [diff] [blame] | 180 | "original": { |
| 181 | "@type": "koral:term", |
| 182 | "foundry": "opennlp", |
| 183 | "key": "PIDAT", |
| 184 | "layer": "p", |
| 185 | "match": "match:eq" |
| 186 | } |
| Akron | 0d9117c | 2025-05-27 15:20:21 +0200 | [diff] [blame] | 187 | } |
| 188 | ] |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 189 | } |
| 190 | }`, |
| 191 | }, |
| 192 | { |
| Akron | 0d9117c | 2025-05-27 15:20:21 +0200 | [diff] [blame] | 193 | name: "B to A direction", |
| 194 | mappingID: "test-mapper", |
| 195 | opts: MappingOptions{ |
| 196 | Direction: BtoA, |
| 197 | }, |
| 198 | input: `{ |
| 199 | "@type": "koral:token", |
| 200 | "wrap": { |
| 201 | "@type": "koral:term", |
| 202 | "foundry": "opennlp", |
| 203 | "key": "PIDAT", |
| 204 | "layer": "p", |
| 205 | "match": "match:eq" |
| 206 | } |
| 207 | }`, |
| 208 | expected: `{ |
| 209 | "@type": "koral:token", |
| 210 | "wrap": { |
| 211 | "@type": "koral:term", |
| 212 | "foundry": "opennlp", |
| 213 | "key": "PIDAT", |
| 214 | "layer": "p", |
| 215 | "match": "match:eq" |
| 216 | } |
| 217 | }`, |
| 218 | expectError: false, |
| 219 | }, |
| 220 | { |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 221 | name: "Invalid mapping ID", |
| 222 | mappingID: "nonexistent", |
| 223 | opts: MappingOptions{ |
| 224 | Direction: AtoB, |
| 225 | }, |
| 226 | input: `{ |
| 227 | "@type": "koral:token", |
| 228 | "wrap": { |
| 229 | "@type": "koral:term", |
| 230 | "foundry": "opennlp", |
| 231 | "key": "PIDAT", |
| 232 | "layer": "p", |
| 233 | "match": "match:eq" |
| 234 | } |
| 235 | }`, |
| 236 | expectError: true, |
| 237 | }, |
| 238 | { |
| 239 | name: "Invalid direction", |
| 240 | mappingID: "test-mapper", |
| 241 | opts: MappingOptions{ |
| Akron | a1a183f | 2025-05-26 17:47:33 +0200 | [diff] [blame] | 242 | Direction: Direction(false), |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 243 | }, |
| 244 | input: `{ |
| 245 | "@type": "koral:token", |
| 246 | "wrap": { |
| 247 | "@type": "koral:term", |
| 248 | "foundry": "opennlp", |
| 249 | "key": "PIDAT", |
| 250 | "layer": "p", |
| 251 | "match": "match:eq" |
| 252 | } |
| 253 | }`, |
| Akron | a1a183f | 2025-05-26 17:47:33 +0200 | [diff] [blame] | 254 | expected: `{ |
| 255 | "@type": "koral:token", |
| 256 | "wrap": { |
| 257 | "@type": "koral:term", |
| 258 | "foundry": "opennlp", |
| 259 | "key": "PIDAT", |
| 260 | "layer": "p", |
| 261 | "match": "match:eq" |
| 262 | } |
| 263 | }`, |
| 264 | expectError: false, |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 265 | }, |
| Akron | 8f1970f | 2025-05-30 12:52:03 +0200 | [diff] [blame] | 266 | { |
| 267 | name: "Query with legacy rewrite field names", |
| 268 | mappingID: "test-mapper", |
| 269 | opts: MappingOptions{ |
| 270 | Direction: AtoB, |
| 271 | }, |
| 272 | input: `{ |
| 273 | "@type": "koral:token", |
| 274 | "rewrites": [ |
| 275 | { |
| 276 | "@type": "koral:rewrite", |
| 277 | "_comment": "Legacy rewrite with source instead of editor", |
| 278 | "source": "LegacyEditor", |
| 279 | "operation": "operation:legacy", |
| 280 | "origin": "LegacySource" |
| 281 | } |
| 282 | ], |
| 283 | "wrap": { |
| 284 | "@type": "koral:term", |
| 285 | "foundry": "opennlp", |
| 286 | "key": "PIDAT", |
| 287 | "layer": "p", |
| 288 | "match": "match:eq" |
| 289 | } |
| 290 | }`, |
| 291 | expected: `{ |
| 292 | "@type": "koral:token", |
| 293 | "rewrites": [ |
| 294 | { |
| 295 | "@type": "koral:rewrite", |
| 296 | "_comment": "Legacy rewrite with source instead of editor", |
| 297 | "editor": "LegacyEditor", |
| 298 | "operation": "operation:legacy", |
| 299 | "src": "LegacySource" |
| 300 | } |
| 301 | ], |
| 302 | "wrap": { |
| 303 | "@type": "koral:termGroup", |
| 304 | "operands": [ |
| 305 | { |
| 306 | "@type": "koral:term", |
| 307 | "foundry": "opennlp", |
| 308 | "key": "PIDAT", |
| 309 | "layer": "p", |
| 310 | "match": "match:eq" |
| 311 | }, |
| 312 | { |
| 313 | "@type": "koral:term", |
| 314 | "foundry": "opennlp", |
| 315 | "key": "AdjType", |
| 316 | "layer": "p", |
| 317 | "match": "match:eq", |
| 318 | "value": "Pdt" |
| 319 | } |
| 320 | ], |
| 321 | "relation": "relation:and" |
| 322 | } |
| 323 | }`, |
| 324 | }, |
| 325 | { |
| 326 | name: "Query with mixed legacy and modern rewrite fields", |
| 327 | mappingID: "test-mapper", |
| 328 | opts: MappingOptions{ |
| 329 | Direction: AtoB, |
| 330 | }, |
| 331 | input: `{ |
| 332 | "@type": "koral:token", |
| 333 | "rewrites": [ |
| 334 | { |
| 335 | "@type": "koral:rewrite", |
| 336 | "_comment": "Modern rewrite", |
| 337 | "editor": "ModernEditor", |
| 338 | "operation": "operation:modern", |
| 339 | "original": { |
| 340 | "@type": "koral:term", |
| 341 | "foundry": "original", |
| 342 | "key": "original-key" |
| 343 | } |
| 344 | }, |
| 345 | { |
| 346 | "@type": "koral:rewrite", |
| 347 | "_comment": "Legacy rewrite with precedence test", |
| 348 | "editor": "PreferredEditor", |
| 349 | "source": "IgnoredSource", |
| 350 | "operation": "operation:precedence", |
| 351 | "original": "PreferredOriginal", |
| 352 | "src": "IgnoredSrc", |
| 353 | "origin": "IgnoredOrigin" |
| 354 | } |
| 355 | ], |
| 356 | "wrap": { |
| 357 | "@type": "koral:term", |
| 358 | "foundry": "opennlp", |
| 359 | "key": "PIDAT", |
| 360 | "layer": "p", |
| 361 | "match": "match:eq" |
| 362 | } |
| 363 | }`, |
| 364 | expected: `{ |
| 365 | "@type": "koral:token", |
| 366 | "rewrites": [ |
| 367 | { |
| 368 | "@type": "koral:rewrite", |
| 369 | "_comment": "Modern rewrite", |
| 370 | "editor": "ModernEditor", |
| 371 | "operation": "operation:modern", |
| 372 | "original": { |
| 373 | "@type": "koral:term", |
| 374 | "foundry": "original", |
| 375 | "key": "original-key" |
| 376 | } |
| 377 | }, |
| 378 | { |
| 379 | "@type": "koral:rewrite", |
| 380 | "_comment": "Legacy rewrite with precedence test", |
| 381 | "editor": "PreferredEditor", |
| 382 | "operation": "operation:precedence", |
| 383 | "original": "PreferredOriginal" |
| 384 | } |
| 385 | ], |
| 386 | "wrap": { |
| 387 | "@type": "koral:termGroup", |
| 388 | "operands": [ |
| 389 | { |
| 390 | "@type": "koral:term", |
| 391 | "foundry": "opennlp", |
| 392 | "key": "PIDAT", |
| 393 | "layer": "p", |
| 394 | "match": "match:eq" |
| 395 | }, |
| 396 | { |
| 397 | "@type": "koral:term", |
| 398 | "foundry": "opennlp", |
| 399 | "key": "AdjType", |
| 400 | "layer": "p", |
| 401 | "match": "match:eq", |
| 402 | "value": "Pdt" |
| 403 | } |
| 404 | ], |
| 405 | "relation": "relation:and" |
| 406 | } |
| 407 | }`, |
| 408 | }, |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 409 | } |
| 410 | |
| 411 | for _, tt := range tests { |
| 412 | t.Run(tt.name, func(t *testing.T) { |
| 413 | // Parse input JSON |
| Akron | 121c66e | 2025-06-02 16:34:05 +0200 | [diff] [blame] | 414 | var inputData any |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 415 | err := json.Unmarshal([]byte(tt.input), &inputData) |
| 416 | require.NoError(t, err) |
| 417 | |
| 418 | // Apply mappings |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 419 | result, err := m.ApplyQueryMappings(tt.mappingID, tt.opts, inputData) |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 420 | if tt.expectError { |
| 421 | assert.Error(t, err) |
| 422 | return |
| 423 | } |
| 424 | require.NoError(t, err) |
| 425 | |
| 426 | // Parse expected JSON |
| Akron | 121c66e | 2025-06-02 16:34:05 +0200 | [diff] [blame] | 427 | var expectedData any |
| Akron | 32d53de | 2025-05-22 13:45:32 +0200 | [diff] [blame] | 428 | err = json.Unmarshal([]byte(tt.expected), &expectedData) |
| 429 | require.NoError(t, err) |
| 430 | |
| 431 | // Compare results |
| 432 | assert.Equal(t, expectedData, result) |
| 433 | }) |
| 434 | } |
| 435 | } |
| Akron | d5850f8 | 2025-05-23 16:44:44 +0200 | [diff] [blame] | 436 | |
| Akron | cc83eb5 | 2025-05-27 14:39:12 +0200 | [diff] [blame] | 437 | func TestTokenToTermGroupWithRewrites(t *testing.T) { |
| 438 | // Create test mapping list specifically for token to termGroup test |
| 439 | mappingList := config.MappingList{ |
| 440 | ID: "test-token-to-termgroup", |
| 441 | FoundryA: "opennlp", |
| 442 | LayerA: "p", |
| Akron | 422cd25 | 2026-05-19 16:31:19 +0200 | [diff] [blame] | 443 | FoundryB: "tt", |
| 444 | LayerB: "pos", |
| Akron | cc83eb5 | 2025-05-27 14:39:12 +0200 | [diff] [blame] | 445 | Mappings: []config.MappingRule{ |
| 446 | "[PIDAT] <> [opennlp/p=PIDAT & opennlp/p=AdjType:Pdt]", |
| 447 | }, |
| 448 | } |
| 449 | |
| 450 | // Create a new mapper |
| 451 | m, err := NewMapper([]config.MappingList{mappingList}) |
| 452 | require.NoError(t, err) |
| 453 | |
| 454 | input := `{ |
| 455 | "@type": "koral:token", |
| 456 | "rewrites": [ |
| 457 | { |
| 458 | "@type": "koral:rewrite", |
| 459 | "_comment": "This rewrite should be preserved", |
| 460 | "editor": "TestEditor", |
| 461 | "operation": "operation:test", |
| 462 | "src": "TestSource" |
| 463 | } |
| 464 | ], |
| 465 | "wrap": { |
| 466 | "@type": "koral:term", |
| 467 | "foundry": "opennlp", |
| 468 | "key": "PIDAT", |
| 469 | "layer": "p", |
| 470 | "match": "match:eq" |
| 471 | } |
| 472 | }` |
| 473 | |
| 474 | expected := `{ |
| 475 | "@type": "koral:token", |
| 476 | "rewrites": [ |
| 477 | { |
| 478 | "@type": "koral:rewrite", |
| 479 | "_comment": "This rewrite should be preserved", |
| 480 | "editor": "TestEditor", |
| 481 | "operation": "operation:test", |
| 482 | "src": "TestSource" |
| 483 | } |
| 484 | ], |
| 485 | "wrap": { |
| 486 | "@type": "koral:termGroup", |
| 487 | "operands": [ |
| 488 | { |
| 489 | "@type": "koral:term", |
| 490 | "foundry": "opennlp", |
| 491 | "key": "PIDAT", |
| 492 | "layer": "p", |
| 493 | "match": "match:eq" |
| 494 | }, |
| 495 | { |
| 496 | "@type": "koral:term", |
| 497 | "foundry": "opennlp", |
| 498 | "key": "AdjType", |
| 499 | "layer": "p", |
| 500 | "match": "match:eq", |
| 501 | "value": "Pdt" |
| 502 | } |
| 503 | ], |
| 504 | "relation": "relation:and" |
| 505 | } |
| 506 | }` |
| 507 | |
| 508 | // Parse input JSON |
| Akron | 121c66e | 2025-06-02 16:34:05 +0200 | [diff] [blame] | 509 | var inputData any |
| Akron | cc83eb5 | 2025-05-27 14:39:12 +0200 | [diff] [blame] | 510 | err = json.Unmarshal([]byte(input), &inputData) |
| 511 | require.NoError(t, err) |
| 512 | |
| 513 | // Apply mappings |
| 514 | result, err := m.ApplyQueryMappings("test-token-to-termgroup", MappingOptions{Direction: AtoB}, inputData) |
| 515 | require.NoError(t, err) |
| 516 | |
| 517 | // Parse expected JSON |
| Akron | 121c66e | 2025-06-02 16:34:05 +0200 | [diff] [blame] | 518 | var expectedData any |
| Akron | cc83eb5 | 2025-05-27 14:39:12 +0200 | [diff] [blame] | 519 | err = json.Unmarshal([]byte(expected), &expectedData) |
| 520 | require.NoError(t, err) |
| 521 | |
| 522 | // Compare results |
| 523 | assert.Equal(t, expectedData, result) |
| 524 | } |
| 525 | |
| Akron | d5850f8 | 2025-05-23 16:44:44 +0200 | [diff] [blame] | 526 | func TestMatchComplexPatterns(t *testing.T) { |
| 527 | tests := []struct { |
| 528 | name string |
| 529 | pattern ast.Pattern |
| 530 | replacement ast.Replacement |
| 531 | input ast.Node |
| 532 | expected ast.Node |
| 533 | }{ |
| 534 | { |
| 535 | name: "Deep nested pattern with mixed operators", |
| 536 | pattern: ast.Pattern{ |
| 537 | Root: &ast.TermGroup{ |
| 538 | Operands: []ast.Node{ |
| 539 | &ast.Term{ |
| 540 | Key: "A", |
| 541 | Match: ast.MatchEqual, |
| 542 | }, |
| 543 | &ast.TermGroup{ |
| 544 | Operands: []ast.Node{ |
| 545 | &ast.Term{ |
| 546 | Key: "B", |
| 547 | Match: ast.MatchEqual, |
| 548 | }, |
| 549 | &ast.TermGroup{ |
| 550 | Operands: []ast.Node{ |
| 551 | &ast.Term{ |
| 552 | Key: "C", |
| 553 | Match: ast.MatchEqual, |
| 554 | }, |
| 555 | &ast.Term{ |
| 556 | Key: "D", |
| 557 | Match: ast.MatchEqual, |
| 558 | }, |
| 559 | }, |
| 560 | Relation: ast.AndRelation, |
| 561 | }, |
| 562 | }, |
| 563 | Relation: ast.OrRelation, |
| 564 | }, |
| 565 | }, |
| 566 | Relation: ast.AndRelation, |
| 567 | }, |
| 568 | }, |
| 569 | replacement: ast.Replacement{ |
| 570 | Root: &ast.Term{ |
| 571 | Key: "RESULT", |
| 572 | Match: ast.MatchEqual, |
| 573 | }, |
| 574 | }, |
| 575 | input: &ast.TermGroup{ |
| 576 | Operands: []ast.Node{ |
| 577 | &ast.Term{ |
| 578 | Key: "A", |
| 579 | Match: ast.MatchEqual, |
| 580 | }, |
| 581 | &ast.TermGroup{ |
| 582 | Operands: []ast.Node{ |
| 583 | &ast.Term{ |
| 584 | Key: "C", |
| 585 | Match: ast.MatchEqual, |
| 586 | }, |
| 587 | &ast.Term{ |
| 588 | Key: "D", |
| 589 | Match: ast.MatchEqual, |
| 590 | }, |
| 591 | }, |
| 592 | Relation: ast.AndRelation, |
| 593 | }, |
| 594 | }, |
| 595 | Relation: ast.AndRelation, |
| 596 | }, |
| 597 | expected: &ast.Term{ |
| 598 | Key: "RESULT", |
| 599 | Match: ast.MatchEqual, |
| 600 | }, |
| 601 | }, |
| 602 | } |
| 603 | |
| 604 | for _, tt := range tests { |
| 605 | t.Run(tt.name, func(t *testing.T) { |
| 606 | m, err := matcher.NewMatcher(tt.pattern, tt.replacement) |
| 607 | require.NoError(t, err) |
| 608 | result := m.Replace(tt.input) |
| 609 | assert.Equal(t, tt.expected, result) |
| 610 | }) |
| 611 | } |
| 612 | } |
| 613 | |
| 614 | func TestInvalidPatternReplacement(t *testing.T) { |
| Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 615 | // Create test mapping list |
| 616 | mappingList := config.MappingList{ |
| 617 | ID: "test-mapper", |
| 618 | FoundryA: "opennlp", |
| 619 | LayerA: "p", |
| 620 | FoundryB: "upos", |
| 621 | LayerB: "p", |
| 622 | Mappings: []config.MappingRule{ |
| 623 | "[PIDAT] <> [opennlp/p=PIDAT & opennlp/p=AdjType:Pdt]", |
| 624 | }, |
| 625 | } |
| Akron | d5850f8 | 2025-05-23 16:44:44 +0200 | [diff] [blame] | 626 | |
| 627 | // Create a new mapper |
| Akron | a00d475 | 2025-05-26 17:34:36 +0200 | [diff] [blame] | 628 | m, err := NewMapper([]config.MappingList{mappingList}) |
| Akron | d5850f8 | 2025-05-23 16:44:44 +0200 | [diff] [blame] | 629 | require.NoError(t, err) |
| 630 | |
| 631 | tests := []struct { |
| 632 | name string |
| 633 | input string |
| 634 | expectError bool |
| 635 | errorMsg string |
| 636 | }{ |
| 637 | { |
| 638 | name: "Invalid input - empty term group", |
| 639 | input: `{ |
| 640 | "@type": "koral:token", |
| 641 | "wrap": { |
| 642 | "@type": "koral:termGroup", |
| 643 | "operands": [], |
| 644 | "relation": "relation:and" |
| 645 | } |
| 646 | }`, |
| 647 | expectError: true, |
| 648 | errorMsg: "failed to parse JSON into AST: error parsing wrapped node: term group must have at least one operand", |
| 649 | }, |
| 650 | } |
| 651 | |
| 652 | for _, tt := range tests { |
| 653 | t.Run(tt.name, func(t *testing.T) { |
| 654 | var inputData any |
| 655 | err := json.Unmarshal([]byte(tt.input), &inputData) |
| 656 | require.NoError(t, err) |
| 657 | |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 658 | result, err := m.ApplyQueryMappings("test-mapper", MappingOptions{Direction: AtoB}, inputData) |
| Akron | d5850f8 | 2025-05-23 16:44:44 +0200 | [diff] [blame] | 659 | if tt.expectError { |
| 660 | assert.Error(t, err) |
| 661 | assert.Equal(t, tt.errorMsg, err.Error()) |
| 662 | assert.Nil(t, result) |
| 663 | } else { |
| 664 | assert.NoError(t, err) |
| 665 | assert.NotNil(t, result) |
| 666 | } |
| 667 | }) |
| 668 | } |
| 669 | } |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 670 | |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 671 | func TestMultiFieldRewritesAreReversible(t *testing.T) { |
| 672 | mappingList := config.MappingList{ |
| 673 | ID: "multi-field", |
| 674 | FoundryA: "opennlp", |
| 675 | LayerA: "p", |
| 676 | FoundryB: "upos", |
| 677 | LayerB: "pos", |
| 678 | Mappings: []config.MappingRule{ |
| 679 | "[DET] <> [PRON]", |
| 680 | }, |
| 681 | } |
| 682 | |
| 683 | m, err := NewMapper([]config.MappingList{mappingList}) |
| 684 | require.NoError(t, err) |
| 685 | |
| 686 | tests := []struct { |
| 687 | name string |
| 688 | opts MappingOptions |
| 689 | input string |
| 690 | expected string |
| 691 | }{ |
| 692 | { |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 693 | name: "Multi-field change: single rewrite with full original", |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 694 | opts: MappingOptions{ |
| 695 | Direction: AtoB, |
| 696 | AddRewrites: true, |
| 697 | }, |
| 698 | input: `{ |
| 699 | "@type": "koral:token", |
| 700 | "wrap": { |
| 701 | "@type": "koral:term", |
| 702 | "foundry": "opennlp", |
| 703 | "key": "DET", |
| 704 | "layer": "p", |
| 705 | "match": "match:eq" |
| 706 | } |
| 707 | }`, |
| 708 | expected: `{ |
| 709 | "@type": "koral:token", |
| 710 | "wrap": { |
| 711 | "@type": "koral:term", |
| 712 | "foundry": "upos", |
| 713 | "key": "PRON", |
| 714 | "layer": "pos", |
| 715 | "match": "match:eq", |
| 716 | "rewrites": [ |
| 717 | { |
| 718 | "@type": "koral:rewrite", |
| 719 | "editor": "Koral-Mapper", |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 720 | "original": { |
| 721 | "@type": "koral:term", |
| 722 | "foundry": "opennlp", |
| 723 | "key": "DET", |
| 724 | "layer": "p", |
| 725 | "match": "match:eq" |
| 726 | } |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 727 | } |
| 728 | ] |
| 729 | } |
| 730 | }`, |
| 731 | }, |
| 732 | { |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 733 | name: "Reverse direction: single rewrite with full original", |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 734 | opts: MappingOptions{ |
| 735 | Direction: BtoA, |
| 736 | AddRewrites: true, |
| 737 | }, |
| 738 | input: `{ |
| 739 | "@type": "koral:token", |
| 740 | "wrap": { |
| 741 | "@type": "koral:term", |
| 742 | "foundry": "upos", |
| 743 | "key": "PRON", |
| 744 | "layer": "pos", |
| 745 | "match": "match:eq" |
| 746 | } |
| 747 | }`, |
| 748 | expected: `{ |
| 749 | "@type": "koral:token", |
| 750 | "wrap": { |
| 751 | "@type": "koral:term", |
| 752 | "foundry": "opennlp", |
| 753 | "key": "DET", |
| 754 | "layer": "p", |
| 755 | "match": "match:eq", |
| 756 | "rewrites": [ |
| 757 | { |
| 758 | "@type": "koral:rewrite", |
| 759 | "editor": "Koral-Mapper", |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 760 | "original": { |
| 761 | "@type": "koral:term", |
| 762 | "foundry": "upos", |
| 763 | "key": "PRON", |
| 764 | "layer": "pos", |
| 765 | "match": "match:eq" |
| 766 | } |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 767 | } |
| 768 | ] |
| 769 | } |
| 770 | }`, |
| 771 | }, |
| 772 | } |
| 773 | |
| 774 | for _, tt := range tests { |
| 775 | t.Run(tt.name, func(t *testing.T) { |
| 776 | var inputData any |
| 777 | err := json.Unmarshal([]byte(tt.input), &inputData) |
| 778 | require.NoError(t, err) |
| 779 | |
| 780 | result, err := m.ApplyQueryMappings("multi-field", tt.opts, inputData) |
| 781 | require.NoError(t, err) |
| 782 | |
| 783 | var expectedData any |
| 784 | err = json.Unmarshal([]byte(tt.expected), &expectedData) |
| 785 | require.NoError(t, err) |
| 786 | |
| 787 | assert.Equal(t, expectedData, result) |
| 788 | }) |
| 789 | } |
| 790 | } |
| 791 | |
| 792 | func TestSingleFieldRewrite(t *testing.T) { |
| 793 | mappingList := config.MappingList{ |
| 794 | ID: "same-fl", |
| 795 | FoundryA: "opennlp", |
| 796 | LayerA: "p", |
| 797 | FoundryB: "opennlp", |
| Akron | 422cd25 | 2026-05-19 16:31:19 +0200 | [diff] [blame] | 798 | LayerB: "pos", |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 799 | Mappings: []config.MappingRule{ |
| 800 | "[DET] <> [PRON]", |
| 801 | }, |
| 802 | } |
| 803 | |
| 804 | m, err := NewMapper([]config.MappingList{mappingList}) |
| 805 | require.NoError(t, err) |
| 806 | |
| 807 | var inputData any |
| 808 | err = json.Unmarshal([]byte(`{ |
| 809 | "@type": "koral:token", |
| 810 | "wrap": { |
| 811 | "@type": "koral:term", |
| 812 | "foundry": "opennlp", |
| 813 | "key": "DET", |
| 814 | "layer": "p", |
| 815 | "match": "match:eq" |
| 816 | } |
| 817 | }`), &inputData) |
| 818 | require.NoError(t, err) |
| 819 | |
| 820 | result, err := m.ApplyQueryMappings("same-fl", MappingOptions{ |
| 821 | Direction: AtoB, |
| 822 | AddRewrites: true, |
| 823 | }, inputData) |
| 824 | require.NoError(t, err) |
| 825 | |
| 826 | var expectedData any |
| 827 | err = json.Unmarshal([]byte(`{ |
| 828 | "@type": "koral:token", |
| 829 | "wrap": { |
| 830 | "@type": "koral:term", |
| 831 | "foundry": "opennlp", |
| 832 | "key": "PRON", |
| Akron | 422cd25 | 2026-05-19 16:31:19 +0200 | [diff] [blame] | 833 | "layer": "pos", |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 834 | "match": "match:eq", |
| 835 | "rewrites": [ |
| 836 | { |
| 837 | "@type": "koral:rewrite", |
| 838 | "editor": "Koral-Mapper", |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 839 | "original": { |
| 840 | "@type": "koral:term", |
| 841 | "foundry": "opennlp", |
| 842 | "key": "DET", |
| 843 | "layer": "p", |
| 844 | "match": "match:eq" |
| 845 | } |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 846 | } |
| 847 | ] |
| 848 | } |
| 849 | }`), &expectedData) |
| 850 | require.NoError(t, err) |
| 851 | |
| 852 | assert.Equal(t, expectedData, result) |
| 853 | } |
| 854 | |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 855 | func TestBuildRewritesSingleObjectRewrite(t *testing.T) { |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 856 | tests := []struct { |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 857 | name string |
| 858 | original *ast.Term |
| 859 | new_ *ast.Term |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 860 | }{ |
| 861 | { |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 862 | name: "All fields change", |
| 863 | original: &ast.Term{Foundry: "a", Layer: "l1", Key: "k1", Value: "v1", Match: ast.MatchEqual}, |
| 864 | new_: &ast.Term{Foundry: "b", Layer: "l2", Key: "k2", Value: "v2", Match: ast.MatchEqual}, |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 865 | }, |
| 866 | { |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 867 | name: "Single field injection: empty value becomes non-empty", |
| 868 | original: &ast.Term{Foundry: "a", Layer: "l", Key: "k", Match: ast.MatchEqual}, |
| 869 | new_: &ast.Term{Foundry: "a", Layer: "l", Key: "k", Value: "v", Match: ast.MatchEqual}, |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 870 | }, |
| 871 | { |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 872 | name: "Single field deletion: non-empty value becomes empty", |
| 873 | original: &ast.Term{Foundry: "a", Layer: "l", Key: "k", Value: "v", Match: ast.MatchEqual}, |
| 874 | new_: &ast.Term{Foundry: "a", Layer: "l", Key: "k", Match: ast.MatchEqual}, |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 875 | }, |
| 876 | } |
| 877 | |
| 878 | for _, tt := range tests { |
| 879 | t.Run(tt.name, func(t *testing.T) { |
| 880 | rewrites := buildRewrites(tt.original, tt.new_) |
| Akron | a017435 | 2026-05-19 17:04:42 +0200 | [diff] [blame] | 881 | require.Len(t, rewrites, 1, "one rule application should produce exactly one rewrite") |
| 882 | rw := rewrites[0] |
| 883 | assert.Equal(t, RewriteEditor, rw.Editor) |
| 884 | assert.Empty(t, rw.Scope, "object-level rewrite should have no scope") |
| 885 | assert.NotNil(t, rw.Original, "rewrite should contain the full original") |
| 886 | originalMap, ok := rw.Original.(map[string]any) |
| 887 | require.True(t, ok) |
| 888 | assert.Equal(t, "koral:term", originalMap["@type"]) |
| Akron | 958fc47 | 2026-05-19 13:58:52 +0200 | [diff] [blame] | 889 | }) |
| 890 | } |
| 891 | } |
| 892 | |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 893 | func TestQueryWrapperMappings(t *testing.T) { |
| 894 | |
| 895 | mappingList := config.MappingList{ |
| 896 | ID: "test-wrapper", |
| 897 | FoundryA: "opennlp", |
| 898 | LayerA: "orth", |
| 899 | FoundryB: "upos", |
| 900 | LayerB: "orth", |
| 901 | Mappings: []config.MappingRule{ |
| 902 | "[opennlp/orth=Baum] <> [opennlp/orth=X]", |
| 903 | }, |
| 904 | } |
| 905 | |
| 906 | // Create a new mapper |
| 907 | m, err := NewMapper([]config.MappingList{mappingList}) |
| 908 | require.NoError(t, err) |
| 909 | |
| 910 | tests := []struct { |
| 911 | name string |
| 912 | mappingID string |
| 913 | opts MappingOptions |
| 914 | input string |
| 915 | expected string |
| 916 | expectError bool |
| 917 | }{ |
| 918 | { |
| Akron | cc83eb5 | 2025-05-27 14:39:12 +0200 | [diff] [blame] | 919 | name: "Query wrapper case with rewrites preservation", |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 920 | mappingID: "test-wrapper", |
| 921 | opts: MappingOptions{ |
| 922 | Direction: AtoB, |
| 923 | }, |
| 924 | input: `{ |
| 925 | "@context": "http://korap.ids-mannheim.de/ns/KoralQuery/v0.3/context.jsonld", |
| 926 | "collection": { |
| 927 | "@type": "koral:doc", |
| 928 | "key": "availability", |
| 929 | "match": "match:eq", |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 930 | "type": "type:regex", |
| 931 | "value": "CC.*" |
| 932 | }, |
| 933 | "query": { |
| 934 | "@type": "koral:token", |
| Akron | cc83eb5 | 2025-05-27 14:39:12 +0200 | [diff] [blame] | 935 | "rewrites": [ |
| 936 | { |
| 937 | "@type": "koral:rewrite", |
| 938 | "_comment": "Original rewrite that should be preserved", |
| 939 | "editor": "Original", |
| 940 | "operation": "operation:original", |
| 941 | "src": "Original" |
| 942 | } |
| 943 | ], |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 944 | "wrap": { |
| 945 | "@type": "koral:term", |
| 946 | "foundry": "opennlp", |
| 947 | "key": "Baum", |
| 948 | "layer": "orth", |
| 949 | "match": "match:eq" |
| 950 | } |
| 951 | } |
| 952 | }`, |
| 953 | expected: `{ |
| 954 | "@context": "http://korap.ids-mannheim.de/ns/KoralQuery/v0.3/context.jsonld", |
| 955 | "collection": { |
| 956 | "@type": "koral:doc", |
| 957 | "key": "availability", |
| 958 | "match": "match:eq", |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 959 | "type": "type:regex", |
| 960 | "value": "CC.*" |
| 961 | }, |
| 962 | "query": { |
| 963 | "@type": "koral:token", |
| Akron | cc83eb5 | 2025-05-27 14:39:12 +0200 | [diff] [blame] | 964 | "rewrites": [ |
| 965 | { |
| 966 | "@type": "koral:rewrite", |
| 967 | "_comment": "Original rewrite that should be preserved", |
| 968 | "editor": "Original", |
| 969 | "operation": "operation:original", |
| 970 | "src": "Original" |
| 971 | } |
| 972 | ], |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 973 | "wrap": { |
| 974 | "@type": "koral:term", |
| 975 | "foundry": "opennlp", |
| 976 | "key": "X", |
| 977 | "layer": "orth", |
| 978 | "match": "match:eq" |
| 979 | } |
| 980 | } |
| 981 | }`, |
| 982 | }, |
| 983 | { |
| 984 | name: "Empty query field", |
| 985 | mappingID: "test-wrapper", |
| 986 | opts: MappingOptions{ |
| 987 | Direction: AtoB, |
| 988 | }, |
| 989 | input: `{ |
| 990 | "@context": "http://korap.ids-mannheim.de/ns/KoralQuery/v0.3/context.jsonld", |
| 991 | "query": null |
| 992 | }`, |
| 993 | expected: `{ |
| 994 | "@context": "http://korap.ids-mannheim.de/ns/KoralQuery/v0.3/context.jsonld", |
| 995 | "query": null |
| 996 | }`, |
| 997 | }, |
| 998 | { |
| 999 | name: "Missing query field", |
| 1000 | mappingID: "test-wrapper", |
| 1001 | opts: MappingOptions{ |
| 1002 | Direction: AtoB, |
| 1003 | }, |
| 1004 | input: `{ |
| 1005 | "@context": "http://korap.ids-mannheim.de/ns/KoralQuery/v0.3/context.jsonld", |
| 1006 | "collection": { |
| 1007 | "@type": "koral:doc" |
| 1008 | } |
| 1009 | }`, |
| 1010 | expected: `{ |
| 1011 | "@context": "http://korap.ids-mannheim.de/ns/KoralQuery/v0.3/context.jsonld", |
| 1012 | "collection": { |
| 1013 | "@type": "koral:doc" |
| 1014 | } |
| 1015 | }`, |
| 1016 | }, |
| 1017 | { |
| 1018 | name: "Query field with non-object value", |
| 1019 | mappingID: "test-wrapper", |
| 1020 | opts: MappingOptions{ |
| 1021 | Direction: AtoB, |
| 1022 | }, |
| 1023 | input: `{ |
| 1024 | "@context": "http://korap.ids-mannheim.de/ns/KoralQuery/v0.3/context.jsonld", |
| 1025 | "query": "invalid" |
| 1026 | }`, |
| 1027 | expected: `{ |
| 1028 | "@context": "http://korap.ids-mannheim.de/ns/KoralQuery/v0.3/context.jsonld", |
| 1029 | "query": "invalid" |
| 1030 | }`, |
| 1031 | }, |
| Akron | cc83eb5 | 2025-05-27 14:39:12 +0200 | [diff] [blame] | 1032 | { |
| 1033 | name: "Query with rewrites in nested token", |
| 1034 | mappingID: "test-wrapper", |
| 1035 | opts: MappingOptions{ |
| 1036 | Direction: AtoB, |
| 1037 | }, |
| 1038 | input: `{ |
| 1039 | "@type": "koral:token", |
| 1040 | "rewrites": [ |
| 1041 | { |
| 1042 | "@type": "koral:rewrite", |
| 1043 | "_comment": "Nested rewrite that should be preserved", |
| 1044 | "editor": "Nested", |
| 1045 | "operation": "operation:nested", |
| 1046 | "src": "Nested" |
| 1047 | } |
| 1048 | ], |
| 1049 | "wrap": { |
| 1050 | "@type": "koral:term", |
| 1051 | "foundry": "opennlp", |
| 1052 | "key": "Baum", |
| 1053 | "layer": "orth", |
| 1054 | "match": "match:eq" |
| 1055 | } |
| 1056 | }`, |
| 1057 | expected: `{ |
| 1058 | "@type": "koral:token", |
| 1059 | "rewrites": [ |
| 1060 | { |
| 1061 | "@type": "koral:rewrite", |
| 1062 | "_comment": "Nested rewrite that should be preserved", |
| 1063 | "editor": "Nested", |
| 1064 | "operation": "operation:nested", |
| 1065 | "src": "Nested" |
| 1066 | } |
| 1067 | ], |
| 1068 | "wrap": { |
| 1069 | "@type": "koral:term", |
| 1070 | "foundry": "opennlp", |
| 1071 | "key": "X", |
| 1072 | "layer": "orth", |
| 1073 | "match": "match:eq" |
| 1074 | } |
| 1075 | }`, |
| 1076 | }, |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 1077 | } |
| 1078 | |
| 1079 | for _, tt := range tests { |
| 1080 | t.Run(tt.name, func(t *testing.T) { |
| 1081 | // Parse input JSON |
| Akron | 121c66e | 2025-06-02 16:34:05 +0200 | [diff] [blame] | 1082 | var inputData any |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 1083 | err := json.Unmarshal([]byte(tt.input), &inputData) |
| 1084 | require.NoError(t, err) |
| 1085 | |
| 1086 | // Apply mappings |
| 1087 | result, err := m.ApplyQueryMappings(tt.mappingID, tt.opts, inputData) |
| 1088 | if tt.expectError { |
| 1089 | assert.Error(t, err) |
| 1090 | return |
| 1091 | } |
| 1092 | require.NoError(t, err) |
| 1093 | |
| 1094 | // Parse expected JSON |
| Akron | 121c66e | 2025-06-02 16:34:05 +0200 | [diff] [blame] | 1095 | var expectedData any |
| Akron | 7b4984e | 2025-05-26 19:12:20 +0200 | [diff] [blame] | 1096 | err = json.Unmarshal([]byte(tt.expected), &expectedData) |
| 1097 | require.NoError(t, err) |
| 1098 | |
| 1099 | // Compare results |
| 1100 | assert.Equal(t, expectedData, result) |
| 1101 | }) |
| 1102 | } |
| 1103 | } |
| Akron | 422cd25 | 2026-05-19 16:31:19 +0200 | [diff] [blame] | 1104 | |
| 1105 | func TestIdenticalEffectiveFoundryLayerRejected(t *testing.T) { |
| 1106 | tests := []struct { |
| 1107 | name string |
| 1108 | list config.MappingList |
| 1109 | opts MappingOptions |
| 1110 | wantErr string |
| 1111 | }{ |
| 1112 | { |
| 1113 | name: "YAML defaults identical", |
| 1114 | list: config.MappingList{ |
| 1115 | ID: "test", FoundryA: "opennlp", LayerA: "p", |
| 1116 | FoundryB: "opennlp", LayerB: "p", |
| 1117 | Mappings: []config.MappingRule{"[A] <> [B]"}, |
| 1118 | }, |
| 1119 | opts: MappingOptions{Direction: AtoB}, |
| 1120 | wantErr: "identical source and target", |
| 1121 | }, |
| 1122 | { |
| 1123 | name: "Query param override makes them identical", |
| 1124 | list: config.MappingList{ |
| 1125 | ID: "test", FoundryA: "opennlp", LayerA: "p", |
| 1126 | FoundryB: "upos", LayerB: "p", |
| 1127 | Mappings: []config.MappingRule{"[A] <> [B]"}, |
| 1128 | }, |
| 1129 | opts: MappingOptions{Direction: AtoB, FoundryB: "opennlp"}, |
| 1130 | wantErr: "identical source and target", |
| 1131 | }, |
| 1132 | { |
| 1133 | name: "Query param override resolves the conflict", |
| 1134 | list: config.MappingList{ |
| 1135 | ID: "test", FoundryA: "opennlp", LayerA: "p", |
| 1136 | FoundryB: "opennlp", LayerB: "p", |
| 1137 | Mappings: []config.MappingRule{"[A] <> [B]"}, |
| 1138 | }, |
| 1139 | opts: MappingOptions{Direction: AtoB, FoundryB: "upos"}, |
| 1140 | wantErr: "", |
| 1141 | }, |
| 1142 | { |
| 1143 | name: "Different foundry same layer is allowed", |
| 1144 | list: config.MappingList{ |
| 1145 | ID: "test", FoundryA: "opennlp", LayerA: "p", |
| 1146 | FoundryB: "upos", LayerB: "p", |
| 1147 | Mappings: []config.MappingRule{"[A] <> [B]"}, |
| 1148 | }, |
| 1149 | opts: MappingOptions{Direction: AtoB}, |
| 1150 | wantErr: "", |
| 1151 | }, |
| 1152 | { |
| 1153 | name: "Both foundries empty is allowed", |
| 1154 | list: config.MappingList{ |
| 1155 | ID: "test", |
| 1156 | Mappings: []config.MappingRule{"[A] <> [B]"}, |
| 1157 | }, |
| 1158 | opts: MappingOptions{Direction: AtoB}, |
| 1159 | wantErr: "", |
| 1160 | }, |
| 1161 | } |
| 1162 | |
| 1163 | for _, tt := range tests { |
| 1164 | t.Run(tt.name, func(t *testing.T) { |
| 1165 | m, err := NewMapper([]config.MappingList{tt.list}) |
| 1166 | require.NoError(t, err) |
| 1167 | |
| 1168 | input := map[string]any{ |
| 1169 | "@type": "koral:token", |
| 1170 | "wrap": map[string]any{ |
| 1171 | "@type": "koral:term", |
| 1172 | "key": "A", |
| 1173 | }, |
| 1174 | } |
| 1175 | |
| 1176 | _, err = m.ApplyQueryMappings("test", tt.opts, input) |
| 1177 | if tt.wantErr != "" { |
| 1178 | require.Error(t, err) |
| 1179 | assert.Contains(t, err.Error(), tt.wantErr) |
| 1180 | } else { |
| 1181 | assert.NoError(t, err) |
| 1182 | } |
| 1183 | }) |
| 1184 | } |
| 1185 | } |
| 1186 | |
| 1187 | func TestIdenticalEffectiveFieldRejected(t *testing.T) { |
| 1188 | tests := []struct { |
| 1189 | name string |
| 1190 | list config.MappingList |
| 1191 | opts MappingOptions |
| 1192 | wantErr string |
| 1193 | }{ |
| 1194 | { |
| 1195 | name: "YAML defaults identical", |
| 1196 | list: config.MappingList{ |
| 1197 | ID: "test", Type: "corpus", |
| 1198 | FieldA: "textClass", FieldB: "textClass", |
| 1199 | Mappings: []config.MappingRule{"novel <> fiction"}, |
| 1200 | }, |
| 1201 | opts: MappingOptions{Direction: AtoB}, |
| 1202 | wantErr: "identical source and target field", |
| 1203 | }, |
| 1204 | { |
| 1205 | name: "Query param override makes them identical", |
| 1206 | list: config.MappingList{ |
| 1207 | ID: "test", Type: "corpus", |
| 1208 | FieldA: "textClass", FieldB: "genre", |
| 1209 | Mappings: []config.MappingRule{"novel <> fiction"}, |
| 1210 | }, |
| 1211 | opts: MappingOptions{Direction: AtoB, FieldB: "textClass"}, |
| 1212 | wantErr: "identical source and target field", |
| 1213 | }, |
| 1214 | { |
| 1215 | name: "Query param override resolves the conflict", |
| 1216 | list: config.MappingList{ |
| 1217 | ID: "test", Type: "corpus", |
| 1218 | FieldA: "textClass", FieldB: "textClass", |
| 1219 | Mappings: []config.MappingRule{"novel <> fiction"}, |
| 1220 | }, |
| 1221 | opts: MappingOptions{Direction: AtoB, FieldB: "genre"}, |
| 1222 | wantErr: "", |
| 1223 | }, |
| 1224 | { |
| 1225 | name: "Different fields is allowed", |
| 1226 | list: config.MappingList{ |
| 1227 | ID: "test", Type: "corpus", |
| 1228 | FieldA: "textClass", FieldB: "genre", |
| 1229 | Mappings: []config.MappingRule{"novel <> fiction"}, |
| 1230 | }, |
| 1231 | opts: MappingOptions{Direction: AtoB}, |
| 1232 | wantErr: "", |
| 1233 | }, |
| 1234 | { |
| 1235 | name: "Both fields empty is allowed", |
| 1236 | list: config.MappingList{ |
| 1237 | ID: "test", Type: "corpus", |
| 1238 | Mappings: []config.MappingRule{"textClass=novel <> genre=fiction"}, |
| 1239 | }, |
| 1240 | opts: MappingOptions{Direction: AtoB}, |
| 1241 | wantErr: "", |
| 1242 | }, |
| 1243 | } |
| 1244 | |
| 1245 | for _, tt := range tests { |
| 1246 | t.Run(tt.name, func(t *testing.T) { |
| 1247 | m, err := NewMapper([]config.MappingList{tt.list}) |
| 1248 | require.NoError(t, err) |
| 1249 | |
| 1250 | input := map[string]any{ |
| 1251 | "collection": map[string]any{ |
| 1252 | "@type": "koral:doc", |
| 1253 | "key": "textClass", |
| 1254 | "value": "novel", |
| 1255 | "match": "match:eq", |
| 1256 | }, |
| 1257 | } |
| 1258 | |
| 1259 | _, err = m.ApplyQueryMappings("test", tt.opts, input) |
| 1260 | if tt.wantErr != "" { |
| 1261 | require.Error(t, err) |
| 1262 | assert.Contains(t, err.Error(), tt.wantErr) |
| 1263 | } else { |
| 1264 | assert.NoError(t, err) |
| 1265 | } |
| 1266 | }) |
| 1267 | } |
| 1268 | } |
| 1269 | |
| 1270 | func TestIdenticalEffectiveValuesResponseEndpoint(t *testing.T) { |
| 1271 | t.Run("annotation response rejects identical effective foundry/layer", func(t *testing.T) { |
| 1272 | m, err := NewMapper([]config.MappingList{{ |
| 1273 | ID: "test", FoundryA: "marmot", LayerA: "p", |
| 1274 | FoundryB: "marmot", LayerB: "p", |
| 1275 | Mappings: []config.MappingRule{"[DET] <> [PRON]"}, |
| 1276 | }}) |
| 1277 | require.NoError(t, err) |
| 1278 | |
| 1279 | input := map[string]any{ |
| 1280 | "snippet": `<span title="marmot/p:DET">Der</span>`, |
| 1281 | } |
| 1282 | |
| 1283 | _, err = m.ApplyResponseMappings("test", MappingOptions{Direction: AtoB}, input) |
| 1284 | require.Error(t, err) |
| 1285 | assert.Contains(t, err.Error(), "identical source and target") |
| 1286 | }) |
| 1287 | |
| 1288 | t.Run("corpus response rejects identical effective field", func(t *testing.T) { |
| 1289 | m, err := NewMapper([]config.MappingList{{ |
| 1290 | ID: "test", Type: "corpus", |
| 1291 | FieldA: "textClass", FieldB: "textClass", |
| 1292 | Mappings: []config.MappingRule{"novel <> fiction"}, |
| 1293 | }}) |
| 1294 | require.NoError(t, err) |
| 1295 | |
| 1296 | input := map[string]any{ |
| 1297 | "fields": []any{ |
| 1298 | map[string]any{ |
| 1299 | "@type": "koral:field", |
| 1300 | "key": "textClass", |
| 1301 | "value": "novel", |
| 1302 | "type": "type:string", |
| 1303 | }, |
| 1304 | }, |
| 1305 | } |
| 1306 | |
| 1307 | _, err = m.ApplyResponseMappings("test", MappingOptions{Direction: AtoB}, input) |
| 1308 | require.Error(t, err) |
| 1309 | assert.Contains(t, err.Error(), "identical source and target field") |
| 1310 | }) |
| 1311 | } |