| Akron | e4f570d | 2026-02-20 08:18:06 +0100 | [diff] [blame] | 1 | package mapper |
| 2 | |
| 3 | import ( |
| 4 | "encoding/json" |
| 5 | "testing" |
| 6 | |
| 7 | "github.com/KorAP/Koral-Mapper/config" |
| 8 | "github.com/stretchr/testify/assert" |
| 9 | "github.com/stretchr/testify/require" |
| 10 | ) |
| 11 | |
| 12 | // parseJSON is a test helper that unmarshals a JSON string. |
| 13 | func parseJSON(t *testing.T, s string) any { |
| 14 | t.Helper() |
| 15 | var v any |
| 16 | require.NoError(t, json.Unmarshal([]byte(s), &v)) |
| 17 | return v |
| 18 | } |
| 19 | |
| 20 | func TestCascadeQueryTwoAnnotationMappings(t *testing.T) { |
| 21 | m, err := NewMapper([]config.MappingList{ |
| 22 | { |
| 23 | ID: "ann-step1", FoundryA: "opennlp", LayerA: "p", |
| 24 | FoundryB: "opennlp", LayerB: "p", |
| 25 | Mappings: []config.MappingRule{`[PIDAT] <> [DET]`}, |
| 26 | }, |
| 27 | { |
| 28 | ID: "ann-step2", FoundryA: "opennlp", LayerA: "p", |
| 29 | FoundryB: "upos", LayerB: "p", |
| 30 | Mappings: []config.MappingRule{`[DET] <> [PRON]`}, |
| 31 | }, |
| 32 | }) |
| 33 | require.NoError(t, err) |
| 34 | |
| 35 | input := parseJSON(t, `{ |
| 36 | "@type": "koral:token", |
| 37 | "wrap": { |
| 38 | "@type": "koral:term", |
| 39 | "foundry": "opennlp", |
| 40 | "key": "PIDAT", |
| 41 | "layer": "p", |
| 42 | "match": "match:eq" |
| 43 | } |
| 44 | }`) |
| 45 | |
| 46 | result, err := m.CascadeQueryMappings( |
| 47 | []string{"ann-step1", "ann-step2"}, |
| 48 | []MappingOptions{{Direction: AtoB}, {Direction: AtoB}}, |
| 49 | input, |
| 50 | ) |
| 51 | require.NoError(t, err) |
| 52 | |
| 53 | expected := parseJSON(t, `{ |
| 54 | "@type": "koral:token", |
| 55 | "wrap": { |
| 56 | "@type": "koral:term", |
| 57 | "foundry": "upos", |
| 58 | "key": "PRON", |
| 59 | "layer": "p", |
| 60 | "match": "match:eq" |
| 61 | } |
| 62 | }`) |
| 63 | assert.Equal(t, expected, result) |
| 64 | } |
| 65 | |
| 66 | func TestCascadeQueryMixAnnotationAndCorpus(t *testing.T) { |
| 67 | m, err := NewMapper([]config.MappingList{ |
| 68 | { |
| 69 | ID: "ann-mapper", FoundryA: "opennlp", LayerA: "p", |
| 70 | FoundryB: "upos", LayerB: "p", |
| 71 | Mappings: []config.MappingRule{`[PIDAT] <> [DET]`}, |
| 72 | }, |
| 73 | { |
| 74 | ID: "corpus-mapper", |
| 75 | Type: "corpus", |
| 76 | Mappings: []config.MappingRule{`textClass=novel <> genre=fiction`}, |
| 77 | }, |
| 78 | }) |
| 79 | require.NoError(t, err) |
| 80 | |
| 81 | input := parseJSON(t, `{ |
| 82 | "query": { |
| 83 | "@type": "koral:token", |
| 84 | "wrap": { |
| 85 | "@type": "koral:term", |
| 86 | "foundry": "opennlp", |
| 87 | "key": "PIDAT", |
| 88 | "layer": "p", |
| 89 | "match": "match:eq" |
| 90 | } |
| 91 | }, |
| 92 | "collection": { |
| 93 | "@type": "koral:doc", |
| 94 | "key": "textClass", |
| 95 | "value": "novel", |
| 96 | "match": "match:eq" |
| 97 | } |
| 98 | }`) |
| 99 | |
| 100 | result, err := m.CascadeQueryMappings( |
| 101 | []string{"ann-mapper", "corpus-mapper"}, |
| 102 | []MappingOptions{{Direction: AtoB}, {Direction: AtoB}}, |
| 103 | input, |
| 104 | ) |
| 105 | require.NoError(t, err) |
| 106 | |
| 107 | resultMap := result.(map[string]any) |
| 108 | |
| 109 | query := resultMap["query"].(map[string]any) |
| 110 | wrap := query["wrap"].(map[string]any) |
| 111 | assert.Equal(t, "DET", wrap["key"]) |
| 112 | assert.Equal(t, "upos", wrap["foundry"]) |
| 113 | |
| 114 | collection := resultMap["collection"].(map[string]any) |
| 115 | assert.Equal(t, "genre", collection["key"]) |
| 116 | assert.Equal(t, "fiction", collection["value"]) |
| 117 | } |
| 118 | |
| 119 | func TestCascadeQuerySingleElement(t *testing.T) { |
| 120 | m, err := NewMapper([]config.MappingList{{ |
| 121 | ID: "single", FoundryA: "opennlp", LayerA: "p", |
| 122 | FoundryB: "upos", LayerB: "p", |
| 123 | Mappings: []config.MappingRule{`[PIDAT] <> [DET]`}, |
| 124 | }}) |
| 125 | require.NoError(t, err) |
| 126 | |
| 127 | makeInput := func() any { |
| 128 | return parseJSON(t, `{ |
| 129 | "@type": "koral:token", |
| 130 | "wrap": { |
| 131 | "@type": "koral:term", |
| 132 | "foundry": "opennlp", |
| 133 | "key": "PIDAT", |
| 134 | "layer": "p", |
| 135 | "match": "match:eq" |
| 136 | } |
| 137 | }`) |
| 138 | } |
| 139 | |
| 140 | opts := MappingOptions{Direction: AtoB} |
| 141 | |
| 142 | cascadeResult, err := m.CascadeQueryMappings( |
| 143 | []string{"single"}, []MappingOptions{opts}, makeInput(), |
| 144 | ) |
| 145 | require.NoError(t, err) |
| 146 | |
| 147 | directResult, err := m.ApplyQueryMappings("single", opts, makeInput()) |
| 148 | require.NoError(t, err) |
| 149 | |
| 150 | assert.Equal(t, directResult, cascadeResult) |
| 151 | } |
| 152 | |
| 153 | func TestCascadeQueryEmptyList(t *testing.T) { |
| 154 | m, err := NewMapper([]config.MappingList{{ |
| 155 | ID: "dummy", FoundryA: "x", LayerA: "y", |
| 156 | FoundryB: "a", LayerB: "b", |
| 157 | Mappings: []config.MappingRule{`[X] <> [Y]`}, |
| 158 | }}) |
| 159 | require.NoError(t, err) |
| 160 | |
| 161 | input := parseJSON(t, `{ |
| 162 | "@type": "koral:token", |
| 163 | "wrap": {"@type": "koral:term", "key": "Z"} |
| 164 | }`) |
| 165 | |
| 166 | result, err := m.CascadeQueryMappings(nil, nil, input) |
| 167 | require.NoError(t, err) |
| 168 | assert.Equal(t, input, result) |
| 169 | } |
| 170 | |
| 171 | func TestCascadeQueryUnknownID(t *testing.T) { |
| 172 | m, err := NewMapper([]config.MappingList{{ |
| 173 | ID: "known", FoundryA: "x", LayerA: "y", |
| 174 | FoundryB: "a", LayerB: "b", |
| 175 | Mappings: []config.MappingRule{`[X] <> [Y]`}, |
| 176 | }}) |
| 177 | require.NoError(t, err) |
| 178 | |
| 179 | input := parseJSON(t, `{ |
| 180 | "@type": "koral:token", |
| 181 | "wrap": {"@type": "koral:term", "key": "X"} |
| 182 | }`) |
| 183 | |
| 184 | _, err = m.CascadeQueryMappings( |
| 185 | []string{"known", "nonexistent"}, |
| 186 | []MappingOptions{{Direction: AtoB}, {Direction: AtoB}}, |
| 187 | input, |
| 188 | ) |
| 189 | assert.Error(t, err) |
| 190 | assert.Contains(t, err.Error(), "nonexistent") |
| 191 | } |
| 192 | |
| 193 | // --- Response cascade tests --- |
| 194 | |
| 195 | func TestCascadeResponseTwoCorpusMappings(t *testing.T) { |
| 196 | m, err := NewMapper([]config.MappingList{ |
| 197 | { |
| 198 | ID: "corpus-step1", Type: "corpus", |
| 199 | Mappings: []config.MappingRule{`textClass=novel <> genre=fiction`}, |
| 200 | }, |
| 201 | { |
| 202 | ID: "corpus-step2", Type: "corpus", |
| 203 | Mappings: []config.MappingRule{`genre=fiction <> category=lit`}, |
| 204 | }, |
| 205 | }) |
| 206 | require.NoError(t, err) |
| 207 | |
| 208 | input := parseJSON(t, `{ |
| 209 | "fields": [{ |
| 210 | "@type": "koral:field", |
| 211 | "key": "textClass", |
| 212 | "value": "novel", |
| 213 | "type": "type:string" |
| 214 | }] |
| 215 | }`) |
| 216 | |
| 217 | result, err := m.CascadeResponseMappings( |
| 218 | []string{"corpus-step1", "corpus-step2"}, |
| 219 | []MappingOptions{{Direction: AtoB}, {Direction: AtoB}}, |
| 220 | input, |
| 221 | ) |
| 222 | require.NoError(t, err) |
| 223 | |
| 224 | fields := result.(map[string]any)["fields"].([]any) |
| 225 | require.GreaterOrEqual(t, len(fields), 3) |
| 226 | |
| 227 | assert.Equal(t, "textClass", fields[0].(map[string]any)["key"]) |
| 228 | |
| 229 | assert.Equal(t, "genre", fields[1].(map[string]any)["key"]) |
| 230 | assert.Equal(t, "fiction", fields[1].(map[string]any)["value"]) |
| 231 | |
| 232 | assert.Equal(t, "category", fields[2].(map[string]any)["key"]) |
| 233 | assert.Equal(t, "lit", fields[2].(map[string]any)["value"]) |
| 234 | } |
| 235 | |
| 236 | func TestCascadeResponseMixAnnotationAndCorpus(t *testing.T) { |
| 237 | m, err := NewMapper([]config.MappingList{ |
| 238 | { |
| 239 | ID: "ann-resp", FoundryA: "opennlp", LayerA: "p", |
| 240 | FoundryB: "upos", LayerB: "p", |
| 241 | Mappings: []config.MappingRule{`[DET] <> [PRON]`}, |
| 242 | }, |
| 243 | { |
| 244 | ID: "corpus-resp", |
| 245 | Type: "corpus", |
| 246 | Mappings: []config.MappingRule{`textClass=novel <> genre=fiction`}, |
| 247 | }, |
| 248 | }) |
| 249 | require.NoError(t, err) |
| 250 | |
| 251 | input := parseJSON(t, `{ |
| 252 | "snippet": "<span title=\"opennlp/p:DET\">Der</span>", |
| 253 | "fields": [{ |
| 254 | "@type": "koral:field", |
| 255 | "key": "textClass", |
| 256 | "value": "novel", |
| 257 | "type": "type:string" |
| 258 | }] |
| 259 | }`) |
| 260 | |
| 261 | result, err := m.CascadeResponseMappings( |
| 262 | []string{"ann-resp", "corpus-resp"}, |
| 263 | []MappingOptions{{Direction: AtoB}, {Direction: AtoB}}, |
| 264 | input, |
| 265 | ) |
| 266 | require.NoError(t, err) |
| 267 | |
| 268 | resultMap := result.(map[string]any) |
| 269 | |
| 270 | snippet := resultMap["snippet"].(string) |
| 271 | assert.Contains(t, snippet, "opennlp/p:DET") |
| 272 | assert.Contains(t, snippet, "upos/p:PRON") |
| 273 | |
| 274 | fields := resultMap["fields"].([]any) |
| 275 | require.GreaterOrEqual(t, len(fields), 2) |
| 276 | assert.Equal(t, "genre", fields[1].(map[string]any)["key"]) |
| 277 | } |
| 278 | |
| 279 | func TestCascadeResponseSingleElement(t *testing.T) { |
| 280 | m, err := NewMapper([]config.MappingList{{ |
| 281 | ID: "corpus-single", Type: "corpus", |
| 282 | Mappings: []config.MappingRule{`textClass=novel <> genre=fiction`}, |
| 283 | }}) |
| 284 | require.NoError(t, err) |
| 285 | |
| 286 | makeInput := func() any { |
| 287 | return parseJSON(t, `{ |
| 288 | "fields": [{ |
| 289 | "@type": "koral:field", |
| 290 | "key": "textClass", |
| 291 | "value": "novel", |
| 292 | "type": "type:string" |
| 293 | }] |
| 294 | }`) |
| 295 | } |
| 296 | |
| 297 | opts := MappingOptions{Direction: AtoB} |
| 298 | |
| 299 | cascadeResult, err := m.CascadeResponseMappings( |
| 300 | []string{"corpus-single"}, []MappingOptions{opts}, makeInput(), |
| 301 | ) |
| 302 | require.NoError(t, err) |
| 303 | |
| 304 | directResult, err := m.ApplyResponseMappings("corpus-single", opts, makeInput()) |
| 305 | require.NoError(t, err) |
| 306 | |
| 307 | assert.Equal(t, directResult, cascadeResult) |
| 308 | } |
| 309 | |
| 310 | func TestCascadeResponseEmptyList(t *testing.T) { |
| 311 | m, err := NewMapper([]config.MappingList{{ |
| 312 | ID: "dummy", Type: "corpus", |
| 313 | Mappings: []config.MappingRule{`x=y <> a=b`}, |
| 314 | }}) |
| 315 | require.NoError(t, err) |
| 316 | |
| 317 | input := parseJSON(t, `{"fields": []}`) |
| 318 | |
| 319 | result, err := m.CascadeResponseMappings(nil, nil, input) |
| 320 | require.NoError(t, err) |
| 321 | assert.Equal(t, input, result) |
| 322 | } |
| 323 | |
| 324 | func TestCascadeResponseUnknownID(t *testing.T) { |
| 325 | m, err := NewMapper([]config.MappingList{{ |
| 326 | ID: "known", Type: "corpus", |
| 327 | Mappings: []config.MappingRule{`x=y <> a=b`}, |
| 328 | }}) |
| 329 | require.NoError(t, err) |
| 330 | |
| 331 | _, err = m.CascadeResponseMappings( |
| 332 | []string{"nonexistent"}, |
| 333 | []MappingOptions{{Direction: AtoB}}, |
| 334 | parseJSON(t, `{"fields": []}`), |
| 335 | ) |
| 336 | assert.Error(t, err) |
| 337 | assert.Contains(t, err.Error(), "nonexistent") |
| 338 | } |