Akron | 57ee558 | 2025-05-21 15:25:13 +0200 | [diff] [blame] | 1 | package config |
| 2 | |
| 3 | import ( |
| 4 | "os" |
| 5 | "testing" |
| 6 | |
Akron | fa55bb2 | 2025-05-26 15:10:42 +0200 | [diff] [blame] | 7 | "github.com/KorAP/KoralPipe-TermMapper/ast" |
Akron | 57ee558 | 2025-05-21 15:25:13 +0200 | [diff] [blame] | 8 | "github.com/stretchr/testify/assert" |
| 9 | "github.com/stretchr/testify/require" |
| 10 | ) |
| 11 | |
| 12 | func TestLoadConfig(t *testing.T) { |
| 13 | // Create a temporary YAML file |
| 14 | content := ` |
| 15 | - id: opennlp-mapper |
| 16 | foundryA: opennlp |
| 17 | layerA: p |
| 18 | foundryB: upos |
| 19 | layerB: p |
| 20 | mappings: |
| 21 | - "[PIDAT] <> [opennlp/p=PIDAT & opennlp/p=AdjType:Pdt]" |
| 22 | - "[PAV] <> [ADV & PronType:Dem]" |
| 23 | |
| 24 | - id: simple-mapper |
| 25 | mappings: |
| 26 | - "[A] <> [B]" |
| 27 | ` |
| 28 | tmpfile, err := os.CreateTemp("", "config-*.yaml") |
| 29 | require.NoError(t, err) |
| 30 | defer os.Remove(tmpfile.Name()) |
| 31 | |
| 32 | _, err = tmpfile.WriteString(content) |
| 33 | require.NoError(t, err) |
| 34 | err = tmpfile.Close() |
| 35 | require.NoError(t, err) |
| 36 | |
| 37 | // Test loading the configuration |
| 38 | config, err := LoadConfig(tmpfile.Name()) |
| 39 | require.NoError(t, err) |
| 40 | |
| 41 | // Verify the configuration |
| 42 | require.Len(t, config.Lists, 2) |
| 43 | |
| 44 | // Check first mapping list |
| 45 | list1 := config.Lists[0] |
| 46 | assert.Equal(t, "opennlp-mapper", list1.ID) |
| 47 | assert.Equal(t, "opennlp", list1.FoundryA) |
| 48 | assert.Equal(t, "p", list1.LayerA) |
| 49 | assert.Equal(t, "upos", list1.FoundryB) |
| 50 | assert.Equal(t, "p", list1.LayerB) |
| 51 | require.Len(t, list1.Mappings, 2) |
| 52 | assert.Equal(t, "[PIDAT] <> [opennlp/p=PIDAT & opennlp/p=AdjType:Pdt]", string(list1.Mappings[0])) |
| 53 | assert.Equal(t, "[PAV] <> [ADV & PronType:Dem]", string(list1.Mappings[1])) |
| 54 | |
| 55 | // Check second mapping list |
| 56 | list2 := config.Lists[1] |
| 57 | assert.Equal(t, "simple-mapper", list2.ID) |
| 58 | assert.Empty(t, list2.FoundryA) |
| 59 | assert.Empty(t, list2.LayerA) |
| 60 | assert.Empty(t, list2.FoundryB) |
| 61 | assert.Empty(t, list2.LayerB) |
| 62 | require.Len(t, list2.Mappings, 1) |
| 63 | assert.Equal(t, "[A] <> [B]", string(list2.Mappings[0])) |
| 64 | } |
| 65 | |
| 66 | func TestParseMappings(t *testing.T) { |
| 67 | list := &MappingList{ |
| 68 | ID: "test-mapper", |
| 69 | FoundryA: "opennlp", |
| 70 | LayerA: "p", |
| 71 | FoundryB: "upos", |
| 72 | LayerB: "p", |
| 73 | Mappings: []MappingRule{ |
| 74 | "[PIDAT] <> [opennlp/p=PIDAT & opennlp/p=AdjType:Pdt]", |
| 75 | }, |
| 76 | } |
| 77 | |
| 78 | results, err := list.ParseMappings() |
| 79 | require.NoError(t, err) |
| 80 | require.Len(t, results, 1) |
| 81 | |
| 82 | // Check the parsed upper pattern |
| 83 | upper := results[0].Upper |
| 84 | require.NotNil(t, upper) |
| 85 | require.IsType(t, &ast.Token{}, upper) |
| 86 | upperTerm := upper.Wrap.(*ast.Term) |
| 87 | assert.Equal(t, "opennlp", upperTerm.Foundry) |
| 88 | assert.Equal(t, "p", upperTerm.Layer) |
| 89 | assert.Equal(t, "PIDAT", upperTerm.Key) |
| 90 | |
| 91 | // Check the parsed lower pattern |
| 92 | lower := results[0].Lower |
| 93 | require.NotNil(t, lower) |
| 94 | require.IsType(t, &ast.Token{}, lower) |
| 95 | lowerGroup := lower.Wrap.(*ast.TermGroup) |
| 96 | require.Len(t, lowerGroup.Operands, 2) |
| 97 | assert.Equal(t, ast.AndRelation, lowerGroup.Relation) |
| 98 | |
| 99 | // Check first operand |
| 100 | term1 := lowerGroup.Operands[0].(*ast.Term) |
| 101 | assert.Equal(t, "opennlp", term1.Foundry) |
| 102 | assert.Equal(t, "p", term1.Layer) |
| 103 | assert.Equal(t, "PIDAT", term1.Key) |
| 104 | |
| 105 | // Check second operand |
| 106 | term2 := lowerGroup.Operands[1].(*ast.Term) |
| 107 | assert.Equal(t, "opennlp", term2.Foundry) |
| 108 | assert.Equal(t, "p", term2.Layer) |
| 109 | assert.Equal(t, "AdjType", term2.Key) |
| 110 | assert.Equal(t, "Pdt", term2.Value) |
| 111 | } |
| 112 | |
| 113 | func TestLoadConfigValidation(t *testing.T) { |
| 114 | tests := []struct { |
| 115 | name string |
| 116 | content string |
| 117 | wantErr string |
| 118 | }{ |
| 119 | { |
| 120 | name: "Missing ID", |
| 121 | content: ` |
| 122 | - foundryA: opennlp |
| 123 | mappings: |
| 124 | - "[A] <> [B]" |
| 125 | `, |
| 126 | wantErr: "mapping list at index 0 is missing an ID", |
| 127 | }, |
| 128 | { |
| 129 | name: "Empty mappings", |
| 130 | content: ` |
| 131 | - id: test |
| 132 | foundryA: opennlp |
| 133 | mappings: [] |
| 134 | `, |
| 135 | wantErr: "mapping list 'test' has no mapping rules", |
| 136 | }, |
| 137 | { |
| 138 | name: "Empty rule", |
| 139 | content: ` |
| 140 | - id: test |
| 141 | mappings: |
| 142 | - "" |
| 143 | `, |
| 144 | wantErr: "mapping list 'test' rule at index 0 is empty", |
| 145 | }, |
| 146 | } |
| 147 | |
| 148 | for _, tt := range tests { |
| 149 | t.Run(tt.name, func(t *testing.T) { |
| 150 | tmpfile, err := os.CreateTemp("", "config-*.yaml") |
| 151 | require.NoError(t, err) |
| 152 | defer os.Remove(tmpfile.Name()) |
| 153 | |
| 154 | _, err = tmpfile.WriteString(tt.content) |
| 155 | require.NoError(t, err) |
| 156 | err = tmpfile.Close() |
| 157 | require.NoError(t, err) |
| 158 | |
| 159 | _, err = LoadConfig(tmpfile.Name()) |
| 160 | require.Error(t, err) |
| 161 | assert.Contains(t, err.Error(), tt.wantErr) |
| 162 | }) |
| 163 | } |
| 164 | } |
Akron | a5d8814 | 2025-05-22 14:42:09 +0200 | [diff] [blame] | 165 | |
| 166 | func TestLoadConfigEdgeCases(t *testing.T) { |
| 167 | tests := []struct { |
| 168 | name string |
| 169 | content string |
| 170 | wantErr string |
| 171 | }{ |
| 172 | { |
| 173 | name: "Duplicate mapping list IDs", |
| 174 | content: ` |
| 175 | - id: test |
| 176 | mappings: |
| 177 | - "[A] <> [B]" |
| 178 | - id: test |
| 179 | mappings: |
| 180 | - "[C] <> [D]"`, |
| 181 | wantErr: "duplicate mapping list ID found: test", |
| 182 | }, |
| 183 | { |
| 184 | name: "Invalid YAML syntax", |
| 185 | content: ` |
| 186 | - id: test |
| 187 | mappings: |
| 188 | - [A] <> [B] # Unquoted special characters |
| 189 | `, |
| 190 | wantErr: "yaml", |
| 191 | }, |
| 192 | { |
| 193 | name: "Empty file", |
| 194 | content: "", |
| 195 | wantErr: "EOF", |
| 196 | }, |
| 197 | { |
| 198 | name: "Non-list YAML", |
| 199 | content: ` |
| 200 | id: test |
| 201 | mappings: |
| 202 | - "[A] <> [B]"`, |
| 203 | wantErr: "cannot unmarshal", |
| 204 | }, |
| 205 | { |
| 206 | name: "Missing required fields", |
| 207 | content: ` |
| 208 | - mappings: |
| 209 | - "[A] <> [B]" |
| 210 | - id: test2 |
| 211 | foundryA: opennlp`, |
| 212 | wantErr: "missing an ID", |
| 213 | }, |
| 214 | { |
| 215 | name: "Empty mappings list", |
| 216 | content: ` |
| 217 | - id: test |
| 218 | foundryA: opennlp |
| 219 | mappings: []`, |
| 220 | wantErr: "has no mapping rules", |
| 221 | }, |
| 222 | { |
| 223 | name: "Null values in optional fields", |
| 224 | content: ` |
| 225 | - id: test |
| 226 | foundryA: null |
| 227 | layerA: null |
| 228 | foundryB: null |
| 229 | layerB: null |
| 230 | mappings: |
| 231 | - "[A] <> [B]"`, |
| 232 | wantErr: "", |
| 233 | }, |
| 234 | { |
| 235 | name: "Special characters in IDs", |
| 236 | content: ` |
| 237 | - id: "test/special@chars#1" |
| 238 | mappings: |
| 239 | - "[A] <> [B]"`, |
| 240 | wantErr: "", |
| 241 | }, |
| 242 | { |
| 243 | name: "Unicode characters in mappings", |
| 244 | content: ` |
| 245 | - id: test |
| 246 | mappings: |
| 247 | - "[ß] <> [ss]" |
| 248 | - "[é] <> [e]"`, |
| 249 | wantErr: "", |
| 250 | }, |
| 251 | } |
| 252 | |
| 253 | for _, tt := range tests { |
| 254 | t.Run(tt.name, func(t *testing.T) { |
| 255 | tmpfile, err := os.CreateTemp("", "config-*.yaml") |
| 256 | require.NoError(t, err) |
| 257 | defer os.Remove(tmpfile.Name()) |
| 258 | |
| 259 | _, err = tmpfile.WriteString(tt.content) |
| 260 | require.NoError(t, err) |
| 261 | err = tmpfile.Close() |
| 262 | require.NoError(t, err) |
| 263 | |
| 264 | config, err := LoadConfig(tmpfile.Name()) |
| 265 | if tt.wantErr != "" { |
| 266 | require.Error(t, err) |
| 267 | assert.Contains(t, err.Error(), tt.wantErr) |
| 268 | return |
| 269 | } |
| 270 | require.NoError(t, err) |
| 271 | require.NotNil(t, config) |
| 272 | }) |
| 273 | } |
| 274 | } |
| 275 | |
| 276 | func TestParseMappingsEdgeCases(t *testing.T) { |
| 277 | tests := []struct { |
| 278 | name string |
| 279 | list *MappingList |
| 280 | wantErr bool |
| 281 | errCheck func(t *testing.T, err error) |
| 282 | }{ |
| 283 | { |
| 284 | name: "Empty mapping rule", |
| 285 | list: &MappingList{ |
| 286 | ID: "test", |
| 287 | Mappings: []MappingRule{""}, |
| 288 | }, |
| 289 | wantErr: true, |
| 290 | errCheck: func(t *testing.T, err error) { |
| 291 | assert.Contains(t, err.Error(), "empty") |
| 292 | }, |
| 293 | }, |
| 294 | { |
| 295 | name: "Invalid mapping syntax", |
| 296 | list: &MappingList{ |
| 297 | ID: "test", |
| 298 | Mappings: []MappingRule{"[A] -> [B]"}, |
| 299 | }, |
| 300 | wantErr: true, |
| 301 | errCheck: func(t *testing.T, err error) { |
| 302 | assert.Contains(t, err.Error(), "failed to parse") |
| 303 | }, |
| 304 | }, |
| 305 | { |
| 306 | name: "Missing brackets", |
| 307 | list: &MappingList{ |
| 308 | ID: "test", |
| 309 | Mappings: []MappingRule{"A <> B"}, |
| 310 | }, |
| 311 | wantErr: true, |
| 312 | errCheck: func(t *testing.T, err error) { |
| 313 | assert.Contains(t, err.Error(), "failed to parse") |
| 314 | }, |
| 315 | }, |
| 316 | { |
| 317 | name: "Complex nested expressions", |
| 318 | list: &MappingList{ |
| 319 | ID: "test", |
| 320 | Mappings: []MappingRule{ |
| 321 | "[A & (B | C) & (D | (E & F))] <> [X & (Y | Z)]", |
| 322 | }, |
| 323 | }, |
| 324 | wantErr: false, |
| 325 | }, |
| 326 | { |
| 327 | name: "Multiple foundry/layer combinations", |
| 328 | list: &MappingList{ |
| 329 | ID: "test", |
| 330 | Mappings: []MappingRule{ |
| 331 | "[foundry1/layer1=A & foundry2/layer2=B] <> [foundry3/layer3=C]", |
| 332 | }, |
| 333 | }, |
| 334 | wantErr: false, |
| 335 | }, |
| 336 | { |
| 337 | name: "Default foundry/layer override", |
| 338 | list: &MappingList{ |
| 339 | ID: "test", |
| 340 | FoundryA: "defaultFoundry", |
| 341 | LayerA: "defaultLayer", |
| 342 | Mappings: []MappingRule{ |
| 343 | "[A] <> [B]", // Should use defaults |
| 344 | }, |
| 345 | }, |
| 346 | wantErr: false, |
| 347 | }, |
| 348 | } |
| 349 | |
| 350 | for _, tt := range tests { |
| 351 | t.Run(tt.name, func(t *testing.T) { |
| 352 | results, err := tt.list.ParseMappings() |
| 353 | if tt.wantErr { |
| 354 | require.Error(t, err) |
| 355 | if tt.errCheck != nil { |
| 356 | tt.errCheck(t, err) |
| 357 | } |
| 358 | return |
| 359 | } |
| 360 | require.NoError(t, err) |
| 361 | require.NotNil(t, results) |
| 362 | }) |
| 363 | } |
| 364 | } |