Akron | e562ad6 | 2025-06-25 11:15:56 +0200 | [diff] [blame] | 1 | package parser |
| 2 | |
| 3 | import ( |
| 4 | "testing" |
| 5 | |
Akron | 2ef703c | 2025-07-03 15:57:42 +0200 | [diff] [blame] | 6 | "github.com/KorAP/Koral-Mapper/ast" |
Akron | e562ad6 | 2025-06-25 11:15:56 +0200 | [diff] [blame] | 7 | "github.com/stretchr/testify/assert" |
| 8 | "github.com/stretchr/testify/require" |
| 9 | ) |
| 10 | |
Akron | 6b4c9eb | 2025-07-03 14:31:58 +0200 | [diff] [blame] | 11 | // TestTitleAttributeParser_ParseTitleAttribute was removed as ParseTitleAttribute is no longer exported |
| 12 | // The functionality is now only available through ParseTitleAttributesToTerms method |
Akron | e562ad6 | 2025-06-25 11:15:56 +0200 | [diff] [blame] | 13 | |
| 14 | func TestTitleAttributeParser_ParseTitleAttributesToTerms(t *testing.T) { |
| 15 | parser := NewTitleAttributeParser() |
| 16 | |
| 17 | tests := []struct { |
| 18 | name string |
| 19 | input []string |
| 20 | expected []ast.Node |
| 21 | wantErr bool |
| 22 | }{ |
| 23 | { |
| 24 | name: "Parse multiple title attributes", |
Akron | 7de5b61 | 2025-06-26 16:15:03 +0200 | [diff] [blame] | 25 | input: []string{"corenlp/p:ART", "marmot/m:case:nom", "tt/l:die"}, |
Akron | e562ad6 | 2025-06-25 11:15:56 +0200 | [diff] [blame] | 26 | expected: []ast.Node{ |
| 27 | &ast.Term{ |
| 28 | Foundry: "corenlp", |
| 29 | Layer: "p", |
| 30 | Key: "ART", |
| 31 | Value: "", |
| 32 | Match: ast.MatchEqual, |
| 33 | }, |
| 34 | &ast.Term{ |
| 35 | Foundry: "marmot", |
| 36 | Layer: "m", |
| 37 | Key: "case", |
| 38 | Value: "nom", |
| 39 | Match: ast.MatchEqual, |
| 40 | }, |
| 41 | &ast.Term{ |
| 42 | Foundry: "tt", |
| 43 | Layer: "l", |
| 44 | Key: "die", |
| 45 | Value: "", |
| 46 | Match: ast.MatchEqual, |
| 47 | }, |
| 48 | }, |
| 49 | wantErr: false, |
| 50 | }, |
| 51 | { |
| 52 | name: "Empty input should return empty slice", |
| 53 | input: []string{}, |
| 54 | expected: []ast.Node{}, |
| 55 | wantErr: false, |
| 56 | }, |
| 57 | { |
| 58 | name: "Invalid title should cause error", |
| 59 | input: []string{"corenlp/p:ART", "invalid_title", "tt/l:die"}, |
| 60 | wantErr: true, |
| 61 | }, |
Akron | 6b4c9eb | 2025-07-03 14:31:58 +0200 | [diff] [blame] | 62 | // Additional tests to cover functionality from removed TestTitleAttributeParser_ParseTitleAttribute |
| 63 | { |
| 64 | name: "Parse simple title with key only", |
| 65 | input: []string{"corenlp/p:ART"}, |
| 66 | expected: []ast.Node{ |
| 67 | &ast.Term{ |
| 68 | Foundry: "corenlp", |
| 69 | Layer: "p", |
| 70 | Key: "ART", |
| 71 | Value: "", |
| 72 | Match: ast.MatchEqual, |
| 73 | }, |
| 74 | }, |
| 75 | wantErr: false, |
| 76 | }, |
| 77 | { |
| 78 | name: "Parse title with key and value", |
| 79 | input: []string{"marmot/m:case:nom"}, |
| 80 | expected: []ast.Node{ |
| 81 | &ast.Term{ |
| 82 | Foundry: "marmot", |
| 83 | Layer: "m", |
| 84 | Key: "case", |
| 85 | Value: "nom", |
| 86 | Match: ast.MatchEqual, |
| 87 | }, |
| 88 | }, |
| 89 | wantErr: false, |
| 90 | }, |
| 91 | { |
| 92 | name: "Parse title with colon separator for value", |
| 93 | input: []string{"marmot/m:gender:masc"}, |
| 94 | expected: []ast.Node{ |
| 95 | &ast.Term{ |
| 96 | Foundry: "marmot", |
| 97 | Layer: "m", |
| 98 | Key: "gender", |
| 99 | Value: "masc", |
| 100 | Match: ast.MatchEqual, |
| 101 | }, |
| 102 | }, |
| 103 | wantErr: false, |
| 104 | }, |
| 105 | { |
| 106 | name: "Parse title with equals separator for value", |
| 107 | input: []string{"marmot/m:degree:pos"}, |
| 108 | expected: []ast.Node{ |
| 109 | &ast.Term{ |
| 110 | Foundry: "marmot", |
| 111 | Layer: "m", |
| 112 | Key: "degree", |
| 113 | Value: "pos", |
| 114 | Match: ast.MatchEqual, |
| 115 | }, |
| 116 | }, |
| 117 | wantErr: false, |
| 118 | }, |
| 119 | { |
| 120 | name: "Parse title with lemma layer", |
| 121 | input: []string{"tt/l:die"}, |
| 122 | expected: []ast.Node{ |
| 123 | &ast.Term{ |
| 124 | Foundry: "tt", |
| 125 | Layer: "l", |
| 126 | Key: "die", |
| 127 | Value: "", |
| 128 | Match: ast.MatchEqual, |
| 129 | }, |
| 130 | }, |
| 131 | wantErr: false, |
| 132 | }, |
| 133 | { |
| 134 | name: "Parse title with special characters in value", |
| 135 | input: []string{"tt/l:@card@"}, |
| 136 | expected: []ast.Node{ |
| 137 | &ast.Term{ |
| 138 | Foundry: "tt", |
| 139 | Layer: "l", |
| 140 | Key: "@card@", |
| 141 | Value: "", |
| 142 | Match: ast.MatchEqual, |
| 143 | }, |
| 144 | }, |
| 145 | wantErr: false, |
| 146 | }, |
| 147 | { |
| 148 | name: "Parse complex key-value with colon", |
| 149 | input: []string{"opennlp/p:PronType:Ind"}, |
| 150 | expected: []ast.Node{ |
| 151 | &ast.Term{ |
| 152 | Foundry: "opennlp", |
| 153 | Layer: "p", |
| 154 | Key: "PronType", |
| 155 | Value: "Ind", |
| 156 | Match: ast.MatchEqual, |
| 157 | }, |
| 158 | }, |
| 159 | wantErr: false, |
| 160 | }, |
| 161 | { |
| 162 | name: "Parse complex key-value with equals", |
| 163 | input: []string{"opennlp/p:AdjType:Pdt"}, |
| 164 | expected: []ast.Node{ |
| 165 | &ast.Term{ |
| 166 | Foundry: "opennlp", |
| 167 | Layer: "p", |
| 168 | Key: "AdjType", |
| 169 | Value: "Pdt", |
| 170 | Match: ast.MatchEqual, |
| 171 | }, |
| 172 | }, |
| 173 | wantErr: false, |
| 174 | }, |
| 175 | { |
| 176 | name: "Parse complex nested pattern", |
| 177 | input: []string{"stts/p:ADJA"}, |
| 178 | expected: []ast.Node{ |
| 179 | &ast.Term{ |
| 180 | Foundry: "stts", |
| 181 | Layer: "p", |
| 182 | Key: "ADJA", |
| 183 | Value: "", |
| 184 | Match: ast.MatchEqual, |
| 185 | }, |
| 186 | }, |
| 187 | wantErr: false, |
| 188 | }, |
| 189 | // Error cases |
| 190 | { |
| 191 | name: "Empty title should fail", |
| 192 | input: []string{""}, |
| 193 | wantErr: true, |
| 194 | }, |
| 195 | { |
| 196 | name: "Missing foundry separator should fail", |
| 197 | input: []string{"corenlp_p:ART"}, |
| 198 | wantErr: true, |
| 199 | }, |
| 200 | { |
| 201 | name: "Missing layer separator should fail", |
| 202 | input: []string{"corenlp/p_ART"}, |
| 203 | wantErr: true, |
| 204 | }, |
| 205 | { |
| 206 | name: "Only foundry should fail", |
| 207 | input: []string{"corenlp"}, |
| 208 | wantErr: true, |
| 209 | }, |
| 210 | { |
| 211 | name: "Only foundry and layer should fail", |
| 212 | input: []string{"corenlp/p"}, |
| 213 | wantErr: true, |
| 214 | }, |
| 215 | { |
| 216 | name: "Missing key should fail", |
| 217 | input: []string{"corenlp/p:"}, |
| 218 | wantErr: true, |
| 219 | }, |
Akron | e562ad6 | 2025-06-25 11:15:56 +0200 | [diff] [blame] | 220 | } |
| 221 | |
| 222 | for _, tt := range tests { |
| 223 | t.Run(tt.name, func(t *testing.T) { |
| 224 | result, err := parser.ParseTitleAttributesToTerms(tt.input) |
| 225 | |
| 226 | if tt.wantErr { |
| 227 | assert.Error(t, err) |
| 228 | } else { |
| 229 | require.NoError(t, err) |
| 230 | require.Len(t, result, len(tt.expected)) |
| 231 | |
| 232 | for i, expectedTerm := range tt.expected { |
| 233 | expectedTermNode := expectedTerm.(*ast.Term) |
| 234 | actualTermNode := result[i].(*ast.Term) |
| 235 | |
| 236 | assert.Equal(t, expectedTermNode.Foundry, actualTermNode.Foundry) |
| 237 | assert.Equal(t, expectedTermNode.Layer, actualTermNode.Layer) |
| 238 | assert.Equal(t, expectedTermNode.Key, actualTermNode.Key) |
| 239 | assert.Equal(t, expectedTermNode.Value, actualTermNode.Value) |
| 240 | assert.Equal(t, expectedTermNode.Match, actualTermNode.Match) |
| 241 | } |
| 242 | } |
| 243 | }) |
| 244 | } |
| 245 | } |
| 246 | |
Akron | 6b4c9eb | 2025-07-03 14:31:58 +0200 | [diff] [blame] | 247 | // TestTitleAttribute_ToAST was removed as ToAST method is no longer available |
| 248 | // TestTitleAttribute_String was removed as String method is no longer available |
| 249 | // TestTitleAttributeParser_RealWorldExample was removed as it used the removed methods |