clerk/internal/lsp/textdocument_semantic_tokens_test.go (view raw)
| 1 | package lsp |
| 2 | |
| 3 | import ( |
| 4 | "fmt" |
| 5 | "os" |
| 6 | "slices" |
| 7 | "strings" |
| 8 | "testing" |
| 9 | |
| 10 | "go.lsp.dev/protocol" |
| 11 | "go.lsp.dev/uri" |
| 12 | |
| 13 | "olexsmir.xyz/clerk/internal/testutil/golden" |
| 14 | ) |
| 15 | |
| 16 | func TestEncodeSemTokens(t *testing.T) { |
| 17 | tests := map[string]struct { |
| 18 | tokens []semanticToken |
| 19 | want []uint32 |
| 20 | }{ |
| 21 | "nil": {nil, nil}, |
| 22 | "empty": {[]semanticToken{}, nil}, |
| 23 | "single": {[]semanticToken{ |
| 24 | {line: 0, col: 0, length: 4, tokenType: semDate}, |
| 25 | }, []uint32{0, 0, 4, semDate, 0}}, |
| 26 | "line 0 col 0": {[]semanticToken{ |
| 27 | {line: 0, col: 0, length: 1, tokenType: semDirective}, |
| 28 | }, []uint32{0, 0, 1, semDirective, 0}}, |
| 29 | "multiple": {[]semanticToken{ |
| 30 | {line: 0, col: 0, length: 10, tokenType: semDate}, |
| 31 | {line: 0, col: 11, length: 5, tokenType: semString}, |
| 32 | {line: 1, col: 4, length: 10, tokenType: semAccount}, |
| 33 | }, []uint32{ |
| 34 | 0, 0, 10, semDate, 0, |
| 35 | 0, 11, 5, semString, 0, |
| 36 | 1, 4, 10, semAccount, 0, |
| 37 | }}, |
| 38 | // input must be sorted by line and column (rawToSemanticTokens output); |
| 39 | // deltas would underflow otherwise |
| 40 | "sorted": {[]semanticToken{ |
| 41 | {line: 0, col: 0, length: 10, tokenType: semDate}, |
| 42 | {line: 0, col: 32, length: 2, tokenType: semAmount}, |
| 43 | {line: 0, col: 36, length: 3, tokenType: semCommodity}, |
| 44 | {line: 1, col: 4, length: 6, tokenType: semAccount}, |
| 45 | }, []uint32{ |
| 46 | 0, 0, 10, semDate, 0, |
| 47 | 0, 32, 2, semAmount, 0, |
| 48 | 0, 4, 3, semCommodity, 0, |
| 49 | 1, 4, 6, semAccount, 0, |
| 50 | }}, |
| 51 | } |
| 52 | |
| 53 | for tname, tt := range tests { |
| 54 | t.Run(tname, func(t *testing.T) { |
| 55 | if got := encodeSemTokens(tt.tokens); !slices.Equal(got, tt.want) { |
| 56 | t.Errorf("encodeSemTokens() = %v, want %v", got, tt.want) |
| 57 | } |
| 58 | }) |
| 59 | } |
| 60 | } |
| 61 | |
| 62 | func TestSemanticTokensEdits(t *testing.T) { |
| 63 | tests := map[string]struct{ old, new []uint32 }{ |
| 64 | "identical": {[]uint32{1, 2, 3}, []uint32{1, 2, 3}}, |
| 65 | "both empty": {}, |
| 66 | "empty to non": {nil, []uint32{1, 2}}, |
| 67 | "non to empty": {[]uint32{1, 2}, nil}, |
| 68 | "pure insert": {[]uint32{1, 2}, []uint32{1, 2, 3, 4}}, |
| 69 | "pure delete": {[]uint32{1, 2, 3, 4}, []uint32{1, 2}}, |
| 70 | "replace middle": {[]uint32{1, 2, 3, 4, 5}, []uint32{1, 2, 9, 4, 5}}, |
| 71 | "replace all": {[]uint32{1, 2}, []uint32{3, 4}}, |
| 72 | "replace tail": {[]uint32{1, 2, 3, 4, 5}, []uint32{1, 2, 3, 4, 6}}, |
| 73 | } |
| 74 | for tname, tt := range tests { |
| 75 | t.Run(tname, func(t *testing.T) { |
| 76 | edits := semanticTokensEdits(tt.old, tt.new) |
| 77 | if got := applySemEdits(tt.old, edits); !slices.Equal(got, tt.new) { |
| 78 | t.Errorf("apply(%v, %v) = %v, want %v", tt.old, edits, got, tt.new) |
| 79 | } |
| 80 | for _, e := range edits { |
| 81 | if e.Start+e.DeleteCount > uint32(len(tt.old)) { |
| 82 | t.Errorf("edit %+v out of bounds for old length %d", e, len(tt.old)) |
| 83 | } |
| 84 | } |
| 85 | }) |
| 86 | } |
| 87 | } |
| 88 | |
| 89 | func TestServer_Semantic_EmptyDocument(t *testing.T) { |
| 90 | srv := NewServer("test") |
| 91 | srv.server.openDoc(uri.URI("file:///empty.journal"), "", 1, "journal") |
| 92 | result, err := srv.server.SemanticTokensFull(t.Context(), &protocol.SemanticTokensParams{ |
| 93 | TextDocument: protocol.TextDocumentIdentifier{URI: uri.URI("file:///empty.journal")}, |
| 94 | }) |
| 95 | if err != nil { |
| 96 | t.Fatal(err) |
| 97 | } |
| 98 | if len(result.Data) != 0 { |
| 99 | t.Errorf("expected empty data for empty doc, got %d values", len(result.Data)) |
| 100 | } |
| 101 | } |
| 102 | |
| 103 | func TestServer_Semantic_DocumentNotFound(t *testing.T) { |
| 104 | result, err := NewServer("test").server.SemanticTokensFull(t.Context(), &protocol.SemanticTokensParams{ |
| 105 | TextDocument: protocol.TextDocumentIdentifier{URI: uri.URI("file:///unknown.journal")}, |
| 106 | }) |
| 107 | if err != nil { |
| 108 | t.Fatal(err) |
| 109 | } |
| 110 | if result == nil { |
| 111 | t.Fatal("result is nil") |
| 112 | } |
| 113 | if result.Data != nil { |
| 114 | t.Errorf("expected nil Data for unknown doc, got %v", result.Data) |
| 115 | } |
| 116 | } |
| 117 | |
| 118 | func TestGolden_SemanticTokens(t *testing.T) { |
| 119 | for _, tt := range []string{"semantic-empty", "semantic-journal", "semantic-directives", "semantic-unparseable", "semantic-with-errors"} { |
| 120 | ar := golden.Read(t, tt) |
| 121 | |
| 122 | t.Run(tt+"_golden", func(t *testing.T) { |
| 123 | toks := renderSemanticTokens(tokSem(ar.Get("in.journal"))) |
| 124 | golden.Assert(t, ar, toks) |
| 125 | }) |
| 126 | |
| 127 | t.Run(tt+"_no-overlap", func(t *testing.T) { |
| 128 | assertGoldenNoOverlap(t, tt, ar) |
| 129 | }) |
| 130 | } |
| 131 | } |
| 132 | |
| 133 | func TestGolden_SemanticTokensRange(t *testing.T) { |
| 134 | ar := golden.Read(t, "semantic-range") |
| 135 | in := ar.Get("in.journal") |
| 136 | |
| 137 | t.Run("no-overlap", func(t *testing.T) { |
| 138 | assertGoldenNoOverlap(t, "semantic-range", ar) |
| 139 | }) |
| 140 | |
| 141 | t.Run("golden", func(t *testing.T) { |
| 142 | u := uri.URI("file:///test.journal") |
| 143 | srv := NewServer("test") |
| 144 | srv.server.openDoc(u, string(in), 1, "journal") |
| 145 | |
| 146 | var out strings.Builder |
| 147 | for line := range strings.SplitSeq(string(ar.Get("ranges.txt")), "\n") { |
| 148 | if line == "" { |
| 149 | continue |
| 150 | } |
| 151 | var name string |
| 152 | var start, end uint32 |
| 153 | if _, err := fmt.Sscanf(line, "%s %d %d", &name, &start, &end); err != nil { |
| 154 | t.Fatalf("ranges.txt: %q: %v", line, err) |
| 155 | } |
| 156 | |
| 157 | res, err := srv.server.SemanticTokensRange(t.Context(), &protocol.SemanticTokensRangeParams{ |
| 158 | TextDocument: protocol.TextDocumentIdentifier{URI: u}, |
| 159 | Range: protocol.Range{ |
| 160 | Start: protocol.Position{Line: start}, |
| 161 | End: protocol.Position{Line: end}, |
| 162 | }, |
| 163 | }) |
| 164 | if err != nil { |
| 165 | t.Fatal(err) |
| 166 | } |
| 167 | tokens := decodeSemTokens(res.Data) |
| 168 | for _, tok := range tokens { |
| 169 | if tok.line < start || tok.line > end { |
| 170 | t.Fatalf("%s: token %d:%d outside range lines [%d,%d]", name, tok.line, tok.col, start, end) |
| 171 | } |
| 172 | } |
| 173 | fmt.Fprintf(&out, "== %s ==\n", name) |
| 174 | out.WriteString(renderSemanticTokens(tokens)) |
| 175 | } |
| 176 | golden.Assert(t, ar, out.String()) |
| 177 | }) |
| 178 | } |
| 179 | |
| 180 | func TestGolden_SemanticTokensDelta(t *testing.T) { |
| 181 | for _, tt := range []string{"semantic-delta-edit", "semantic-delta-nochange", "semantic-delta-stale"} { |
| 182 | ar := golden.Read(t, tt) |
| 183 | |
| 184 | t.Run(tt+"_no-overlap", func(t *testing.T) { |
| 185 | assertGoldenNoOverlap(t, tt, ar) |
| 186 | }) |
| 187 | |
| 188 | t.Run(tt, func(t *testing.T) { |
| 189 | in := ar.Get("in.journal") |
| 190 | |
| 191 | u := uri.URI("file:///test.journal") |
| 192 | srv := NewServer("test") |
| 193 | srv.server.openDoc(u, string(in), 1, "journal") |
| 194 | |
| 195 | full, err := srv.server.SemanticTokensFull(t.Context(), &protocol.SemanticTokensParams{ |
| 196 | TextDocument: protocol.TextDocumentIdentifier{URI: u}, |
| 197 | }) |
| 198 | if err != nil { |
| 199 | t.Fatal(err) |
| 200 | } |
| 201 | if full.ResultID == nil || *full.ResultID == "" { |
| 202 | t.Fatal("expected a resultId on the full result") |
| 203 | } |
| 204 | |
| 205 | finalText := in |
| 206 | if ed := ar.Get("edited.journal"); ed != nil { |
| 207 | srv.server.updateDoc(u, 2, []protocol.TextDocumentContentChangeEvent{ |
| 208 | &protocol.TextDocumentContentChangeWholeDocument{Text: string(ed)}, |
| 209 | }) |
| 210 | finalText = ed |
| 211 | } |
| 212 | prev := *full.ResultID |
| 213 | if p := ar.Get("prev-id"); p != nil { |
| 214 | prev = strings.TrimSpace(string(p)) |
| 215 | } |
| 216 | |
| 217 | res, err := srv.server.SemanticTokensFullDelta(t.Context(), &protocol.SemanticTokensDeltaParams{ |
| 218 | TextDocument: protocol.TextDocumentIdentifier{URI: u}, |
| 219 | PreviousResultID: prev, |
| 220 | }) |
| 221 | if err != nil { |
| 222 | t.Fatal(err) |
| 223 | } |
| 224 | |
| 225 | var client []uint32 |
| 226 | var out strings.Builder |
| 227 | switch r := res.(type) { |
| 228 | case *protocol.SemanticTokensDelta: |
| 229 | fmt.Fprintf(&out, "delta resultId %s\n", *r.ResultID) |
| 230 | client = applySemEdits(full.Data, r.Edits) |
| 231 | case *protocol.SemanticTokens: |
| 232 | fmt.Fprintf(&out, "full resultId %s\n", *r.ResultID) |
| 233 | client = r.Data |
| 234 | default: |
| 235 | t.Fatalf("unexpected result type %T", res) |
| 236 | } |
| 237 | |
| 238 | // The client's token state must equal the final text's tokens: an |
| 239 | // independent reference that keeps the golden from capturing bugs. |
| 240 | if want := encodeSemTokens(tokSem(finalText)); !slices.Equal(client, want) { |
| 241 | t.Errorf("delta flow produced %d elems, want %d", len(client), len(want)) |
| 242 | } |
| 243 | |
| 244 | out.WriteString(renderSemanticTokens(tokSem(finalText))) |
| 245 | golden.Assert(t, ar, out.String()) |
| 246 | }) |
| 247 | } |
| 248 | } |
| 249 | |
| 250 | func assertGoldenNoOverlap(t *testing.T, tt string, ar *golden.Archive) { |
| 251 | t.Helper() |
| 252 | for _, f := range ar.Files { |
| 253 | if strings.HasSuffix(f.Name, ".journal") { |
| 254 | assertNoOverlap(t, tt, tokSem(f.Data)) |
| 255 | } |
| 256 | } |
| 257 | } |
| 258 | |
| 259 | func assertNoOverlap(t *testing.T, tt string, toks []semanticToken) { |
| 260 | t.Helper() |
| 261 | slices.SortFunc(toks, func(a, b semanticToken) int { |
| 262 | if a.line != b.line { |
| 263 | return int(a.line) - int(b.line) |
| 264 | } |
| 265 | return int(a.col) - int(b.col) |
| 266 | }) |
| 267 | for i := 1; i < len(toks); i++ { |
| 268 | prev, cur := toks[i-1], toks[i] |
| 269 | if prev.line != cur.line { |
| 270 | continue |
| 271 | } |
| 272 | if cur.col < prev.col+prev.length { |
| 273 | t.Errorf("%s: overlapping tokens on line %d: %s@%d+%d then %s@%d+%d", |
| 274 | tt, prev.line, tokenTypeStrings[prev.tokenType], prev.col, prev.length, |
| 275 | tokenTypeStrings[cur.tokenType], cur.col, cur.length) |
| 276 | } |
| 277 | } |
| 278 | } |
| 279 | |
| 280 | func renderSemanticTokens(tokens []semanticToken) string { |
| 281 | slices.SortFunc(tokens, func(a, b semanticToken) int { |
| 282 | if a.line != b.line { |
| 283 | return int(a.line) - int(b.line) |
| 284 | } |
| 285 | return int(a.col) - int(b.col) |
| 286 | }) |
| 287 | var b strings.Builder |
| 288 | for _, tok := range tokens { |
| 289 | fmt.Fprintf(&b, "%d:%d+%d %s", tok.line, tok.col, tok.length, tokenTypeStrings[tok.tokenType]) |
| 290 | for i, m := range modifierStrings { |
| 291 | if tok.modifiers&(1<<uint(i)) != 0 { |
| 292 | b.WriteByte(' ') |
| 293 | b.WriteString(m) |
| 294 | } |
| 295 | } |
| 296 | b.WriteByte('\n') |
| 297 | } |
| 298 | return b.String() |
| 299 | } |
| 300 | |
| 301 | func decodeSemTokens(data []uint32) []semanticToken { |
| 302 | var out []semanticToken |
| 303 | line, col := 0, 0 |
| 304 | for i := 0; i+4 < len(data); i += 5 { |
| 305 | if data[i] > 0 { |
| 306 | line += int(data[i]) |
| 307 | col = int(data[i+1]) |
| 308 | } else { |
| 309 | col += int(data[i+1]) |
| 310 | } |
| 311 | out = append(out, semanticToken{ |
| 312 | line: uint32(line), |
| 313 | col: uint32(col), |
| 314 | length: data[i+2], |
| 315 | tokenType: data[i+3], |
| 316 | modifiers: data[i+4], |
| 317 | }) |
| 318 | } |
| 319 | return out |
| 320 | } |
| 321 | |
| 322 | func tokSem(content []byte) []semanticToken { |
| 323 | c := string(content) |
| 324 | return tokenizeForSemantics(c, parseJournalStr(c)) |
| 325 | } |
| 326 | |
| 327 | func BenchmarkSemanticTokens(b *testing.B) { |
| 328 | content := openJournal(b, "../../journal/testdata/journals/actual-1ktxns-100accts.journal") |
| 329 | |
| 330 | // Cold path: each iteration re-parses and re-encodes, as after an edit. |
| 331 | b.ReportAllocs() |
| 332 | b.ResetTimer() |
| 333 | for b.Loop() { |
| 334 | tokens := tokenizeForSemantics(content, parseJournalStr(content)) |
| 335 | _ = encodeSemTokens(tokens) |
| 336 | } |
| 337 | } |
| 338 | |
| 339 | // BenchmarkSemanticTokensDelta measures the cost of one dela response after an edit. |
| 340 | func BenchmarkSemanticTokensDelta(b *testing.B) { |
| 341 | content := openJournal(b, "../../journal/testdata/journals/actual-1ktxns-100accts.journal") |
| 342 | prev := encodeSemTokens(tokenizeForSemantics(content, parseJournalStr(content))) |
| 343 | edited := content + "\n2000-06-15 transaction 2501\n expenses:new 1 C\n assets:cash\n" |
| 344 | |
| 345 | b.ReportAllocs() |
| 346 | b.ResetTimer() |
| 347 | for b.Loop() { |
| 348 | data := encodeSemTokens(tokenizeForSemantics(edited, parseJournalStr(edited))) |
| 349 | _ = semanticTokensEdits(prev, data) |
| 350 | } |
| 351 | } |
| 352 | |
| 353 | func BenchmarkSemanticTokensEdits(b *testing.B) { |
| 354 | content := openJournal(b, "../../journal/testdata/journals/actual-1ktxns-100accts.journal") |
| 355 | old := encodeSemTokens(tokenizeForSemantics(content, parseJournalStr(content))) |
| 356 | new := encodeSemTokens(tokenizeForSemantics( |
| 357 | content+"\n2000-06-15 transaction 2501\n expenses:new 1 C\n assets:cash\n", |
| 358 | parseJournalStr(content+"\n2000-06-15 transaction 2501\n expenses:new 1 C\n assets:cash\n"), |
| 359 | )) |
| 360 | |
| 361 | b.ReportAllocs() |
| 362 | b.ResetTimer() |
| 363 | for b.Loop() { |
| 364 | _ = semanticTokensEdits(old, new) |
| 365 | } |
| 366 | } |
| 367 | |
| 368 | func applySemEdits(data []uint32, edits []protocol.SemanticTokensEdit) []uint32 { |
| 369 | out := slices.Clone(data) |
| 370 | for _, e := range edits { |
| 371 | out = append(append(out[:e.Start], e.Data...), out[e.Start+e.DeleteCount:]...) |
| 372 | } |
| 373 | return out |
| 374 | } |
| 375 | |
| 376 | func openJournal(t testing.TB, path string) string { |
| 377 | t.Helper() |
| 378 | src, err := os.ReadFile(path) |
| 379 | if err != nil { |
| 380 | t.Fatal(err) |
| 381 | } |
| 382 | return string(src) |
| 383 | } |