| 1 |
// Copyright (C) 2019 Google Inc. |
| 2 |
// |
| 3 |
// Licensed under the Apache License, Version 2.0 (the "License"); |
| 4 |
// you may not use this file except in compliance with the License. |
| 5 |
// You may obtain a copy of the License at |
| 6 |
// |
| 7 |
// http://www.apache.org/licenses/LICENSE-2.0 |
| 8 |
// |
| 9 |
// Unless required by applicable law or agreed to in writing, software |
| 10 |
// distributed under the License is distributed on an "AS IS" BASIS, |
| 11 |
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 12 |
// See the License for the specific language governing permissions and |
| 13 |
// limitations under the License. |
| 14 |
|
| 15 |
// Package parser implements a SPIR-V assembly parser. |
| 16 |
package parser |
| 17 |
|
| 18 |
import ( |
| 19 |
"fmt" |
| 20 |
"io" |
| 21 |
"log" |
| 22 |
"strings" |
| 23 |
"unicode" |
| 24 |
"unicode/utf8" |
| 25 |
|
| 26 |
"../schema" |
| 27 |
) |
| 28 |
|
| 29 |
// Type is an enumerator of token types. |
| 30 |
type Type int |
| 31 |
|
| 32 |
// Type enumerators |
| 33 |
const ( |
| 34 |
Ident Type = iota // Foo |
| 35 |
PIdent // %32, %foo |
| 36 |
Integer |
| 37 |
Float |
| 38 |
String |
| 39 |
Operator |
| 40 |
Comment |
| 41 |
Newline |
| 42 |
) |
| 43 |
|
| 44 |
func (t Type) String() string { |
| 45 |
switch t { |
| 46 |
case Ident: |
| 47 |
return "Ident" |
| 48 |
case PIdent: |
| 49 |
return "PIdent" |
| 50 |
case Integer: |
| 51 |
return "Integer" |
| 52 |
case Float: |
| 53 |
return "Float" |
| 54 |
case String: |
| 55 |
return "String" |
| 56 |
case Operator: |
| 57 |
return "Operator" |
| 58 |
case Comment: |
| 59 |
return "Comment" |
| 60 |
default: |
| 61 |
return "<unknown>" |
| 62 |
} |
| 63 |
} |
| 64 |
|
| 65 |
// Token represents a single lexed token. |
| 66 |
type Token struct { |
| 67 |
Type Type |
| 68 |
Range Range |
| 69 |
} |
| 70 |
|
| 71 |
func (t Token) String() string { return fmt.Sprintf("{%v %v}", t.Type, t.Range) } |
| 72 |
|
| 73 |
// Text returns the tokens text from the source. |
| 74 |
func (t Token) Text(lines []string) string { return t.Range.Text(lines) } |
| 75 |
|
| 76 |
// Range represents an interval in a text file. |
| 77 |
type Range struct { |
| 78 |
Start Position |
| 79 |
End Position |
| 80 |
} |
| 81 |
|
| 82 |
func (r Range) String() string { return fmt.Sprintf("[%v %v]", r.Start, r.End) } |
| 83 |
|
| 84 |
// Text returns the text for the given Range in the provided lines. |
| 85 |
func (r Range) Text(lines []string) string { |
| 86 |
sl, sc := r.Start.Line-1, r.Start.Column-1 |
| 87 |
if sl < 0 || sc < 0 || sl > len(lines) || sc > len(lines[sl]) { |
| 88 |
return fmt.Sprintf("<invalid start position %v>", r.Start) |
| 89 |
} |
| 90 |
el, ec := r.End.Line-1, r.End.Column-1 |
| 91 |
if el < 0 || ec < 0 || el > len(lines) || ec > len(lines[sl]) { |
| 92 |
return fmt.Sprintf("<invalid end position %v>", r.End) |
| 93 |
} |
| 94 |
|
| 95 |
sb := strings.Builder{} |
| 96 |
if sl != el { |
| 97 |
sb.WriteString(lines[sl][sc:]) |
| 98 |
for l := sl + 1; l < el; l++ { |
| 99 |
sb.WriteString(lines[l]) |
| 100 |
} |
| 101 |
sb.WriteString(lines[el][:ec]) |
| 102 |
} else { |
| 103 |
sb.WriteString(lines[sl][sc:ec]) |
| 104 |
} |
| 105 |
return sb.String() |
| 106 |
} |
| 107 |
|
| 108 |
// Contains returns true if p is in r. |
| 109 |
func (r Range) Contains(p Position) bool { |
| 110 |
return !(p.LessThan(r.Start) || p.GreaterThan(r.End)) |
| 111 |
} |
| 112 |
|
| 113 |
func (r *Range) grow(o Range) { |
| 114 |
if !r.Start.IsValid() || o.Start.LessThan(r.Start) { |
| 115 |
r.Start = o.Start |
| 116 |
} |
| 117 |
if !r.End.IsValid() || o.End.GreaterThan(r.End) { |
| 118 |
r.End = o.End |
| 119 |
} |
| 120 |
} |
| 121 |
|
| 122 |
// Position holds a line and column position in a text file. |
| 123 |
type Position struct { |
| 124 |
Line, Column int |
| 125 |
} |
| 126 |
|
| 127 |
func (p Position) String() string { return fmt.Sprintf("%v:%v", p.Line, p.Column) } |
| 128 |
|
| 129 |
// IsValid returns true if the position has a line and column greater than 1. |
| 130 |
func (p Position) IsValid() bool { return p.Line > 0 && p.Column > 0 } |
| 131 |
|
| 132 |
// LessThan returns true iff o is before p. |
| 133 |
func (p Position) LessThan(o Position) bool { |
| 134 |
switch { |
| 135 |
case !p.IsValid() || !o.IsValid(): |
| 136 |
return false |
| 137 |
case p.Line < o.Line: |
| 138 |
return true |
| 139 |
case p.Line > o.Line: |
| 140 |
return false |
| 141 |
case p.Column < o.Column: |
| 142 |
return true |
| 143 |
default: |
| 144 |
return false |
| 145 |
} |
| 146 |
} |
| 147 |
|
| 148 |
// GreaterThan returns true iff o is greater than p. |
| 149 |
func (p Position) GreaterThan(o Position) bool { |
| 150 |
switch { |
| 151 |
case !p.IsValid() || !o.IsValid(): |
| 152 |
return false |
| 153 |
case p.Line > o.Line: |
| 154 |
return true |
| 155 |
case p.Line < o.Line: |
| 156 |
return false |
| 157 |
case p.Column > o.Column: |
| 158 |
return true |
| 159 |
default: |
| 160 |
return false |
| 161 |
} |
| 162 |
} |
| 163 |
|
| 164 |
type lexer struct { |
| 165 |
source string |
| 166 |
lexerState |
| 167 |
diags []Diagnostic |
| 168 |
e error |
| 169 |
} |
| 170 |
|
| 171 |
type lexerState struct { |
| 172 |
offset int // byte offset in source |
| 173 |
toks []*Token // all the lexed tokens |
| 174 |
pos Position // current position |
| 175 |
} |
| 176 |
|
| 177 |
// err appends an fmt.Printf style error into l.diags for the given token. |
| 178 |
func (l *lexer) err(tok *Token, msg string, args ...interface{}) { |
| 179 |
rng := Range{} |
| 180 |
if tok != nil { |
| 181 |
rng = tok.Range |
| 182 |
} |
| 183 |
l.diags = append(l.diags, Diagnostic{ |
| 184 |
Range: rng, |
| 185 |
Severity: SeverityError, |
| 186 |
Message: fmt.Sprintf(msg, args...), |
| 187 |
}) |
| 188 |
} |
| 189 |
|
| 190 |
// next returns the next rune, or io.EOF if the last rune has already been |
| 191 |
// consumed. |
| 192 |
func (l *lexer) next() rune { |
| 193 |
if l.offset >= len(l.source) { |
| 194 |
l.e = io.EOF |
| 195 |
return 0 |
| 196 |
} |
| 197 |
r, n := utf8.DecodeRuneInString(l.source[l.offset:]) |
| 198 |
l.offset += n |
| 199 |
if n == 0 { |
| 200 |
l.e = io.EOF |
| 201 |
return 0 |
| 202 |
} |
| 203 |
if r == '\n' { |
| 204 |
l.pos.Line++ |
| 205 |
l.pos.Column = 1 |
| 206 |
} else { |
| 207 |
l.pos.Column++ |
| 208 |
} |
| 209 |
return r |
| 210 |
} |
| 211 |
|
| 212 |
// save returns the current lexerState. |
| 213 |
func (l *lexer) save() lexerState { |
| 214 |
return l.lexerState |
| 215 |
} |
| 216 |
|
| 217 |
// restore restores the current lexer state with s. |
| 218 |
func (l *lexer) restore(s lexerState) { |
| 219 |
l.lexerState = s |
| 220 |
} |
| 221 |
|
| 222 |
// pident processes the PIdent token at the current position. |
| 223 |
// The lexer *must* know the next token is a PIdent before calling. |
| 224 |
func (l *lexer) pident() { |
| 225 |
tok := &Token{Type: PIdent, Range: Range{Start: l.pos, End: l.pos}} |
| 226 |
if r := l.next(); r != '%' { |
| 227 |
log.Fatalf("lexer expected '%%', got '%v'", r) |
| 228 |
return |
| 229 |
} |
| 230 |
for l.e == nil { |
| 231 |
s := l.save() |
| 232 |
r := l.next() |
| 233 |
if !isAlphaNumeric(r) && r != '_' { |
| 234 |
l.restore(s) |
| 235 |
break |
| 236 |
} |
| 237 |
} |
| 238 |
tok.Range.End = l.pos |
| 239 |
l.toks = append(l.toks, tok) |
| 240 |
} |
| 241 |
|
| 242 |
// numberOrIdent processes the Ident, Float or Integer token at the current |
| 243 |
// position. |
| 244 |
func (l *lexer) numberOrIdent() { |
| 245 |
const Unknown Type = -1 |
| 246 |
tok := &Token{Type: Unknown, Range: Range{Start: l.pos, End: l.pos}} |
| 247 |
loop: |
| 248 |
for l.e == nil { |
| 249 |
s := l.save() |
| 250 |
r := l.next() |
| 251 |
switch { |
| 252 |
case r == '-', r == '+', isNumeric(r): |
| 253 |
continue |
| 254 |
case isAlpha(r), r == '_': |
| 255 |
switch tok.Type { |
| 256 |
case Unknown: |
| 257 |
tok.Type = Ident |
| 258 |
case Float, Integer: |
| 259 |
l.err(tok, "invalid number") |
| 260 |
return |
| 261 |
} |
| 262 |
case r == '.': |
| 263 |
switch tok.Type { |
| 264 |
case Unknown: |
| 265 |
tok.Type = Float |
| 266 |
default: |
| 267 |
l.restore(s) |
| 268 |
break loop |
| 269 |
} |
| 270 |
default: |
| 271 |
if tok.Type == Unknown { |
| 272 |
tok.Type = Integer |
| 273 |
} |
| 274 |
l.restore(s) |
| 275 |
break loop |
| 276 |
} |
| 277 |
} |
| 278 |
tok.Range.End = l.pos |
| 279 |
l.toks = append(l.toks, tok) |
| 280 |
} |
| 281 |
|
| 282 |
// string processes the String token at the current position. |
| 283 |
// The lexer *must* know the next token is a String before calling. |
| 284 |
func (l *lexer) string() { |
| 285 |
tok := &Token{Type: String, Range: Range{Start: l.pos, End: l.pos}} |
| 286 |
if r := l.next(); r != '"' { |
| 287 |
log.Fatalf("lexer expected '\"', got '%v'", r) |
| 288 |
return |
| 289 |
} |
| 290 |
escape := false |
| 291 |
for l.e == nil { |
| 292 |
switch l.next() { |
| 293 |
case '"': |
| 294 |
if !escape { |
| 295 |
tok.Range.End = l.pos |
| 296 |
l.toks = append(l.toks, tok) |
| 297 |
return |
| 298 |
} |
| 299 |
case '\\': |
| 300 |
escape = !escape |
| 301 |
default: |
| 302 |
escape = false |
| 303 |
} |
| 304 |
} |
| 305 |
} |
| 306 |
|
| 307 |
// operator processes the Operator token at the current position. |
| 308 |
// The lexer *must* know the next token is a Operator before calling. |
| 309 |
func (l *lexer) operator() { |
| 310 |
tok := &Token{Type: Operator, Range: Range{Start: l.pos, End: l.pos}} |
| 311 |
for l.e == nil { |
| 312 |
switch l.next() { |
| 313 |
case '=', '|': |
| 314 |
tok.Range.End = l.pos |
| 315 |
l.toks = append(l.toks, tok) |
| 316 |
return |
| 317 |
} |
| 318 |
} |
| 319 |
} |
| 320 |
|
| 321 |
// lineComment processes the Comment token at the current position. |
| 322 |
// The lexer *must* know the next token is a Comment before calling. |
| 323 |
func (l *lexer) lineComment() { |
| 324 |
tok := &Token{Type: Comment, Range: Range{Start: l.pos, End: l.pos}} |
| 325 |
if r := l.next(); r != ';' { |
| 326 |
log.Fatalf("lexer expected ';', got '%v'", r) |
| 327 |
return |
| 328 |
} |
| 329 |
for l.e == nil { |
| 330 |
s := l.save() |
| 331 |
switch l.next() { |
| 332 |
case '\n': |
| 333 |
l.restore(s) |
| 334 |
tok.Range.End = l.pos |
| 335 |
l.toks = append(l.toks, tok) |
| 336 |
return |
| 337 |
} |
| 338 |
} |
| 339 |
} |
| 340 |
|
| 341 |
// newline processes the Newline token at the current position. |
| 342 |
// The lexer *must* know the next token is a Newline before calling. |
| 343 |
func (l *lexer) newline() { |
| 344 |
tok := &Token{Type: Newline, Range: Range{Start: l.pos, End: l.pos}} |
| 345 |
if r := l.next(); r != '\n' { |
| 346 |
log.Fatalf("lexer expected '\n', got '%v'", r) |
| 347 |
return |
| 348 |
} |
| 349 |
tok.Range.End = l.pos |
| 350 |
l.toks = append(l.toks, tok) |
| 351 |
} |
| 352 |
|
| 353 |
// lex returns all the tokens and diagnostics after lexing source. |
| 354 |
func lex(source string) ([]*Token, []Diagnostic, error) { |
| 355 |
l := lexer{source: source, lexerState: lexerState{pos: Position{1, 1}}} |
| 356 |
|
| 357 |
lastPos := Position{} |
| 358 |
for l.e == nil { |
| 359 |
// Integrity check that the parser is making progress |
| 360 |
if l.pos == lastPos { |
| 361 |
log.Panicf("Parsing stuck at %v", l.pos) |
| 362 |
} |
| 363 |
lastPos = l.pos |
| 364 |
|
| 365 |
s := l.save() |
| 366 |
r := l.next() |
| 367 |
switch { |
| 368 |
case r == '%': |
| 369 |
l.restore(s) |
| 370 |
l.pident() |
| 371 |
case r == '+' || r == '-' || r == '_' || isAlphaNumeric(r): |
| 372 |
l.restore(s) |
| 373 |
l.numberOrIdent() |
| 374 |
case r == '"': |
| 375 |
l.restore(s) |
| 376 |
l.string() |
| 377 |
case r == '=', r == '|': |
| 378 |
l.restore(s) |
| 379 |
l.operator() |
| 380 |
case r == ';': |
| 381 |
l.restore(s) |
| 382 |
l.lineComment() |
| 383 |
case r == '\n': |
| 384 |
l.restore(s) |
| 385 |
l.newline() |
| 386 |
} |
| 387 |
} |
| 388 |
if l.e != nil && l.e != io.EOF { |
| 389 |
return nil, nil, l.e |
| 390 |
} |
| 391 |
return l.toks, l.diags, nil |
| 392 |
} |
| 393 |
|
| 394 |
func isNumeric(r rune) bool { return unicode.IsDigit(r) } |
| 395 |
func isAlpha(r rune) bool { return unicode.IsLetter(r) } |
| 396 |
func isAlphaNumeric(r rune) bool { return isAlpha(r) || isNumeric(r) } |
| 397 |
|
| 398 |
type parser struct { |
| 399 |
lines []string // all source lines |
| 400 |
toks []*Token // all tokens |
| 401 |
diags []Diagnostic // parser emitted diagnostics |
| 402 |
idents map[string]*Identifier // identifiers by name |
| 403 |
mappings map[*Token]interface{} // tokens to semantic map |
| 404 |
extInstImports map[string]schema.OpcodeMap // extension imports by identifier |
| 405 |
insts []*Instruction // all instructions |
| 406 |
} |
| 407 |
|
| 408 |
func (p *parser) parse() error { |
| 409 |
for i := 0; i < len(p.toks); { |
| 410 |
if p.newline(i) || p.comment(i) { |
| 411 |
i++ |
| 412 |
continue |
| 413 |
} |
| 414 |
if n := p.instruction(i); n > 0 { |
| 415 |
i += n |
| 416 |
} else { |
| 417 |
p.unexpected(i) |
| 418 |
i++ |
| 419 |
} |
| 420 |
} |
| 421 |
return nil |
| 422 |
} |
| 423 |
|
| 424 |
// instruction parses the instruction starting at the i'th token. |
| 425 |
func (p *parser) instruction(i int) (n int) { |
| 426 |
inst := &Instruction{} |
| 427 |
|
| 428 |
switch { |
| 429 |
case p.opcode(i) != nil: |
| 430 |
inst.Opcode = p.opcode(i) |
| 431 |
inst.Tokens = []*Token{p.tok(i)} |
| 432 |
p.mappings[p.tok(i)] = inst |
| 433 |
n++ |
| 434 |
case p.opcode(i+2) != nil: // try '%id' '=' |
| 435 |
inst.Result, inst.Opcode = p.pident(i), p.opcode(i+2) |
| 436 |
if inst.Result == nil || p.operator(i+1) != "=" { |
| 437 |
return 0 |
| 438 |
} |
| 439 |
n += 3 |
| 440 |
inst.Tokens = []*Token{p.tok(i), p.tok(i + 1), p.tok(i + 2)} |
| 441 |
p.mappings[p.tok(i+2)] = inst |
| 442 |
default: |
| 443 |
return |
| 444 |
} |
| 445 |
|
| 446 |
expectsResult := len(inst.Opcode.Operands) > 0 && IsResult(inst.Opcode.Operands[0].Kind) |
| 447 |
operands := inst.Opcode.Operands |
| 448 |
switch { |
| 449 |
case inst.Result != nil && !expectsResult: |
| 450 |
p.err(inst.Result, "'%s' does not have a result", inst.Opcode.Opname) |
| 451 |
return |
| 452 |
case inst.Result == nil && expectsResult: |
| 453 |
p.err(p.tok(i), "'%s' expects a result", inst.Opcode.Opname) |
| 454 |
return |
| 455 |
case inst.Result != nil && expectsResult: |
| 456 |
// Check the result is of the correct type |
| 457 |
o := inst.Opcode.Operands[0] |
| 458 |
p.operand(o.Name, o.Kind, i, false) |
| 459 |
operands = operands[1:] |
| 460 |
p.addIdentDef(inst.Result.Text(p.lines), inst, p.tok(i)) |
| 461 |
} |
| 462 |
|
| 463 |
processOperand := func(o schema.Operand) bool { |
| 464 |
if p.newline(i + n) { |
| 465 |
return false |
| 466 |
} |
| 467 |
|
| 468 |
switch o.Quantifier { |
| 469 |
case schema.Once: |
| 470 |
if op, c := p.operand(o.Name, o.Kind, i+n, false); op != nil { |
| 471 |
inst.Tokens = append(inst.Tokens, op.Tokens...) |
| 472 |
n += c |
| 473 |
} |
| 474 |
case schema.ZeroOrOnce: |
| 475 |
if op, c := p.operand(o.Name, o.Kind, i+n, true); op != nil { |
| 476 |
inst.Tokens = append(inst.Tokens, op.Tokens...) |
| 477 |
n += c |
| 478 |
} |
| 479 |
case schema.ZeroOrMany: |
| 480 |
for !p.newline(i + n) { |
| 481 |
if op, c := p.operand(o.Name, o.Kind, i+n, true); op != nil { |
| 482 |
inst.Tokens = append(inst.Tokens, op.Tokens...) |
| 483 |
n += c |
| 484 |
} else { |
| 485 |
return false |
| 486 |
} |
| 487 |
} |
| 488 |
} |
| 489 |
return true |
| 490 |
} |
| 491 |
|
| 492 |
for _, o := range operands { |
| 493 |
if !processOperand(o) { |
| 494 |
break |
| 495 |
} |
| 496 |
|
| 497 |
if inst.Opcode == schema.OpExtInst && n == 4 { |
| 498 |
extImportTok, extNameTok := p.tok(i+n), p.tok(i+n+1) |
| 499 |
extImport := extImportTok.Text(p.lines) |
| 500 |
if extOpcodes, ok := p.extInstImports[extImport]; ok { |
| 501 |
extName := extNameTok.Text(p.lines) |
| 502 |
if extOpcode, ok := extOpcodes[extName]; ok { |
| 503 |
n += 2 // skip ext import, ext name |
| 504 |
for _, o := range extOpcode.Operands { |
| 505 |
if !processOperand(o) { |
| 506 |
break |
| 507 |
} |
| 508 |
} |
| 509 |
} else { |
| 510 |
p.err(extNameTok, "Unknown extension opcode '%s'", extName) |
| 511 |
} |
| 512 |
} else { |
| 513 |
p.err(extImportTok, "Expected identifier to OpExtInstImport") |
| 514 |
} |
| 515 |
} |
| 516 |
} |
| 517 |
|
| 518 |
for _, t := range inst.Tokens { |
| 519 |
inst.Range.grow(t.Range) |
| 520 |
} |
| 521 |
|
| 522 |
p.insts = append(p.insts, inst) |
| 523 |
|
| 524 |
if inst.Opcode == schema.OpExtInstImport && len(inst.Tokens) >= 4 { |
| 525 |
// Instruction is a OpExtInstImport. Keep track of this. |
| 526 |
extTok := inst.Tokens[3] |
| 527 |
extName := strings.Trim(extTok.Text(p.lines), `"`) |
| 528 |
extOpcodes, ok := schema.ExtOpcodes[extName] |
| 529 |
if !ok { |
| 530 |
p.err(extTok, "Unknown extension '%s'", extName) |
| 531 |
} |
| 532 |
extImport := inst.Result.Text(p.lines) |
| 533 |
p.extInstImports[extImport] = extOpcodes |
| 534 |
} |
| 535 |
|
| 536 |
return |
| 537 |
} |
| 538 |
|
| 539 |
// operand parses the operand with the name n, kind k, starting at the i'th |
| 540 |
// token. |
| 541 |
func (p *parser) operand(n string, k *schema.OperandKind, i int, optional bool) (*Operand, int) { |
| 542 |
tok := p.tok(i) |
| 543 |
if tok == nil { |
| 544 |
return nil, 0 |
| 545 |
} |
| 546 |
|
| 547 |
op := &Operand{ |
| 548 |
Name: n, |
| 549 |
Kind: k, |
| 550 |
Tokens: []*Token{tok}, |
| 551 |
} |
| 552 |
p.mappings[tok] = op |
| 553 |
|
| 554 |
switch k.Category { |
| 555 |
case schema.OperandCategoryBitEnum, schema.OperandCategoryValueEnum: |
| 556 |
s := tok.Text(p.lines) |
| 557 |
for _, e := range k.Enumerants { |
| 558 |
if e.Enumerant == s { |
| 559 |
count := 1 |
| 560 |
for _, param := range e.Parameters { |
| 561 |
p, c := p.operand(param.Name, param.Kind, i+count, false) |
| 562 |
if p != nil { |
| 563 |
op.Tokens = append(op.Tokens, p.Tokens...) |
| 564 |
op.Parameters = append(op.Parameters, p) |
| 565 |
} |
| 566 |
count += c |
| 567 |
} |
| 568 |
|
| 569 |
// Handle bitfield '|' chains |
| 570 |
if p.tok(i+count).Text(p.lines) == "|" { |
| 571 |
count++ // '|' |
| 572 |
p, c := p.operand(n, k, i+count, false) |
| 573 |
if p != nil { |
| 574 |
op.Tokens = append(op.Tokens, p.Tokens...) |
| 575 |
op.Parameters = append(op.Parameters, p) |
| 576 |
} |
| 577 |
count += c |
| 578 |
} |
| 579 |
|
| 580 |
return op, count |
| 581 |
} |
| 582 |
} |
| 583 |
if !optional { |
| 584 |
p.err(p.tok(i), "invalid operand value '%s'", s) |
| 585 |
} |
| 586 |
|
| 587 |
return nil, 0 |
| 588 |
|
| 589 |
case schema.OperandCategoryID: |
| 590 |
id := p.pident(i) |
| 591 |
if id != nil { |
| 592 |
p.addIdentRef(p.tok(i)) |
| 593 |
return op, 1 |
| 594 |
} |
| 595 |
if !optional { |
| 596 |
p.err(p.tok(i), "operand requires id, got '%s'", tok.Text(p.lines)) |
| 597 |
} |
| 598 |
return nil, 0 |
| 599 |
|
| 600 |
case schema.OperandCategoryLiteral: |
| 601 |
switch tok.Type { |
| 602 |
case String, Integer, Float, Ident: |
| 603 |
return op, 1 |
| 604 |
} |
| 605 |
if !optional { |
| 606 |
p.err(p.tok(i), "operand requires literal, got '%s'", tok.Text(p.lines)) |
| 607 |
} |
| 608 |
return nil, 0 |
| 609 |
|
| 610 |
case schema.OperandCategoryComposite: |
| 611 |
n := 1 |
| 612 |
for _, b := range k.Bases { |
| 613 |
o, c := p.operand(b.Kind, b, i+n, optional) |
| 614 |
if o != nil { |
| 615 |
op.Tokens = append(op.Tokens, o.Tokens...) |
| 616 |
} |
| 617 |
n += c |
| 618 |
} |
| 619 |
return op, n |
| 620 |
|
| 621 |
default: |
| 622 |
p.err(p.tok(i), "OperandKind '%s' has unexpected category '%s'", k.Kind, k.Category) |
| 623 |
return nil, 0 |
| 624 |
} |
| 625 |
} |
| 626 |
|
| 627 |
// tok returns the i'th token, or nil if i is out of bounds. |
| 628 |
func (p *parser) tok(i int) *Token { |
| 629 |
if i < 0 || i >= len(p.toks) { |
| 630 |
return nil |
| 631 |
} |
| 632 |
return p.toks[i] |
| 633 |
} |
| 634 |
|
| 635 |
// opcode returns the schema.Opcode for the i'th token, or nil if the i'th token |
| 636 |
// does not represent an opcode. |
| 637 |
func (p *parser) opcode(i int) *schema.Opcode { |
| 638 |
if tok := p.ident(i); tok != nil { |
| 639 |
name := tok.Text(p.lines) |
| 640 |
if inst, found := schema.Opcodes[name]; found { |
| 641 |
return inst |
| 642 |
} |
| 643 |
} |
| 644 |
return nil |
| 645 |
} |
| 646 |
|
| 647 |
// operator returns the operator for the i'th token, or and empty string if the |
| 648 |
// i'th token is not an operator. |
| 649 |
func (p *parser) operator(i int) string { |
| 650 |
if tok := p.tok(i); tok != nil && tok.Type == Operator { |
| 651 |
return tok.Text(p.lines) |
| 652 |
} |
| 653 |
return "" |
| 654 |
} |
| 655 |
|
| 656 |
// ident returns the i'th token if it is an Ident, otherwise nil. |
| 657 |
func (p *parser) ident(i int) *Token { |
| 658 |
if tok := p.tok(i); tok != nil && tok.Type == Ident { |
| 659 |
return tok |
| 660 |
} |
| 661 |
return nil |
| 662 |
} |
| 663 |
|
| 664 |
// pident returns the i'th token if it is an PIdent, otherwise nil. |
| 665 |
func (p *parser) pident(i int) *Token { |
| 666 |
if tok := p.tok(i); tok != nil && tok.Type == PIdent { |
| 667 |
return tok |
| 668 |
} |
| 669 |
return nil |
| 670 |
} |
| 671 |
|
| 672 |
// comment returns true if the i'th token is a Comment, otherwise false. |
| 673 |
func (p *parser) comment(i int) bool { |
| 674 |
if tok := p.tok(i); tok != nil && tok.Type == Comment { |
| 675 |
return true |
| 676 |
} |
| 677 |
return false |
| 678 |
} |
| 679 |
|
| 680 |
// newline returns true if the i'th token is a Newline, otherwise false. |
| 681 |
func (p *parser) newline(i int) bool { |
| 682 |
if tok := p.tok(i); tok != nil && tok.Type == Newline { |
| 683 |
return true |
| 684 |
} |
| 685 |
return false |
| 686 |
} |
| 687 |
|
| 688 |
// unexpected emits an 'unexpected token error' for the i'th token. |
| 689 |
func (p *parser) unexpected(i int) { |
| 690 |
p.err(p.toks[i], "syntax error: unexpected '%s'", p.toks[i].Text(p.lines)) |
| 691 |
} |
| 692 |
|
| 693 |
// addIdentDef records the token definition for the instruction inst with the |
| 694 |
// given id. |
| 695 |
func (p *parser) addIdentDef(id string, inst *Instruction, def *Token) { |
| 696 |
i, existing := p.idents[id] |
| 697 |
if !existing { |
| 698 |
i = &Identifier{} |
| 699 |
p.idents[id] = i |
| 700 |
} |
| 701 |
if i.Definition == nil { |
| 702 |
i.Definition = inst |
| 703 |
} else { |
| 704 |
p.err(def, "id '%v' redeclared", id) |
| 705 |
} |
| 706 |
} |
| 707 |
|
| 708 |
// addIdentRef adds a identifier reference for the token ref. |
| 709 |
func (p *parser) addIdentRef(ref *Token) { |
| 710 |
id := ref.Text(p.lines) |
| 711 |
i, existing := p.idents[id] |
| 712 |
if !existing { |
| 713 |
i = &Identifier{} |
| 714 |
p.idents[id] = i |
| 715 |
} |
| 716 |
i.References = append(i.References, ref) |
| 717 |
} |
| 718 |
|
| 719 |
// err appends an fmt.Printf style error into l.diags for the given token. |
| 720 |
func (p *parser) err(tok *Token, msg string, args ...interface{}) { |
| 721 |
rng := Range{} |
| 722 |
if tok != nil { |
| 723 |
rng = tok.Range |
| 724 |
} |
| 725 |
p.diags = append(p.diags, Diagnostic{ |
| 726 |
Range: rng, |
| 727 |
Severity: SeverityError, |
| 728 |
Message: fmt.Sprintf(msg, args...), |
| 729 |
}) |
| 730 |
} |
| 731 |
|
| 732 |
// Parse parses the SPIR-V assembly string source, returning the parse results. |
| 733 |
func Parse(source string) (Results, error) { |
| 734 |
toks, diags, err := lex(source) |
| 735 |
if err != nil { |
| 736 |
return Results{}, err |
| 737 |
} |
| 738 |
lines := strings.SplitAfter(source, "\n") |
| 739 |
p := parser{ |
| 740 |
lines: lines, |
| 741 |
toks: toks, |
| 742 |
idents: map[string]*Identifier{}, |
| 743 |
mappings: map[*Token]interface{}{}, |
| 744 |
extInstImports: map[string]schema.OpcodeMap{}, |
| 745 |
} |
| 746 |
if err := p.parse(); err != nil { |
| 747 |
return Results{}, err |
| 748 |
} |
| 749 |
diags = append(diags, p.diags...) |
| 750 |
return Results{ |
| 751 |
Lines: lines, |
| 752 |
Tokens: toks, |
| 753 |
Diagnostics: p.diags, |
| 754 |
Identifiers: p.idents, |
| 755 |
Mappings: p.mappings, |
| 756 |
}, nil |
| 757 |
} |
| 758 |
|
| 759 |
// IsResult returns true if k is used to store the result of an instruction. |
| 760 |
func IsResult(k *schema.OperandKind) bool { |
| 761 |
switch k { |
| 762 |
case schema.OperandKindIdResult, schema.OperandKindIdResultType: |
| 763 |
return true |
| 764 |
default: |
| 765 |
return false |
| 766 |
} |
| 767 |
} |
| 768 |
|
| 769 |
// Results holds the output of Parse(). |
| 770 |
type Results struct { |
| 771 |
Lines []string |
| 772 |
Tokens []*Token |
| 773 |
Diagnostics []Diagnostic |
| 774 |
Identifiers map[string]*Identifier // identifiers by name |
| 775 |
Mappings map[*Token]interface{} // tokens to semantic map |
| 776 |
} |
| 777 |
|
| 778 |
// Instruction describes a single instruction instance |
| 779 |
type Instruction struct { |
| 780 |
Tokens []*Token // all the tokens that make up the instruction |
| 781 |
Result *Token // the token that represents the result of the instruction, or nil |
| 782 |
Operands []*Operand // the operands of the instruction |
| 783 |
Range Range // the textual range of the instruction |
| 784 |
Opcode *schema.Opcode // the opcode for the instruction |
| 785 |
} |
| 786 |
|
| 787 |
// Operand describes a single operand instance |
| 788 |
type Operand struct { |
| 789 |
Name string // name of the operand |
| 790 |
Kind *schema.OperandKind // kind of the operand |
| 791 |
Tokens []*Token // all the tokens that make up the operand |
| 792 |
Parameters []*Operand // all the parameters for the operand |
| 793 |
} |
| 794 |
|
| 795 |
// Identifier describes a single, unique SPIR-V identifier (i.e. %32) |
| 796 |
type Identifier struct { |
| 797 |
Definition *Instruction // where the identifier was defined |
| 798 |
References []*Token // all the places the identifier was referenced |
| 799 |
} |
| 800 |
|
| 801 |
// Severity is an enumerator of diagnositc seeverities |
| 802 |
type Severity int |
| 803 |
|
| 804 |
// Severity levels |
| 805 |
const ( |
| 806 |
SeverityError Severity = iota |
| 807 |
SeverityWarning |
| 808 |
SeverityInformation |
| 809 |
SeverityHint |
| 810 |
) |
| 811 |
|
| 812 |
// Diagnostic holds a single diagnostic message that was generated while |
| 813 |
// parsing. |
| 814 |
type Diagnostic struct { |
| 815 |
Range Range |
| 816 |
Severity Severity |
| 817 |
Message string |
| 818 |
} |