Coverage for hdl_registers/parser/parser.py: 97%

1# --------------------------------------------------------------------------------------------------

4# This file is part of the hdl-registers project, an HDL register generator fast enough to run

5# in real time.

6# https://hdl-registers.com

7# https://github.com/hdl-registers/hdl-registers

8# --------------------------------------------------------------------------------------------------

10from __future__ import annotations

12import copy

13import json

14from typing import TYPE_CHECKING, Any, ClassVar

16import tomli_w

17import yaml

18from tsfpga import DEFAULT_FILE_ENCODING

20from hdl_registers.about import WEBSITE_URL

21from hdl_registers.constant.bit_vector_constant import UnsignedVector

22from hdl_registers.register_list import RegisterList

23from hdl_registers.register_modes import REGISTER_MODES

25if TYPE_CHECKING:

26 from pathlib import Path

28 from hdl_registers.register import Register

29 from hdl_registers.register_mode import RegisterMode

32class RegisterParser:

33 """

34 Parse register data in the form of a dictionary into a :class:`.RegisterList` object.

35 See :ref:`toml_format` for further documentation.

37 A note on sanity check strategy:

38 The parser performs only the basic sanity checks related to the data file format.

39 For example, missing properties, unknown properties, etc.

40 A lot of other sanity checks are performed in the :class:`.Register`, :class:`.RegisterArray`,

41 :class:`.RegisterField`, etc, classes themselves.

43 For example, the default value of a bit field should be a string with the value "0" or "1".

44 This is checked in the constructor of the :class:`.Bit` class, not here in the parser.

45 Similar for a lot of other things.

47 This is because these objects can be created from the Python API also, without involving

48 the parser.

49 Hence these sanity checks have to be present there.

50 Having them also in the parser would enable better error messages, but would be redundant

51 and would slow down the parser.

52 Since the parser is run in real time, the performance is critical, and we can not afford

53 to slow it down.

54 """

56 # Attributes of the constant.

57 recognized_constant_items: ClassVar = {"type", "value", "description", "data_type"}

58 # Note that "type" being present is implied. We would not be parsing a constant unless we

59 # know it to be a "constant" type.

60 # So we save some CPU cycles by not checking for it.

61 required_constant_items: ClassVar = ["value"]

63 # Attributes of the register.

64 # Anything apart from these are names of fields.

65 default_register_items: ClassVar = {

66 "type",

67 "mode",

68 "description",

69 }

70 # While a 'mode' is required for a register, it may NOT be specified/changed in the data file

71 # for a default register.

72 # Hence this property is handled separately.

73 # And hence, registers have no required items.

75 # Attributes of the register array.

76 # Anything apart from these are names of registers.

77 default_register_array_items: ClassVar = {"type", "array_length", "description"}

78 # Note that "type" being present is implied.

79 # We would not be parsing a register array unless we know it to be a "register_array" type.

80 # So we save some CPU cycles by not checking for it.

81 required_register_array_items: ClassVar = ["array_length"]

83 # Attributes of the "bit" register field.

84 recognized_bit_items: ClassVar = {"type", "description", "default_value"}

85 # Note that "type" being present is implied.

86 # We would not be parsing a bit unless we know it to be a "bit" type.

87 # So we save some CPU cycles by not checking for it.

88 required_bit_items: ClassVar[list[str]] = []

90 # Attributes of the "bit_vector" register field.

91 recognized_bit_vector_items: ClassVar = {"type", "description", "width", "default_value"}

92 # Note that "type" being present is implied.

93 # We would not be parsing a bit_vector unless we know it to be a "bit_vector" type.

94 # So we save some CPU cycles by not checking for it.

95 required_bit_vector_items: ClassVar = ["width"]

97 # Attributes of the "enumeration" register field.

98 recognized_enumeration_items: ClassVar = {"type", "description", "default_value", "element"}

99 # Note that "type" being present is implied.

100 # We would not be parsing a enumeration unless we know it to be a "enumeration" type.

101 # So we save some CPU cycles by not checking for it.

102 required_enumeration_items: ClassVar = ["element"]

103

104 # Attributes of the "integer" register field.

105 recognized_integer_items: ClassVar = {

106 "type",

107 "description",

108 "min_value",

109 "max_value",

110 "default_value",

111 }

112 # Note that "type" being present is implied.

113 # We would not be parsing a integer unless we know it to be a "integer" type.

114 # So we save some CPU cycles by not checking for it.

115 required_integer_items: ClassVar = ["max_value"]

116

117 def __init__(

118 self,

119 name: str,

120 source_definition_file: Path,

121 default_registers: list[Register] | None = None,

122 ) -> None:

123 """

124 Arguments:

125 name: The name of the register list.

126 source_definition_file: The source file that defined this register list.

127 Will be displayed in generated source code and documentation

128 for traceability.

129 default_registers: List of default registers.

130 Note that this list with :class:`.Register` objects will be deep copied, so you can

131 use the same list many times without worrying about mutability.

132 """

133 self._register_list = RegisterList(name=name, source_definition_file=source_definition_file)

134 self._source_definition_file = source_definition_file

135

136 self._default_register_names = []

137 if default_registers:

138 # Perform deep copy of the mutable register objects.

139 self._register_list.register_objects = copy.deepcopy(default_registers)

140 for register in default_registers:

141 self._default_register_names.append(register.name)

142

143 def parse(self, register_data: dict[str, Any]) -> RegisterList:

144 """

145 Parse the register data.

146

147 Arguments:

148 register_data: Register data as a dictionary.

149 Preferably read by the :func:`.from_toml`, :func:`.from_json` or

150 :func:`.from_yaml` functions.

151

152 Return:

153 The resulting register list.

154 """

155 for old_top_level_key_name in ["constant", "register", "register_array"]:

156 if old_top_level_key_name in register_data:

157 source_file = self._source_definition_file

158 output_file = (

159 source_file.parent.resolve()

160 / f"{source_file.stem}_version_6_format{source_file.suffix}"

161 )

162

163 print(

164 f"""

165ERROR: Parsing register data that appears to be in the old pre-6.0.0 format.

166ERROR: For more information, see: {WEBSITE_URL}/rst/about/new_data_file_format.html

167ERROR: Your data will be automatically converted to the new format and saved to: {output_file}

168ERROR: Please inspect that file and update your data file to the new format.

169"""

170 )

171 _save_to_new_format(old_data=register_data, output_file=output_file)

172 raise ValueError("Found register data in old format. See message above.")

173

174 parser_methods = {

175 "constant": self._parse_constant,

176 "register": self._parse_plain_register,

177 "register_array": self._parse_register_array,

178 }

179

180 for top_level_name, top_level_items in register_data.items():

181 if not isinstance(top_level_items, dict):

182 message = (

183 f"Error while parsing {self._source_definition_file}: "

184 f'Got unknown top-level property "{top_level_name}".'

185 )

186 # Seems to the linter like a type error, but it is actually the user specifying

187 # a property/value that they shouldn't.

188 # Corresponds better to a 'ValueError' than a 'TypeError'.

189 raise ValueError(message) # noqa: TRY004

190

191 top_level_type = top_level_items.get("type", "register")

192

193 if top_level_type not in parser_methods:

194 valid_types_str = ", ".join(f'"{parser_key}"' for parser_key in parser_methods)

195 message = (

196 f'Error while parsing "{top_level_name}" in {self._source_definition_file}: '

197 f'Got unknown type "{top_level_type}". Expected one of {valid_types_str}.'

198 )

199 raise ValueError(message)

200

201 parser_methods[top_level_type](name=top_level_name, items=top_level_items)

202

203 return self._register_list

204

205 def _parse_constant(self, name: str, items: dict[str, Any]) -> None:

206 for item_name in self.required_constant_items:

207 if item_name not in items:

208 message = (

209 f'Error while parsing constant "{name}" in {self._source_definition_file}: '

210 f'Missing required property "{item_name}".'

211 )

212 raise ValueError(message)

213

214 for item_name in items:

215 if item_name not in self.recognized_constant_items:

216 message = (

217 f'Error while parsing constant "{name}" in {self._source_definition_file}: '

218 f'Got unknown property "{item_name}".'

219 )

220 raise ValueError(message)

221

222 value = items["value"]

223 description = items.get("description", "")

224 data_type_str = items.get("data_type")

225

226 if data_type_str is not None:

227 if not isinstance(value, str):

228 raise ValueError(

229 f'Error while parsing constant "{name}" in '

230 f"{self._source_definition_file}: "

231 'May not set "data_type" for non-string constant.'

232 )

233

234 if data_type_str == "unsigned":

235 value = UnsignedVector(value)

236 else:

237 raise ValueError(

238 f'Error while parsing constant "{name}" in '

239 f"{self._source_definition_file}: "

240 f'Invalid data type "{data_type_str}".'

241 )

242

243 self._register_list.add_constant(name=name, value=value, description=description)

244

245 def _parse_plain_register(self, name: str, items: dict[str, Any]) -> None:

246 description = items.get("description", "")

247

248 if name in self._default_register_names:

249 # Default registers can be "updated" in the sense that the user can set a custom

250 # 'description' and add whatever fields they want in the current register list.

251 # They may not, however, change the 'mode' which is part of the default definition.

252 if "mode" in items:

253 message = (

254 f'Error while parsing register "{name}" in {self._source_definition_file}: '

255 'A "mode" may not be specified for a default register.'

256 )

257 raise ValueError(message)

258

259 register = self._register_list.get_register(register_name=name)

260 register.description = description

261

262 else:

263 # If it is a new register however, the 'mode' has to be specified.

264 if "mode" not in items:

265 message = (

266 f'Error while parsing register "{name}" in {self._source_definition_file}: '

267 f'Missing required property "mode".'

268 )

269 raise ValueError(message)

270

271 mode = self._get_mode(mode_name=items["mode"], register_name=name)

272

273 register = self._register_list.append_register(

274 name=name, mode=mode, description=description

275 )

276

277 self._parse_register_fields(register=register, register_items=items, register_array_note="")

278

279 def _get_mode(self, mode_name: str, register_name: str) -> RegisterMode:

280 if mode_name in REGISTER_MODES:

281 return REGISTER_MODES[mode_name]

282

283 valid_modes_str = ", ".join(f'"{mode_key}"' for mode_key in REGISTER_MODES)

284 message = (

285 f'Error while parsing register "{register_name}" in {self._source_definition_file}: '

286 f'Got unknown mode "{mode_name}". Expected one of {valid_modes_str}.'

287 )

288 raise ValueError(message)

289

290 def _parse_register_fields(

291 self,

292 register_items: dict[str, Any],

293 register: Register,

294 register_array_note: str,

295 ) -> None:

296 # Add any fields that are specified.

297 for item_name, item_value in register_items.items():

298 # Skip default items so we only get the fields.

299 if item_name in self.default_register_items:

300 continue

301

302 if not isinstance(item_value, dict):

303 message = (

304 f'Error while parsing register "{register.name}"{register_array_note} '

305 f"in {self._source_definition_file}: "

306 f'Got unknown property "{item_name}".'

307 )

308 # Seems to the linter like a type error, but it is actually the user specifying

309 # a property/value that they shouldn't.

310 # Corresponds better to a 'ValueError' than a 'TypeError'.

311 raise ValueError(message) # noqa: TRY004

312

313 if "type" not in item_value:

314 message = (

315 f'Error while parsing field "{item_name}" in register '

316 f'"{register.name}"{register_array_note} in {self._source_definition_file}: '

317 'Missing required property "type".'

318 )

319 raise ValueError(message)

320

321 field_type = item_value["type"]

322

323 parser_methods = {

324 "bit": self._parse_bit,

325 "bit_vector": self._parse_bit_vector,

326 "enumeration": self._parse_enumeration,

327 "integer": self._parse_integer,

328 }

329

330 if field_type not in parser_methods:

331 valid_types_str = ", ".join(f'"{parser_key}"' for parser_key in parser_methods)

332 message = (

333 f'Error while parsing field "{item_name}" in register '

334 f'"{register.name}"{register_array_note} in {self._source_definition_file}: '

335 f'Unknown field type "{field_type}". Expected one of {valid_types_str}.'

336 )

337 raise ValueError(message)

338

339 parser_methods[field_type](

340 register=register, field_name=item_name, field_items=item_value

341 )

342

343 def _parse_register_array(self, name: str, items: dict[str, Any]) -> None:

344 for required_property in self.required_register_array_items:

345 if required_property not in items:

346 message = (

347 f'Error while parsing register array "{name}" in '

348 f"{self._source_definition_file}: "

349 f'Missing required property "{required_property}".'

350 )

351 raise ValueError(message)

352

353 register_array_length = items["array_length"]

354 register_array_description = items.get("description", "")

355 register_array = self._register_list.append_register_array(

356 name=name, length=register_array_length, description=register_array_description

357 )

358

359 # Add all registers that are specified.

360 found_at_least_one_register = False

361 for item_name, item_value in items.items():

362 # Skip default items so we only get the registers.

363 if item_name in self.default_register_array_items:

364 continue

365

366 found_at_least_one_register = True

367

368 if not isinstance(item_value, dict):

369 message = (

370 f'Error while parsing register array "{name}" in '

371 f"{self._source_definition_file}: "

372 f'Got unknown property "{item_name}".'

373 )

374 # Seems to the linter like a type error, but it is actually the user specifying

375 # a property/value that they shouldn't.

376 # Corresponds better to a 'ValueError' than a 'TypeError'.

377 raise ValueError(message) # noqa: TRY004

378

379 item_type = item_value.get("type", "register")

380 if item_type != "register":

381 message = (

382 f'Error while parsing register "{item_name}" within array "{name}" in '

383 f"{self._source_definition_file}: "

384 f'Got unknown type "{item_type}". Expected "register".'

385 )

386 raise ValueError(message)

387

388 # A 'mode' is semi-required for plain registers, but always required for

389 # array registers.

390 if "mode" not in item_value:

391 raise ValueError(

392 f'Error while parsing register "{item_name}" within array "{name}" in '

393 f"{self._source_definition_file}: "

394 f'Missing required property "mode".'

395 )

396 register_mode = self._get_mode(mode_name=item_value["mode"], register_name=item_name)

397

398 register_description = item_value.get("description", "")

399

400 register = register_array.append_register(

401 name=item_name, mode=register_mode, description=register_description

402 )

403

404 self._parse_register_fields(

405 register_items=item_value,

406 register=register,

407 register_array_note=f' within array "{name}"',

408 )

409

410 if not found_at_least_one_register:

411 message = (

412 f'Error while parsing register array "{name}" in {self._source_definition_file}: '

413 "Array must contain at least one register."

414 )

415 raise ValueError(message)

416

417 def _check_field_items(

418 self,

419 register_name: str,

420 field_name: str,

421 field_items: dict[str, Any],

422 recognized_items: set[str],

423 required_items: list[str],

424 ) -> None:

425 """

426 Will raise exception if anything is wrong.

427 """

428 for item_name in required_items:

429 if item_name not in field_items:

430 message = (

431 f'Error while parsing field "{field_name}" in register "{register_name}" in '

432 f"{self._source_definition_file}: "

433 f'Missing required property "{item_name}".'

434 )

435 raise ValueError(message)

436

437 for item_name in field_items:

438 if item_name not in recognized_items:

439 message = (

440 f'Error while parsing field "{field_name}" in register '

441 f'"{register_name}" in {self._source_definition_file}: '

442 f'Unknown property "{item_name}".'

443 )

444 raise ValueError(message)

445

446 def _parse_bit(self, register: Register, field_name: str, field_items: dict[str, Any]) -> None:

447 self._check_field_items(

448 register_name=register.name,

449 field_name=field_name,

450 field_items=field_items,

451 recognized_items=self.recognized_bit_items,

452 required_items=self.required_bit_items,

453 )

454

455 description = field_items.get("description", "")

456 default_value = field_items.get("default_value", "0")

457

458 register.append_bit(name=field_name, description=description, default_value=default_value)

459

460 def _parse_bit_vector(

461 self, register: Register, field_name: str, field_items: dict[str, Any]

462 ) -> None:

463 self._check_field_items(

464 register_name=register.name,

465 field_name=field_name,

466 field_items=field_items,

467 recognized_items=self.recognized_bit_vector_items,

468 required_items=self.required_bit_vector_items,

469 )

470

471 width = field_items["width"]

472

473 description = field_items.get("description", "")

474 default_value = field_items.get("default_value", 0)

475

476 register.append_bit_vector(

477 name=field_name, description=description, width=width, default_value=default_value

478 )

479

480 def _parse_enumeration(

481 self, register: Register, field_name: str, field_items: dict[str, Any]

482 ) -> None:

483 self._check_field_items(

484 register_name=register.name,

485 field_name=field_name,

486 field_items=field_items,

487 recognized_items=self.recognized_enumeration_items,

488 # Check that we have at least one element.

489 # This is checked also in the Enumeration class, which is needed if the user

490 # is working directly with the Python API.

491 # That is where we usually sanity check, to avoid duplication.

492 # However, this particular check is needed here also since the logic for default

493 # value below does not work if there are no elements.

494 required_items=self.required_enumeration_items,

495 )

496

497 description = field_items.get("description", "")

498 # We assert above that the enumeration has at least one element.

499 # Meaning that the result of this get can not be None.

500 elements: dict[str, str] = field_items.get("element")

501

502 # The default "default value" is the first declared enumeration element.

503 # Note that this works because dictionaries in Python are guaranteed ordered since

504 # Python 3.7.

505 default_value = field_items.get("default_value", next(iter(elements)))

506

507 register.append_enumeration(

508 name=field_name,

509 description=description,

510 elements=elements,

511 default_value=default_value,

512 )

513

514 def _parse_integer(

515 self, register: Register, field_name: str, field_items: dict[str, Any]

516 ) -> None:

517 self._check_field_items(

518 register_name=register.name,

519 field_name=field_name,

520 field_items=field_items,

521 recognized_items=self.recognized_integer_items,

522 required_items=self.required_integer_items,

523 )

524

525 max_value = field_items["max_value"]

526

527 description = field_items.get("description", "")

528 min_value = field_items.get("min_value", 0)

529 default_value = field_items.get("default_value", min_value)

530

531 register.append_integer(

532 name=field_name,

533 description=description,

534 min_value=min_value,

535 max_value=max_value,

536 default_value=default_value,

537 )

538

539

540def _convert_to_new_format( # noqa: C901

541 old_data: dict[str, Any],

542) -> dict[str, Any]:

543 """

544 Convert pre-6.0.0 format to the new format.

545 This is a semi-trash function that will be removed in the future.

546 """

547

548 def _get_register_dict(register_items: dict[str, Any]) -> dict[str, Any]:

549 register_dict = {}

550

551 for register_item_name, register_item_value in register_items.items():

552 if register_item_name in RegisterParser.default_register_items:

553 register_dict[register_item_name] = register_item_value

554

555 elif register_item_name in ["bit", "bit_vector", "enumeration", "integer"]:

556 for field_name, field_items in register_item_value.items():

557 field_dict = {"type": register_item_name}

558 field_dict.update(dict(field_items.items()))

559

560 register_dict[field_name] = field_dict

561

562 else:

563 raise ValueError(

564 f"Unknown item {register_item_name}. Looks like an error in the user data file."

565 )

566

567 return register_dict

568

569 result = {}

570

571 def _add_item(name: str, items: dict[str, Any]) -> None:

572 if name in result:

573 raise ValueError(f"Duplicate item {name}")

574

575 result[name] = items

576

577 if "register" in old_data:

578 for register_name, register_items in old_data["register"].items():

579 register_dict = _get_register_dict(register_items=register_items)

580 _add_item(name=register_name, items=register_dict)

581

582 if "register_array" in old_data:

583 for register_array_name, register_array_items in old_data["register_array"].items():

584 register_array_dict: dict[str, Any] = {"type": "register_array"}

585

586 for register_array_item_name, register_array_item_value in register_array_items.items():

587 if register_array_item_name in RegisterParser.default_register_array_items:

588 register_array_dict[register_array_item_name] = register_array_item_value

589

590 elif register_array_item_name == "register":

591 for register_name, register_items in register_array_item_value.items():

592 register_array_dict[register_name] = _get_register_dict(

593 register_items=register_items

594 )

595

596 else:

597 raise ValueError(

598 f"Unknown item {register_array_item_name}. "

599 "Looks like an error in the user data file."

600 )

601

602 _add_item(name=register_array_name, items=register_array_dict)

603

604 if "constant" in old_data:

605 for constant_name, constant_items in old_data["constant"].items():

606 constant_dict = {"type": "constant"}

607 constant_dict.update(dict(constant_items.items()))

608

609 _add_item(name=constant_name, items=constant_dict)

610

611 return result

612

613

614def _save_to_new_format(old_data: dict[str, Any], output_file: Path) -> None:

615 """

616 Save the old data to the new format.

617 """

618 new_data = _convert_to_new_format(old_data=old_data)

619

620 if output_file.suffix == ".toml":

621 with output_file.open("wb") as file_handle:

622 tomli_w.dump(new_data, file_handle, multiline_strings=True)

623

624 return

625

626 if output_file.suffix == ".json":

627 with output_file.open("w", encoding=DEFAULT_FILE_ENCODING) as file_handle:

628 json.dump(new_data, file_handle, indent=4)

629

630 return

631

632 if output_file.suffix == ".yaml":

633 with output_file.open("w", encoding=DEFAULT_FILE_ENCODING) as file_handle:

634 yaml.dump(new_data, file_handle)

635

636 return

637

638 raise ValueError(f"Unknown file format {output_file}")