Commit | Line | Data |
---|---|---|
71aaa3f7 PP |
1 | # The MIT License (MIT) |
2 | # | |
3 | # Copyright (c) 2023 Philippe Proulx <eeppeliteloop@gmail.com> | |
4 | # | |
5 | # Permission is hereby granted, free of charge, to any person obtaining | |
6 | # a copy of this software and associated documentation files (the | |
7 | # "Software"), to deal in the Software without restriction, including | |
8 | # without limitation the rights to use, copy, modify, merge, publish, | |
9 | # distribute, sublicense, and/or sell copies of the Software, and to | |
10 | # permit persons to whom the Software is furnished to do so, subject to | |
11 | # the following conditions: | |
12 | # | |
13 | # The above copyright notice and this permission notice shall be | |
14 | # included in all copies or substantial portions of the Software. | |
15 | # | |
16 | # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, | |
17 | # EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF | |
18 | # MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. | |
19 | # IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY | |
20 | # CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, | |
21 | # TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE | |
22 | # SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. | |
23 | ||
5a993698 PP |
24 | # This module is the portable Normand processor. It offers both the |
25 | # parse() function and the command-line tool (run the module itself) | |
26 | # without external dependencies except a `typing` module for Python 3.4. | |
27 | # | |
28 | # Feel free to copy this module file to your own project to use Normand. | |
29 | # | |
30 | # Upstream repository: <https://github.com/efficios/normand>. | |
31 | ||
71aaa3f7 | 32 | __author__ = "Philippe Proulx" |
f5dcb24c | 33 | __version__ = "0.15.0" |
71aaa3f7 | 34 | __all__ = [ |
320644e2 PP |
35 | "__author__", |
36 | "__version__", | |
71aaa3f7 | 37 | "ByteOrder", |
320644e2 | 38 | "LabelsT", |
71aaa3f7 PP |
39 | "parse", |
40 | "ParseError", | |
f5dcb24c | 41 | "ParseErrorMessage", |
71aaa3f7 | 42 | "ParseResult", |
e57a18e1 | 43 | "TextLocation", |
e57a18e1 | 44 | "VariablesT", |
71aaa3f7 PP |
45 | ] |
46 | ||
47 | import re | |
48 | import abc | |
49 | import ast | |
50 | import sys | |
320644e2 | 51 | import copy |
71aaa3f7 | 52 | import enum |
05f81895 | 53 | import math |
71aaa3f7 | 54 | import struct |
e57a18e1 PP |
55 | import typing |
56 | from typing import Any, Set, Dict, List, Union, Pattern, Callable, NoReturn, Optional | |
71aaa3f7 PP |
57 | |
58 | ||
59 | # Text location (line and column numbers). | |
e57a18e1 | 60 | class TextLocation: |
71aaa3f7 PP |
61 | @classmethod |
62 | def _create(cls, line_no: int, col_no: int): | |
63 | self = cls.__new__(cls) | |
64 | self._init(line_no, col_no) | |
65 | return self | |
66 | ||
67 | def __init__(*args, **kwargs): # type: ignore | |
68 | raise NotImplementedError | |
69 | ||
70 | def _init(self, line_no: int, col_no: int): | |
71 | self._line_no = line_no | |
72 | self._col_no = col_no | |
73 | ||
74 | # Line number. | |
75 | @property | |
76 | def line_no(self): | |
77 | return self._line_no | |
78 | ||
79 | # Column number. | |
80 | @property | |
81 | def col_no(self): | |
82 | return self._col_no | |
83 | ||
2adf4336 | 84 | def __repr__(self): |
e57a18e1 | 85 | return "TextLocation({}, {})".format(self._line_no, self._col_no) |
2adf4336 | 86 | |
71aaa3f7 PP |
87 | |
88 | # Any item. | |
89 | class _Item: | |
e57a18e1 | 90 | def __init__(self, text_loc: TextLocation): |
71aaa3f7 PP |
91 | self._text_loc = text_loc |
92 | ||
93 | # Source text location. | |
94 | @property | |
95 | def text_loc(self): | |
96 | return self._text_loc | |
97 | ||
2adf4336 PP |
98 | |
99 | # Scalar item. | |
100 | class _ScalarItem(_Item): | |
71aaa3f7 PP |
101 | # Returns the size, in bytes, of this item. |
102 | @property | |
103 | @abc.abstractmethod | |
104 | def size(self) -> int: | |
105 | ... | |
106 | ||
107 | ||
108 | # A repeatable item. | |
2adf4336 | 109 | class _RepableItem: |
71aaa3f7 PP |
110 | pass |
111 | ||
112 | ||
113 | # Single byte. | |
2adf4336 | 114 | class _Byte(_ScalarItem, _RepableItem): |
e57a18e1 | 115 | def __init__(self, val: int, text_loc: TextLocation): |
71aaa3f7 PP |
116 | super().__init__(text_loc) |
117 | self._val = val | |
118 | ||
119 | # Byte value. | |
120 | @property | |
121 | def val(self): | |
122 | return self._val | |
123 | ||
124 | @property | |
125 | def size(self): | |
126 | return 1 | |
127 | ||
128 | def __repr__(self): | |
676f6189 | 129 | return "_Byte({}, {})".format(hex(self._val), repr(self._text_loc)) |
71aaa3f7 PP |
130 | |
131 | ||
132 | # String. | |
2adf4336 | 133 | class _Str(_ScalarItem, _RepableItem): |
e57a18e1 | 134 | def __init__(self, data: bytes, text_loc: TextLocation): |
71aaa3f7 PP |
135 | super().__init__(text_loc) |
136 | self._data = data | |
137 | ||
138 | # Encoded bytes. | |
139 | @property | |
140 | def data(self): | |
141 | return self._data | |
142 | ||
143 | @property | |
144 | def size(self): | |
145 | return len(self._data) | |
146 | ||
147 | def __repr__(self): | |
676f6189 | 148 | return "_Str({}, {})".format(repr(self._data), repr(self._text_loc)) |
71aaa3f7 PP |
149 | |
150 | ||
151 | # Byte order. | |
152 | @enum.unique | |
153 | class ByteOrder(enum.Enum): | |
154 | # Big endian. | |
155 | BE = "be" | |
156 | ||
157 | # Little endian. | |
158 | LE = "le" | |
159 | ||
160 | ||
2adf4336 PP |
161 | # Byte order setting. |
162 | class _SetBo(_Item): | |
e57a18e1 | 163 | def __init__(self, bo: ByteOrder, text_loc: TextLocation): |
0e8e3169 | 164 | super().__init__(text_loc) |
71aaa3f7 PP |
165 | self._bo = bo |
166 | ||
167 | @property | |
168 | def bo(self): | |
169 | return self._bo | |
170 | ||
2adf4336 | 171 | def __repr__(self): |
676f6189 | 172 | return "_SetBo({}, {})".format(repr(self._bo), repr(self._text_loc)) |
71aaa3f7 PP |
173 | |
174 | ||
175 | # Label. | |
176 | class _Label(_Item): | |
e57a18e1 | 177 | def __init__(self, name: str, text_loc: TextLocation): |
71aaa3f7 PP |
178 | super().__init__(text_loc) |
179 | self._name = name | |
180 | ||
181 | # Label name. | |
182 | @property | |
183 | def name(self): | |
184 | return self._name | |
185 | ||
71aaa3f7 | 186 | def __repr__(self): |
676f6189 | 187 | return "_Label({}, {})".format(repr(self._name), repr(self._text_loc)) |
71aaa3f7 PP |
188 | |
189 | ||
2adf4336 PP |
190 | # Offset setting. |
191 | class _SetOffset(_Item): | |
e57a18e1 | 192 | def __init__(self, val: int, text_loc: TextLocation): |
71aaa3f7 PP |
193 | super().__init__(text_loc) |
194 | self._val = val | |
195 | ||
676f6189 | 196 | # Offset value (bytes). |
71aaa3f7 PP |
197 | @property |
198 | def val(self): | |
199 | return self._val | |
200 | ||
71aaa3f7 | 201 | def __repr__(self): |
676f6189 PP |
202 | return "_SetOffset({}, {})".format(repr(self._val), repr(self._text_loc)) |
203 | ||
204 | ||
205 | # Offset alignment. | |
206 | class _AlignOffset(_Item): | |
e57a18e1 | 207 | def __init__(self, val: int, pad_val: int, text_loc: TextLocation): |
676f6189 PP |
208 | super().__init__(text_loc) |
209 | self._val = val | |
210 | self._pad_val = pad_val | |
211 | ||
212 | # Alignment value (bits). | |
213 | @property | |
214 | def val(self): | |
215 | return self._val | |
216 | ||
217 | # Padding byte value. | |
218 | @property | |
219 | def pad_val(self): | |
220 | return self._pad_val | |
221 | ||
222 | def __repr__(self): | |
223 | return "_AlignOffset({}, {}, {})".format( | |
224 | repr(self._val), repr(self._pad_val), repr(self._text_loc) | |
225 | ) | |
71aaa3f7 PP |
226 | |
227 | ||
228 | # Mixin of containing an AST expression and its string. | |
229 | class _ExprMixin: | |
230 | def __init__(self, expr_str: str, expr: ast.Expression): | |
231 | self._expr_str = expr_str | |
232 | self._expr = expr | |
233 | ||
234 | # Expression string. | |
235 | @property | |
236 | def expr_str(self): | |
237 | return self._expr_str | |
238 | ||
239 | # Expression node to evaluate. | |
240 | @property | |
241 | def expr(self): | |
242 | return self._expr | |
243 | ||
244 | ||
25ca454b PP |
245 | # Fill until some offset. |
246 | class _FillUntil(_Item, _ExprMixin): | |
247 | def __init__( | |
248 | self, expr_str: str, expr: ast.Expression, pad_val: int, text_loc: TextLocation | |
249 | ): | |
250 | super().__init__(text_loc) | |
251 | _ExprMixin.__init__(self, expr_str, expr) | |
252 | self._pad_val = pad_val | |
253 | ||
254 | # Padding byte value. | |
255 | @property | |
256 | def pad_val(self): | |
257 | return self._pad_val | |
258 | ||
259 | def __repr__(self): | |
260 | return "_FillUntil({}, {}, {}, {})".format( | |
261 | repr(self._expr_str), | |
262 | repr(self._expr), | |
263 | repr(self._pad_val), | |
264 | repr(self._text_loc), | |
265 | ) | |
266 | ||
267 | ||
2adf4336 PP |
268 | # Variable assignment. |
269 | class _VarAssign(_Item, _ExprMixin): | |
71aaa3f7 | 270 | def __init__( |
e57a18e1 | 271 | self, name: str, expr_str: str, expr: ast.Expression, text_loc: TextLocation |
71aaa3f7 PP |
272 | ): |
273 | super().__init__(text_loc) | |
274 | _ExprMixin.__init__(self, expr_str, expr) | |
275 | self._name = name | |
276 | ||
277 | # Name. | |
278 | @property | |
279 | def name(self): | |
280 | return self._name | |
281 | ||
71aaa3f7 | 282 | def __repr__(self): |
2adf4336 | 283 | return "_VarAssign({}, {}, {}, {})".format( |
676f6189 PP |
284 | repr(self._name), |
285 | repr(self._expr_str), | |
286 | repr(self._expr), | |
287 | repr(self._text_loc), | |
71aaa3f7 PP |
288 | ) |
289 | ||
290 | ||
269f6eb3 PP |
291 | # Fixed-length number, possibly needing more than one byte. |
292 | class _FlNum(_ScalarItem, _RepableItem, _ExprMixin): | |
71aaa3f7 | 293 | def __init__( |
e57a18e1 | 294 | self, expr_str: str, expr: ast.Expression, len: int, text_loc: TextLocation |
71aaa3f7 PP |
295 | ): |
296 | super().__init__(text_loc) | |
297 | _ExprMixin.__init__(self, expr_str, expr) | |
298 | self._len = len | |
299 | ||
300 | # Length (bits). | |
301 | @property | |
302 | def len(self): | |
303 | return self._len | |
304 | ||
305 | @property | |
306 | def size(self): | |
307 | return self._len // 8 | |
308 | ||
309 | def __repr__(self): | |
269f6eb3 | 310 | return "_FlNum({}, {}, {}, {})".format( |
676f6189 PP |
311 | repr(self._expr_str), |
312 | repr(self._expr), | |
313 | repr(self._len), | |
314 | repr(self._text_loc), | |
71aaa3f7 PP |
315 | ) |
316 | ||
317 | ||
05f81895 PP |
318 | # LEB128 integer. |
319 | class _Leb128Int(_Item, _RepableItem, _ExprMixin): | |
e57a18e1 | 320 | def __init__(self, expr_str: str, expr: ast.Expression, text_loc: TextLocation): |
05f81895 PP |
321 | super().__init__(text_loc) |
322 | _ExprMixin.__init__(self, expr_str, expr) | |
323 | ||
324 | def __repr__(self): | |
325 | return "{}({}, {}, {})".format( | |
326 | self.__class__.__name__, | |
327 | repr(self._expr_str), | |
328 | repr(self._expr), | |
676f6189 | 329 | repr(self._text_loc), |
05f81895 PP |
330 | ) |
331 | ||
332 | ||
333 | # Unsigned LEB128 integer. | |
334 | class _ULeb128Int(_Leb128Int, _RepableItem, _ExprMixin): | |
335 | pass | |
336 | ||
337 | ||
338 | # Signed LEB128 integer. | |
339 | class _SLeb128Int(_Leb128Int, _RepableItem, _ExprMixin): | |
340 | pass | |
341 | ||
342 | ||
71aaa3f7 | 343 | # Group of items. |
2adf4336 | 344 | class _Group(_Item, _RepableItem): |
e57a18e1 | 345 | def __init__(self, items: List[_Item], text_loc: TextLocation): |
71aaa3f7 PP |
346 | super().__init__(text_loc) |
347 | self._items = items | |
71aaa3f7 PP |
348 | |
349 | # Contained items. | |
350 | @property | |
351 | def items(self): | |
352 | return self._items | |
353 | ||
71aaa3f7 | 354 | def __repr__(self): |
676f6189 | 355 | return "_Group({}, {})".format(repr(self._items), repr(self._text_loc)) |
71aaa3f7 PP |
356 | |
357 | ||
358 | # Repetition item. | |
2adf4336 PP |
359 | class _Rep(_Item, _ExprMixin): |
360 | def __init__( | |
e57a18e1 | 361 | self, item: _Item, expr_str: str, expr: ast.Expression, text_loc: TextLocation |
2adf4336 | 362 | ): |
71aaa3f7 | 363 | super().__init__(text_loc) |
2adf4336 | 364 | _ExprMixin.__init__(self, expr_str, expr) |
71aaa3f7 | 365 | self._item = item |
71aaa3f7 PP |
366 | |
367 | # Item to repeat. | |
368 | @property | |
369 | def item(self): | |
370 | return self._item | |
371 | ||
71aaa3f7 | 372 | def __repr__(self): |
2adf4336 | 373 | return "_Rep({}, {}, {}, {})".format( |
676f6189 PP |
374 | repr(self._item), |
375 | repr(self._expr_str), | |
376 | repr(self._expr), | |
377 | repr(self._text_loc), | |
71aaa3f7 PP |
378 | ) |
379 | ||
380 | ||
27d52a19 PP |
381 | # Conditional item. |
382 | class _Cond(_Item, _ExprMixin): | |
383 | def __init__( | |
12b5dbc0 PP |
384 | self, |
385 | true_item: _Item, | |
386 | false_item: _Item, | |
387 | expr_str: str, | |
388 | expr: ast.Expression, | |
389 | text_loc: TextLocation, | |
27d52a19 PP |
390 | ): |
391 | super().__init__(text_loc) | |
392 | _ExprMixin.__init__(self, expr_str, expr) | |
12b5dbc0 PP |
393 | self._true_item = true_item |
394 | self._false_item = false_item | |
27d52a19 | 395 | |
12b5dbc0 | 396 | # Item when condition is true. |
27d52a19 | 397 | @property |
12b5dbc0 PP |
398 | def true_item(self): |
399 | return self._true_item | |
400 | ||
401 | # Item when condition is false. | |
402 | @property | |
403 | def false_item(self): | |
404 | return self._false_item | |
27d52a19 PP |
405 | |
406 | def __repr__(self): | |
12b5dbc0 PP |
407 | return "_Cond({}, {}, {}, {}, {})".format( |
408 | repr(self._true_item), | |
409 | repr(self._false_item), | |
27d52a19 PP |
410 | repr(self._expr_str), |
411 | repr(self._expr), | |
412 | repr(self._text_loc), | |
413 | ) | |
414 | ||
415 | ||
320644e2 PP |
416 | # Macro definition item. |
417 | class _MacroDef(_Item): | |
418 | def __init__( | |
419 | self, name: str, param_names: List[str], group: _Group, text_loc: TextLocation | |
420 | ): | |
421 | super().__init__(text_loc) | |
422 | self._name = name | |
423 | self._param_names = param_names | |
424 | self._group = group | |
425 | ||
426 | # Name. | |
427 | @property | |
428 | def name(self): | |
429 | return self._name | |
430 | ||
431 | # Parameters. | |
432 | @property | |
433 | def param_names(self): | |
434 | return self._param_names | |
435 | ||
436 | # Contained items. | |
437 | @property | |
438 | def group(self): | |
439 | return self._group | |
440 | ||
441 | def __repr__(self): | |
442 | return "_MacroDef({}, {}, {}, {})".format( | |
443 | repr(self._name), | |
444 | repr(self._param_names), | |
445 | repr(self._group), | |
446 | repr(self._text_loc), | |
447 | ) | |
448 | ||
449 | ||
450 | # Macro expansion parameter. | |
451 | class _MacroExpParam: | |
452 | def __init__(self, expr_str: str, expr: ast.Expression, text_loc: TextLocation): | |
453 | self._expr_str = expr_str | |
454 | self._expr = expr | |
455 | self._text_loc = text_loc | |
456 | ||
457 | # Expression string. | |
458 | @property | |
459 | def expr_str(self): | |
460 | return self._expr_str | |
461 | ||
462 | # Expression. | |
463 | @property | |
464 | def expr(self): | |
465 | return self._expr | |
466 | ||
467 | # Source text location. | |
468 | @property | |
469 | def text_loc(self): | |
470 | return self._text_loc | |
471 | ||
472 | def __repr__(self): | |
473 | return "_MacroExpParam({}, {}, {})".format( | |
474 | repr(self._expr_str), repr(self._expr), repr(self._text_loc) | |
475 | ) | |
476 | ||
477 | ||
478 | # Macro expansion item. | |
479 | class _MacroExp(_Item, _RepableItem): | |
480 | def __init__( | |
481 | self, | |
482 | name: str, | |
483 | params: List[_MacroExpParam], | |
484 | text_loc: TextLocation, | |
485 | ): | |
486 | super().__init__(text_loc) | |
487 | self._name = name | |
488 | self._params = params | |
489 | ||
490 | # Name. | |
491 | @property | |
492 | def name(self): | |
493 | return self._name | |
494 | ||
495 | # Parameters. | |
496 | @property | |
497 | def params(self): | |
498 | return self._params | |
499 | ||
500 | def __repr__(self): | |
501 | return "_MacroExp({}, {}, {})".format( | |
502 | repr(self._name), | |
503 | repr(self._params), | |
504 | repr(self._text_loc), | |
505 | ) | |
2adf4336 PP |
506 | |
507 | ||
f5dcb24c PP |
508 | # A parsing error message: a string and a text location. |
509 | class ParseErrorMessage: | |
510 | @classmethod | |
511 | def _create(cls, text: str, text_loc: TextLocation): | |
512 | self = cls.__new__(cls) | |
513 | self._init(text, text_loc) | |
514 | return self | |
515 | ||
516 | def __init__(self, *args, **kwargs): # type: ignore | |
517 | raise NotImplementedError | |
518 | ||
519 | def _init(self, text: str, text_loc: TextLocation): | |
520 | self._text = text | |
521 | self._text_loc = text_loc | |
522 | ||
523 | # Message text. | |
524 | @property | |
525 | def text(self): | |
526 | return self._text | |
527 | ||
528 | # Source text location. | |
529 | @property | |
530 | def text_location(self): | |
531 | return self._text_loc | |
532 | ||
533 | ||
534 | # A parsing error containing one or more messages (`ParseErrorMessage`). | |
71aaa3f7 PP |
535 | class ParseError(RuntimeError): |
536 | @classmethod | |
e57a18e1 | 537 | def _create(cls, msg: str, text_loc: TextLocation): |
71aaa3f7 PP |
538 | self = cls.__new__(cls) |
539 | self._init(msg, text_loc) | |
540 | return self | |
541 | ||
542 | def __init__(self, *args, **kwargs): # type: ignore | |
543 | raise NotImplementedError | |
544 | ||
e57a18e1 | 545 | def _init(self, msg: str, text_loc: TextLocation): |
71aaa3f7 | 546 | super().__init__(msg) |
f5dcb24c PP |
547 | self._msgs = [] # type: List[ParseErrorMessage] |
548 | self._add_msg(msg, text_loc) | |
71aaa3f7 | 549 | |
f5dcb24c PP |
550 | def _add_msg(self, msg: str, text_loc: TextLocation): |
551 | self._msgs.append( | |
552 | ParseErrorMessage._create( # pyright: ignore[reportPrivateUsage] | |
553 | msg, text_loc | |
554 | ) | |
555 | ) | |
556 | ||
557 | # Parsing error messages. | |
558 | # | |
559 | # The first message is the most specific one. | |
71aaa3f7 | 560 | @property |
f5dcb24c PP |
561 | def messages(self): |
562 | return self._msgs | |
71aaa3f7 PP |
563 | |
564 | ||
565 | # Raises a parsing error, forwarding the parameters to the constructor. | |
e57a18e1 | 566 | def _raise_error(msg: str, text_loc: TextLocation) -> NoReturn: |
71aaa3f7 PP |
567 | raise ParseError._create(msg, text_loc) # pyright: ignore[reportPrivateUsage] |
568 | ||
569 | ||
f5dcb24c PP |
570 | # Adds a message to the parsing error `exc`. |
571 | def _add_error_msg(exc: ParseError, msg: str, text_loc: TextLocation): | |
572 | exc._add_msg(msg, text_loc) # pyright: ignore[reportPrivateUsage] | |
573 | ||
574 | ||
575 | # Appends a message to the parsing error `exc` and reraises it. | |
576 | def _augment_error(exc: ParseError, msg: str, text_loc: TextLocation) -> NoReturn: | |
577 | _add_error_msg(exc, msg, text_loc) | |
578 | raise exc | |
579 | ||
580 | ||
e57a18e1 PP |
581 | # Variables dictionary type (for type hints). |
582 | VariablesT = Dict[str, Union[int, float]] | |
583 | ||
584 | ||
585 | # Labels dictionary type (for type hints). | |
586 | LabelsT = Dict[str, int] | |
71aaa3f7 PP |
587 | |
588 | ||
589 | # Python name pattern. | |
590 | _py_name_pat = re.compile(r"[a-zA-Z_][a-zA-Z0-9_]*") | |
591 | ||
592 | ||
320644e2 PP |
593 | # Macro definition dictionary. |
594 | _MacroDefsT = Dict[str, _MacroDef] | |
595 | ||
596 | ||
71aaa3f7 PP |
597 | # Normand parser. |
598 | # | |
599 | # The constructor accepts a Normand input. After building, use the `res` | |
600 | # property to get the resulting main group. | |
601 | class _Parser: | |
602 | # Builds a parser to parse the Normand input `normand`, parsing | |
603 | # immediately. | |
e57a18e1 | 604 | def __init__(self, normand: str, variables: VariablesT, labels: LabelsT): |
71aaa3f7 PP |
605 | self._normand = normand |
606 | self._at = 0 | |
607 | self._line_no = 1 | |
608 | self._col_no = 1 | |
609 | self._label_names = set(labels.keys()) | |
610 | self._var_names = set(variables.keys()) | |
320644e2 | 611 | self._macro_defs = {} # type: _MacroDefsT |
71aaa3f7 PP |
612 | self._parse() |
613 | ||
614 | # Result (main group). | |
615 | @property | |
616 | def res(self): | |
617 | return self._res | |
618 | ||
320644e2 PP |
619 | # Macro definitions. |
620 | @property | |
621 | def macro_defs(self): | |
622 | return self._macro_defs | |
623 | ||
71aaa3f7 PP |
624 | # Current text location. |
625 | @property | |
626 | def _text_loc(self): | |
e57a18e1 | 627 | return TextLocation._create( # pyright: ignore[reportPrivateUsage] |
71aaa3f7 PP |
628 | self._line_no, self._col_no |
629 | ) | |
630 | ||
631 | # Returns `True` if this parser is done parsing. | |
632 | def _is_done(self): | |
633 | return self._at == len(self._normand) | |
634 | ||
635 | # Returns `True` if this parser isn't done parsing. | |
636 | def _isnt_done(self): | |
637 | return not self._is_done() | |
638 | ||
639 | # Raises a parse error, creating it using the message `msg` and the | |
640 | # current text location. | |
641 | def _raise_error(self, msg: str) -> NoReturn: | |
642 | _raise_error(msg, self._text_loc) | |
643 | ||
644 | # Tries to make the pattern `pat` match the current substring, | |
645 | # returning the match object and updating `self._at`, | |
646 | # `self._line_no`, and `self._col_no` on success. | |
647 | def _try_parse_pat(self, pat: Pattern[str]): | |
648 | m = pat.match(self._normand, self._at) | |
649 | ||
650 | if m is None: | |
651 | return | |
652 | ||
653 | # Skip matched string | |
654 | self._at += len(m.group(0)) | |
655 | ||
656 | # Update line number | |
657 | self._line_no += m.group(0).count("\n") | |
658 | ||
659 | # Update column number | |
660 | for i in reversed(range(self._at)): | |
661 | if self._normand[i] == "\n" or i == 0: | |
662 | if i == 0: | |
663 | self._col_no = self._at + 1 | |
664 | else: | |
665 | self._col_no = self._at - i | |
666 | ||
667 | break | |
668 | ||
669 | # Return match object | |
670 | return m | |
671 | ||
672 | # Expects the pattern `pat` to match the current substring, | |
673 | # returning the match object and updating `self._at`, | |
674 | # `self._line_no`, and `self._col_no` on success, or raising a parse | |
675 | # error with the message `error_msg` on error. | |
676 | def _expect_pat(self, pat: Pattern[str], error_msg: str): | |
677 | # Match | |
678 | m = self._try_parse_pat(pat) | |
679 | ||
680 | if m is None: | |
681 | # No match: error | |
682 | self._raise_error(error_msg) | |
683 | ||
684 | # Return match object | |
685 | return m | |
686 | ||
687 | # Pattern for _skip_ws_and_comments() | |
688 | _ws_or_syms_or_comments_pat = re.compile( | |
25ca454b | 689 | r"(?:[\s/\\?&:;.,[\]_=|-]|#[^#]*?(?:\n|#))*" |
71aaa3f7 PP |
690 | ) |
691 | ||
692 | # Skips as many whitespaces, insignificant symbol characters, and | |
693 | # comments as possible. | |
694 | def _skip_ws_and_comments(self): | |
695 | self._try_parse_pat(self._ws_or_syms_or_comments_pat) | |
696 | ||
320644e2 PP |
697 | # Pattern for _skip_ws() |
698 | _ws_pat = re.compile(r"\s*") | |
699 | ||
700 | # Skips as many whitespaces as possible. | |
701 | def _skip_ws(self): | |
702 | self._try_parse_pat(self._ws_pat) | |
703 | ||
71aaa3f7 PP |
704 | # Pattern for _try_parse_hex_byte() |
705 | _nibble_pat = re.compile(r"[A-Fa-f0-9]") | |
706 | ||
707 | # Tries to parse a hexadecimal byte, returning a byte item on | |
708 | # success. | |
709 | def _try_parse_hex_byte(self): | |
0e8e3169 PP |
710 | begin_text_loc = self._text_loc |
711 | ||
71aaa3f7 PP |
712 | # Match initial nibble |
713 | m_high = self._try_parse_pat(self._nibble_pat) | |
714 | ||
715 | if m_high is None: | |
716 | # No match | |
717 | return | |
718 | ||
719 | # Expect another nibble | |
720 | self._skip_ws_and_comments() | |
721 | m_low = self._expect_pat( | |
722 | self._nibble_pat, "Expecting another hexadecimal nibble" | |
723 | ) | |
724 | ||
725 | # Return item | |
0e8e3169 | 726 | return _Byte(int(m_high.group(0) + m_low.group(0), 16), begin_text_loc) |
71aaa3f7 PP |
727 | |
728 | # Patterns for _try_parse_bin_byte() | |
729 | _bin_byte_bit_pat = re.compile(r"[01]") | |
730 | _bin_byte_prefix_pat = re.compile(r"%") | |
731 | ||
732 | # Tries to parse a binary byte, returning a byte item on success. | |
733 | def _try_parse_bin_byte(self): | |
0e8e3169 PP |
734 | begin_text_loc = self._text_loc |
735 | ||
71aaa3f7 PP |
736 | # Match prefix |
737 | if self._try_parse_pat(self._bin_byte_prefix_pat) is None: | |
738 | # No match | |
739 | return | |
740 | ||
741 | # Expect eight bits | |
742 | bits = [] # type: List[str] | |
743 | ||
744 | for _ in range(8): | |
745 | self._skip_ws_and_comments() | |
746 | m = self._expect_pat(self._bin_byte_bit_pat, "Expecting a bit (`0` or `1`)") | |
747 | bits.append(m.group(0)) | |
748 | ||
749 | # Return item | |
0e8e3169 | 750 | return _Byte(int("".join(bits), 2), begin_text_loc) |
71aaa3f7 PP |
751 | |
752 | # Patterns for _try_parse_dec_byte() | |
320644e2 | 753 | _dec_byte_prefix_pat = re.compile(r"\$") |
71aaa3f7 PP |
754 | _dec_byte_val_pat = re.compile(r"(?P<neg>-?)(?P<val>\d+)") |
755 | ||
756 | # Tries to parse a decimal byte, returning a byte item on success. | |
757 | def _try_parse_dec_byte(self): | |
0e8e3169 PP |
758 | begin_text_loc = self._text_loc |
759 | ||
71aaa3f7 PP |
760 | # Match prefix |
761 | if self._try_parse_pat(self._dec_byte_prefix_pat) is None: | |
762 | # No match | |
763 | return | |
764 | ||
765 | # Expect the value | |
320644e2 | 766 | self._skip_ws() |
71aaa3f7 PP |
767 | m = self._expect_pat(self._dec_byte_val_pat, "Expecting a decimal constant") |
768 | ||
769 | # Compute value | |
770 | val = int(m.group("val")) * (-1 if m.group("neg") == "-" else 1) | |
771 | ||
772 | # Validate | |
773 | if val < -128 or val > 255: | |
0e8e3169 | 774 | _raise_error("Invalid decimal byte value {}".format(val), begin_text_loc) |
71aaa3f7 PP |
775 | |
776 | # Two's complement | |
05f81895 | 777 | val %= 256 |
71aaa3f7 PP |
778 | |
779 | # Return item | |
0e8e3169 | 780 | return _Byte(val, begin_text_loc) |
71aaa3f7 PP |
781 | |
782 | # Tries to parse a byte, returning a byte item on success. | |
783 | def _try_parse_byte(self): | |
784 | # Hexadecimal | |
785 | item = self._try_parse_hex_byte() | |
786 | ||
787 | if item is not None: | |
788 | return item | |
789 | ||
790 | # Binary | |
791 | item = self._try_parse_bin_byte() | |
792 | ||
793 | if item is not None: | |
794 | return item | |
795 | ||
796 | # Decimal | |
797 | item = self._try_parse_dec_byte() | |
798 | ||
799 | if item is not None: | |
800 | return item | |
801 | ||
802 | # Patterns for _try_parse_str() | |
803 | _str_prefix_pat = re.compile(r'(?:u(?P<len>16|32)(?P<bo>be|le))?\s*"') | |
804 | _str_suffix_pat = re.compile(r'"') | |
805 | _str_str_pat = re.compile(r'(?:(?:\\.)|[^"])*') | |
806 | ||
807 | # Strings corresponding to escape sequence characters | |
808 | _str_escape_seq_strs = { | |
809 | "0": "\0", | |
810 | "a": "\a", | |
811 | "b": "\b", | |
812 | "e": "\x1b", | |
813 | "f": "\f", | |
814 | "n": "\n", | |
815 | "r": "\r", | |
816 | "t": "\t", | |
817 | "v": "\v", | |
818 | "\\": "\\", | |
819 | '"': '"', | |
820 | } | |
821 | ||
822 | # Tries to parse a string, returning a string item on success. | |
823 | def _try_parse_str(self): | |
0e8e3169 PP |
824 | begin_text_loc = self._text_loc |
825 | ||
71aaa3f7 PP |
826 | # Match prefix |
827 | m = self._try_parse_pat(self._str_prefix_pat) | |
828 | ||
829 | if m is None: | |
830 | # No match | |
831 | return | |
832 | ||
833 | # Get encoding | |
834 | encoding = "utf8" | |
835 | ||
836 | if m.group("len") is not None: | |
837 | encoding = "utf_{}_{}".format(m.group("len"), m.group("bo")) | |
838 | ||
839 | # Actual string | |
840 | m = self._expect_pat(self._str_str_pat, "Expecting a literal string") | |
841 | ||
842 | # Expect end of string | |
843 | self._expect_pat(self._str_suffix_pat, 'Expecting `"` (end of literal string)') | |
844 | ||
845 | # Replace escape sequences | |
846 | val = m.group(0) | |
847 | ||
848 | for ec in '0abefnrtv"\\': | |
849 | val = val.replace(r"\{}".format(ec), self._str_escape_seq_strs[ec]) | |
850 | ||
851 | # Encode | |
852 | data = val.encode(encoding) | |
853 | ||
854 | # Return item | |
0e8e3169 | 855 | return _Str(data, begin_text_loc) |
71aaa3f7 | 856 | |
320644e2 PP |
857 | # Common right parenthesis pattern |
858 | _right_paren_pat = re.compile(r"\)") | |
859 | ||
71aaa3f7 | 860 | # Patterns for _try_parse_group() |
320644e2 | 861 | _group_prefix_pat = re.compile(r"\(|!g(?:roup)?\b") |
71aaa3f7 PP |
862 | |
863 | # Tries to parse a group, returning a group item on success. | |
864 | def _try_parse_group(self): | |
0e8e3169 PP |
865 | begin_text_loc = self._text_loc |
866 | ||
71aaa3f7 | 867 | # Match prefix |
261c5ecf PP |
868 | m_open = self._try_parse_pat(self._group_prefix_pat) |
869 | ||
870 | if m_open is None: | |
71aaa3f7 PP |
871 | # No match |
872 | return | |
873 | ||
874 | # Parse items | |
875 | items = self._parse_items() | |
876 | ||
877 | # Expect end of group | |
878 | self._skip_ws_and_comments() | |
261c5ecf PP |
879 | |
880 | if m_open.group(0) == "(": | |
320644e2 | 881 | pat = self._right_paren_pat |
261c5ecf PP |
882 | exp = ")" |
883 | else: | |
884 | pat = self._block_end_pat | |
885 | exp = "!end" | |
886 | ||
887 | self._expect_pat(pat, "Expecting an item or `{}` (end of group)".format(exp)) | |
71aaa3f7 PP |
888 | |
889 | # Return item | |
0e8e3169 | 890 | return _Group(items, begin_text_loc) |
71aaa3f7 PP |
891 | |
892 | # Returns a stripped expression string and an AST expression node | |
893 | # from the expression string `expr_str` at text location `text_loc`. | |
e57a18e1 | 894 | def _ast_expr_from_str(self, expr_str: str, text_loc: TextLocation): |
71aaa3f7 PP |
895 | # Create an expression node from the expression string |
896 | expr_str = expr_str.strip().replace("\n", " ") | |
897 | ||
898 | try: | |
899 | expr = ast.parse(expr_str, mode="eval") | |
900 | except SyntaxError: | |
901 | _raise_error( | |
902 | "Invalid expression `{}`: invalid syntax".format(expr_str), | |
903 | text_loc, | |
904 | ) | |
905 | ||
906 | return expr_str, expr | |
907 | ||
269f6eb3 | 908 | # Patterns for _try_parse_num_and_attr() |
05f81895 | 909 | _val_expr_pat = re.compile(r"([^}:]+):\s*") |
269f6eb3 | 910 | _fl_num_len_attr_pat = re.compile(r"8|16|24|32|40|48|56|64") |
05f81895 | 911 | _leb128_int_attr_pat = re.compile(r"(u|s)leb128") |
71aaa3f7 | 912 | |
05f81895 PP |
913 | # Tries to parse a value and attribute (fixed length in bits or |
914 | # `leb128`), returning a value item on success. | |
269f6eb3 | 915 | def _try_parse_num_and_attr(self): |
71aaa3f7 PP |
916 | begin_text_loc = self._text_loc |
917 | ||
918 | # Match | |
919 | m_expr = self._try_parse_pat(self._val_expr_pat) | |
920 | ||
921 | if m_expr is None: | |
922 | # No match | |
923 | return | |
924 | ||
71aaa3f7 PP |
925 | # Create an expression node from the expression string |
926 | expr_str, expr = self._ast_expr_from_str(m_expr.group(1), begin_text_loc) | |
927 | ||
05f81895 | 928 | # Length? |
269f6eb3 | 929 | m_attr = self._try_parse_pat(self._fl_num_len_attr_pat) |
05f81895 PP |
930 | |
931 | if m_attr is None: | |
932 | # LEB128? | |
933 | m_attr = self._try_parse_pat(self._leb128_int_attr_pat) | |
934 | ||
935 | if m_attr is None: | |
936 | # At this point it's invalid | |
937 | self._raise_error( | |
938 | "Expecting a length (multiple of eight bits), `uleb128`, or `sleb128`" | |
939 | ) | |
940 | ||
941 | # Return LEB128 integer item | |
942 | cls = _ULeb128Int if m_attr.group(1) == "u" else _SLeb128Int | |
943 | return cls(expr_str, expr, begin_text_loc) | |
944 | else: | |
269f6eb3 PP |
945 | # Return fixed-length number item |
946 | return _FlNum( | |
05f81895 PP |
947 | expr_str, |
948 | expr, | |
949 | int(m_attr.group(0)), | |
950 | begin_text_loc, | |
951 | ) | |
71aaa3f7 | 952 | |
320644e2 PP |
953 | # Patterns for _try_parse_var_assign() |
954 | _var_assign_name_equal_pat = re.compile(r"({})\s*=".format(_py_name_pat.pattern)) | |
955 | _var_assign_expr_pat = re.compile(r"[^}]+") | |
71aaa3f7 | 956 | |
2adf4336 PP |
957 | # Tries to parse a variable assignment, returning a variable |
958 | # assignment item on success. | |
959 | def _try_parse_var_assign(self): | |
71aaa3f7 PP |
960 | begin_text_loc = self._text_loc |
961 | ||
962 | # Match | |
320644e2 | 963 | m = self._try_parse_pat(self._var_assign_name_equal_pat) |
71aaa3f7 PP |
964 | |
965 | if m is None: | |
966 | # No match | |
967 | return | |
968 | ||
969 | # Validate name | |
320644e2 | 970 | name = m.group(1) |
71aaa3f7 PP |
971 | |
972 | if name == _icitte_name: | |
0e8e3169 PP |
973 | _raise_error( |
974 | "`{}` is a reserved variable name".format(_icitte_name), begin_text_loc | |
975 | ) | |
71aaa3f7 PP |
976 | |
977 | if name in self._label_names: | |
0e8e3169 | 978 | _raise_error("Existing label named `{}`".format(name), begin_text_loc) |
71aaa3f7 | 979 | |
320644e2 PP |
980 | # Expect an expression |
981 | self._skip_ws() | |
982 | m = self._expect_pat(self._var_assign_expr_pat, "Expecting an expression") | |
71aaa3f7 PP |
983 | |
984 | # Create an expression node from the expression string | |
320644e2 PP |
985 | expr_str, expr = self._ast_expr_from_str(m.group(0), begin_text_loc) |
986 | ||
987 | # Add to known variable names | |
988 | self._var_names.add(name) | |
71aaa3f7 PP |
989 | |
990 | # Return item | |
2adf4336 | 991 | return _VarAssign( |
71aaa3f7 PP |
992 | name, |
993 | expr_str, | |
994 | expr, | |
0e8e3169 | 995 | begin_text_loc, |
71aaa3f7 PP |
996 | ) |
997 | ||
2adf4336 | 998 | # Pattern for _try_parse_set_bo() |
71aaa3f7 PP |
999 | _bo_pat = re.compile(r"[bl]e") |
1000 | ||
2adf4336 PP |
1001 | # Tries to parse a byte order name, returning a byte order setting |
1002 | # item on success. | |
1003 | def _try_parse_set_bo(self): | |
0e8e3169 PP |
1004 | begin_text_loc = self._text_loc |
1005 | ||
71aaa3f7 PP |
1006 | # Match |
1007 | m = self._try_parse_pat(self._bo_pat) | |
1008 | ||
1009 | if m is None: | |
1010 | # No match | |
1011 | return | |
1012 | ||
1013 | # Return corresponding item | |
1014 | if m.group(0) == "be": | |
2adf4336 | 1015 | return _SetBo(ByteOrder.BE, begin_text_loc) |
71aaa3f7 PP |
1016 | else: |
1017 | assert m.group(0) == "le" | |
2adf4336 | 1018 | return _SetBo(ByteOrder.LE, begin_text_loc) |
71aaa3f7 PP |
1019 | |
1020 | # Patterns for _try_parse_val_or_bo() | |
320644e2 PP |
1021 | _val_var_assign_set_bo_prefix_pat = re.compile(r"\{") |
1022 | _val_var_assign_set_bo_suffix_pat = re.compile(r"\}") | |
71aaa3f7 | 1023 | |
2adf4336 PP |
1024 | # Tries to parse a value, a variable assignment, or a byte order |
1025 | # setting, returning an item on success. | |
1026 | def _try_parse_val_or_var_assign_or_set_bo(self): | |
71aaa3f7 | 1027 | # Match prefix |
2adf4336 | 1028 | if self._try_parse_pat(self._val_var_assign_set_bo_prefix_pat) is None: |
71aaa3f7 PP |
1029 | # No match |
1030 | return | |
1031 | ||
320644e2 PP |
1032 | self._skip_ws() |
1033 | ||
2adf4336 PP |
1034 | # Variable assignment item? |
1035 | item = self._try_parse_var_assign() | |
71aaa3f7 PP |
1036 | |
1037 | if item is None: | |
269f6eb3 PP |
1038 | # Number item? |
1039 | item = self._try_parse_num_and_attr() | |
71aaa3f7 PP |
1040 | |
1041 | if item is None: | |
2adf4336 PP |
1042 | # Byte order setting item? |
1043 | item = self._try_parse_set_bo() | |
71aaa3f7 PP |
1044 | |
1045 | if item is None: | |
1046 | # At this point it's invalid | |
2adf4336 | 1047 | self._raise_error( |
269f6eb3 | 1048 | "Expecting a fixed-length number, a variable assignment, or a byte order setting" |
2adf4336 | 1049 | ) |
71aaa3f7 PP |
1050 | |
1051 | # Expect suffix | |
320644e2 | 1052 | self._skip_ws() |
2adf4336 | 1053 | self._expect_pat(self._val_var_assign_set_bo_suffix_pat, "Expecting `}`") |
71aaa3f7 PP |
1054 | return item |
1055 | ||
fc21bb27 PP |
1056 | # Returns a normalized version (so as to be parseable by int()) of |
1057 | # the constant integer string `s`, possibly negative, dealing with | |
1058 | # any radix suffix. | |
1059 | @staticmethod | |
1060 | def _norm_const_int(s: str): | |
1061 | neg = "" | |
1062 | pos = s | |
1063 | ||
1064 | if s.startswith("-"): | |
1065 | neg = "-" | |
1066 | pos = s[1:] | |
1067 | ||
1068 | for r in "xXoObB": | |
1069 | if pos.startswith("0" + r): | |
1070 | # Already correct | |
1071 | return s | |
1072 | ||
1073 | # Try suffix | |
1074 | asm_suf_base = { | |
1075 | "h": "x", | |
1076 | "H": "x", | |
1077 | "q": "o", | |
1078 | "Q": "o", | |
1079 | "o": "o", | |
1080 | "O": "o", | |
1081 | "b": "b", | |
1082 | "B": "B", | |
1083 | } | |
1084 | ||
1085 | for suf in asm_suf_base: | |
1086 | if pos[-1] == suf: | |
1087 | s = "{}0{}{}".format(neg, asm_suf_base[suf], pos.rstrip(suf)) | |
1088 | ||
1089 | return s | |
1090 | ||
320644e2 | 1091 | # Common constant integer patterns |
fc21bb27 PP |
1092 | _pos_const_int_pat = re.compile( |
1093 | r"0[Xx][A-Fa-f0-9]+|0[Oo][0-7]+|0[Bb][01]+|[A-Fa-f0-9]+[hH]|[0-7]+[qQoO]|[01]+[bB]|\d+" | |
1094 | ) | |
320644e2 | 1095 | _const_int_pat = re.compile(r"(?P<neg>-)?(?:{})".format(_pos_const_int_pat.pattern)) |
71aaa3f7 | 1096 | |
2adf4336 PP |
1097 | # Tries to parse an offset setting value (after the initial `<`), |
1098 | # returning an offset item on success. | |
1099 | def _try_parse_set_offset_val(self): | |
0e8e3169 PP |
1100 | begin_text_loc = self._text_loc |
1101 | ||
71aaa3f7 PP |
1102 | # Match |
1103 | m = self._try_parse_pat(self._pos_const_int_pat) | |
1104 | ||
1105 | if m is None: | |
1106 | # No match | |
1107 | return | |
1108 | ||
1109 | # Return item | |
fc21bb27 | 1110 | return _SetOffset(int(self._norm_const_int(m.group(0)), 0), begin_text_loc) |
71aaa3f7 PP |
1111 | |
1112 | # Tries to parse a label name (after the initial `<`), returning a | |
1113 | # label item on success. | |
1114 | def _try_parse_label_name(self): | |
0e8e3169 PP |
1115 | begin_text_loc = self._text_loc |
1116 | ||
71aaa3f7 PP |
1117 | # Match |
1118 | m = self._try_parse_pat(_py_name_pat) | |
1119 | ||
1120 | if m is None: | |
1121 | # No match | |
1122 | return | |
1123 | ||
1124 | # Validate | |
1125 | name = m.group(0) | |
1126 | ||
1127 | if name == _icitte_name: | |
0e8e3169 PP |
1128 | _raise_error( |
1129 | "`{}` is a reserved label name".format(_icitte_name), begin_text_loc | |
1130 | ) | |
71aaa3f7 PP |
1131 | |
1132 | if name in self._label_names: | |
0e8e3169 | 1133 | _raise_error("Duplicate label name `{}`".format(name), begin_text_loc) |
71aaa3f7 PP |
1134 | |
1135 | if name in self._var_names: | |
0e8e3169 | 1136 | _raise_error("Existing variable named `{}`".format(name), begin_text_loc) |
71aaa3f7 PP |
1137 | |
1138 | # Add to known label names | |
1139 | self._label_names.add(name) | |
1140 | ||
1141 | # Return item | |
0e8e3169 | 1142 | return _Label(name, begin_text_loc) |
71aaa3f7 | 1143 | |
2adf4336 | 1144 | # Patterns for _try_parse_label_or_set_offset() |
320644e2 PP |
1145 | _label_set_offset_prefix_pat = re.compile(r"<") |
1146 | _label_set_offset_suffix_pat = re.compile(r">") | |
71aaa3f7 | 1147 | |
2adf4336 PP |
1148 | # Tries to parse a label or an offset setting, returning an item on |
1149 | # success. | |
1150 | def _try_parse_label_or_set_offset(self): | |
71aaa3f7 | 1151 | # Match prefix |
2adf4336 | 1152 | if self._try_parse_pat(self._label_set_offset_prefix_pat) is None: |
71aaa3f7 PP |
1153 | # No match |
1154 | return | |
1155 | ||
2adf4336 | 1156 | # Offset setting item? |
320644e2 | 1157 | self._skip_ws() |
2adf4336 | 1158 | item = self._try_parse_set_offset_val() |
71aaa3f7 PP |
1159 | |
1160 | if item is None: | |
1161 | # Label item? | |
1162 | item = self._try_parse_label_name() | |
1163 | ||
1164 | if item is None: | |
1165 | # At this point it's invalid | |
2adf4336 | 1166 | self._raise_error("Expecting a label name or an offset setting value") |
71aaa3f7 PP |
1167 | |
1168 | # Expect suffix | |
320644e2 | 1169 | self._skip_ws() |
2adf4336 | 1170 | self._expect_pat(self._label_set_offset_suffix_pat, "Expecting `>`") |
71aaa3f7 PP |
1171 | return item |
1172 | ||
25ca454b PP |
1173 | # Pattern for _parse_pad_val() |
1174 | _pad_val_prefix_pat = re.compile(r"~") | |
1175 | ||
1176 | # Tries to parse a padding value, returning the padding value, or 0 | |
1177 | # if none. | |
1178 | def _parse_pad_val(self): | |
1179 | # Padding value? | |
1180 | self._skip_ws() | |
1181 | pad_val = 0 | |
1182 | ||
1183 | if self._try_parse_pat(self._pad_val_prefix_pat) is not None: | |
1184 | self._skip_ws() | |
1185 | pad_val_text_loc = self._text_loc | |
1186 | m = self._expect_pat( | |
1187 | self._pos_const_int_pat, | |
1188 | "Expecting a positive constant integer (byte value)", | |
1189 | ) | |
1190 | ||
1191 | # Validate | |
fc21bb27 | 1192 | pad_val = int(self._norm_const_int(m.group(0)), 0) |
25ca454b PP |
1193 | |
1194 | if pad_val > 255: | |
1195 | _raise_error( | |
1196 | "Invalid padding byte value {}".format(pad_val), | |
1197 | pad_val_text_loc, | |
1198 | ) | |
1199 | ||
1200 | return pad_val | |
1201 | ||
676f6189 | 1202 | # Patterns for _try_parse_align_offset() |
320644e2 PP |
1203 | _align_offset_prefix_pat = re.compile(r"@") |
1204 | _align_offset_val_pat = re.compile(r"\d+") | |
676f6189 PP |
1205 | |
1206 | # Tries to parse an offset alignment, returning an offset alignment | |
1207 | # item on success. | |
1208 | def _try_parse_align_offset(self): | |
1209 | begin_text_loc = self._text_loc | |
1210 | ||
1211 | # Match prefix | |
1212 | if self._try_parse_pat(self._align_offset_prefix_pat) is None: | |
1213 | # No match | |
1214 | return | |
1215 | ||
320644e2 | 1216 | # Expect an alignment |
25ca454b | 1217 | self._skip_ws() |
676f6189 PP |
1218 | align_text_loc = self._text_loc |
1219 | m = self._expect_pat( | |
1220 | self._align_offset_val_pat, | |
1221 | "Expecting an alignment (positive multiple of eight bits)", | |
1222 | ) | |
1223 | ||
1224 | # Validate alignment | |
320644e2 | 1225 | val = int(m.group(0)) |
676f6189 PP |
1226 | |
1227 | if val <= 0 or (val % 8) != 0: | |
1228 | _raise_error( | |
1229 | "Invalid alignment value {} (not a positive multiple of eight)".format( | |
1230 | val | |
1231 | ), | |
1232 | align_text_loc, | |
1233 | ) | |
1234 | ||
25ca454b PP |
1235 | # Padding value |
1236 | pad_val = self._parse_pad_val() | |
676f6189 | 1237 | |
25ca454b PP |
1238 | # Return item |
1239 | return _AlignOffset(val, pad_val, begin_text_loc) | |
676f6189 | 1240 | |
25ca454b PP |
1241 | # Patterns for _try_parse_fill_until() |
1242 | _fill_until_prefix_pat = re.compile(r"\+") | |
1243 | _fill_until_pad_val_prefix_pat = re.compile(r"~") | |
676f6189 | 1244 | |
25ca454b PP |
1245 | # Tries to parse a filling, returning a filling item on success. |
1246 | def _try_parse_fill_until(self): | |
1247 | begin_text_loc = self._text_loc | |
1248 | ||
1249 | # Match prefix | |
1250 | if self._try_parse_pat(self._fill_until_prefix_pat) is None: | |
1251 | # No match | |
1252 | return | |
1253 | ||
1254 | # Expect expression | |
1255 | self._skip_ws() | |
1256 | expr_str, expr = self._expect_const_int_name_expr(True) | |
1257 | ||
1258 | # Padding value | |
1259 | pad_val = self._parse_pad_val() | |
676f6189 PP |
1260 | |
1261 | # Return item | |
25ca454b | 1262 | return _FillUntil(expr_str, expr, pad_val, begin_text_loc) |
676f6189 | 1263 | |
e57a18e1 | 1264 | # Patterns for _expect_rep_mul_expr() |
320644e2 PP |
1265 | _inner_expr_prefix_pat = re.compile(r"\{") |
1266 | _inner_expr_pat = re.compile(r"[^}]+") | |
1267 | _inner_expr_suffix_pat = re.compile(r"\}") | |
1268 | ||
1269 | # Parses a constant integer if `accept_const_int` is `True` | |
1270 | # (possibly negative if `allow_neg` is `True`), a name, or an | |
1271 | # expression within `{` and `}`. | |
1272 | def _expect_const_int_name_expr( | |
1273 | self, accept_const_int: bool, allow_neg: bool = False | |
1274 | ): | |
e57a18e1 PP |
1275 | expr_text_loc = self._text_loc |
1276 | ||
1277 | # Constant integer? | |
27d52a19 PP |
1278 | m = None |
1279 | ||
320644e2 PP |
1280 | if accept_const_int: |
1281 | m = self._try_parse_pat(self._const_int_pat) | |
e57a18e1 PP |
1282 | |
1283 | if m is None: | |
1284 | # Name? | |
1285 | m = self._try_parse_pat(_py_name_pat) | |
1286 | ||
1287 | if m is None: | |
1288 | # Expression? | |
320644e2 PP |
1289 | if self._try_parse_pat(self._inner_expr_prefix_pat) is None: |
1290 | pos_msg = "" if allow_neg else "positive " | |
1291 | ||
1292 | if accept_const_int: | |
1293 | mid_msg = "a {}constant integer, a name, or `{{`".format( | |
1294 | pos_msg | |
1295 | ) | |
27d52a19 PP |
1296 | else: |
1297 | mid_msg = "a name or `{`" | |
1298 | ||
e57a18e1 | 1299 | # At this point it's invalid |
27d52a19 | 1300 | self._raise_error("Expecting {}".format(mid_msg)) |
e57a18e1 PP |
1301 | |
1302 | # Expect an expression | |
320644e2 | 1303 | self._skip_ws() |
e57a18e1 | 1304 | expr_text_loc = self._text_loc |
320644e2 | 1305 | m = self._expect_pat(self._inner_expr_pat, "Expecting an expression") |
e57a18e1 PP |
1306 | expr_str = m.group(0) |
1307 | ||
1308 | # Expect `}` | |
320644e2 PP |
1309 | self._skip_ws() |
1310 | self._expect_pat(self._inner_expr_suffix_pat, "Expecting `}`") | |
e57a18e1 PP |
1311 | else: |
1312 | expr_str = m.group(0) | |
1313 | else: | |
320644e2 PP |
1314 | if m.group("neg") == "-" and not allow_neg: |
1315 | _raise_error("Expecting a positive constant integer", expr_text_loc) | |
1316 | ||
fc21bb27 | 1317 | expr_str = self._norm_const_int(m.group(0)) |
e57a18e1 PP |
1318 | |
1319 | return self._ast_expr_from_str(expr_str, expr_text_loc) | |
1320 | ||
27d52a19 PP |
1321 | # Parses the multiplier expression of a repetition (block or |
1322 | # post-item) and returns the expression string and AST node. | |
1323 | def _expect_rep_mul_expr(self): | |
320644e2 | 1324 | return self._expect_const_int_name_expr(True) |
27d52a19 PP |
1325 | |
1326 | # Common block end pattern | |
320644e2 | 1327 | _block_end_pat = re.compile(r"!end\b") |
27d52a19 | 1328 | |
e57a18e1 | 1329 | # Pattern for _try_parse_rep_block() |
320644e2 | 1330 | _rep_block_prefix_pat = re.compile(r"!r(?:epeat)?\b") |
e57a18e1 PP |
1331 | |
1332 | # Tries to parse a repetition block, returning a repetition item on | |
1333 | # success. | |
1334 | def _try_parse_rep_block(self): | |
1335 | begin_text_loc = self._text_loc | |
1336 | ||
1337 | # Match prefix | |
1338 | if self._try_parse_pat(self._rep_block_prefix_pat) is None: | |
1339 | # No match | |
1340 | return | |
1341 | ||
1342 | # Expect expression | |
1343 | self._skip_ws_and_comments() | |
1344 | expr_str, expr = self._expect_rep_mul_expr() | |
1345 | ||
1346 | # Parse items | |
1347 | self._skip_ws_and_comments() | |
1348 | items_text_loc = self._text_loc | |
1349 | items = self._parse_items() | |
1350 | ||
1351 | # Expect end of block | |
1352 | self._skip_ws_and_comments() | |
1353 | self._expect_pat( | |
27d52a19 | 1354 | self._block_end_pat, "Expecting an item or `!end` (end of repetition block)" |
e57a18e1 PP |
1355 | ) |
1356 | ||
1357 | # Return item | |
1358 | return _Rep(_Group(items, items_text_loc), expr_str, expr, begin_text_loc) | |
1359 | ||
27d52a19 | 1360 | # Pattern for _try_parse_cond_block() |
320644e2 | 1361 | _cond_block_prefix_pat = re.compile(r"!if\b") |
12b5dbc0 | 1362 | _cond_block_else_pat = re.compile(r"!else\b") |
27d52a19 PP |
1363 | |
1364 | # Tries to parse a conditional block, returning a conditional item | |
1365 | # on success. | |
1366 | def _try_parse_cond_block(self): | |
1367 | begin_text_loc = self._text_loc | |
1368 | ||
1369 | # Match prefix | |
1370 | if self._try_parse_pat(self._cond_block_prefix_pat) is None: | |
1371 | # No match | |
1372 | return | |
1373 | ||
1374 | # Expect expression | |
1375 | self._skip_ws_and_comments() | |
320644e2 | 1376 | expr_str, expr = self._expect_const_int_name_expr(False) |
27d52a19 | 1377 | |
12b5dbc0 | 1378 | # Parse "true" items |
27d52a19 | 1379 | self._skip_ws_and_comments() |
12b5dbc0 PP |
1380 | true_items_text_loc = self._text_loc |
1381 | true_items = self._parse_items() | |
1382 | false_items = [] # type: List[_Item] | |
1383 | false_items_text_loc = begin_text_loc | |
27d52a19 | 1384 | |
12b5dbc0 | 1385 | # `!else`? |
27d52a19 | 1386 | self._skip_ws_and_comments() |
12b5dbc0 PP |
1387 | |
1388 | if self._try_parse_pat(self._cond_block_else_pat) is not None: | |
1389 | # Parse "false" items | |
1390 | self._skip_ws_and_comments() | |
1391 | false_items_text_loc = self._text_loc | |
1392 | false_items = self._parse_items() | |
1393 | ||
1394 | # Expect end of block | |
27d52a19 PP |
1395 | self._expect_pat( |
1396 | self._block_end_pat, | |
12b5dbc0 | 1397 | "Expecting an item, `!else`, or `!end` (end of conditional block)", |
27d52a19 PP |
1398 | ) |
1399 | ||
1400 | # Return item | |
12b5dbc0 PP |
1401 | return _Cond( |
1402 | _Group(true_items, true_items_text_loc), | |
1403 | _Group(false_items, false_items_text_loc), | |
1404 | expr_str, | |
1405 | expr, | |
1406 | begin_text_loc, | |
1407 | ) | |
27d52a19 | 1408 | |
320644e2 PP |
1409 | # Common left parenthesis pattern |
1410 | _left_paren_pat = re.compile(r"\(") | |
1411 | ||
1412 | # Patterns for _try_parse_macro_def() and _try_parse_macro_exp() | |
1413 | _macro_params_comma_pat = re.compile(",") | |
1414 | ||
1415 | # Patterns for _try_parse_macro_def() | |
1416 | _macro_def_prefix_pat = re.compile(r"!m(?:acro)?\b") | |
1417 | ||
1418 | # Tries to parse a macro definition, adding it to `self._macro_defs` | |
1419 | # and returning `True` on success. | |
1420 | def _try_parse_macro_def(self): | |
1421 | begin_text_loc = self._text_loc | |
1422 | ||
1423 | # Match prefix | |
1424 | if self._try_parse_pat(self._macro_def_prefix_pat) is None: | |
1425 | # No match | |
1426 | return False | |
1427 | ||
1428 | # Expect a name | |
1429 | self._skip_ws() | |
1430 | name_text_loc = self._text_loc | |
1431 | m = self._expect_pat(_py_name_pat, "Expecting a valid macro name") | |
1432 | ||
1433 | # Validate name | |
1434 | name = m.group(0) | |
1435 | ||
1436 | if name in self._macro_defs: | |
1437 | _raise_error("Duplicate macro named `{}`".format(name), name_text_loc) | |
1438 | ||
1439 | # Expect `(` | |
1440 | self._skip_ws() | |
1441 | self._expect_pat(self._left_paren_pat, "Expecting `(`") | |
1442 | ||
1443 | # Try to parse comma-separated parameter names | |
1444 | param_names = [] # type: List[str] | |
1445 | expect_comma = False | |
1446 | ||
1447 | while True: | |
1448 | self._skip_ws() | |
1449 | ||
1450 | # End? | |
1451 | if self._try_parse_pat(self._right_paren_pat) is not None: | |
1452 | # End | |
1453 | break | |
1454 | ||
1455 | # Comma? | |
1456 | if expect_comma: | |
1457 | self._expect_pat(self._macro_params_comma_pat, "Expecting `,`") | |
1458 | ||
1459 | # Expect parameter name | |
1460 | self._skip_ws() | |
1461 | param_text_loc = self._text_loc | |
1462 | m = self._expect_pat(_py_name_pat, "Expecting valid parameter name") | |
1463 | ||
1464 | if m.group(0) in param_names: | |
1465 | _raise_error( | |
1466 | "Duplicate macro parameter named `{}`".format(m.group(0)), | |
1467 | param_text_loc, | |
1468 | ) | |
1469 | ||
1470 | param_names.append(m.group(0)) | |
1471 | expect_comma = True | |
1472 | ||
1473 | # Expect items | |
1474 | self._skip_ws_and_comments() | |
1475 | items_text_loc = self._text_loc | |
1476 | old_var_names = self._var_names.copy() | |
1477 | old_label_names = self._label_names.copy() | |
1478 | self._var_names = set() # type: Set[str] | |
1479 | self._label_names = set() # type: Set[str] | |
1480 | items = self._parse_items() | |
1481 | self._var_names = old_var_names | |
1482 | self._label_names = old_label_names | |
1483 | ||
1484 | # Expect suffix | |
1485 | self._expect_pat( | |
1486 | self._block_end_pat, "Expecting an item or `!end` (end of macro block)" | |
1487 | ) | |
1488 | ||
1489 | # Register macro | |
1490 | self._macro_defs[name] = _MacroDef( | |
1491 | name, param_names, _Group(items, items_text_loc), begin_text_loc | |
1492 | ) | |
1493 | ||
1494 | return True | |
1495 | ||
1496 | # Patterns for _try_parse_macro_exp() | |
1497 | _macro_exp_prefix_pat = re.compile(r"m\b") | |
1498 | _macro_exp_colon_pat = re.compile(r":") | |
1499 | ||
1500 | # Tries to parse a macro expansion, returning a macro expansion item | |
1501 | # on success. | |
1502 | def _try_parse_macro_exp(self): | |
1503 | begin_text_loc = self._text_loc | |
1504 | ||
1505 | # Match prefix | |
1506 | if self._try_parse_pat(self._macro_exp_prefix_pat) is None: | |
1507 | # No match | |
1508 | return | |
1509 | ||
1510 | # Expect `:` | |
1511 | self._skip_ws() | |
1512 | self._expect_pat(self._macro_exp_colon_pat, "Expecting `:`") | |
1513 | ||
1514 | # Expect a macro name | |
1515 | self._skip_ws() | |
1516 | name_text_loc = self._text_loc | |
1517 | m = self._expect_pat(_py_name_pat, "Expecting a valid macro name") | |
1518 | ||
1519 | # Validate name | |
1520 | name = m.group(0) | |
1521 | macro_def = self._macro_defs.get(name) | |
1522 | ||
1523 | if macro_def is None: | |
1524 | _raise_error("Unknown macro name `{}`".format(name), name_text_loc) | |
1525 | ||
1526 | # Expect `(` | |
1527 | self._skip_ws() | |
1528 | self._expect_pat(self._left_paren_pat, "Expecting `(`") | |
1529 | ||
1530 | # Try to parse comma-separated parameter values | |
1531 | params_text_loc = self._text_loc | |
1532 | params = [] # type: List[_MacroExpParam] | |
1533 | expect_comma = False | |
1534 | ||
1535 | while True: | |
1536 | self._skip_ws() | |
1537 | ||
1538 | # End? | |
1539 | if self._try_parse_pat(self._right_paren_pat) is not None: | |
1540 | # End | |
1541 | break | |
1542 | ||
1543 | # Expect a Value | |
1544 | if expect_comma: | |
1545 | self._expect_pat(self._macro_params_comma_pat, "Expecting `,`") | |
1546 | ||
1547 | self._skip_ws() | |
1548 | param_text_loc = self._text_loc | |
1549 | params.append( | |
1550 | _MacroExpParam( | |
1551 | *self._expect_const_int_name_expr(True, True), param_text_loc | |
1552 | ) | |
1553 | ) | |
1554 | expect_comma = True | |
1555 | ||
1556 | # Validate parameter values | |
1557 | if len(params) != len(macro_def.param_names): | |
1558 | sing_plur = "" if len(params) == 1 else "s" | |
1559 | _raise_error( | |
1560 | "Macro expansion passes {} parameter{} while the definition expects {}".format( | |
1561 | len(params), sing_plur, len(macro_def.param_names) | |
1562 | ), | |
1563 | params_text_loc, | |
1564 | ) | |
1565 | ||
1566 | # Return item | |
1567 | return _MacroExp(name, params, begin_text_loc) | |
1568 | ||
71aaa3f7 PP |
1569 | # Tries to parse a base item (anything except a repetition), |
1570 | # returning it on success. | |
1571 | def _try_parse_base_item(self): | |
1572 | # Byte item? | |
1573 | item = self._try_parse_byte() | |
1574 | ||
1575 | if item is not None: | |
1576 | return item | |
1577 | ||
1578 | # String item? | |
1579 | item = self._try_parse_str() | |
1580 | ||
1581 | if item is not None: | |
1582 | return item | |
1583 | ||
2adf4336 PP |
1584 | # Value, variable assignment, or byte order setting item? |
1585 | item = self._try_parse_val_or_var_assign_or_set_bo() | |
71aaa3f7 PP |
1586 | |
1587 | if item is not None: | |
1588 | return item | |
1589 | ||
2adf4336 PP |
1590 | # Label or offset setting item? |
1591 | item = self._try_parse_label_or_set_offset() | |
71aaa3f7 | 1592 | |
676f6189 PP |
1593 | if item is not None: |
1594 | return item | |
1595 | ||
1596 | # Offset alignment item? | |
1597 | item = self._try_parse_align_offset() | |
1598 | ||
25ca454b PP |
1599 | if item is not None: |
1600 | return item | |
1601 | ||
1602 | # Filling item? | |
1603 | item = self._try_parse_fill_until() | |
1604 | ||
71aaa3f7 PP |
1605 | if item is not None: |
1606 | return item | |
1607 | ||
1608 | # Group item? | |
1609 | item = self._try_parse_group() | |
1610 | ||
1611 | if item is not None: | |
1612 | return item | |
1613 | ||
320644e2 | 1614 | # Repetition block item? |
e57a18e1 | 1615 | item = self._try_parse_rep_block() |
71aaa3f7 | 1616 | |
e57a18e1 PP |
1617 | if item is not None: |
1618 | return item | |
1619 | ||
27d52a19 PP |
1620 | # Conditional block item? |
1621 | item = self._try_parse_cond_block() | |
1622 | ||
1623 | if item is not None: | |
1624 | return item | |
1625 | ||
320644e2 PP |
1626 | # Macro expansion? |
1627 | item = self._try_parse_macro_exp() | |
1628 | ||
1629 | if item is not None: | |
1630 | return item | |
1631 | ||
e57a18e1 PP |
1632 | # Pattern for _try_parse_rep_post() |
1633 | _rep_post_prefix_pat = re.compile(r"\*") | |
1634 | ||
1635 | # Tries to parse a post-item repetition, returning the expression | |
1636 | # string and AST expression node on success. | |
1637 | def _try_parse_rep_post(self): | |
71aaa3f7 | 1638 | # Match prefix |
e57a18e1 | 1639 | if self._try_parse_pat(self._rep_post_prefix_pat) is None: |
71aaa3f7 | 1640 | # No match |
2adf4336 | 1641 | return |
71aaa3f7 | 1642 | |
e57a18e1 | 1643 | # Return expression string and AST expression |
71aaa3f7 | 1644 | self._skip_ws_and_comments() |
e57a18e1 | 1645 | return self._expect_rep_mul_expr() |
71aaa3f7 | 1646 | |
1ca7b5e1 PP |
1647 | # Tries to parse an item, possibly followed by a repetition, |
1648 | # returning `True` on success. | |
1649 | # | |
1650 | # Appends any parsed item to `items`. | |
1651 | def _try_append_item(self, items: List[_Item]): | |
71aaa3f7 PP |
1652 | self._skip_ws_and_comments() |
1653 | ||
320644e2 | 1654 | # Base item |
71aaa3f7 PP |
1655 | item = self._try_parse_base_item() |
1656 | ||
1657 | if item is None: | |
320644e2 | 1658 | return |
71aaa3f7 PP |
1659 | |
1660 | # Parse repetition if the base item is repeatable | |
1661 | if isinstance(item, _RepableItem): | |
0e8e3169 PP |
1662 | self._skip_ws_and_comments() |
1663 | rep_text_loc = self._text_loc | |
e57a18e1 | 1664 | rep_ret = self._try_parse_rep_post() |
71aaa3f7 | 1665 | |
2adf4336 | 1666 | if rep_ret is not None: |
320644e2 | 1667 | item = _Rep(item, *rep_ret, rep_text_loc) |
71aaa3f7 | 1668 | |
1ca7b5e1 PP |
1669 | items.append(item) |
1670 | return True | |
71aaa3f7 PP |
1671 | |
1672 | # Parses and returns items, skipping whitespaces, insignificant | |
1673 | # symbols, and comments when allowed, and stopping at the first | |
1674 | # unknown character. | |
320644e2 PP |
1675 | # |
1676 | # Accepts and registers macro definitions if `accept_macro_defs` | |
1677 | # is `True`. | |
1678 | def _parse_items(self, accept_macro_defs: bool = False) -> List[_Item]: | |
71aaa3f7 PP |
1679 | items = [] # type: List[_Item] |
1680 | ||
1681 | while self._isnt_done(): | |
1ca7b5e1 PP |
1682 | # Try to append item |
1683 | if not self._try_append_item(items): | |
320644e2 PP |
1684 | if accept_macro_defs and self._try_parse_macro_def(): |
1685 | continue | |
1686 | ||
1ca7b5e1 PP |
1687 | # Unknown at this point |
1688 | break | |
71aaa3f7 PP |
1689 | |
1690 | return items | |
1691 | ||
1692 | # Parses the whole Normand input, setting `self._res` to the main | |
1693 | # group item on success. | |
1694 | def _parse(self): | |
1695 | if len(self._normand.strip()) == 0: | |
1696 | # Special case to make sure there's something to consume | |
1697 | self._res = _Group([], self._text_loc) | |
1698 | return | |
1699 | ||
1700 | # Parse first level items | |
320644e2 | 1701 | items = self._parse_items(True) |
71aaa3f7 PP |
1702 | |
1703 | # Make sure there's nothing left | |
1704 | self._skip_ws_and_comments() | |
1705 | ||
1706 | if self._isnt_done(): | |
1707 | self._raise_error( | |
1708 | "Unexpected character `{}`".format(self._normand[self._at]) | |
1709 | ) | |
1710 | ||
1711 | # Set main group item | |
1712 | self._res = _Group(items, self._text_loc) | |
1713 | ||
1714 | ||
1715 | # The return type of parse(). | |
1716 | class ParseResult: | |
1717 | @classmethod | |
1718 | def _create( | |
1719 | cls, | |
1720 | data: bytearray, | |
e57a18e1 PP |
1721 | variables: VariablesT, |
1722 | labels: LabelsT, | |
71aaa3f7 PP |
1723 | offset: int, |
1724 | bo: Optional[ByteOrder], | |
1725 | ): | |
1726 | self = cls.__new__(cls) | |
1727 | self._init(data, variables, labels, offset, bo) | |
1728 | return self | |
1729 | ||
1730 | def __init__(self, *args, **kwargs): # type: ignore | |
1731 | raise NotImplementedError | |
1732 | ||
1733 | def _init( | |
1734 | self, | |
1735 | data: bytearray, | |
e57a18e1 PP |
1736 | variables: VariablesT, |
1737 | labels: LabelsT, | |
71aaa3f7 PP |
1738 | offset: int, |
1739 | bo: Optional[ByteOrder], | |
1740 | ): | |
1741 | self._data = data | |
1742 | self._vars = variables | |
1743 | self._labels = labels | |
1744 | self._offset = offset | |
1745 | self._bo = bo | |
1746 | ||
1747 | # Generated data. | |
1748 | @property | |
1749 | def data(self): | |
1750 | return self._data | |
1751 | ||
1752 | # Dictionary of updated variable names to their last computed value. | |
1753 | @property | |
1754 | def variables(self): | |
1755 | return self._vars | |
1756 | ||
1757 | # Dictionary of updated main group label names to their computed | |
1758 | # value. | |
1759 | @property | |
1760 | def labels(self): | |
1761 | return self._labels | |
1762 | ||
1763 | # Updated offset. | |
1764 | @property | |
1765 | def offset(self): | |
1766 | return self._offset | |
1767 | ||
1768 | # Updated byte order. | |
1769 | @property | |
1770 | def byte_order(self): | |
1771 | return self._bo | |
1772 | ||
1773 | ||
1774 | # Raises a parse error for the item `item`, creating it using the | |
1775 | # message `msg`. | |
1776 | def _raise_error_for_item(msg: str, item: _Item) -> NoReturn: | |
1777 | _raise_error(msg, item.text_loc) | |
1778 | ||
1779 | ||
1780 | # The `ICITTE` reserved name. | |
1781 | _icitte_name = "ICITTE" | |
1782 | ||
1783 | ||
2adf4336 PP |
1784 | # Base node visitor. |
1785 | # | |
1786 | # Calls the _visit_name() method for each name node which isn't the name | |
1787 | # of a call. | |
1788 | class _NodeVisitor(ast.NodeVisitor): | |
1789 | def __init__(self): | |
71aaa3f7 PP |
1790 | self._parent_is_call = False |
1791 | ||
1792 | def generic_visit(self, node: ast.AST): | |
1793 | if type(node) is ast.Call: | |
1794 | self._parent_is_call = True | |
1795 | elif type(node) is ast.Name and not self._parent_is_call: | |
2adf4336 | 1796 | self._visit_name(node.id) |
71aaa3f7 PP |
1797 | |
1798 | super().generic_visit(node) | |
1799 | self._parent_is_call = False | |
1800 | ||
2adf4336 PP |
1801 | @abc.abstractmethod |
1802 | def _visit_name(self, name: str): | |
1803 | ... | |
1804 | ||
71aaa3f7 | 1805 | |
2adf4336 PP |
1806 | # Expression validator: validates that all the names within the |
1807 | # expression are allowed. | |
1808 | class _ExprValidator(_NodeVisitor): | |
320644e2 | 1809 | def __init__(self, expr_str: str, text_loc: TextLocation, allowed_names: Set[str]): |
2adf4336 | 1810 | super().__init__() |
320644e2 PP |
1811 | self._expr_str = expr_str |
1812 | self._text_loc = text_loc | |
2adf4336 | 1813 | self._allowed_names = allowed_names |
2adf4336 PP |
1814 | |
1815 | def _visit_name(self, name: str): | |
1816 | # Make sure the name refers to a known and reachable | |
1817 | # variable/label name. | |
e57a18e1 | 1818 | if name != _icitte_name and name not in self._allowed_names: |
2adf4336 | 1819 | msg = "Illegal (unknown or unreachable) variable/label name `{}` in expression `{}`".format( |
320644e2 | 1820 | name, self._expr_str |
2adf4336 PP |
1821 | ) |
1822 | ||
05f81895 | 1823 | allowed_names = self._allowed_names.copy() |
e57a18e1 | 1824 | allowed_names.add(_icitte_name) |
2adf4336 | 1825 | |
05f81895 | 1826 | if len(allowed_names) > 0: |
2adf4336 PP |
1827 | allowed_names_str = ", ".join( |
1828 | sorted(["`{}`".format(name) for name in allowed_names]) | |
1829 | ) | |
1830 | msg += "; the legal names are {{{}}}".format(allowed_names_str) | |
1831 | ||
1832 | _raise_error( | |
1833 | msg, | |
320644e2 | 1834 | self._text_loc, |
2adf4336 PP |
1835 | ) |
1836 | ||
1837 | ||
2adf4336 PP |
1838 | # Generator state. |
1839 | class _GenState: | |
1840 | def __init__( | |
1b8aa84a | 1841 | self, |
e57a18e1 PP |
1842 | variables: VariablesT, |
1843 | labels: LabelsT, | |
1b8aa84a PP |
1844 | offset: int, |
1845 | bo: Optional[ByteOrder], | |
2adf4336 PP |
1846 | ): |
1847 | self.variables = variables.copy() | |
1848 | self.labels = labels.copy() | |
1849 | self.offset = offset | |
1850 | self.bo = bo | |
71aaa3f7 | 1851 | |
320644e2 PP |
1852 | def __repr__(self): |
1853 | return "_GenState({}, {}, {}, {})".format( | |
1854 | repr(self.variables), repr(self.labels), repr(self.offset), repr(self.bo) | |
1855 | ) | |
1856 | ||
1857 | ||
1858 | # Fixed-length number item instance. | |
1859 | class _FlNumItemInst: | |
f5dcb24c PP |
1860 | def __init__( |
1861 | self, | |
1862 | item: _FlNum, | |
1863 | offset_in_data: int, | |
1864 | state: _GenState, | |
1865 | parse_error_msgs: List[ParseErrorMessage], | |
1866 | ): | |
320644e2 PP |
1867 | self._item = item |
1868 | self._offset_in_data = offset_in_data | |
1869 | self._state = state | |
f5dcb24c | 1870 | self._parse_error_msgs = parse_error_msgs |
320644e2 PP |
1871 | |
1872 | @property | |
1873 | def item(self): | |
1874 | return self._item | |
1875 | ||
1876 | @property | |
1877 | def offset_in_data(self): | |
1878 | return self._offset_in_data | |
1879 | ||
1880 | @property | |
1881 | def state(self): | |
1882 | return self._state | |
1883 | ||
f5dcb24c PP |
1884 | @property |
1885 | def parse_error_msgs(self): | |
1886 | return self._parse_error_msgs | |
1887 | ||
71aaa3f7 | 1888 | |
2adf4336 | 1889 | # Generator of data and final state from a group item. |
71aaa3f7 PP |
1890 | # |
1891 | # Generation happens in memory at construction time. After building, use | |
1892 | # the `data`, `variables`, `labels`, `offset`, and `bo` properties to | |
1893 | # get the resulting context. | |
2adf4336 PP |
1894 | # |
1895 | # The steps of generation are: | |
1896 | # | |
320644e2 PP |
1897 | # 1. Handle each item in prefix order. |
1898 | # | |
1899 | # The handlers append bytes to `self._data` and update some current | |
1900 | # state object (`_GenState` instance). | |
1901 | # | |
1902 | # When handling a fixed-length number item, try to evaluate its | |
1903 | # expression using the current state. If this fails, then it might be | |
1904 | # because the expression refers to a "future" label: save the current | |
1905 | # offset in `self._data` (generated data) and a snapshot of the | |
1906 | # current state within `self._fl_num_item_insts` (`_FlNumItemInst` | |
f5dcb24c PP |
1907 | # object). _gen_fl_num_item_insts() will deal with this later. A |
1908 | # `_FlNumItemInst` instance also contains a snapshot of the current | |
1909 | # parsing error messages (`self._parse_error_msgs`) which need to be | |
1910 | # taken into account when handling the instance later. | |
2adf4336 | 1911 | # |
320644e2 PP |
1912 | # When handling the items of a group, keep a map of immediate label |
1913 | # names to their offset. Then, after having processed all the items, | |
1914 | # update the relevant saved state snapshots in | |
1915 | # `self._fl_num_item_insts` with those immediate label values. | |
1916 | # _gen_fl_num_item_insts() will deal with this later. | |
2adf4336 | 1917 | # |
320644e2 PP |
1918 | # 2. Handle all the fixed-length number item instances of which the |
1919 | # expression evaluation failed before. | |
2adf4336 | 1920 | # |
320644e2 PP |
1921 | # At this point, `self._fl_num_item_insts` contains everything that's |
1922 | # needed to evaluate the expressions, including the values of | |
1923 | # "future" labels from the point of view of some fixed-length number | |
1924 | # item instance. | |
2adf4336 | 1925 | # |
f5dcb24c PP |
1926 | # If an evaluation fails at this point, then it's a user error. Add |
1927 | # to the parsing error all the saved parsing error messages of the | |
1928 | # instance. Those additional messages add precious context to the | |
1929 | # error. | |
71aaa3f7 PP |
1930 | class _Gen: |
1931 | def __init__( | |
1932 | self, | |
1933 | group: _Group, | |
320644e2 | 1934 | macro_defs: _MacroDefsT, |
e57a18e1 PP |
1935 | variables: VariablesT, |
1936 | labels: LabelsT, | |
71aaa3f7 PP |
1937 | offset: int, |
1938 | bo: Optional[ByteOrder], | |
1939 | ): | |
320644e2 PP |
1940 | self._macro_defs = macro_defs |
1941 | self._fl_num_item_insts = [] # type: List[_FlNumItemInst] | |
f5dcb24c | 1942 | self._parse_error_msgs = [] # type: List[ParseErrorMessage] |
2adf4336 | 1943 | self._gen(group, _GenState(variables, labels, offset, bo)) |
71aaa3f7 PP |
1944 | |
1945 | # Generated bytes. | |
1946 | @property | |
1947 | def data(self): | |
1948 | return self._data | |
1949 | ||
1950 | # Updated variables. | |
1951 | @property | |
1952 | def variables(self): | |
2adf4336 | 1953 | return self._final_state.variables |
71aaa3f7 PP |
1954 | |
1955 | # Updated main group labels. | |
1956 | @property | |
1957 | def labels(self): | |
2adf4336 | 1958 | return self._final_state.labels |
71aaa3f7 PP |
1959 | |
1960 | # Updated offset. | |
1961 | @property | |
1962 | def offset(self): | |
2adf4336 | 1963 | return self._final_state.offset |
71aaa3f7 PP |
1964 | |
1965 | # Updated byte order. | |
1966 | @property | |
1967 | def bo(self): | |
2adf4336 PP |
1968 | return self._final_state.bo |
1969 | ||
320644e2 PP |
1970 | # Evaluates the expression `expr` of which the original string is |
1971 | # `expr_str` at the location `text_loc` considering the current | |
2adf4336 PP |
1972 | # generation state `state`. |
1973 | # | |
269f6eb3 PP |
1974 | # If `allow_float` is `True`, then the type of the result may be |
1975 | # `float` too. | |
2adf4336 | 1976 | @staticmethod |
320644e2 PP |
1977 | def _eval_expr( |
1978 | expr_str: str, | |
1979 | expr: ast.Expression, | |
1980 | text_loc: TextLocation, | |
269f6eb3 | 1981 | state: _GenState, |
269f6eb3 PP |
1982 | allow_float: bool = False, |
1983 | ): | |
e57a18e1 PP |
1984 | syms = {} # type: VariablesT |
1985 | syms.update(state.labels) | |
71aaa3f7 | 1986 | |
e57a18e1 PP |
1987 | # Set the `ICITTE` name to the current offset |
1988 | syms[_icitte_name] = state.offset | |
71aaa3f7 PP |
1989 | |
1990 | # Add the current variables | |
2adf4336 | 1991 | syms.update(state.variables) |
71aaa3f7 PP |
1992 | |
1993 | # Validate the node and its children | |
320644e2 | 1994 | _ExprValidator(expr_str, text_loc, set(syms.keys())).visit(expr) |
71aaa3f7 PP |
1995 | |
1996 | # Compile and evaluate expression node | |
1997 | try: | |
320644e2 | 1998 | val = eval(compile(expr, "", "eval"), None, syms) |
71aaa3f7 | 1999 | except Exception as exc: |
320644e2 PP |
2000 | _raise_error( |
2001 | "Failed to evaluate expression `{}`: {}".format(expr_str, exc), | |
2002 | text_loc, | |
71aaa3f7 PP |
2003 | ) |
2004 | ||
27d52a19 PP |
2005 | # Convert `bool` result type to `int` to normalize |
2006 | if type(val) is bool: | |
2007 | val = int(val) | |
2008 | ||
269f6eb3 PP |
2009 | # Validate result type |
2010 | expected_types = {int} # type: Set[type] | |
2011 | type_msg = "`int`" | |
2012 | ||
2013 | if allow_float: | |
2014 | expected_types.add(float) | |
2015 | type_msg += " or `float`" | |
2016 | ||
2017 | if type(val) not in expected_types: | |
320644e2 | 2018 | _raise_error( |
269f6eb3 | 2019 | "Invalid expression `{}`: expecting result type {}, not `{}`".format( |
320644e2 | 2020 | expr_str, type_msg, type(val).__name__ |
71aaa3f7 | 2021 | ), |
320644e2 | 2022 | text_loc, |
71aaa3f7 PP |
2023 | ) |
2024 | ||
2025 | return val | |
2026 | ||
320644e2 PP |
2027 | # Evaluates the expression of `item` considering the current |
2028 | # generation state `state`. | |
2029 | # | |
2030 | # If `allow_float` is `True`, then the type of the result may be | |
2031 | # `float` too. | |
2032 | @staticmethod | |
2033 | def _eval_item_expr( | |
25ca454b | 2034 | item: Union[_FlNum, _Leb128Int, _FillUntil, _VarAssign, _Rep, _Cond], |
320644e2 PP |
2035 | state: _GenState, |
2036 | allow_float: bool = False, | |
2037 | ): | |
2038 | return _Gen._eval_expr( | |
2039 | item.expr_str, item.expr, item.text_loc, state, allow_float | |
2040 | ) | |
2041 | ||
2042 | # Handles the byte item `item`. | |
2043 | def _handle_byte_item(self, item: _Byte, state: _GenState): | |
2044 | self._data.append(item.val) | |
2045 | state.offset += item.size | |
2046 | ||
2047 | # Handles the string item `item`. | |
2048 | def _handle_str_item(self, item: _Str, state: _GenState): | |
2049 | self._data += item.data | |
2050 | state.offset += item.size | |
2051 | ||
2052 | # Handles the byte order setting item `item`. | |
2053 | def _handle_set_bo_item(self, item: _SetBo, state: _GenState): | |
2054 | # Update current byte order | |
2055 | state.bo = item.bo | |
2056 | ||
2057 | # Handles the variable assignment item `item`. | |
2058 | def _handle_var_assign_item(self, item: _VarAssign, state: _GenState): | |
2059 | # Update variable | |
2060 | state.variables[item.name] = self._eval_item_expr(item, state, True) | |
2061 | ||
2062 | # Handles the fixed-length number item `item`. | |
2063 | def _handle_fl_num_item(self, item: _FlNum, state: _GenState): | |
2064 | # Validate current byte order | |
2065 | if state.bo is None and item.len > 8: | |
2066 | _raise_error_for_item( | |
2067 | "Current byte order isn't defined at first fixed-length number (`{}`) to encode on more than 8 bits".format( | |
2068 | item.expr_str | |
2069 | ), | |
2070 | item, | |
2071 | ) | |
2072 | ||
2073 | # Try an immediate evaluation. If it fails, then keep everything | |
2074 | # needed to (try to) generate the bytes of this item later. | |
2075 | try: | |
2076 | data = self._gen_fl_num_item_inst_data(item, state) | |
2077 | except Exception: | |
2078 | self._fl_num_item_insts.append( | |
f5dcb24c PP |
2079 | _FlNumItemInst( |
2080 | item, | |
2081 | len(self._data), | |
2082 | copy.deepcopy(state), | |
2083 | copy.deepcopy(self._parse_error_msgs), | |
2084 | ) | |
320644e2 PP |
2085 | ) |
2086 | ||
2087 | # Reserve space in `self._data` for this instance | |
2088 | data = bytes([0] * (item.len // 8)) | |
2089 | ||
2090 | # Append bytes | |
2091 | self._data += data | |
2092 | ||
2093 | # Update offset | |
2094 | state.offset += len(data) | |
2095 | ||
05f81895 PP |
2096 | # Returns the size, in bytes, required to encode the value `val` |
2097 | # with LEB128 (signed version if `is_signed` is `True`). | |
2098 | @staticmethod | |
2099 | def _leb128_size_for_val(val: int, is_signed: bool): | |
2100 | if val < 0: | |
2101 | # Equivalent upper bound. | |
2102 | # | |
2103 | # For example, if `val` is -128, then the full integer for | |
2104 | # this number of bits would be [-128, 127]. | |
2105 | val = -val - 1 | |
2106 | ||
2107 | # Number of bits (add one for the sign if needed) | |
2108 | bits = val.bit_length() + int(is_signed) | |
2109 | ||
2110 | if bits == 0: | |
2111 | bits = 1 | |
2112 | ||
2113 | # Seven bits per byte | |
2114 | return math.ceil(bits / 7) | |
2115 | ||
320644e2 PP |
2116 | # Handles the LEB128 integer item `item`. |
2117 | def _handle_leb128_int_item(self, item: _Leb128Int, state: _GenState): | |
2118 | # Compute value | |
2119 | val = self._eval_item_expr(item, state, False) | |
676f6189 | 2120 | |
320644e2 PP |
2121 | # Size in bytes |
2122 | size = self._leb128_size_for_val(val, type(item) is _SLeb128Int) | |
05f81895 | 2123 | |
320644e2 PP |
2124 | # For each byte |
2125 | for _ in range(size): | |
2126 | # Seven LSBs, MSB of the byte set (continue) | |
2127 | self._data.append((val & 0x7F) | 0x80) | |
2128 | val >>= 7 | |
2adf4336 | 2129 | |
320644e2 PP |
2130 | # Clear MSB of last byte (stop) |
2131 | self._data[-1] &= ~0x80 | |
2adf4336 | 2132 | |
320644e2 PP |
2133 | # Update offset |
2134 | state.offset += size | |
27d52a19 | 2135 | |
320644e2 PP |
2136 | # Handles the group item `item`, removing the immediate labels from |
2137 | # `state` at the end if `remove_immediate_labels` is `True`. | |
2138 | def _handle_group_item( | |
2139 | self, item: _Group, state: _GenState, remove_immediate_labels: bool = True | |
2140 | ): | |
2141 | first_fl_num_item_inst_index = len(self._fl_num_item_insts) | |
2142 | immediate_labels = {} # type: LabelsT | |
27d52a19 | 2143 | |
320644e2 PP |
2144 | # Handle each item |
2145 | for subitem in item.items: | |
2146 | if type(subitem) is _Label: | |
2147 | # Add to local immediate labels | |
2148 | immediate_labels[subitem.name] = state.offset | |
2adf4336 | 2149 | |
320644e2 | 2150 | self._handle_item(subitem, state) |
2adf4336 | 2151 | |
320644e2 PP |
2152 | # Remove immediate labels from current state if needed |
2153 | if remove_immediate_labels: | |
2154 | for name in immediate_labels: | |
2155 | del state.labels[name] | |
2adf4336 | 2156 | |
320644e2 PP |
2157 | # Add all immediate labels to all state snapshots since |
2158 | # `first_fl_num_item_inst_index`. | |
2159 | for inst in self._fl_num_item_insts[first_fl_num_item_inst_index:]: | |
2160 | inst.state.labels.update(immediate_labels) | |
2adf4336 | 2161 | |
320644e2 PP |
2162 | # Handles the repetition item `item`. |
2163 | def _handle_rep_item(self, item: _Rep, state: _GenState): | |
2164 | # Compute the repetition count | |
2165 | mul = _Gen._eval_item_expr(item, state) | |
05f81895 | 2166 | |
320644e2 PP |
2167 | # Validate result |
2168 | if mul < 0: | |
2169 | _raise_error_for_item( | |
2170 | "Invalid expression `{}`: unexpected negative result {:,}".format( | |
2171 | item.expr_str, mul | |
2172 | ), | |
2173 | item, | |
2174 | ) | |
2adf4336 | 2175 | |
320644e2 PP |
2176 | # Generate item data `mul` times |
2177 | for _ in range(mul): | |
2178 | self._handle_item(item.item, state) | |
2adf4336 | 2179 | |
320644e2 | 2180 | # Handles the conditional item `item`. |
12b5dbc0 | 2181 | def _handle_cond_item(self, item: _Cond, state: _GenState): |
320644e2 PP |
2182 | # Compute the conditional value |
2183 | val = _Gen._eval_item_expr(item, state) | |
2adf4336 | 2184 | |
320644e2 PP |
2185 | # Generate item data if needed |
2186 | if val: | |
12b5dbc0 PP |
2187 | self._handle_item(item.true_item, state) |
2188 | else: | |
2189 | self._handle_item(item.false_item, state) | |
2adf4336 | 2190 | |
320644e2 PP |
2191 | # Evaluates the parameters of the macro expansion item `item` |
2192 | # considering the initial state `init_state` and returns a new state | |
2193 | # to handle the items of the macro. | |
2194 | def _eval_macro_exp_params(self, item: _MacroExp, init_state: _GenState): | |
2195 | # New state | |
2196 | exp_state = _GenState({}, {}, init_state.offset, init_state.bo) | |
2adf4336 | 2197 | |
320644e2 PP |
2198 | # Evaluate the parameter expressions |
2199 | macro_def = self._macro_defs[item.name] | |
2adf4336 | 2200 | |
320644e2 PP |
2201 | for param_name, param in zip(macro_def.param_names, item.params): |
2202 | exp_state.variables[param_name] = _Gen._eval_expr( | |
2203 | param.expr_str, param.expr, param.text_loc, init_state, True | |
2204 | ) | |
2adf4336 | 2205 | |
320644e2 | 2206 | return exp_state |
2adf4336 | 2207 | |
320644e2 PP |
2208 | # Handles the macro expansion item `item`. |
2209 | def _handle_macro_exp_item(self, item: _MacroExp, state: _GenState): | |
f5dcb24c | 2210 | parse_error_msg_text = "While expanding the macro `{}`:".format(item.name) |
27d52a19 | 2211 | |
f5dcb24c PP |
2212 | try: |
2213 | # New state | |
2214 | exp_state = self._eval_macro_exp_params(item, state) | |
2215 | ||
2216 | # Process the contained group | |
2217 | init_data_size = len(self._data) | |
2218 | parse_error_msg = ( | |
2219 | ParseErrorMessage._create( # pyright: ignore[reportPrivateUsage] | |
2220 | parse_error_msg_text, item.text_loc | |
2221 | ) | |
2222 | ) | |
2223 | self._parse_error_msgs.append(parse_error_msg) | |
2224 | self._handle_item(self._macro_defs[item.name].group, exp_state) | |
2225 | self._parse_error_msgs.pop() | |
2226 | except ParseError as exc: | |
2227 | _augment_error(exc, parse_error_msg_text, item.text_loc) | |
27d52a19 | 2228 | |
320644e2 PP |
2229 | # Update state offset and return |
2230 | state.offset += len(self._data) - init_data_size | |
676f6189 | 2231 | |
320644e2 PP |
2232 | # Handles the offset setting item `item`. |
2233 | def _handle_set_offset_item(self, item: _SetOffset, state: _GenState): | |
676f6189 | 2234 | state.offset = item.val |
2adf4336 | 2235 | |
25ca454b | 2236 | # Handles the offset alignment item `item` (adds padding). |
320644e2 PP |
2237 | def _handle_align_offset_item(self, item: _AlignOffset, state: _GenState): |
2238 | init_offset = state.offset | |
2239 | align_bytes = item.val // 8 | |
2240 | state.offset = (state.offset + align_bytes - 1) // align_bytes * align_bytes | |
2241 | self._data += bytes([item.pad_val] * (state.offset - init_offset)) | |
2adf4336 | 2242 | |
25ca454b PP |
2243 | # Handles the filling item `item` (adds padding). |
2244 | def _handle_fill_until_item(self, item: _FillUntil, state: _GenState): | |
2245 | # Compute the new offset | |
2246 | new_offset = _Gen._eval_item_expr(item, state) | |
2247 | ||
2248 | # Validate the new offset | |
2249 | if new_offset < state.offset: | |
2250 | _raise_error_for_item( | |
2251 | "Invalid expression `{}`: new offset {:,} is less than current offset {:,}".format( | |
2252 | item.expr_str, new_offset, state.offset | |
2253 | ), | |
2254 | item, | |
2255 | ) | |
2256 | ||
2257 | # Fill | |
2258 | self._data += bytes([item.pad_val] * (new_offset - state.offset)) | |
2259 | ||
2260 | # Update offset | |
2261 | state.offset = new_offset | |
2262 | ||
320644e2 PP |
2263 | # Handles the label item `item`. |
2264 | def _handle_label_item(self, item: _Label, state: _GenState): | |
2265 | state.labels[item.name] = state.offset | |
2adf4336 | 2266 | |
320644e2 PP |
2267 | # Handles the item `item`, returning the updated next repetition |
2268 | # instance. | |
2269 | def _handle_item(self, item: _Item, state: _GenState): | |
2270 | return self._item_handlers[type(item)](item, state) | |
71aaa3f7 | 2271 | |
320644e2 PP |
2272 | # Generates the data for a fixed-length integer item instance having |
2273 | # the value `val` and returns it. | |
2274 | def _gen_fl_int_item_inst_data(self, val: int, item: _FlNum, state: _GenState): | |
71aaa3f7 PP |
2275 | # Validate range |
2276 | if val < -(2 ** (item.len - 1)) or val > 2**item.len - 1: | |
2277 | _raise_error_for_item( | |
320644e2 PP |
2278 | "Value {:,} is outside the {}-bit range when evaluating expression `{}`".format( |
2279 | val, item.len, item.expr_str | |
71aaa3f7 PP |
2280 | ), |
2281 | item, | |
2282 | ) | |
2283 | ||
2284 | # Encode result on 64 bits (to extend the sign bit whatever the | |
2285 | # value of `item.len`). | |
71aaa3f7 PP |
2286 | data = struct.pack( |
2287 | "{}{}".format( | |
2adf4336 | 2288 | ">" if state.bo in (None, ByteOrder.BE) else "<", |
71aaa3f7 PP |
2289 | "Q" if val >= 0 else "q", |
2290 | ), | |
2291 | val, | |
2292 | ) | |
2293 | ||
2294 | # Keep only the requested length | |
2295 | len_bytes = item.len // 8 | |
2296 | ||
2adf4336 | 2297 | if state.bo in (None, ByteOrder.BE): |
71aaa3f7 PP |
2298 | # Big endian: keep last bytes |
2299 | data = data[-len_bytes:] | |
2300 | else: | |
2301 | # Little endian: keep first bytes | |
2adf4336 | 2302 | assert state.bo == ByteOrder.LE |
71aaa3f7 PP |
2303 | data = data[:len_bytes] |
2304 | ||
320644e2 PP |
2305 | # Return data |
2306 | return data | |
269f6eb3 | 2307 | |
320644e2 PP |
2308 | # Generates the data for a fixed-length floating point number item |
2309 | # instance having the value `val` and returns it. | |
2310 | def _gen_fl_float_item_inst_data(self, val: float, item: _FlNum, state: _GenState): | |
269f6eb3 PP |
2311 | # Validate length |
2312 | if item.len not in (32, 64): | |
2313 | _raise_error_for_item( | |
2314 | "Invalid {}-bit length for a fixed-length floating point number (value {:,})".format( | |
2315 | item.len, val | |
2316 | ), | |
2317 | item, | |
2318 | ) | |
2319 | ||
320644e2 PP |
2320 | # Encode and return result |
2321 | return struct.pack( | |
269f6eb3 PP |
2322 | "{}{}".format( |
2323 | ">" if state.bo in (None, ByteOrder.BE) else "<", | |
2324 | "f" if item.len == 32 else "d", | |
2325 | ), | |
2326 | val, | |
2327 | ) | |
2328 | ||
320644e2 PP |
2329 | # Generates the data for a fixed-length number item instance and |
2330 | # returns it. | |
2331 | def _gen_fl_num_item_inst_data(self, item: _FlNum, state: _GenState): | |
269f6eb3 | 2332 | # Compute value |
e57a18e1 | 2333 | val = self._eval_item_expr(item, state, True) |
269f6eb3 | 2334 | |
269f6eb3 PP |
2335 | # Handle depending on type |
2336 | if type(val) is int: | |
320644e2 | 2337 | return self._gen_fl_int_item_inst_data(val, item, state) |
269f6eb3 PP |
2338 | else: |
2339 | assert type(val) is float | |
320644e2 | 2340 | return self._gen_fl_float_item_inst_data(val, item, state) |
05f81895 | 2341 | |
320644e2 PP |
2342 | # Generates the data for all the fixed-length number item instances |
2343 | # and writes it at the correct offset within `self._data`. | |
2344 | def _gen_fl_num_item_insts(self): | |
2345 | for inst in self._fl_num_item_insts: | |
2346 | # Generate bytes | |
f5dcb24c PP |
2347 | try: |
2348 | data = self._gen_fl_num_item_inst_data(inst.item, inst.state) | |
2349 | except ParseError as exc: | |
2350 | # Add all the saved parse error messages for this | |
2351 | # instance. | |
2352 | for msg in reversed(inst.parse_error_msgs): | |
2353 | _add_error_msg(exc, msg.text, msg.text_location) | |
2354 | ||
2355 | raise | |
05f81895 | 2356 | |
320644e2 PP |
2357 | # Insert bytes into `self._data` |
2358 | self._data[inst.offset_in_data : inst.offset_in_data + len(data)] = data | |
2adf4336 PP |
2359 | |
2360 | # Generates the data (`self._data`) and final state | |
2361 | # (`self._final_state`) from `group` and the initial state `state`. | |
2362 | def _gen(self, group: _Group, state: _GenState): | |
71aaa3f7 PP |
2363 | # Initial state |
2364 | self._data = bytearray() | |
71aaa3f7 PP |
2365 | |
2366 | # Item handlers | |
2367 | self._item_handlers = { | |
676f6189 | 2368 | _AlignOffset: self._handle_align_offset_item, |
71aaa3f7 | 2369 | _Byte: self._handle_byte_item, |
27d52a19 | 2370 | _Cond: self._handle_cond_item, |
25ca454b | 2371 | _FillUntil: self._handle_fill_until_item, |
269f6eb3 | 2372 | _FlNum: self._handle_fl_num_item, |
71aaa3f7 | 2373 | _Group: self._handle_group_item, |
2adf4336 | 2374 | _Label: self._handle_label_item, |
320644e2 | 2375 | _MacroExp: self._handle_macro_exp_item, |
71aaa3f7 | 2376 | _Rep: self._handle_rep_item, |
2adf4336 PP |
2377 | _SetBo: self._handle_set_bo_item, |
2378 | _SetOffset: self._handle_set_offset_item, | |
05f81895 | 2379 | _SLeb128Int: self._handle_leb128_int_item, |
2adf4336 | 2380 | _Str: self._handle_str_item, |
05f81895 | 2381 | _ULeb128Int: self._handle_leb128_int_item, |
2adf4336 | 2382 | _VarAssign: self._handle_var_assign_item, |
320644e2 | 2383 | } # type: Dict[type, Callable[[Any, _GenState], None]] |
2adf4336 PP |
2384 | |
2385 | # Handle the group item, _not_ removing the immediate labels | |
2386 | # because the `labels` property offers them. | |
320644e2 | 2387 | self._handle_group_item(group, state, False) |
2adf4336 PP |
2388 | |
2389 | # This is actually the final state | |
2390 | self._final_state = state | |
71aaa3f7 | 2391 | |
320644e2 PP |
2392 | # Generate all the fixed-length number bytes now that we know |
2393 | # their full state | |
2394 | self._gen_fl_num_item_insts() | |
2395 | ||
71aaa3f7 PP |
2396 | |
2397 | # Returns a `ParseResult` instance containing the bytes encoded by the | |
2398 | # input string `normand`. | |
2399 | # | |
2400 | # `init_variables` is a dictionary of initial variable names (valid | |
2401 | # Python names) to integral values. A variable name must not be the | |
2402 | # reserved name `ICITTE`. | |
2403 | # | |
2404 | # `init_labels` is a dictionary of initial label names (valid Python | |
2405 | # names) to integral values. A label name must not be the reserved name | |
2406 | # `ICITTE`. | |
2407 | # | |
2408 | # `init_offset` is the initial offset. | |
2409 | # | |
2410 | # `init_byte_order` is the initial byte order. | |
2411 | # | |
2412 | # Raises `ParseError` on any parsing error. | |
2413 | def parse( | |
2414 | normand: str, | |
e57a18e1 PP |
2415 | init_variables: Optional[VariablesT] = None, |
2416 | init_labels: Optional[LabelsT] = None, | |
71aaa3f7 PP |
2417 | init_offset: int = 0, |
2418 | init_byte_order: Optional[ByteOrder] = None, | |
2419 | ): | |
2420 | if init_variables is None: | |
2421 | init_variables = {} | |
2422 | ||
2423 | if init_labels is None: | |
2424 | init_labels = {} | |
2425 | ||
320644e2 | 2426 | parser = _Parser(normand, init_variables, init_labels) |
71aaa3f7 | 2427 | gen = _Gen( |
320644e2 PP |
2428 | parser.res, |
2429 | parser.macro_defs, | |
71aaa3f7 PP |
2430 | init_variables, |
2431 | init_labels, | |
2432 | init_offset, | |
2433 | init_byte_order, | |
2434 | ) | |
2435 | return ParseResult._create( # pyright: ignore[reportPrivateUsage] | |
2436 | gen.data, gen.variables, gen.labels, gen.offset, gen.bo | |
2437 | ) | |
2438 | ||
2439 | ||
f5dcb24c PP |
2440 | # Raises a command-line error with the message `msg`. |
2441 | def _raise_cli_error(msg: str) -> NoReturn: | |
2442 | raise RuntimeError("Command-line error: {}".format(msg)) | |
2443 | ||
2444 | ||
2445 | # Returns a dictionary of string to integers from the list of strings | |
2446 | # `args` containing `NAME=VAL` entries. | |
2447 | def _dict_from_arg(args: Optional[List[str]]): | |
2448 | d = {} # type: LabelsT | |
2449 | ||
2450 | if args is None: | |
2451 | return d | |
2452 | ||
2453 | for arg in args: | |
2454 | m = re.match(r"({})=(\d+)$".format(_py_name_pat.pattern), arg) | |
2455 | ||
2456 | if m is None: | |
2457 | _raise_cli_error("Invalid assignment {}".format(arg)) | |
2458 | ||
2459 | d[m.group(1)] = int(m.group(2)) | |
2460 | ||
2461 | return d | |
2462 | ||
2463 | ||
2464 | # Parses the command-line arguments and returns, in this order: | |
2465 | # | |
2466 | # 1. The input file path, or `None` if none. | |
2467 | # 2. The Normand input text. | |
2468 | # 3. The initial offset. | |
2469 | # 4. The initial byte order. | |
2470 | # 5. The initial variables. | |
2471 | # 6. The initial labels. | |
71aaa3f7 PP |
2472 | def _parse_cli_args(): |
2473 | import argparse | |
2474 | ||
2475 | # Build parser | |
2476 | ap = argparse.ArgumentParser() | |
2477 | ap.add_argument( | |
2478 | "--offset", | |
2479 | metavar="OFFSET", | |
2480 | action="store", | |
2481 | type=int, | |
2482 | default=0, | |
2483 | help="initial offset (positive)", | |
2484 | ) | |
2485 | ap.add_argument( | |
2486 | "-b", | |
2487 | "--byte-order", | |
2488 | metavar="BO", | |
2489 | choices=["be", "le"], | |
2490 | type=str, | |
2491 | help="initial byte order (`be` or `le`)", | |
2492 | ) | |
2493 | ap.add_argument( | |
2494 | "--var", | |
2495 | metavar="NAME=VAL", | |
2496 | action="append", | |
2497 | help="add an initial variable (may be repeated)", | |
2498 | ) | |
2499 | ap.add_argument( | |
2500 | "-l", | |
2501 | "--label", | |
2502 | metavar="NAME=VAL", | |
2503 | action="append", | |
2504 | help="add an initial label (may be repeated)", | |
2505 | ) | |
2506 | ap.add_argument( | |
2507 | "--version", action="version", version="Normand {}".format(__version__) | |
2508 | ) | |
2509 | ap.add_argument( | |
2510 | "path", | |
2511 | metavar="PATH", | |
2512 | action="store", | |
2513 | nargs="?", | |
2514 | help="input path (none means standard input)", | |
2515 | ) | |
2516 | ||
2517 | # Parse | |
f5dcb24c | 2518 | args = ap.parse_args() |
71aaa3f7 PP |
2519 | |
2520 | # Read input | |
2521 | if args.path is None: | |
2522 | normand = sys.stdin.read() | |
2523 | else: | |
2524 | with open(args.path) as f: | |
2525 | normand = f.read() | |
2526 | ||
2527 | # Variables and labels | |
e57a18e1 | 2528 | variables = typing.cast(VariablesT, _dict_from_arg(args.var)) |
71aaa3f7 PP |
2529 | labels = _dict_from_arg(args.label) |
2530 | ||
2531 | # Validate offset | |
2532 | if args.offset < 0: | |
2533 | _raise_cli_error("Invalid negative offset {}") | |
2534 | ||
2535 | # Validate and set byte order | |
2536 | bo = None # type: Optional[ByteOrder] | |
2537 | ||
2538 | if args.byte_order is not None: | |
2539 | if args.byte_order == "be": | |
2540 | bo = ByteOrder.BE | |
2541 | else: | |
2542 | assert args.byte_order == "le" | |
2543 | bo = ByteOrder.LE | |
2544 | ||
f5dcb24c PP |
2545 | # Return input and initial state |
2546 | return args.path, normand, args.offset, bo, variables, labels | |
71aaa3f7 | 2547 | |
71aaa3f7 | 2548 | |
f5dcb24c PP |
2549 | # CLI entry point without exception handling. |
2550 | def _run_cli_with_args( | |
2551 | normand: str, | |
2552 | offset: int, | |
2553 | bo: Optional[ByteOrder], | |
2554 | variables: VariablesT, | |
2555 | labels: LabelsT, | |
2556 | ): | |
2557 | sys.stdout.buffer.write(parse(normand, variables, labels, offset, bo).data) | |
71aaa3f7 PP |
2558 | |
2559 | ||
2560 | # Prints the exception message `msg` and exits with status 1. | |
2561 | def _fail(msg: str) -> NoReturn: | |
2562 | if not msg.endswith("."): | |
2563 | msg += "." | |
2564 | ||
f5dcb24c | 2565 | print(msg.strip(), file=sys.stderr) |
71aaa3f7 PP |
2566 | sys.exit(1) |
2567 | ||
2568 | ||
2569 | # CLI entry point. | |
2570 | def _run_cli(): | |
2571 | try: | |
f5dcb24c PP |
2572 | args = _parse_cli_args() |
2573 | except Exception as exc: | |
2574 | _fail(str(exc)) | |
2575 | ||
2576 | try: | |
2577 | _run_cli_with_args(*args[1:]) | |
2578 | except ParseError as exc: | |
2579 | import os.path | |
2580 | ||
2581 | prefix = "" if args[0] is None else "{}:".format(os.path.abspath(args[0])) | |
2582 | fail_msg = "" | |
2583 | ||
2584 | for msg in reversed(exc.messages): | |
2585 | fail_msg += "{}{}:{} - {}".format( | |
2586 | prefix, | |
2587 | msg.text_location.line_no, | |
2588 | msg.text_location.col_no, | |
2589 | msg.text, | |
2590 | ) | |
2591 | ||
2592 | if fail_msg[-1] not in ".:;": | |
2593 | fail_msg += "." | |
2594 | ||
2595 | fail_msg += "\n" | |
2596 | ||
2597 | _fail(fail_msg.strip()) | |
71aaa3f7 PP |
2598 | except Exception as exc: |
2599 | _fail(str(exc)) | |
2600 | ||
2601 | ||
2602 | if __name__ == "__main__": | |
2603 | _run_cli() |