| 1 |
# SPDX-License-Identifier: MIT |
| 2 |
# SPDX-FileCopyrightText: 2021 Taneli Hukkinen |
| 3 |
# Licensed to PSF under a Contributor Agreement. |
| 4 |
|
| 5 |
from __future__ import annotations |
| 6 |
|
| 7 |
import sys |
| 8 |
from types import MappingProxyType |
| 9 |
|
| 10 |
from ._re import ( |
| 11 |
RE_DATETIME, |
| 12 |
RE_LOCALTIME, |
| 13 |
RE_NUMBER, |
| 14 |
match_to_datetime, |
| 15 |
match_to_localtime, |
| 16 |
match_to_number, |
| 17 |
) |
| 18 |
|
| 19 |
TYPE_CHECKING = False |
| 20 |
if TYPE_CHECKING: |
| 21 |
from collections.abc import Iterable |
| 22 |
from typing import IO, Any, Final |
| 23 |
|
| 24 |
from ._types import Key, ParseFloat, Pos |
| 25 |
|
| 26 |
# Pathologically excessive number of parts in a key runs into quadratic |
| 27 |
# behavior (e.g. in Flags.is_). |
| 28 |
# Even if keys aren't currently parsed using recursion, they name a |
| 29 |
# recursive structure, so it makes sense to limit it using getrecursionlimit() |
| 30 |
# and RecursionError. |
| 31 |
MAX_KEY_PARTS: Final = sys.getrecursionlimit() |
| 32 |
|
| 33 |
ASCII_CTRL = frozenset(chr(i) for i in range(32)) | frozenset(chr(127)) |
| 34 |
|
| 35 |
# Neither of these sets include quotation mark or backslash. They are |
| 36 |
# currently handled as separate cases in the parser functions. |
| 37 |
ILLEGAL_BASIC_STR_CHARS = ASCII_CTRL - frozenset("\t") |
| 38 |
ILLEGAL_MULTILINE_BASIC_STR_CHARS = ASCII_CTRL - frozenset("\t\n") |
| 39 |
|
| 40 |
ILLEGAL_LITERAL_STR_CHARS = ILLEGAL_BASIC_STR_CHARS |
| 41 |
ILLEGAL_MULTILINE_LITERAL_STR_CHARS = ILLEGAL_MULTILINE_BASIC_STR_CHARS |
| 42 |
|
| 43 |
ILLEGAL_COMMENT_CHARS = ILLEGAL_BASIC_STR_CHARS |
| 44 |
|
| 45 |
TOML_WS = frozenset(" \t") |
| 46 |
TOML_WS_AND_NEWLINE = TOML_WS | frozenset("\n") |
| 47 |
BARE_KEY_CHARS = frozenset( |
| 48 |
"abcdefghijklmnopqrstuvwxyz" "ABCDEFGHIJKLMNOPQRSTUVWXYZ" "0123456789" "-_" |
| 49 |
) |
| 50 |
KEY_INITIAL_CHARS = BARE_KEY_CHARS | frozenset("\"'") |
| 51 |
HEXDIGIT_CHARS = frozenset("abcdef" "ABCDEF" "0123456789") |
| 52 |
|
| 53 |
BASIC_STR_ESCAPE_REPLACEMENTS = MappingProxyType( |
| 54 |
{ |
| 55 |
"\\b": "\u0008", # backspace |
| 56 |
"\\t": "\u0009", # tab |
| 57 |
"\\n": "\u000A", # linefeed |
| 58 |
"\\f": "\u000C", # form feed |
| 59 |
"\\r": "\u000D", # carriage return |
| 60 |
'\\"': "\u0022", # quote |
| 61 |
"\\\\": "\u005C", # backslash |
| 62 |
} |
| 63 |
) |
| 64 |
|
| 65 |
|
| 66 |
class DEPRECATED_DEFAULT: |
| 67 |
"""Sentinel to be used as default arg during deprecation |
| 68 |
period of TOMLDecodeError's free-form arguments.""" |
| 69 |
|
| 70 |
|
| 71 |
class TOMLDecodeError(ValueError): |
| 72 |
"""An error raised if a document is not valid TOML. |
| 73 |
|
| 74 |
Adds the following attributes to ValueError: |
| 75 |
msg: The unformatted error message |
| 76 |
doc: The TOML document being parsed |
| 77 |
pos: The index of doc where parsing failed |
| 78 |
lineno: The line corresponding to pos |
| 79 |
colno: The column corresponding to pos |
| 80 |
""" |
| 81 |
|
| 82 |
def __init__( |
| 83 |
self, |
| 84 |
msg: str = DEPRECATED_DEFAULT, # type: ignore[assignment] |
| 85 |
doc: str = DEPRECATED_DEFAULT, # type: ignore[assignment] |
| 86 |
pos: Pos = DEPRECATED_DEFAULT, # type: ignore[assignment] |
| 87 |
*args: Any, |
| 88 |
): |
| 89 |
if ( |
| 90 |
args |
| 91 |
or not isinstance(msg, str) |
| 92 |
or not isinstance(doc, str) |
| 93 |
or not isinstance(pos, int) |
| 94 |
): |
| 95 |
import warnings |
| 96 |
|
| 97 |
warnings.warn( |
| 98 |
"Free-form arguments for TOMLDecodeError are deprecated. " |
| 99 |
"Please set 'msg' (str), 'doc' (str) and 'pos' (int) arguments only.", |
| 100 |
DeprecationWarning, |
| 101 |
stacklevel=2, |
| 102 |
) |
| 103 |
if pos is not DEPRECATED_DEFAULT: # type: ignore[comparison-overlap] |
| 104 |
args = pos, *args |
| 105 |
if doc is not DEPRECATED_DEFAULT: # type: ignore[comparison-overlap] |
| 106 |
args = doc, *args |
| 107 |
if msg is not DEPRECATED_DEFAULT: # type: ignore[comparison-overlap] |
| 108 |
args = msg, *args |
| 109 |
ValueError.__init__(self, *args) |
| 110 |
return |
| 111 |
|
| 112 |
lineno = doc.count("\n", 0, pos) + 1 |
| 113 |
if lineno == 1: |
| 114 |
colno = pos + 1 |
| 115 |
else: |
| 116 |
colno = pos - doc.rindex("\n", 0, pos) |
| 117 |
|
| 118 |
if pos >= len(doc): |
| 119 |
coord_repr = "end of document" |
| 120 |
else: |
| 121 |
coord_repr = f"line {lineno}, column {colno}" |
| 122 |
errmsg = f"{msg} (at {coord_repr})" |
| 123 |
ValueError.__init__(self, errmsg) |
| 124 |
|
| 125 |
self.msg = msg |
| 126 |
self.doc = doc |
| 127 |
self.pos = pos |
| 128 |
self.lineno = lineno |
| 129 |
self.colno = colno |
| 130 |
|
| 131 |
|
| 132 |
def load(fp: IO[bytes], /, *, parse_float: ParseFloat = float) -> dict[str, Any]: |
| 133 |
"""Parse TOML from a binary file object.""" |
| 134 |
b = fp.read() |
| 135 |
try: |
| 136 |
s = b.decode() |
| 137 |
except AttributeError: |
| 138 |
raise TypeError( |
| 139 |
"File must be opened in binary mode, e.g. use `open('foo.toml', 'rb')`" |
| 140 |
) from None |
| 141 |
return loads(s, parse_float=parse_float) |
| 142 |
|
| 143 |
|
| 144 |
def loads(s: str, /, *, parse_float: ParseFloat = float) -> dict[str, Any]: # noqa: C901 |
| 145 |
"""Parse TOML from a string.""" |
| 146 |
|
| 147 |
# The spec allows converting "\r\n" to "\n", even in string |
| 148 |
# literals. Let's do so to simplify parsing. |
| 149 |
try: |
| 150 |
src = s.replace("\r\n", "\n") |
| 151 |
except (AttributeError, TypeError): |
| 152 |
raise TypeError( |
| 153 |
f"Expected str object, not '{type(s).__qualname__}'" |
| 154 |
) from None |
| 155 |
pos = 0 |
| 156 |
out = Output() |
| 157 |
header: Key = () |
| 158 |
parse_float = make_safe_parse_float(parse_float) |
| 159 |
|
| 160 |
# Parse one statement at a time |
| 161 |
# (typically means one line in TOML source) |
| 162 |
while True: |
| 163 |
# 1. Skip line leading whitespace |
| 164 |
pos = skip_chars(src, pos, TOML_WS) |
| 165 |
|
| 166 |
# 2. Parse rules. Expect one of the following: |
| 167 |
# - end of file |
| 168 |
# - end of line |
| 169 |
# - comment |
| 170 |
# - key/value pair |
| 171 |
# - append dict to list (and move to its namespace) |
| 172 |
# - create dict (and move to its namespace) |
| 173 |
# Skip trailing whitespace when applicable. |
| 174 |
try: |
| 175 |
char = src[pos] |
| 176 |
except IndexError: |
| 177 |
break |
| 178 |
if char == "\n": |
| 179 |
pos += 1 |
| 180 |
continue |
| 181 |
if char in KEY_INITIAL_CHARS: |
| 182 |
pos = key_value_rule(src, pos, out, header, parse_float) |
| 183 |
pos = skip_chars(src, pos, TOML_WS) |
| 184 |
elif char == "[": |
| 185 |
try: |
| 186 |
second_char: str | None = src[pos + 1] |
| 187 |
except IndexError: |
| 188 |
second_char = None |
| 189 |
out.flags.finalize_pending() |
| 190 |
if second_char == "[": |
| 191 |
pos, header = create_list_rule(src, pos, out) |
| 192 |
else: |
| 193 |
pos, header = create_dict_rule(src, pos, out) |
| 194 |
pos = skip_chars(src, pos, TOML_WS) |
| 195 |
elif char != "#": |
| 196 |
raise TOMLDecodeError("Invalid statement", src, pos) |
| 197 |
|
| 198 |
# 3. Skip comment |
| 199 |
pos = skip_comment(src, pos) |
| 200 |
|
| 201 |
# 4. Expect end of line or end of file |
| 202 |
try: |
| 203 |
char = src[pos] |
| 204 |
except IndexError: |
| 205 |
break |
| 206 |
if char != "\n": |
| 207 |
raise TOMLDecodeError( |
| 208 |
"Expected newline or end of document after a statement", src, pos |
| 209 |
) |
| 210 |
pos += 1 |
| 211 |
|
| 212 |
return out.data.dict |
| 213 |
|
| 214 |
|
| 215 |
class Flags: |
| 216 |
"""Flags that map to parsed keys/namespaces.""" |
| 217 |
|
| 218 |
# Marks an immutable namespace (inline array or inline table). |
| 219 |
FROZEN = 0 |
| 220 |
# Marks a nest that has been explicitly created and can no longer |
| 221 |
# be opened using the "[table]" syntax. |
| 222 |
EXPLICIT_NEST = 1 |
| 223 |
|
| 224 |
def __init__(self) -> None: |
| 225 |
self._flags: dict[str, dict[Any, Any]] = {} |
| 226 |
self._pending_flags: set[tuple[Key, int]] = set() |
| 227 |
|
| 228 |
def add_pending(self, key: Key, flag: int) -> None: |
| 229 |
self._pending_flags.add((key, flag)) |
| 230 |
|
| 231 |
def finalize_pending(self) -> None: |
| 232 |
for key, flag in self._pending_flags: |
| 233 |
self.set(key, flag, recursive=False) |
| 234 |
self._pending_flags.clear() |
| 235 |
|
| 236 |
def unset_all(self, key: Key) -> None: |
| 237 |
cont = self._flags |
| 238 |
for k in key[:-1]: |
| 239 |
if k not in cont: |
| 240 |
return |
| 241 |
cont = cont[k]["nested"] |
| 242 |
cont.pop(key[-1], None) |
| 243 |
|
| 244 |
def set(self, key: Key, flag: int, *, recursive: bool) -> None: # noqa: A003 |
| 245 |
cont = self._flags |
| 246 |
key_parent, key_stem = key[:-1], key[-1] |
| 247 |
for k in key_parent: |
| 248 |
if k not in cont: |
| 249 |
cont[k] = {"flags": set(), "recursive_flags": set(), "nested": {}} |
| 250 |
cont = cont[k]["nested"] |
| 251 |
if key_stem not in cont: |
| 252 |
cont[key_stem] = {"flags": set(), "recursive_flags": set(), "nested": {}} |
| 253 |
cont[key_stem]["recursive_flags" if recursive else "flags"].add(flag) |
| 254 |
|
| 255 |
def is_(self, key: Key, flag: int) -> bool: |
| 256 |
if not key: |
| 257 |
return False # document root has no flags |
| 258 |
cont = self._flags |
| 259 |
for k in key[:-1]: |
| 260 |
if k not in cont: |
| 261 |
return False |
| 262 |
inner_cont = cont[k] |
| 263 |
if flag in inner_cont["recursive_flags"]: |
| 264 |
return True |
| 265 |
cont = inner_cont["nested"] |
| 266 |
key_stem = key[-1] |
| 267 |
if key_stem in cont: |
| 268 |
cont = cont[key_stem] |
| 269 |
return flag in cont["flags"] or flag in cont["recursive_flags"] |
| 270 |
return False |
| 271 |
|
| 272 |
|
| 273 |
class NestedDict: |
| 274 |
def __init__(self) -> None: |
| 275 |
# The parsed content of the TOML document |
| 276 |
self.dict: dict[str, Any] = {} |
| 277 |
|
| 278 |
def get_or_create_nest( |
| 279 |
self, |
| 280 |
key: Key, |
| 281 |
*, |
| 282 |
access_lists: bool = True, |
| 283 |
) -> dict[str, Any]: |
| 284 |
cont: Any = self.dict |
| 285 |
for k in key: |
| 286 |
if k not in cont: |
| 287 |
cont[k] = {} |
| 288 |
cont = cont[k] |
| 289 |
if access_lists and isinstance(cont, list): |
| 290 |
cont = cont[-1] |
| 291 |
if not isinstance(cont, dict): |
| 292 |
raise KeyError("There is no nest behind this key") |
| 293 |
return cont # type: ignore[no-any-return] |
| 294 |
|
| 295 |
def append_nest_to_list(self, key: Key) -> None: |
| 296 |
cont = self.get_or_create_nest(key[:-1]) |
| 297 |
last_key = key[-1] |
| 298 |
if last_key in cont: |
| 299 |
list_ = cont[last_key] |
| 300 |
if not isinstance(list_, list): |
| 301 |
raise KeyError("An object other than list found behind this key") |
| 302 |
list_.append({}) |
| 303 |
else: |
| 304 |
cont[last_key] = [{}] |
| 305 |
|
| 306 |
|
| 307 |
class Output: |
| 308 |
def __init__(self) -> None: |
| 309 |
self.data = NestedDict() |
| 310 |
self.flags = Flags() |
| 311 |
|
| 312 |
|
| 313 |
def skip_chars(src: str, pos: Pos, chars: Iterable[str]) -> Pos: |
| 314 |
try: |
| 315 |
while src[pos] in chars: |
| 316 |
pos += 1 |
| 317 |
except IndexError: |
| 318 |
pass |
| 319 |
return pos |
| 320 |
|
| 321 |
|
| 322 |
def skip_until( |
| 323 |
src: str, |
| 324 |
pos: Pos, |
| 325 |
expect: str, |
| 326 |
*, |
| 327 |
error_on: frozenset[str], |
| 328 |
error_on_eof: bool, |
| 329 |
) -> Pos: |
| 330 |
try: |
| 331 |
new_pos = src.index(expect, pos) |
| 332 |
except ValueError: |
| 333 |
new_pos = len(src) |
| 334 |
if error_on_eof: |
| 335 |
raise TOMLDecodeError(f"Expected {expect!r}", src, new_pos) from None |
| 336 |
|
| 337 |
if not error_on.isdisjoint(src[pos:new_pos]): |
| 338 |
while src[pos] not in error_on: |
| 339 |
pos += 1 |
| 340 |
raise TOMLDecodeError(f"Found invalid character {src[pos]!r}", src, pos) |
| 341 |
return new_pos |
| 342 |
|
| 343 |
|
| 344 |
def skip_comment(src: str, pos: Pos) -> Pos: |
| 345 |
try: |
| 346 |
char: str | None = src[pos] |
| 347 |
except IndexError: |
| 348 |
char = None |
| 349 |
if char == "#": |
| 350 |
return skip_until( |
| 351 |
src, pos + 1, "\n", error_on=ILLEGAL_COMMENT_CHARS, error_on_eof=False |
| 352 |
) |
| 353 |
return pos |
| 354 |
|
| 355 |
|
| 356 |
def skip_comments_and_array_ws(src: str, pos: Pos) -> Pos: |
| 357 |
while True: |
| 358 |
pos_before_skip = pos |
| 359 |
pos = skip_chars(src, pos, TOML_WS_AND_NEWLINE) |
| 360 |
pos = skip_comment(src, pos) |
| 361 |
if pos == pos_before_skip: |
| 362 |
return pos |
| 363 |
|
| 364 |
|
| 365 |
def create_dict_rule(src: str, pos: Pos, out: Output) -> tuple[Pos, Key]: |
| 366 |
pos += 1 # Skip "[" |
| 367 |
pos = skip_chars(src, pos, TOML_WS) |
| 368 |
pos, key = parse_key(src, pos) |
| 369 |
|
| 370 |
if out.flags.is_(key, Flags.EXPLICIT_NEST) or out.flags.is_(key, Flags.FROZEN): |
| 371 |
raise TOMLDecodeError(f"Cannot declare {key} twice", src, pos) |
| 372 |
out.flags.set(key, Flags.EXPLICIT_NEST, recursive=False) |
| 373 |
try: |
| 374 |
out.data.get_or_create_nest(key) |
| 375 |
except KeyError: |
| 376 |
raise TOMLDecodeError("Cannot overwrite a value", src, pos) from None |
| 377 |
|
| 378 |
if not src.startswith("]", pos): |
| 379 |
raise TOMLDecodeError( |
| 380 |
"Expected ']' at the end of a table declaration", src, pos |
| 381 |
) |
| 382 |
return pos + 1, key |
| 383 |
|
| 384 |
|
| 385 |
def create_list_rule(src: str, pos: Pos, out: Output) -> tuple[Pos, Key]: |
| 386 |
pos += 2 # Skip "[[" |
| 387 |
pos = skip_chars(src, pos, TOML_WS) |
| 388 |
pos, key = parse_key(src, pos) |
| 389 |
|
| 390 |
if out.flags.is_(key, Flags.FROZEN): |
| 391 |
raise TOMLDecodeError(f"Cannot mutate immutable namespace {key}", src, pos) |
| 392 |
# Free the namespace now that it points to another empty list item... |
| 393 |
out.flags.unset_all(key) |
| 394 |
# ...but this key precisely is still prohibited from table declaration |
| 395 |
out.flags.set(key, Flags.EXPLICIT_NEST, recursive=False) |
| 396 |
try: |
| 397 |
out.data.append_nest_to_list(key) |
| 398 |
except KeyError: |
| 399 |
raise TOMLDecodeError("Cannot overwrite a value", src, pos) from None |
| 400 |
|
| 401 |
if not src.startswith("]]", pos): |
| 402 |
raise TOMLDecodeError( |
| 403 |
"Expected ']]' at the end of an array declaration", src, pos |
| 404 |
) |
| 405 |
return pos + 2, key |
| 406 |
|
| 407 |
|
| 408 |
def key_value_rule( |
| 409 |
src: str, pos: Pos, out: Output, header: Key, parse_float: ParseFloat |
| 410 |
) -> Pos: |
| 411 |
pos, key, value = parse_key_value_pair(src, pos, parse_float) |
| 412 |
key_parent, key_stem = key[:-1], key[-1] |
| 413 |
abs_key_parent = header + key_parent |
| 414 |
|
| 415 |
relative_path_cont_keys = (header + key[:i] for i in range(1, len(key))) |
| 416 |
for cont_key in relative_path_cont_keys: |
| 417 |
# Check that dotted key syntax does not redefine an existing table |
| 418 |
if out.flags.is_(cont_key, Flags.EXPLICIT_NEST): |
| 419 |
raise TOMLDecodeError(f"Cannot redefine namespace {cont_key}", src, pos) |
| 420 |
# Containers in the relative path can't be opened with the table syntax or |
| 421 |
# dotted key/value syntax in following table sections. |
| 422 |
out.flags.add_pending(cont_key, Flags.EXPLICIT_NEST) |
| 423 |
|
| 424 |
if out.flags.is_(abs_key_parent, Flags.FROZEN): |
| 425 |
raise TOMLDecodeError( |
| 426 |
f"Cannot mutate immutable namespace {abs_key_parent}", src, pos |
| 427 |
) |
| 428 |
|
| 429 |
try: |
| 430 |
nest = out.data.get_or_create_nest(abs_key_parent) |
| 431 |
except KeyError: |
| 432 |
raise TOMLDecodeError("Cannot overwrite a value", src, pos) from None |
| 433 |
if key_stem in nest: |
| 434 |
raise TOMLDecodeError("Cannot overwrite a value", src, pos) |
| 435 |
# Mark inline table and array namespaces recursively immutable |
| 436 |
if isinstance(value, (dict, list)): |
| 437 |
out.flags.set(header + key, Flags.FROZEN, recursive=True) |
| 438 |
nest[key_stem] = value |
| 439 |
return pos |
| 440 |
|
| 441 |
|
| 442 |
def parse_key_value_pair( |
| 443 |
src: str, pos: Pos, parse_float: ParseFloat |
| 444 |
) -> tuple[Pos, Key, Any]: |
| 445 |
pos, key = parse_key(src, pos) |
| 446 |
try: |
| 447 |
char: str | None = src[pos] |
| 448 |
except IndexError: |
| 449 |
char = None |
| 450 |
if char != "=": |
| 451 |
raise TOMLDecodeError("Expected '=' after a key in a key/value pair", src, pos) |
| 452 |
pos += 1 |
| 453 |
pos = skip_chars(src, pos, TOML_WS) |
| 454 |
pos, value = parse_value(src, pos, parse_float) |
| 455 |
return pos, key, value |
| 456 |
|
| 457 |
|
| 458 |
def parse_key(src: str, pos: Pos) -> tuple[Pos, Key]: |
| 459 |
pos, key_part = parse_key_part(src, pos) |
| 460 |
key: Key = (key_part,) |
| 461 |
pos = skip_chars(src, pos, TOML_WS) |
| 462 |
while True: |
| 463 |
try: |
| 464 |
char: str | None = src[pos] |
| 465 |
except IndexError: |
| 466 |
char = None |
| 467 |
if char != ".": |
| 468 |
return pos, key |
| 469 |
pos += 1 |
| 470 |
pos = skip_chars(src, pos, TOML_WS) |
| 471 |
pos, key_part = parse_key_part(src, pos) |
| 472 |
key += (key_part,) |
| 473 |
if len(key) > MAX_KEY_PARTS: |
| 474 |
raise RecursionError( |
| 475 |
f"TOML key has more than the allowed {MAX_KEY_PARTS} parts" |
| 476 |
) |
| 477 |
pos = skip_chars(src, pos, TOML_WS) |
| 478 |
|
| 479 |
|
| 480 |
def parse_key_part(src: str, pos: Pos) -> tuple[Pos, str]: |
| 481 |
try: |
| 482 |
char: str | None = src[pos] |
| 483 |
except IndexError: |
| 484 |
char = None |
| 485 |
if char in BARE_KEY_CHARS: |
| 486 |
start_pos = pos |
| 487 |
pos = skip_chars(src, pos, BARE_KEY_CHARS) |
| 488 |
return pos, src[start_pos:pos] |
| 489 |
if char == "'": |
| 490 |
return parse_literal_str(src, pos) |
| 491 |
if char == '"': |
| 492 |
return parse_one_line_basic_str(src, pos) |
| 493 |
raise TOMLDecodeError("Invalid initial character for a key part", src, pos) |
| 494 |
|
| 495 |
|
| 496 |
def parse_one_line_basic_str(src: str, pos: Pos) -> tuple[Pos, str]: |
| 497 |
pos += 1 |
| 498 |
return parse_basic_str(src, pos, multiline=False) |
| 499 |
|
| 500 |
|
| 501 |
def parse_array(src: str, pos: Pos, parse_float: ParseFloat) -> tuple[Pos, list[Any]]: |
| 502 |
pos += 1 |
| 503 |
array: list[Any] = [] |
| 504 |
|
| 505 |
pos = skip_comments_and_array_ws(src, pos) |
| 506 |
if src.startswith("]", pos): |
| 507 |
return pos + 1, array |
| 508 |
while True: |
| 509 |
pos, val = parse_value(src, pos, parse_float) |
| 510 |
array.append(val) |
| 511 |
pos = skip_comments_and_array_ws(src, pos) |
| 512 |
|
| 513 |
c = src[pos : pos + 1] |
| 514 |
if c == "]": |
| 515 |
return pos + 1, array |
| 516 |
if c != ",": |
| 517 |
raise TOMLDecodeError("Unclosed array", src, pos) |
| 518 |
pos += 1 |
| 519 |
|
| 520 |
pos = skip_comments_and_array_ws(src, pos) |
| 521 |
if src.startswith("]", pos): |
| 522 |
return pos + 1, array |
| 523 |
|
| 524 |
|
| 525 |
def parse_inline_table(src: str, pos: Pos, parse_float: ParseFloat) -> tuple[Pos, dict[str, Any]]: |
| 526 |
pos += 1 |
| 527 |
nested_dict = NestedDict() |
| 528 |
flags = Flags() |
| 529 |
|
| 530 |
pos = skip_chars(src, pos, TOML_WS) |
| 531 |
if src.startswith("}", pos): |
| 532 |
return pos + 1, nested_dict.dict |
| 533 |
while True: |
| 534 |
pos, key, value = parse_key_value_pair(src, pos, parse_float) |
| 535 |
key_parent, key_stem = key[:-1], key[-1] |
| 536 |
if flags.is_(key, Flags.FROZEN): |
| 537 |
raise TOMLDecodeError(f"Cannot mutate immutable namespace {key}", src, pos) |
| 538 |
try: |
| 539 |
nest = nested_dict.get_or_create_nest(key_parent, access_lists=False) |
| 540 |
except KeyError: |
| 541 |
raise TOMLDecodeError("Cannot overwrite a value", src, pos) from None |
| 542 |
if key_stem in nest: |
| 543 |
raise TOMLDecodeError(f"Duplicate inline table key {key_stem!r}", src, pos) |
| 544 |
nest[key_stem] = value |
| 545 |
pos = skip_chars(src, pos, TOML_WS) |
| 546 |
c = src[pos : pos + 1] |
| 547 |
if c == "}": |
| 548 |
return pos + 1, nested_dict.dict |
| 549 |
if c != ",": |
| 550 |
raise TOMLDecodeError("Unclosed inline table", src, pos) |
| 551 |
if isinstance(value, (dict, list)): |
| 552 |
flags.set(key, Flags.FROZEN, recursive=True) |
| 553 |
pos += 1 |
| 554 |
pos = skip_chars(src, pos, TOML_WS) |
| 555 |
|
| 556 |
|
| 557 |
def parse_basic_str_escape( |
| 558 |
src: str, pos: Pos, *, multiline: bool = False |
| 559 |
) -> tuple[Pos, str]: |
| 560 |
escape_id = src[pos : pos + 2] |
| 561 |
pos += 2 |
| 562 |
if multiline and escape_id in {"\\ ", "\\\t", "\\\n"}: |
| 563 |
# Skip whitespace until next non-whitespace character or end of |
| 564 |
# the doc. Error if non-whitespace is found before newline. |
| 565 |
if escape_id != "\\\n": |
| 566 |
pos = skip_chars(src, pos, TOML_WS) |
| 567 |
try: |
| 568 |
char = src[pos] |
| 569 |
except IndexError: |
| 570 |
return pos, "" |
| 571 |
if char != "\n": |
| 572 |
raise TOMLDecodeError("Unescaped '\\' in a string", src, pos) |
| 573 |
pos += 1 |
| 574 |
pos = skip_chars(src, pos, TOML_WS_AND_NEWLINE) |
| 575 |
return pos, "" |
| 576 |
if escape_id == "\\u": |
| 577 |
return parse_hex_char(src, pos, 4) |
| 578 |
if escape_id == "\\U": |
| 579 |
return parse_hex_char(src, pos, 8) |
| 580 |
try: |
| 581 |
return pos, BASIC_STR_ESCAPE_REPLACEMENTS[escape_id] |
| 582 |
except KeyError: |
| 583 |
raise TOMLDecodeError("Unescaped '\\' in a string", src, pos) from None |
| 584 |
|
| 585 |
|
| 586 |
def parse_basic_str_escape_multiline(src: str, pos: Pos) -> tuple[Pos, str]: |
| 587 |
return parse_basic_str_escape(src, pos, multiline=True) |
| 588 |
|
| 589 |
|
| 590 |
def parse_hex_char(src: str, pos: Pos, hex_len: int) -> tuple[Pos, str]: |
| 591 |
hex_str = src[pos : pos + hex_len] |
| 592 |
if len(hex_str) != hex_len or not HEXDIGIT_CHARS.issuperset(hex_str): |
| 593 |
raise TOMLDecodeError("Invalid hex value", src, pos) |
| 594 |
pos += hex_len |
| 595 |
hex_int = int(hex_str, 16) |
| 596 |
if not is_unicode_scalar_value(hex_int): |
| 597 |
raise TOMLDecodeError( |
| 598 |
"Escaped character is not a Unicode scalar value", src, pos |
| 599 |
) |
| 600 |
return pos, chr(hex_int) |
| 601 |
|
| 602 |
|
| 603 |
def parse_literal_str(src: str, pos: Pos) -> tuple[Pos, str]: |
| 604 |
pos += 1 # Skip starting apostrophe |
| 605 |
start_pos = pos |
| 606 |
pos = skip_until( |
| 607 |
src, pos, "'", error_on=ILLEGAL_LITERAL_STR_CHARS, error_on_eof=True |
| 608 |
) |
| 609 |
return pos + 1, src[start_pos:pos] # Skip ending apostrophe |
| 610 |
|
| 611 |
|
| 612 |
def parse_multiline_str(src: str, pos: Pos, *, literal: bool) -> tuple[Pos, str]: |
| 613 |
pos += 3 |
| 614 |
if src.startswith("\n", pos): |
| 615 |
pos += 1 |
| 616 |
|
| 617 |
if literal: |
| 618 |
delim = "'" |
| 619 |
end_pos = skip_until( |
| 620 |
src, |
| 621 |
pos, |
| 622 |
"'''", |
| 623 |
error_on=ILLEGAL_MULTILINE_LITERAL_STR_CHARS, |
| 624 |
error_on_eof=True, |
| 625 |
) |
| 626 |
result = src[pos:end_pos] |
| 627 |
pos = end_pos + 3 |
| 628 |
else: |
| 629 |
delim = '"' |
| 630 |
pos, result = parse_basic_str(src, pos, multiline=True) |
| 631 |
|
| 632 |
# Add at maximum two extra apostrophes/quotes if the end sequence |
| 633 |
# is 4 or 5 chars long instead of just 3. |
| 634 |
if not src.startswith(delim, pos): |
| 635 |
return pos, result |
| 636 |
pos += 1 |
| 637 |
if not src.startswith(delim, pos): |
| 638 |
return pos, result + delim |
| 639 |
pos += 1 |
| 640 |
return pos, result + (delim * 2) |
| 641 |
|
| 642 |
|
| 643 |
def parse_basic_str(src: str, pos: Pos, *, multiline: bool) -> tuple[Pos, str]: |
| 644 |
if multiline: |
| 645 |
error_on = ILLEGAL_MULTILINE_BASIC_STR_CHARS |
| 646 |
parse_escapes = parse_basic_str_escape_multiline |
| 647 |
else: |
| 648 |
error_on = ILLEGAL_BASIC_STR_CHARS |
| 649 |
parse_escapes = parse_basic_str_escape |
| 650 |
result = "" |
| 651 |
start_pos = pos |
| 652 |
while True: |
| 653 |
try: |
| 654 |
char = src[pos] |
| 655 |
except IndexError: |
| 656 |
raise TOMLDecodeError("Unterminated string", src, pos) from None |
| 657 |
if char == '"': |
| 658 |
if not multiline: |
| 659 |
return pos + 1, result + src[start_pos:pos] |
| 660 |
if src.startswith('"""', pos): |
| 661 |
return pos + 3, result + src[start_pos:pos] |
| 662 |
pos += 1 |
| 663 |
continue |
| 664 |
if char == "\\": |
| 665 |
result += src[start_pos:pos] |
| 666 |
pos, parsed_escape = parse_escapes(src, pos) |
| 667 |
result += parsed_escape |
| 668 |
start_pos = pos |
| 669 |
continue |
| 670 |
if char in error_on: |
| 671 |
raise TOMLDecodeError(f"Illegal character {char!r}", src, pos) |
| 672 |
pos += 1 |
| 673 |
|
| 674 |
|
| 675 |
def parse_value( # noqa: C901 |
| 676 |
src: str, pos: Pos, parse_float: ParseFloat |
| 677 |
) -> tuple[Pos, Any]: |
| 678 |
try: |
| 679 |
char: str | None = src[pos] |
| 680 |
except IndexError: |
| 681 |
char = None |
| 682 |
|
| 683 |
# IMPORTANT: order conditions based on speed of checking and likelihood |
| 684 |
|
| 685 |
# Basic strings |
| 686 |
if char == '"': |
| 687 |
if src.startswith('"""', pos): |
| 688 |
return parse_multiline_str(src, pos, literal=False) |
| 689 |
return parse_one_line_basic_str(src, pos) |
| 690 |
|
| 691 |
# Literal strings |
| 692 |
if char == "'": |
| 693 |
if src.startswith("'''", pos): |
| 694 |
return parse_multiline_str(src, pos, literal=True) |
| 695 |
return parse_literal_str(src, pos) |
| 696 |
|
| 697 |
# Booleans |
| 698 |
if char == "t": |
| 699 |
if src.startswith("true", pos): |
| 700 |
return pos + 4, True |
| 701 |
if char == "f": |
| 702 |
if src.startswith("false", pos): |
| 703 |
return pos + 5, False |
| 704 |
|
| 705 |
# Arrays |
| 706 |
if char == "[": |
| 707 |
return parse_array(src, pos, parse_float) |
| 708 |
|
| 709 |
# Inline tables |
| 710 |
if char == "{": |
| 711 |
return parse_inline_table(src, pos, parse_float) |
| 712 |
|
| 713 |
# Dates and times |
| 714 |
datetime_match = RE_DATETIME.match(src, pos) |
| 715 |
if datetime_match: |
| 716 |
try: |
| 717 |
datetime_obj = match_to_datetime(datetime_match) |
| 718 |
except ValueError as e: |
| 719 |
raise TOMLDecodeError("Invalid date or datetime", src, pos) from e |
| 720 |
return datetime_match.end(), datetime_obj |
| 721 |
localtime_match = RE_LOCALTIME.match(src, pos) |
| 722 |
if localtime_match: |
| 723 |
return localtime_match.end(), match_to_localtime(localtime_match) |
| 724 |
|
| 725 |
# Integers and "normal" floats. |
| 726 |
# The regex will greedily match any type starting with a decimal |
| 727 |
# char, so needs to be located after handling of dates and times. |
| 728 |
number_match = RE_NUMBER.match(src, pos) |
| 729 |
if number_match: |
| 730 |
return number_match.end(), match_to_number(number_match, parse_float) |
| 731 |
|
| 732 |
# Special floats |
| 733 |
first_three = src[pos : pos + 3] |
| 734 |
if first_three in {"inf", "nan"}: |
| 735 |
return pos + 3, parse_float(first_three) |
| 736 |
first_four = src[pos : pos + 4] |
| 737 |
if first_four in {"-inf", "+inf", "-nan", "+nan"}: |
| 738 |
return pos + 4, parse_float(first_four) |
| 739 |
|
| 740 |
raise TOMLDecodeError("Invalid value", src, pos) |
| 741 |
|
| 742 |
|
| 743 |
def is_unicode_scalar_value(codepoint: int) -> bool: |
| 744 |
return (0 <= codepoint <= 55295) or (57344 <= codepoint <= 1114111) |
| 745 |
|
| 746 |
|
| 747 |
def make_safe_parse_float(parse_float: ParseFloat) -> ParseFloat: |
| 748 |
"""A decorator to make `parse_float` safe. |
| 749 |
|
| 750 |
`parse_float` must not return dicts or lists, because these types |
| 751 |
would be mixed with parsed TOML tables and arrays, thus confusing |
| 752 |
the parser. The returned decorated callable raises `ValueError` |
| 753 |
instead of returning illegal types. |
| 754 |
""" |
| 755 |
# The default `float` callable never returns illegal types. Optimize it. |
| 756 |
if parse_float is float: |
| 757 |
return float |
| 758 |
|
| 759 |
def safe_parse_float(float_str: str) -> Any: |
| 760 |
float_value = parse_float(float_str) |
| 761 |
if isinstance(float_value, (dict, list)): |
| 762 |
raise ValueError("parse_float must not return dicts or lists") |
| 763 |
return float_value |
| 764 |
|
| 765 |
return safe_parse_float |