| 1 | """YSH Str methods"""
|
| 2 |
|
| 3 | from __future__ import print_function
|
| 4 |
|
| 5 | from _devbuild.gen.syntax_asdl import loc_t
|
| 6 | from _devbuild.gen.value_asdl import (value, value_e, value_t, eggex_ops,
|
| 7 | eggex_ops_t, RegexMatch)
|
| 8 | from builtin import pure_ysh
|
| 9 | from core import error
|
| 10 | from core import state
|
| 11 | from core import vm
|
| 12 | from frontend import typed_args
|
| 13 | from mycpp import mops
|
| 14 | from mycpp.mylib import log, tagswitch
|
| 15 | from osh import string_ops
|
| 16 | from ysh import expr_eval
|
| 17 | from ysh import regex_translate
|
| 18 | from ysh import val_ops
|
| 19 |
|
| 20 | import libc
|
| 21 | from libc import REG_NOTBOL
|
| 22 |
|
| 23 | from typing import cast, List, Tuple
|
| 24 |
|
| 25 | _ = log
|
| 26 |
|
| 27 |
|
| 28 | def _StrMatchStart(s, p):
|
| 29 | # type: (str, str) -> Tuple[bool, int, int]
|
| 30 | """Returns the range of bytes in 's' that match string pattern `p`. the
|
| 31 | pattern matches if 's' starts with all the characters in 'p'.
|
| 32 |
|
| 33 | The returned match result is the tuple "(matched, begin, end)". 'matched'
|
| 34 | is true if the pattern matched. 'begin' and 'end' give the half-open range
|
| 35 | "[begin, end)" of byte indices from 's' for the match, and are a valid but
|
| 36 | empty range if 'match' is false.
|
| 37 |
|
| 38 | Used for shell functions like 'trimStart' when trimming a prefix string.
|
| 39 | """
|
| 40 | if s.startswith(p):
|
| 41 | return (True, 0, len(p))
|
| 42 | else:
|
| 43 | return (False, 0, 0)
|
| 44 |
|
| 45 |
|
| 46 | def _StrMatchEnd(s, p):
|
| 47 | # type: (str, str) -> Tuple[bool, int, int]
|
| 48 | """Returns a match result for the bytes in 's' that match string pattern
|
| 49 | `p`. the pattern matches if 's' ends with all the characters in 'p'.
|
| 50 |
|
| 51 | The returned match result is the tuple "(matched, begin, end)". 'matched'
|
| 52 | is true if the pattern matched. 'begin' and 'end' give the half-open range
|
| 53 | "[begin, end)" of byte indices from 's' for the match, and are a valid but
|
| 54 | empty range if 'match' is false.
|
| 55 |
|
| 56 | Used for shell functions like 'trimEnd' when trimming a suffix string.
|
| 57 | """
|
| 58 | len_s = len(s)
|
| 59 | if s.endswith(p):
|
| 60 | return (True, len_s - len(p), len_s)
|
| 61 | else:
|
| 62 | return (False, len_s, len_s)
|
| 63 |
|
| 64 |
|
| 65 | def _EggexMatchCommon(s, p, ere, empty_p):
|
| 66 | # type: (str, value.Eggex, str, int) -> Tuple[bool, int, int]
|
| 67 | cflags = regex_translate.LibcFlags(p.canonical_flags)
|
| 68 | eflags = 0
|
| 69 | indices = libc.regex_search(ere, cflags, s, eflags)
|
| 70 | if indices is None:
|
| 71 | return (False, empty_p, empty_p)
|
| 72 |
|
| 73 | start = indices[0]
|
| 74 | end = indices[1]
|
| 75 |
|
| 76 | return (True, start, end)
|
| 77 |
|
| 78 |
|
| 79 | def _EggexMatchStart(s, p):
|
| 80 | # type: (str, value.Eggex) -> Tuple[bool, int, int]
|
| 81 | """Returns a match result for the bytes in 's' that match Eggex pattern
|
| 82 | `p` when constrained to match at the start of the string.
|
| 83 |
|
| 84 | Any capturing done by the Eggex pattern is ignored.
|
| 85 |
|
| 86 | The returned match result is the tuple "(matched, begin, end)". 'matched'
|
| 87 | is true if the pattern matched. 'begin' and 'end' give the half-open range
|
| 88 | "[begin, end)" of byte indices from 's' for the match, and are a valid but
|
| 89 | empty range if 'match' is false.
|
| 90 |
|
| 91 | Used for shell functions like 'trimStart' when trimming with an Eggex
|
| 92 | pattern.
|
| 93 | """
|
| 94 | ere = regex_translate.AsPosixEre(p)
|
| 95 | if not ere.startswith('^'):
|
| 96 | ere = '^' + ere
|
| 97 | return _EggexMatchCommon(s, p, ere, 0)
|
| 98 |
|
| 99 |
|
| 100 | def _EggexMatchEnd(s, p):
|
| 101 | # type: (str, value.Eggex) -> Tuple[bool, int, int]
|
| 102 | """Like _EggexMatchStart, but matches against the end of the
|
| 103 | string.
|
| 104 | """
|
| 105 | ere = regex_translate.AsPosixEre(p)
|
| 106 | if not ere.endswith('$'):
|
| 107 | ere = ere + '$'
|
| 108 | return _EggexMatchCommon(s, p, ere, len(s))
|
| 109 |
|
| 110 |
|
| 111 | START = 0b01
|
| 112 | END = 0b10
|
| 113 |
|
| 114 |
|
| 115 | class HasAffix(vm._Callable):
|
| 116 | """ Implements `startsWith()`, `endsWith()`. """
|
| 117 |
|
| 118 | def __init__(self, anchor):
|
| 119 | # type: (int) -> None
|
| 120 | assert anchor in (START, END), ("Anchor must be START or END")
|
| 121 | self.anchor = anchor
|
| 122 |
|
| 123 | def Call(self, rd):
|
| 124 | # type: (typed_args.Reader) -> value_t
|
| 125 | """
|
| 126 | string => startsWith(pattern_str) # => bool
|
| 127 | string => startsWith(pattern_eggex) # => bool
|
| 128 | string => endsWith(pattern_str) # => bool
|
| 129 | string => endsWith(pattern_eggex) # => bool
|
| 130 | """
|
| 131 |
|
| 132 | string = rd.PosStr()
|
| 133 | pattern_val = rd.PosValue()
|
| 134 | pattern_str = None # type: str
|
| 135 | pattern_eggex = None # type: value.Eggex
|
| 136 | with tagswitch(pattern_val) as case:
|
| 137 | if case(value_e.Eggex):
|
| 138 | pattern_eggex = cast(value.Eggex, pattern_val)
|
| 139 | elif case(value_e.Str):
|
| 140 | pattern_str = cast(value.Str, pattern_val).s
|
| 141 | else:
|
| 142 | raise error.TypeErr(pattern_val,
|
| 143 | 'expected pattern to be Eggex or Str',
|
| 144 | rd.LeftParenToken())
|
| 145 | rd.Done()
|
| 146 |
|
| 147 | matched = False
|
| 148 | try:
|
| 149 | if pattern_str is not None:
|
| 150 | if self.anchor & START:
|
| 151 | matched, _, _ = _StrMatchStart(string, pattern_str)
|
| 152 | else:
|
| 153 | matched, _, _ = _StrMatchEnd(string, pattern_str)
|
| 154 | else:
|
| 155 | assert pattern_eggex is not None
|
| 156 | if self.anchor & START:
|
| 157 | matched, _, _ = _EggexMatchStart(string, pattern_eggex)
|
| 158 | else:
|
| 159 | matched, _, _ = _EggexMatchEnd(string, pattern_eggex)
|
| 160 | except error.Strict as e:
|
| 161 | raise error.Expr(e.msg, e.location)
|
| 162 |
|
| 163 | return value.Bool(matched)
|
| 164 |
|
| 165 |
|
| 166 | class Trim(vm._Callable):
|
| 167 | """ Implements `trimStart()`, `trimEnd()`, and `trim()` """
|
| 168 |
|
| 169 | def __init__(self, anchor):
|
| 170 | # type: (int) -> None
|
| 171 | assert anchor in (START, END, START
|
| 172 | | END), ("Anchor must be START, END, or START|END")
|
| 173 | self.anchor = anchor
|
| 174 |
|
| 175 | def Call(self, rd):
|
| 176 | # type: (typed_args.Reader) -> value_t
|
| 177 | """
|
| 178 | string => trimStart() # => Str
|
| 179 | string => trimEnd() # => Str
|
| 180 | string => trim() # => Str
|
| 181 | string => trimStart(pattern_str) # => Str
|
| 182 | string => trimEnd(pattern_str) # => Str
|
| 183 | string => trim(pattern_str) # => Str
|
| 184 | string => trimStart(pattern_eggex) # => Str
|
| 185 | string => trimEnd(pattern_eggex) # => Str
|
| 186 | string => trim(pattern_eggex) # => Str
|
| 187 | """
|
| 188 |
|
| 189 | string = rd.PosStr()
|
| 190 | pattern_val = rd.OptionalValue()
|
| 191 | pattern_str = None # type: str
|
| 192 | pattern_eggex = None # type: value.Eggex
|
| 193 | if pattern_val:
|
| 194 | with tagswitch(pattern_val) as case:
|
| 195 | if case(value_e.Eggex):
|
| 196 | pattern_eggex = cast(value.Eggex, pattern_val)
|
| 197 | elif case(value_e.Str):
|
| 198 | pattern_str = cast(value.Str, pattern_val).s
|
| 199 | else:
|
| 200 | raise error.TypeErr(pattern_val,
|
| 201 | 'expected pattern to be Eggex or Str',
|
| 202 | rd.LeftParenToken())
|
| 203 | rd.Done()
|
| 204 |
|
| 205 | start = 0
|
| 206 | end = len(string)
|
| 207 | try:
|
| 208 | if pattern_str is not None:
|
| 209 | if self.anchor & START:
|
| 210 | _, _, start = _StrMatchStart(string, pattern_str)
|
| 211 | if self.anchor & END:
|
| 212 | _, end, _ = _StrMatchEnd(string, pattern_str)
|
| 213 | elif pattern_eggex is not None:
|
| 214 | if self.anchor & START:
|
| 215 | _, _, start = _EggexMatchStart(string, pattern_eggex)
|
| 216 | if self.anchor & END:
|
| 217 | _, end, _ = _EggexMatchEnd(string, pattern_eggex)
|
| 218 | else:
|
| 219 | if self.anchor & START:
|
| 220 | _, start = string_ops.StartsWithWhitespaceByteRange(string)
|
| 221 | if self.anchor & END:
|
| 222 | end, _ = string_ops.EndsWithWhitespaceByteRange(string)
|
| 223 | except error.Strict as e:
|
| 224 | raise error.Expr(e.msg, e.location)
|
| 225 |
|
| 226 | res = string[start:end]
|
| 227 | return value.Str(res)
|
| 228 |
|
| 229 |
|
| 230 | class Upper(vm._Callable):
|
| 231 |
|
| 232 | def __init__(self):
|
| 233 | # type: () -> None
|
| 234 | pass
|
| 235 |
|
| 236 | def Call(self, rd):
|
| 237 | # type: (typed_args.Reader) -> value_t
|
| 238 |
|
| 239 | s = rd.PosStr()
|
| 240 | rd.Done()
|
| 241 |
|
| 242 | # TODO: unicode support
|
| 243 | return value.Str(s.upper())
|
| 244 |
|
| 245 |
|
| 246 | class Lower(vm._Callable):
|
| 247 |
|
| 248 | def __init__(self):
|
| 249 | # type: () -> None
|
| 250 | pass
|
| 251 |
|
| 252 | def Call(self, rd):
|
| 253 | # type: (typed_args.Reader) -> value_t
|
| 254 |
|
| 255 | s = rd.PosStr()
|
| 256 | rd.Done()
|
| 257 |
|
| 258 | # TODO: unicode support
|
| 259 | return value.Str(s.lower())
|
| 260 |
|
| 261 |
|
| 262 | SEARCH = 0
|
| 263 | LEFT_MATCH = 1
|
| 264 |
|
| 265 |
|
| 266 | class SearchMatch(vm._Callable):
|
| 267 |
|
| 268 | def __init__(self, which_method):
|
| 269 | # type: (int) -> None
|
| 270 | self.which_method = which_method
|
| 271 |
|
| 272 | def Call(self, rd):
|
| 273 | # type: (typed_args.Reader) -> value_t
|
| 274 | """
|
| 275 | s => search(eggex, pos=0)
|
| 276 | """
|
| 277 | string = rd.PosStr()
|
| 278 |
|
| 279 | pattern = rd.PosValue() # Eggex or ERE Str
|
| 280 | with tagswitch(pattern) as case:
|
| 281 | if case(value_e.Eggex):
|
| 282 | eggex_val = cast(value.Eggex, pattern)
|
| 283 |
|
| 284 | # lazily converts to ERE
|
| 285 | ere = regex_translate.AsPosixEre(eggex_val)
|
| 286 | cflags = regex_translate.LibcFlags(eggex_val.canonical_flags)
|
| 287 | capture = eggex_ops.Yes(
|
| 288 | eggex_val.convert_funcs, eggex_val.convert_toks,
|
| 289 | eggex_val.capture_names) # type: eggex_ops_t
|
| 290 |
|
| 291 | elif case(value_e.Str):
|
| 292 | ere = cast(value.Str, pattern).s
|
| 293 | cflags = 0
|
| 294 | capture = eggex_ops.No
|
| 295 |
|
| 296 | else:
|
| 297 | # TODO: add method name to this error
|
| 298 | raise error.TypeErr(pattern, 'expected Eggex or Str',
|
| 299 | rd.LeftParenToken())
|
| 300 |
|
| 301 | # It's called 'pos', not 'start' like Python. Python has 2 kinds of
|
| 302 | # 'start' in its regex API, which can be confusing.
|
| 303 | pos = mops.BigTruncate(rd.NamedInt('pos', 0))
|
| 304 | rd.Done()
|
| 305 |
|
| 306 | # Make it anchored
|
| 307 | if self.which_method == LEFT_MATCH and not ere.startswith('^'):
|
| 308 | ere = '^' + ere
|
| 309 |
|
| 310 | if self.which_method == LEFT_MATCH:
|
| 311 | eflags = 0 # ^ matches beginning even if pos=5
|
| 312 | else:
|
| 313 | eflags = 0 if pos == 0 else REG_NOTBOL # ^ only matches when pos=0
|
| 314 |
|
| 315 | indices = libc.regex_search(ere, cflags, string, eflags, pos)
|
| 316 |
|
| 317 | if indices is None:
|
| 318 | return value.Null
|
| 319 |
|
| 320 | return RegexMatch(string, indices, capture)
|
| 321 |
|
| 322 |
|
| 323 | class Replace(vm._Callable):
|
| 324 |
|
| 325 | def __init__(self, mem, expr_ev):
|
| 326 | # type: (state.Mem, expr_eval.ExprEvaluator) -> None
|
| 327 | self.mem = mem
|
| 328 | self.expr_ev = expr_ev
|
| 329 |
|
| 330 | def EvalSubstExpr(self, expr, blame_loc):
|
| 331 | # type: (value.Expr, loc_t) -> str
|
| 332 | res = self.expr_ev.EvalExpr(expr.e, blame_loc)
|
| 333 | if res.tag() == value_e.Str:
|
| 334 | return cast(value.Str, res).s
|
| 335 |
|
| 336 | raise error.TypeErr(res, "expected expr to eval to a Str", blame_loc)
|
| 337 |
|
| 338 | def Call(self, rd):
|
| 339 | # type: (typed_args.Reader) -> value_t
|
| 340 | """
|
| 341 | s => replace(string_val, subst_str, count=-1)
|
| 342 | s => replace(string_val, subst_expr, count=-1)
|
| 343 | s => replace(eggex_val, subst_str, count=-1)
|
| 344 | s => replace(eggex_val, subst_expr, count=-1)
|
| 345 |
|
| 346 | For count in [0, MAX_INT], there will be no more than count
|
| 347 | replacements. Any negative count should read as unset, and replace will
|
| 348 | replace all occurances of the pattern.
|
| 349 | """
|
| 350 | string = rd.PosStr()
|
| 351 |
|
| 352 | string_val = None # type: value.Str
|
| 353 | eggex_val = None # type: value.Eggex
|
| 354 | subst_str = None # type: value.Str
|
| 355 | subst_expr = None # type: value.Expr
|
| 356 |
|
| 357 | pattern = rd.PosValue()
|
| 358 | with tagswitch(pattern) as case:
|
| 359 | if case(value_e.Eggex):
|
| 360 | # HACK: mycpp will otherwise generate:
|
| 361 | # value::Eggex* eggex_val ...
|
| 362 | eggex_val_ = cast(value.Eggex, pattern)
|
| 363 | eggex_val = eggex_val_
|
| 364 |
|
| 365 | elif case(value_e.Str):
|
| 366 | string_val_ = cast(value.Str, pattern)
|
| 367 | string_val = string_val_
|
| 368 |
|
| 369 | else:
|
| 370 | raise error.TypeErr(pattern,
|
| 371 | 'expected pattern to be Eggex or Str',
|
| 372 | rd.LeftParenToken())
|
| 373 |
|
| 374 | subst = rd.PosValue()
|
| 375 | with tagswitch(subst) as case:
|
| 376 | if case(value_e.Str):
|
| 377 | subst_str_ = cast(value.Str, subst)
|
| 378 | subst_str = subst_str_
|
| 379 |
|
| 380 | elif case(value_e.Expr):
|
| 381 | subst_expr_ = cast(value.Expr, subst)
|
| 382 | subst_expr = subst_expr_
|
| 383 |
|
| 384 | else:
|
| 385 | raise error.TypeErr(subst,
|
| 386 | 'expected substitution to be Str or Expr',
|
| 387 | rd.LeftParenToken())
|
| 388 |
|
| 389 | count = mops.BigTruncate(rd.NamedInt("count", -1))
|
| 390 | rd.Done()
|
| 391 |
|
| 392 | if count == 0:
|
| 393 | return value.Str(string)
|
| 394 |
|
| 395 | if string_val:
|
| 396 | if subst_str:
|
| 397 | s = subst_str.s
|
| 398 | if subst_expr:
|
| 399 | # Eval with $0 set to string_val (the matched substring)
|
| 400 | with state.ctx_Eval(self.mem, string_val.s, None, None):
|
| 401 | s = self.EvalSubstExpr(subst_expr, rd.LeftParenToken())
|
| 402 | assert s is not None
|
| 403 |
|
| 404 | result = string.replace(string_val.s, s, count)
|
| 405 |
|
| 406 | return value.Str(result)
|
| 407 |
|
| 408 | if eggex_val:
|
| 409 | ere = regex_translate.AsPosixEre(eggex_val)
|
| 410 | cflags = regex_translate.LibcFlags(eggex_val.canonical_flags)
|
| 411 |
|
| 412 | # Walk through the string finding all matches of the compiled ere.
|
| 413 | # Then, collect unmatched substrings and substitutions into the
|
| 414 | # `parts` list.
|
| 415 | pos = 0
|
| 416 | parts = [] # type: List[str]
|
| 417 | replace_count = 0
|
| 418 | while pos < len(string):
|
| 419 | indices = libc.regex_search(ere, cflags, string, 0, pos)
|
| 420 | if indices is None:
|
| 421 | break
|
| 422 |
|
| 423 | # Collect captures
|
| 424 | arg0 = None # type: str
|
| 425 | argv = [] # type: List[str]
|
| 426 | named_vars = [] # type: List[Tuple[str, value_t]]
|
| 427 | num_groups = len(indices) / 2
|
| 428 | for group in xrange(num_groups):
|
| 429 | start = indices[2 * group]
|
| 430 | end = indices[2 * group + 1]
|
| 431 | captured = string[start:end]
|
| 432 | val = value.Str(captured) # type: value_t
|
| 433 |
|
| 434 | if len(eggex_val.convert_funcs) and group != 0:
|
| 435 | convert_func = eggex_val.convert_funcs[group - 1]
|
| 436 | convert_tok = eggex_val.convert_toks[group - 1]
|
| 437 |
|
| 438 | if convert_func:
|
| 439 | val = self.expr_ev.CallConvertFunc(
|
| 440 | convert_func, val, convert_tok,
|
| 441 | rd.LeftParenToken())
|
| 442 |
|
| 443 | # $0, $1, $2 variables are argv values, which must be
|
| 444 | # strings. Furthermore, they can only be used in string
|
| 445 | # contexts
|
| 446 | # eg. "$[1]" != "$1".
|
| 447 | val_str = val_ops.Stringify(val, rd.LeftParenToken())
|
| 448 | if group == 0:
|
| 449 | arg0 = val_str
|
| 450 | else:
|
| 451 | argv.append(val_str)
|
| 452 |
|
| 453 | # $0 cannot be named
|
| 454 | if group != 0:
|
| 455 | name = eggex_val.capture_names[group - 2]
|
| 456 | if name is not None:
|
| 457 | named_vars.append((name, val))
|
| 458 |
|
| 459 | if subst_str:
|
| 460 | s = subst_str.s
|
| 461 | if subst_expr:
|
| 462 | with state.ctx_Eval(self.mem, arg0, argv, None):
|
| 463 | with pure_ysh.ctx_Shvar(self.mem, named_vars):
|
| 464 | s = self.EvalSubstExpr(subst_expr,
|
| 465 | rd.LeftParenToken())
|
| 466 | assert s is not None
|
| 467 |
|
| 468 | start = indices[0]
|
| 469 | end = indices[1]
|
| 470 | parts.append(string[pos:start]) # Unmatched substring
|
| 471 | parts.append(s) # Replacement
|
| 472 | pos = end # Move to end of match
|
| 473 |
|
| 474 | replace_count += 1
|
| 475 | if count != -1 and replace_count == count:
|
| 476 | break
|
| 477 |
|
| 478 | parts.append(string[pos:]) # Remaining unmatched substring
|
| 479 |
|
| 480 | return value.Str("".join(parts))
|
| 481 |
|
| 482 | raise AssertionError()
|