1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696
|
#
#
# Nim's Runtime Library
# (c) Copyright 2016 Andreas Rumpf
#
# See the file "copying.txt", included in this
# distribution, for details about the copyright.
#
##[
This module contains a `scanf`:idx: macro that can be used for extracting
substrings from an input string. This is often easier than regular expressions.
Some examples as an appetizer:
```nim
# check if input string matches a triple of integers:
const input = "(1,2,4)"
var x, y, z: int
if scanf(input, "($i,$i,$i)", x, y, z):
echo "matches and x is ", x, " y is ", y, " z is ", z
# check if input string matches an ISO date followed by an identifier followed
# by whitespace and a floating point number:
var year, month, day: int
var identifier: string
var myfloat: float
if scanf(input, "$i-$i-$i $w$s$f", year, month, day, identifier, myfloat):
echo "yes, we have a match!"
```
As can be seen from the examples, strings are matched verbatim except for
substrings starting with ``$``. These constructions are available:
================= ========================================================
``$b`` Matches a binary integer. This uses ``parseutils.parseBin``.
``$o`` Matches an octal integer. This uses ``parseutils.parseOct``.
``$i`` Matches a decimal integer. This uses ``parseutils.parseInt``.
``$h`` Matches a hex integer. This uses ``parseutils.parseHex``.
``$f`` Matches a floating-point number. Uses ``parseFloat``.
``$w`` Matches an ASCII identifier: ``[A-Za-z_][A-Za-z_0-9]*``.
``$c`` Matches a single ASCII character.
``$s`` Skips optional whitespace.
``$$`` Matches a single dollar sign.
``$.`` Matches if the end of the input string has been reached.
``$*`` Matches until the token following the ``$*`` was found.
The match is allowed to be of 0 length.
``$+`` Matches until the token following the ``$+`` was found.
The match must consist of at least one char.
``${foo}`` User defined matcher. Uses the proc ``foo`` to perform
the match. See below for more details.
``$[foo]`` Call user defined proc ``foo`` to **skip** some optional
parts in the input string. See below for more details.
================= ========================================================
Even though ``$*`` and ``$+`` look similar to the regular expressions ``.*``
and ``.+``, they work quite differently. There is no non-deterministic
state machine involved and the matches are non-greedy. ``[$*]``
matches ``[xyz]`` via ``parseutils.parseUntil``.
Furthermore no backtracking is performed, if parsing fails after a value
has already been bound to a matched subexpression this value is not restored
to its original value. This rarely causes problems in practice and if it does
for you, it's easy enough to bind to a temporary variable first.
Startswith vs full match
========================
``scanf`` returns true if the input string **starts with** the specified
pattern. If instead it should only return true if there is also nothing
left in the input, append ``$.`` to your pattern.
User definable matchers
=======================
One very nice advantage over regular expressions is that ``scanf`` is
extensible with ordinary Nim procs. The proc is either enclosed in ``${}``
or in ``$[]``. ``${}`` matches and binds the result
to a variable (that was passed to the ``scanf`` macro) while ``$[]`` merely
matches optional tokens without any result binding.
In this example, we define a helper proc ``someSep`` that skips some separators
which we then use in our scanf pattern to help us in the matching process:
```nim
proc someSep(input: string; start: int; seps: set[char] = {':','-','.'}): int =
# Note: The parameters and return value must match to what ``scanf`` requires
result = 0
while start+result < input.len and input[start+result] in seps: inc result
if scanf(input, "$w$[someSep]$w", key, value):
...
```
It is also possible to pass arguments to a user definable matcher:
```nim
proc ndigits(input: string; intVal: var int; start: int; n: int): int =
# matches exactly ``n`` digits. Matchers need to return 0 if nothing
# matched or otherwise the number of processed chars.
var x = 0
var i = 0
while i < n and i+start < input.len and input[i+start] in {'0'..'9'}:
x = x * 10 + input[i+start].ord - '0'.ord
inc i
# only overwrite if we had a match
if i == n:
result = n
intVal = x
# match an ISO date extracting year, month, day at the same time.
# Also ensure the input ends after the ISO date:
var year, month, day: int
if scanf("2013-01-03", "${ndigits(4)}-${ndigits(2)}-${ndigits(2)}$.", year, month, day):
...
```
The scanp macro
===============
This module also implements a ``scanp`` macro, which syntax somewhat resembles
an EBNF or PEG grammar, except that it uses Nim's expression syntax and so has
to use prefix instead of postfix operators.
============== ===============================================================
``(E)`` Grouping
``*E`` Zero or more
``+E`` One or more
``?E`` Zero or One
``E{n,m}`` From ``n`` up to ``m`` times ``E``
``~E`` Not predicate
``a ^* b`` Shortcut for ``?(a *(b a))``. Usually used for separators.
``a ^+ b`` Shortcut for ``?(a +(b a))``. Usually used for separators.
``'a'`` Matches a single character
``{'a'..'b'}`` Matches a character set
``"s"`` Matches a string
``E -> a`` Bind matching to some action
``$_`` Access the currently matched character
============== ===============================================================
Note that unordered or ordered choice operators (``/``, ``|``) are
not implemented.
Simple example that parses the ``/etc/passwd`` file line by line:
```nim
const
etc_passwd = """root:x:0:0:root:/root:/bin/bash
daemon:x:1:1:daemon:/usr/sbin:/bin/sh
bin:x:2:2:bin:/bin:/bin/sh
sys:x:3:3:sys:/dev:/bin/sh
nobody:x:65534:65534:nobody:/nonexistent:/bin/sh
messagebus:x:103:107::/var/run/dbus:/bin/false
"""
proc parsePasswd(content: string): seq[string] =
result = @[]
var idx = 0
while true:
var entry = ""
if scanp(content, idx, +(~{'\L', '\0'} -> entry.add($_)), '\L'):
result.add entry
else:
break
```
The ``scanp`` maps the grammar code into Nim code that performs the parsing.
The parsing is performed with the help of 3 helper templates that that can be
implemented for a custom type.
These templates need to be named ``atom`` and ``nxt``. ``atom`` should be
overloaded to handle both `char` and `set[char]`.
```nim
import std/streams
template atom(input: Stream; idx: int; c: char): bool =
## Used in scanp for the matching of atoms (usually chars).
peekChar(input) == c
template atom(input: Stream; idx: int; s: set[char]): bool =
peekChar(input) in s
template nxt(input: Stream; idx, step: int = 1) =
inc(idx, step)
setPosition(input, idx)
if scanp(content, idx, +( ~{'\L', '\0'} -> entry.add(peekChar($input))), '\L'):
result.add entry
```
Calling ordinary Nim procs inside the macro is possible:
```nim
proc digits(s: string; intVal: var int; start: int): int =
var x = 0
while result+start < s.len and s[result+start] in {'0'..'9'} and s[result+start] != ':':
x = x * 10 + s[result+start].ord - '0'.ord
inc result
intVal = x
proc extractUsers(content: string): seq[string] =
# Extracts the username and home directory
# of each entry (with UID greater than 1000)
const
digits = {'0'..'9'}
result = @[]
var idx = 0
while true:
var login = ""
var uid = 0
var homedir = ""
if scanp(content, idx, *(~ {':', '\0'}) -> login.add($_), ':', * ~ ':', ':',
digits($input, uid, $index), ':', *`digits`, ':', * ~ ':', ':',
*('/', * ~{':', '/'}) -> homedir.add($_), ':', *('/', * ~{'\L', '/'}), '\L'):
if uid >= 1000:
result.add login & " " & homedir
else:
break
```
When used for matching, keep in mind that likewise scanf, no backtracking
is performed.
```nim
proc skipUntil(s: string; until: string; unless = '\0'; start: int): int =
# Skips all characters until the string `until` is found. Returns 0
# if the char `unless` is found first or the end is reached.
var i = start
var u = 0
while true:
if i >= s.len or s[i] == unless:
return 0
elif s[i] == until[0]:
u = 1
while i+u < s.len and u < until.len and s[i+u] == until[u]:
inc u
if u >= until.len: break
inc(i)
result = i+u-start
iterator collectLinks(s: string): string =
const quote = {'\'', '"'}
var idx, old = 0
var res = ""
while idx < s.len:
old = idx
if scanp(s, idx, "<a", skipUntil($input, "href=", '>', $index),
`quote`, *( ~`quote`) -> res.add($_)):
yield res
res = ""
idx = old + 1
for r in collectLinks(body):
echo r
```
In this example both macros are combined seamlessly in order to maximise
efficiency and perform different checks.
```nim
iterator parseIps*(soup: string): string =
## ipv4 only!
const digits = {'0'..'9'}
var a, b, c, d: int
var buf = ""
var idx = 0
while idx < soup.len:
if scanp(soup, idx, (`digits`{1,3}, '.', `digits`{1,3}, '.',
`digits`{1,3}, '.', `digits`{1,3}) -> buf.add($_)):
discard buf.scanf("$i.$i.$i.$i", a, b, c, d)
if (a >= 0 and a <= 254) and
(b >= 0 and b <= 254) and
(c >= 0 and c <= 254) and
(d >= 0 and d <= 254):
yield buf
buf.setLen(0) # need to clear `buf` each time, cause it might contain garbage
idx.inc
```
]##
import std/[macros, parseutils]
import std/private/since
when defined(nimPreviewSlimSystem):
import std/assertions
proc conditionsToIfChain(n, idx, res: NimNode; start: int): NimNode =
assert n.kind == nnkStmtList
if start >= n.len: return newAssignment(res, newLit true)
var ifs: NimNode = nil
if n[start+1].kind == nnkEmpty:
ifs = conditionsToIfChain(n, idx, res, start+3)
else:
ifs = newIfStmt((n[start+1],
newTree(nnkStmtList, newCall(bindSym"inc", idx, n[start+2]),
conditionsToIfChain(n, idx, res, start+3))))
result = newTree(nnkStmtList, n[start], ifs)
proc notZero(x: NimNode): NimNode = newCall(bindSym"!=", x, newLit 0)
proc buildUserCall(x: string; args: varargs[NimNode]): NimNode =
let y = parseExpr(x)
result = newTree(nnkCall)
if y.kind in nnkCallKinds: result.add y[0]
else: result.add y
for a in args: result.add a
if y.kind in nnkCallKinds:
for i in 1..<y.len: result.add y[i]
macro scanf*(input: string; pattern: static[string]; results: varargs[typed]): bool =
## See top level documentation of this module about how ``scanf`` works.
template matchBind(parser) {.dirty.} =
var resLen = genSym(nskLet, "resLen")
conds.add newLetStmt(resLen, newCall(bindSym(parser), inp, results[i], idx))
conds.add resLen.notZero
conds.add resLen
template at(s: string; i: int): char = (if i < s.len: s[i] else: '\0')
template matchError() =
error("type mismatch between pattern '$" & pattern[p] & "' (position: " & $p &
") and " & $getTypeInst(results[i]) & " var '" & repr(results[i]) & "'")
var i = 0
var p = 0
var idx = genSym(nskVar, "idx")
var res = genSym(nskVar, "res")
let inp = genSym(nskLet, "inp")
result = newTree(nnkStmtListExpr, newLetStmt(inp, input),
newVarStmt(idx, newLit 0), newVarStmt(res, newLit false))
var conds = newTree(nnkStmtList)
var fullMatch = false
while p < pattern.len:
if pattern[p] == '$':
inc p
case pattern[p]
of '$':
var resLen = genSym(nskLet, "resLen")
conds.add newLetStmt(resLen, newCall(bindSym"skip", inp,
newLit($pattern[p]), idx))
conds.add resLen.notZero
conds.add resLen
of 'w':
if i < results.len and getType(results[i]).typeKind == ntyString:
matchBind "parseIdent"
else:
matchError
inc i
of 'c':
if i < results.len and getType(results[i]).typeKind == ntyChar:
matchBind "parseChar"
else:
matchError
inc i
of 'b':
if i < results.len and getType(results[i]).typeKind == ntyInt:
matchBind "parseBin"
else:
matchError
inc i
of 'o':
if i < results.len and getType(results[i]).typeKind == ntyInt:
matchBind "parseOct"
else:
matchError
inc i
of 'i':
if i < results.len and getType(results[i]).typeKind == ntyInt:
matchBind "parseInt"
else:
matchError
inc i
of 'h':
if i < results.len and getType(results[i]).typeKind == ntyInt:
matchBind "parseHex"
else:
matchError
inc i
of 'f':
if i < results.len and getType(results[i]).typeKind == ntyFloat:
matchBind "parseFloat"
else:
matchError
inc i
of 's':
conds.add newCall(bindSym"inc", idx,
newCall(bindSym"skipWhitespace", inp, idx))
conds.add newEmptyNode()
conds.add newEmptyNode()
of '.':
if p == pattern.len-1:
fullMatch = true
else:
error("invalid format string")
of '*', '+':
if i < results.len and getType(results[i]).typeKind == ntyString:
var min = ord(pattern[p] == '+')
var q = p+1
var token = ""
while q < pattern.len and pattern[q] != '$':
token.add pattern[q]
inc q
var resLen = genSym(nskLet, "resLen")
conds.add newLetStmt(resLen, newCall(bindSym"parseUntil", inp,
results[i], newLit(token), idx))
conds.add newCall(bindSym">=", resLen, newLit min)
conds.add resLen
else:
matchError
inc i
of '{':
inc p
var nesting = 0
let start = p
while true:
case pattern.at(p)
of '{': inc nesting
of '}':
if nesting == 0: break
dec nesting
of '\0': error("expected closing '}'")
else: discard
inc p
let expr = pattern.substr(start, p-1)
if i < results.len:
var resLen = genSym(nskLet, "resLen")
conds.add newLetStmt(resLen, buildUserCall(expr, inp, results[i], idx))
conds.add newCall(bindSym"!=", resLen, newLit 0)
conds.add resLen
else:
error("no var given for $" & expr & " (position: " & $p & ")")
inc i
of '[':
inc p
var nesting = 0
let start = p
while true:
case pattern.at(p)
of '[': inc nesting
of ']':
if nesting == 0: break
dec nesting
of '\0': error("expected closing ']'")
else: discard
inc p
let expr = pattern.substr(start, p-1)
conds.add newCall(bindSym"inc", idx, buildUserCall(expr, inp, idx))
conds.add newEmptyNode()
conds.add newEmptyNode()
else: error("invalid format string")
inc p
else:
var token = ""
while p < pattern.len and pattern[p] != '$':
token.add pattern[p]
inc p
var resLen = genSym(nskLet, "resLen")
conds.add newLetStmt(resLen, newCall(bindSym"skip", inp, newLit(token), idx))
conds.add resLen.notZero
conds.add resLen
result.add conditionsToIfChain(conds, idx, res, 0)
if fullMatch:
result.add newCall(bindSym"and", res,
newCall(bindSym">=", idx, newCall(bindSym"len", inp)))
else:
result.add res
macro scanTuple*(input: untyped; pattern: static[string]; matcherTypes: varargs[untyped]): untyped {.since: (1, 5).}=
## Works identically as scanf, but instead of predeclaring variables it returns a tuple.
## Tuple is started with a bool which indicates if the scan was successful
## followed by the requested data.
## If using a user defined matcher, provide the types in order they appear after pattern:
## `line.scanTuple("${yourMatcher()}", int)`
runnableExamples:
let (success, year, month, day, time) = scanTuple("1000-01-01 00:00:00", "$i-$i-$i$s$+")
if success:
assert year == 1000
assert month == 1
assert day == 1
assert time == "00:00:00"
var
p = 0
userMatches = 0
arguments: seq[NimNode]
result = newStmtList()
template addVar(typ: string) =
let varIdent = ident("temp" & $arguments.len)
result.add(newNimNode(nnkVarSection).add(newIdentDefs(varIdent, ident(typ), newEmptyNode())))
arguments.add(varIdent)
while p < pattern.len:
if pattern[p] == '$':
inc p
case pattern[p]
of 'w', '*', '+':
addVar("string")
of 'c':
addVar("char")
of 'b', 'o', 'i', 'h':
addVar("int")
of 'f':
addVar("float")
of '{':
if userMatches < matcherTypes.len:
let varIdent = ident("temp" & $arguments.len)
result.add(newNimNode(nnkVarSection).add(newIdentDefs(varIdent, matcherTypes[userMatches], newEmptyNode())))
arguments.add(varIdent)
inc userMatches
else: discard
inc p
result.add nnkTupleConstr.newTree(newCall(ident("scanf"), input, newStrLitNode(pattern)))
for arg in arguments:
result[^1][0].add arg
result[^1].add arg
result = newBlockStmt(result)
template atom*(input: string; idx: int; c: char): bool =
## Used in scanp for the matching of atoms (usually chars).
## EOF is matched as ``'\0'``.
(idx < input.len and input[idx] == c) or (idx == input.len and c == '\0')
template atom*(input: string; idx: int; s: set[char]): bool =
(idx < input.len and input[idx] in s) or (idx == input.len and '\0' in s)
template hasNxt*(input: string; idx: int): bool = idx < input.len
#template prepare*(input: string): int = 0
template success*(x: int): bool = x != 0
template nxt*(input: string; idx, step: int = 1) = inc(idx, step)
macro scanp*(input, idx: typed; pattern: varargs[untyped]): bool =
## See top level documentation of this module about how ``scanp`` works.
type StmtTriple = tuple[init, cond, action: NimNode]
template interf(x): untyped = bindSym(x, brForceOpen)
proc toIfChain(n: seq[StmtTriple]; idx, res: NimNode; start: int): NimNode =
if start >= n.len: return newAssignment(res, newLit true)
var ifs: NimNode = nil
if n[start].cond.kind == nnkEmpty:
ifs = toIfChain(n, idx, res, start+1)
else:
ifs = newIfStmt((n[start].cond,
newTree(nnkStmtList, n[start].action,
toIfChain(n, idx, res, start+1))))
result = newTree(nnkStmtList, n[start].init, ifs)
proc attach(x, attached: NimNode): NimNode =
if attached == nil: x
else: newStmtList(attached, x)
proc placeholder(n, x, j: NimNode): NimNode =
if n.kind == nnkPrefix and n[0].eqIdent("$"):
let n1 = n[1]
if n1.eqIdent"_" or n1.eqIdent"current":
result = newTree(nnkBracketExpr, x, j)
elif n1.eqIdent"input":
result = x
elif n1.eqIdent"i" or n1.eqIdent"index":
result = j
else:
error("unknown pattern " & repr(n))
else:
result = copyNimNode(n)
for i in 0 ..< n.len:
result.add placeholder(n[i], x, j)
proc atm(it, input, idx, attached: NimNode): StmtTriple =
template `!!`(x): untyped = attach(x, attached)
case it.kind
of nnkIdent:
var resLen = genSym(nskLet, "resLen")
result = (newLetStmt(resLen, newCall(it, input, idx)),
newCall(interf"success", resLen),
!!newCall(interf"nxt", input, idx, resLen))
of nnkCallKinds:
# *{'A'..'Z'} !! s.add(!_)
template buildWhile(input, idx, init, cond, action): untyped =
mixin hasNxt
while hasNxt(input, idx):
init
if not cond: break
action
# (x) a # bind action a to (x)
if it[0].kind in {nnkPar, nnkTupleConstr} and it.len == 2:
result = atm(it[0], input, idx, placeholder(it[1], input, idx))
elif it.kind == nnkInfix and it[0].eqIdent"->":
# bind matching to some action:
result = atm(it[1], input, idx, placeholder(it[2], input, idx))
elif it.kind == nnkInfix and it[0].eqIdent"as":
let cond = if it[1].kind in nnkCallKinds: placeholder(it[1], input, idx)
else: newCall(it[1], input, idx)
result = (newLetStmt(it[2], cond),
newCall(interf"success", it[2]),
!!newCall(interf"nxt", input, idx, it[2]))
elif it.kind == nnkPrefix and it[0].eqIdent"*":
let (init, cond, action) = atm(it[1], input, idx, attached)
result = (getAst(buildWhile(input, idx, init, cond, action)),
newEmptyNode(), newEmptyNode())
elif it.kind == nnkPrefix and it[0].eqIdent"+":
# x+ is the same as xx*
result = atm(newTree(nnkTupleConstr, it[1], newTree(nnkPrefix, ident"*", it[1])),
input, idx, attached)
elif it.kind == nnkPrefix and it[0].eqIdent"?":
# optional.
let (init, cond, action) = atm(it[1], input, idx, attached)
if cond.kind == nnkEmpty:
error("'?' operator applied to a non-condition")
else:
result = (newTree(nnkStmtList, init, newIfStmt((cond, action))),
newEmptyNode(), newEmptyNode())
elif it.kind == nnkPrefix and it[0].eqIdent"~":
# not operator
let (init, cond, action) = atm(it[1], input, idx, attached)
if cond.kind == nnkEmpty:
error("'~' operator applied to a non-condition")
else:
result = (init, newCall(bindSym"not", cond), action)
elif it.kind == nnkInfix and it[0].eqIdent"|":
let a = atm(it[1], input, idx, attached)
let b = atm(it[2], input, idx, attached)
if a.cond.kind == nnkEmpty or b.cond.kind == nnkEmpty:
error("'|' operator applied to a non-condition")
else:
result = (newStmtList(a.init, newIfStmt((a.cond, a.action),
(newTree(nnkStmtListExpr, b.init, b.cond), b.action))),
newEmptyNode(), newEmptyNode())
elif it.kind == nnkInfix and it[0].eqIdent"^*":
# a ^* b is rewritten to: (a *(b a))?
#exprList = expr ^+ comma
template tmp(a, b): untyped = ?(a, *(b, a))
result = atm(getAst(tmp(it[1], it[2])), input, idx, attached)
elif it.kind == nnkInfix and it[0].eqIdent"^+":
# a ^* b is rewritten to: (a +(b a))?
template tmp(a, b): untyped = (a, *(b, a))
result = atm(getAst(tmp(it[1], it[2])), input, idx, attached)
elif it.kind == nnkCommand and it.len == 2 and it[0].eqIdent"pred":
# enforce that the wrapped call is interpreted as a predicate, not a non-terminal:
result = (newEmptyNode(), placeholder(it[1], input, idx), newEmptyNode())
else:
var resLen = genSym(nskLet, "resLen")
result = (newLetStmt(resLen, placeholder(it, input, idx)),
newCall(interf"success", resLen),
!!newCall(interf"nxt", input, idx, resLen))
of nnkStrLit..nnkTripleStrLit:
var resLen = genSym(nskLet, "resLen")
result = (newLetStmt(resLen, newCall(interf"skip", input, it, idx)),
newCall(interf"success", resLen),
!!newCall(interf"nxt", input, idx, resLen))
of nnkCurly, nnkAccQuoted, nnkCharLit:
result = (newEmptyNode(), newCall(interf"atom", input, idx, it),
!!newCall(interf"nxt", input, idx))
of nnkCurlyExpr:
if it.len == 3 and it[1].kind == nnkIntLit and it[2].kind == nnkIntLit:
var h = newTree(nnkTupleConstr, it[0])
for count in 2i64 .. it[1].intVal: h.add(it[0])
for count in it[1].intVal .. it[2].intVal-1:
h.add(newTree(nnkPrefix, ident"?", it[0]))
result = atm(h, input, idx, attached)
elif it.len == 2 and it[1].kind == nnkIntLit:
var h = newTree(nnkTupleConstr, it[0])
for count in 2i64 .. it[1].intVal: h.add(it[0])
result = atm(h, input, idx, attached)
else:
error("invalid pattern")
of nnkPar, nnkTupleConstr:
if it.len == 1 and it.kind == nnkPar:
result = atm(it[0], input, idx, attached)
else:
# concatenation:
var conds: seq[StmtTriple] = @[]
for x in it: conds.add atm(x, input, idx, attached)
var res = genSym(nskVar, "res")
result = (newStmtList(newVarStmt(res, newLit false),
toIfChain(conds, idx, res, 0)), res, newEmptyNode())
else:
error("invalid pattern")
#var idx = genSym(nskVar, "idx")
var res = genSym(nskVar, "res")
result = newTree(nnkStmtListExpr, #newVarStmt(idx, newCall(interf"prepare", input)),
newVarStmt(res, newLit false))
var conds: seq[StmtTriple] = @[]
for it in pattern:
conds.add atm(it, input, idx, nil)
result.add toIfChain(conds, idx, res, 0)
result.add res
when defined(debugScanp):
echo repr result
|