1;;; SRFI-1 list-processing library -*- Scheme -*- 2;;; Reference implementation 3;;; 4;;; Copyright (c) 1998, 1999 by Olin Shivers. You may do as you please with 5;;; this code as long as you do not remove this copyright notice or 6;;; hold me liable for its use. Please send bug reports to shivers@ai.mit.edu. 7;;; -Olin 8 9;;; Copyright (c) 2007-2008 SigScheme Project <uim-en AT googlegroups.com> 10 11;; ChangeLog 12;; 13;; 2007-06-15 yamaken - Imported from 14;; http://srfi.schemers.org/srfi-1/srfi-1-reference.scm 15;; and adapted to SigScheme 16;; - Add for-each 17;; 2007-06-30 yamaken - Fix broken arguments receiving of delete-duplicates! 18;; - Fix broken lset-difference call of lset-xor and 19;; lset-xor! (as like as Scheme48) 20;; 2007-07-01 yamaken - Fix broken comparison of list= on 3 or more lists 21;; 2007-07-13 yamaken - Change default value for make-list to #<undef> 22 23 24;;; This is a library of list- and pair-processing functions. I wrote it after 25;;; carefully considering the functions provided by the libraries found in 26;;; R4RS/R5RS Scheme, MIT Scheme, Gambit, RScheme, MzScheme, slib, Common 27;;; Lisp, Bigloo, guile, T, APL and the SML standard basis. It is a pretty 28;;; rich toolkit, providing a superset of the functionality found in any of 29;;; the various Schemes I considered. 30 31;;; This implementation is intended as a portable reference implementation 32;;; for SRFI-1. See the porting notes below for more information. 33 34;;; Exported: 35;;; xcons tree-copy make-list list-tabulate cons* list-copy 36;;; proper-list? circular-list? dotted-list? not-pair? null-list? list= 37;;; circular-list length+ 38;;; iota 39;;; first second third fourth fifth sixth seventh eighth ninth tenth 40;;; car+cdr 41;;; take drop 42;;; take-right drop-right 43;;; take! drop-right! 44;;; split-at split-at! 45;;; last last-pair 46;;; zip unzip1 unzip2 unzip3 unzip4 unzip5 47;;; count 48;;; append! append-reverse append-reverse! concatenate concatenate! 49;;; unfold fold pair-fold reduce 50;;; unfold-right fold-right pair-fold-right reduce-right 51;;; append-map append-map! map! pair-for-each filter-map map-in-order 52;;; filter partition remove 53;;; filter! partition! remove! 54;;; find find-tail any every list-index 55;;; take-while drop-while take-while! 56;;; span break span! break! 57;;; delete delete! 58;;; alist-cons alist-copy 59;;; delete-duplicates delete-duplicates! 60;;; alist-delete alist-delete! 61;;; reverse! 62;;; lset<= lset= lset-adjoin 63;;; lset-union lset-intersection lset-difference lset-xor lset-diff+intersection 64;;; lset-union! lset-intersection! lset-difference! lset-xor! lset-diff+intersection! 65;;; 66;;; In principle, the following R4RS list- and pair-processing procedures 67;;; are also part of this package's exports, although they are not defined 68;;; in this file: 69;;; Primitives: cons pair? null? car cdr set-car! set-cdr! 70;;; Non-primitives: list length append reverse cadr ... cddddr list-ref 71;;; memq memv assq assv 72;;; (The non-primitives are defined in this file, but commented out.) 73;;; 74;;; These R4RS procedures have extended definitions in SRFI-1 and are defined 75;;; in this file: 76;;; map for-each member assoc 77;;; 78;;; The remaining two R4RS list-processing procedures are not included: 79;;; list-tail (use drop) 80;;; list? (use proper-list?) 81 82 83;;; A note on recursion and iteration/reversal: 84;;; Many iterative list-processing algorithms naturally compute the elements 85;;; of the answer list in the wrong order (left-to-right or head-to-tail) from 86;;; the order needed to cons them into the proper answer (right-to-left, or 87;;; tail-then-head). One style or idiom of programming these algorithms, then, 88;;; loops, consing up the elements in reverse order, then destructively 89;;; reverses the list at the end of the loop. I do not do this. The natural 90;;; and efficient way to code these algorithms is recursively. This trades off 91;;; intermediate temporary list structure for intermediate temporary stack 92;;; structure. In a stack-based system, this improves cache locality and 93;;; lightens the load on the GC system. Don't stand on your head to iterate! 94;;; Recurse, where natural. Multiple-value returns make this even more 95;;; convenient, when the recursion/iteration has multiple state values. 96 97;;; Porting: 98;;; This is carefully tuned code; do not modify casually. 99;;; - It is careful to share storage when possible; 100;;; - Side-effecting code tries not to perform redundant writes. 101;;; 102;;; That said, a port of this library to a specific Scheme system might wish 103;;; to tune this code to exploit particulars of the implementation. 104;;; The single most important compiler-specific optimisation you could make 105;;; to this library would be to add rewrite rules or transforms to: 106;;; - transform applications of n-ary procedures (e.g. LIST=, CONS*, APPEND, 107;;; LSET-UNION) into multiple applications of a primitive two-argument 108;;; variant. 109;;; - transform applications of the mapping functions (MAP, FOR-EACH, FOLD, 110;;; ANY, EVERY) into open-coded loops. The killer here is that these 111;;; functions are n-ary. Handling the general case is quite inefficient, 112;;; requiring many intermediate data structures to be allocated and 113;;; discarded. 114;;; - transform applications of procedures that take optional arguments 115;;; into calls to variants that do not take optional arguments. This 116;;; eliminates unnecessary consing and parsing of the rest parameter. 117;;; 118;;; These transforms would provide BIG speedups. In particular, the n-ary 119;;; mapping functions are particularly slow and cons-intensive, and are good 120;;; candidates for tuning. I have coded fast paths for the single-list cases, 121;;; but what you really want to do is exploit the fact that the compiler 122;;; usually knows how many arguments are being passed to a particular 123;;; application of these functions -- they are usually explicitly called, not 124;;; passed around as higher-order values. If you can arrange to have your 125;;; compiler produce custom code or custom linkages based on the number of 126;;; arguments in the call, you can speed these functions up a *lot*. But this 127;;; kind of compiler technology no longer exists in the Scheme world as far as 128;;; I can see. 129;;; 130;;; Note that this code is, of course, dependent upon standard bindings for 131;;; the R5RS procedures -- i.e., it assumes that the variable CAR is bound 132;;; to the procedure that takes the car of a list. If your Scheme 133;;; implementation allows user code to alter the bindings of these procedures 134;;; in a manner that would be visible to these definitions, then there might 135;;; be trouble. You could consider horrible kludgery along the lines of 136;;; (define fact 137;;; (let ((= =) (- -) (* *)) 138;;; (letrec ((real-fact (lambda (n) 139;;; (if (= n 0) 1 (* n (real-fact (- n 1))))))) 140;;; real-fact))) 141;;; Or you could consider shifting to a reasonable Scheme system that, say, 142;;; has a module system protecting code from this kind of lossage. 143;;; 144;;; This code does a fair amount of run-time argument checking. If your 145;;; Scheme system has a sophisticated compiler that can eliminate redundant 146;;; error checks, this is no problem. However, if not, these checks incur 147;;; some performance overhead -- and, in a safe Scheme implementation, they 148;;; are in some sense redundant: if we don't check to see that the PROC 149;;; parameter is a procedure, we'll find out anyway three lines later when 150;;; we try to call the value. It's pretty easy to rip all this argument 151;;; checking code out if it's inappropriate for your implementation -- just 152;;; nuke every call to CHECK-ARG. 153;;; 154;;; On the other hand, if you *do* have a sophisticated compiler that will 155;;; actually perform soft-typing and eliminate redundant checks (Rice's systems 156;;; being the only possible candidate of which I'm aware), leaving these checks 157;;; in can *help*, since their presence can be elided in redundant cases, 158;;; and in cases where they are needed, performing the checks early, at 159;;; procedure entry, can "lift" a check out of a loop. 160;;; 161;;; Finally, I have only checked the properties that can portably be checked 162;;; with R5RS Scheme -- and this is not complete. You may wish to alter 163;;; the CHECK-ARG parameter checks to perform extra, implementation-specific 164;;; checks, such as procedure arity for higher-order values. 165;;; 166;;; The code has only these non-R4RS dependencies: 167;;; A few calls to an ERROR procedure; 168;;; Uses of the R5RS multiple-value procedure VALUES and the m-v binding 169;;; RECEIVE macro (which isn't R5RS, but is a trivial macro). 170;;; Many calls to a parameter-checking procedure check-arg: 171;;; (define (check-arg pred val caller) 172;;; (let lp ((val val)) 173;;; (if (pred val) val (lp (error "Bad argument" val pred caller))))) 174;;; A few uses of the LET-OPTIONAL and :OPTIONAL macros for parsing 175;;; optional arguments. 176;;; 177;;; Most of these procedures use the NULL-LIST? test to trigger the 178;;; base case in the inner loop or recursion. The NULL-LIST? function 179;;; is defined to be a careful one -- it raises an error if passed a 180;;; non-nil, non-pair value. The spec allows an implementation to use 181;;; a less-careful implementation that simply defines NULL-LIST? to 182;;; be NOT-PAIR?. This would speed up the inner loops of these procedures 183;;; at the expense of having them silently accept dotted lists. 184 185;;; A note on dotted lists: 186;;; I, personally, take the view that the only consistent view of lists 187;;; in Scheme is the view that *everything* is a list -- values such as 188;;; 3 or "foo" or 'bar are simply empty dotted lists. This is due to the 189;;; fact that Scheme actually has no true list type. It has a pair type, 190;;; and there is an *interpretation* of the trees built using this type 191;;; as lists. 192;;; 193;;; I lobbied to have these list-processing procedures hew to this 194;;; view, and accept any value as a list argument. I was overwhelmingly 195;;; overruled during the SRFI discussion phase. So I am inserting this 196;;; text in the reference lib and the SRFI spec as a sort of "minority 197;;; opinion" dissent. 198;;; 199;;; Many of the procedures in this library can be trivially redefined 200;;; to handle dotted lists, just by changing the NULL-LIST? base-case 201;;; check to NOT-PAIR?, meaning that any non-pair value is taken to be 202;;; an empty list. For most of these procedures, that's all that is 203;;; required. 204;;; 205;;; However, we have to do a little more work for some procedures that 206;;; *produce* lists from other lists. Were we to extend these procedures to 207;;; accept dotted lists, we would have to define how they terminate the lists 208;;; produced as results when passed a dotted list. I designed a coherent set 209;;; of termination rules for these cases; this was posted to the SRFI-1 210;;; discussion list. I additionally wrote an earlier version of this library 211;;; that implemented that spec. It has been discarded during later phases of 212;;; the definition and implementation of this library. 213;;; 214;;; The argument *against* defining these procedures to work on dotted 215;;; lists is that dotted lists are the rare, odd case, and that by 216;;; arranging for the procedures to handle them, we lose error checking 217;;; in the cases where a dotted list is passed by accident -- e.g., when 218;;; the programmer swaps a two arguments to a list-processing function, 219;;; one being a scalar and one being a list. For example, 220;;; (member '(1 3 5 7 9) 7) 221;;; This would quietly return #f if we extended MEMBER to accept dotted 222;;; lists. 223;;; 224;;; The SRFI discussion record contains more discussion on this topic. 225 226;;; SigScheme adaptation 227;;;;;;;;;;;;;;;;;;;;;;;; 228 229(require-extension (srfi 8 23)) 230 231(define %srfi-1:undefined (for-each values '())) 232 233(define (check-arg pred val caller) 234 (let lp ((val val)) 235 (if (pred val) val (lp (error "Bad argument" val pred caller))))) 236;; If you need efficiency, define this once SRFI-1 has been enabled. 237;;(define (check-arg . args) #f) 238 239(define :optional 240 (lambda (opt default) 241 (case (length opt) 242 ((0) default) 243 ((1) (car opt)) 244 (else (error "superfluous arguments"))))) 245 246 247;;; Constructors 248;;;;;;;;;;;;;;;; 249 250;;; Occasionally useful as a value to be passed to a fold or other 251;;; higher-order procedure. 252(define (xcons d a) (cons a d)) 253 254;;;; Recursively copy every cons. 255;(define (tree-copy x) 256; (let recur ((x x)) 257; (if (not (pair? x)) x 258; (cons (recur (car x)) (recur (cdr x)))))) 259 260;;; Make a list of length LEN. 261 262(define (make-list len . maybe-elt) 263 (check-arg (lambda (n) (and (integer? n) (>= n 0))) len make-list) 264 (let ((elt (cond ((null? maybe-elt) %srfi-1:undefined) ; Default value 265 ((null? (cdr maybe-elt)) (car maybe-elt)) 266 (else (error "Too many arguments to MAKE-LIST" 267 (cons len maybe-elt)))))) 268 (do ((i len (- i 1)) 269 (ans '() (cons elt ans))) 270 ((<= i 0) ans)))) 271 272 273;(define (list . ans) ans) ; R4RS 274 275 276;;; Make a list of length LEN. Elt i is (PROC i) for 0 <= i < LEN. 277 278(define (list-tabulate len proc) 279 (check-arg (lambda (n) (and (integer? n) (>= n 0))) len list-tabulate) 280 (check-arg procedure? proc list-tabulate) 281 (do ((i (- len 1) (- i 1)) 282 (ans '() (cons (proc i) ans))) 283 ((< i 0) ans))) 284 285;;; (cons* a1 a2 ... an) = (cons a1 (cons a2 (cons ... an))) 286;;; (cons* a1) = a1 (cons* a1 a2 ...) = (cons a1 (cons* a2 ...)) 287;;; 288;;; (cons first (unfold not-pair? car cdr rest values)) 289 290(define (cons* first . rest) 291 (let recur ((x first) (rest rest)) 292 (if (pair? rest) 293 (cons x (recur (car rest) (cdr rest))) 294 x))) 295 296;;; (unfold not-pair? car cdr lis values) 297 298(define (list-copy lis) 299 (let recur ((lis lis)) 300 (if (pair? lis) 301 (cons (car lis) (recur (cdr lis))) 302 lis))) 303 304;;; IOTA count [start step] (start start+step ... start+(count-1)*step) 305 306(define (iota count . maybe-start+step) 307 (check-arg integer? count iota) 308 (if (< count 0) (error "Negative step count" iota count)) 309 (let-optionals* maybe-start+step ((start 0) (step 1) . must-be-null) 310 (check-arg number? start iota) 311 (check-arg number? step iota) 312 (if (not (null? must-be-null)) (error "superfluous arguments")) 313 (let ((last-val (+ start (* (- count 1) step)))) 314 (do ((count count (- count 1)) 315 (val last-val (- val step)) 316 (ans '() (cons val ans))) 317 ((<= count 0) ans))))) 318 319;;; I thought these were lovely, but the public at large did not share my 320;;; enthusiasm... 321;;; :IOTA to (0 ... to-1) 322;;; :IOTA from to (from ... to-1) 323;;; :IOTA from to step (from from+step ...) 324 325;;; IOTA: to (1 ... to) 326;;; IOTA: from to (from+1 ... to) 327;;; IOTA: from to step (from+step from+2step ...) 328 329;(define (%parse-iota-args arg1 rest-args proc) 330; (let ((check (lambda (n) (check-arg integer? n proc)))) 331; (check arg1) 332; (if (pair? rest-args) 333; (let ((arg2 (check (car rest-args))) 334; (rest (cdr rest-args))) 335; (if (pair? rest) 336; (let ((arg3 (check (car rest))) 337; (rest (cdr rest))) 338; (if (pair? rest) (error "Too many parameters" proc arg1 rest-args) 339; (values arg1 arg2 arg3))) 340; (values arg1 arg2 1))) 341; (values 0 arg1 1)))) 342; 343;(define (iota: arg1 . rest-args) 344; (receive (from to step) (%parse-iota-args arg1 rest-args iota:) 345; (let* ((numsteps (floor (/ (- to from) step))) 346; (last-val (+ from (* step numsteps)))) 347; (if (< numsteps 0) (error "Negative step count" iota: from to step)) 348; (do ((steps-left numsteps (- steps-left 1)) 349; (val last-val (- val step)) 350; (ans '() (cons val ans))) 351; ((<= steps-left 0) ans))))) 352; 353; 354;(define (:iota arg1 . rest-args) 355; (receive (from to step) (%parse-iota-args arg1 rest-args :iota) 356; (let* ((numsteps (ceiling (/ (- to from) step))) 357; (last-val (+ from (* step (- numsteps 1))))) 358; (if (< numsteps 0) (error "Negative step count" :iota from to step)) 359; (do ((steps-left numsteps (- steps-left 1)) 360; (val last-val (- val step)) 361; (ans '() (cons val ans))) 362; ((<= steps-left 0) ans))))) 363 364 365 366(define (circular-list val1 . vals) 367 (let ((ans (cons val1 vals))) 368 (set-cdr! (last-pair ans) ans) 369 ans)) 370 371;;; <proper-list> ::= () ; Empty proper list 372;;; | (cons <x> <proper-list>) ; Proper-list pair 373;;; Note that this definition rules out circular lists -- and this 374;;; function is required to detect this case and return false. 375 376(define (proper-list? x) 377 (let lp ((x x) (lag x)) 378 (if (pair? x) 379 (let ((x (cdr x))) 380 (if (pair? x) 381 (let ((x (cdr x)) 382 (lag (cdr lag))) 383 (and (not (eq? x lag)) (lp x lag))) 384 (null? x))) 385 (null? x)))) 386 387 388;;; A dotted list is a finite list (possibly of length 0) terminated 389;;; by a non-nil value. Any non-cons, non-nil value (e.g., "foo" or 5) 390;;; is a dotted list of length 0. 391;;; 392;;; <dotted-list> ::= <non-nil,non-pair> ; Empty dotted list 393;;; | (cons <x> <dotted-list>) ; Proper-list pair 394 395(define (dotted-list? x) 396 (let lp ((x x) (lag x)) 397 (if (pair? x) 398 (let ((x (cdr x))) 399 (if (pair? x) 400 (let ((x (cdr x)) 401 (lag (cdr lag))) 402 (and (not (eq? x lag)) (lp x lag))) 403 (not (null? x)))) 404 (not (null? x))))) 405 406(define (circular-list? x) 407 (let lp ((x x) (lag x)) 408 (and (pair? x) 409 (let ((x (cdr x))) 410 (and (pair? x) 411 (let ((x (cdr x)) 412 (lag (cdr lag))) 413 (or (eq? x lag) (lp x lag)))))))) 414 415(define (not-pair? x) (not (pair? x))) ; Inline me. 416 417;;; This is a legal definition which is fast and sloppy: 418;;; (define null-list? not-pair?) 419;;; but we'll provide a more careful one: 420(define (null-list? l) 421 (cond ((pair? l) #f) 422 ((null? l) #t) 423 (else (error "null-list?: argument out of domain" l)))) 424 425 426(define (list= = . lists) 427 (or (null? lists) ; special case 428 429 (let lp1 ((list-a (car lists)) (others (cdr lists))) 430 (or (null? others) 431 (let ((list-b (car others)) 432 (others (cdr others))) 433 (if (eq? list-a list-b) ; EQ? => LIST= 434 (lp1 list-b others) 435 (let lp2 ((tail-a list-a) (tail-b list-b)) 436 (if (null-list? tail-a) 437 (and (null-list? tail-b) 438 (lp1 list-b others)) 439 (and (not (null-list? tail-b)) 440 (= (car tail-a) (car tail-b)) 441 (lp2 (cdr tail-a) (cdr tail-b))))))))))) 442 443 444 445;;; R4RS, so commented out. 446;(define (length x) ; LENGTH may diverge or 447; (let lp ((x x) (len 0)) ; raise an error if X is 448; (if (pair? x) ; a circular list. This version 449; (lp (cdr x) (+ len 1)) ; diverges. 450; len))) 451 452(define (length+ x) ; Returns #f if X is circular. 453 (let lp ((x x) (lag x) (len 0)) 454 (if (pair? x) 455 (let ((x (cdr x)) 456 (len (+ len 1))) 457 (if (pair? x) 458 (let ((x (cdr x)) 459 (lag (cdr lag)) 460 (len (+ len 1))) 461 (and (not (eq? x lag)) (lp x lag len))) 462 len)) 463 len))) 464 465(define (zip list1 . more-lists) (apply map list list1 more-lists)) 466 467 468;;; Selectors 469;;;;;;;;;;;;; 470 471;;; R4RS non-primitives: 472;(define (caar x) (car (car x))) 473;(define (cadr x) (car (cdr x))) 474;(define (cdar x) (cdr (car x))) 475;(define (cddr x) (cdr (cdr x))) 476; 477;(define (caaar x) (caar (car x))) 478;(define (caadr x) (caar (cdr x))) 479;(define (cadar x) (cadr (car x))) 480;(define (caddr x) (cadr (cdr x))) 481;(define (cdaar x) (cdar (car x))) 482;(define (cdadr x) (cdar (cdr x))) 483;(define (cddar x) (cddr (car x))) 484;(define (cdddr x) (cddr (cdr x))) 485; 486;(define (caaaar x) (caaar (car x))) 487;(define (caaadr x) (caaar (cdr x))) 488;(define (caadar x) (caadr (car x))) 489;(define (caaddr x) (caadr (cdr x))) 490;(define (cadaar x) (cadar (car x))) 491;(define (cadadr x) (cadar (cdr x))) 492;(define (caddar x) (caddr (car x))) 493;(define (cadddr x) (caddr (cdr x))) 494;(define (cdaaar x) (cdaar (car x))) 495;(define (cdaadr x) (cdaar (cdr x))) 496;(define (cdadar x) (cdadr (car x))) 497;(define (cdaddr x) (cdadr (cdr x))) 498;(define (cddaar x) (cddar (car x))) 499;(define (cddadr x) (cddar (cdr x))) 500;(define (cdddar x) (cdddr (car x))) 501;(define (cddddr x) (cdddr (cdr x))) 502 503 504(define first car) 505(define second cadr) 506(define third caddr) 507(define fourth cadddr) 508(define (fifth x) (car (cddddr x))) 509(define (sixth x) (cadr (cddddr x))) 510(define (seventh x) (caddr (cddddr x))) 511(define (eighth x) (cadddr (cddddr x))) 512(define (ninth x) (car (cddddr (cddddr x)))) 513(define (tenth x) (cadr (cddddr (cddddr x)))) 514 515(define (car+cdr pair) (values (car pair) (cdr pair))) 516 517;;; take & drop 518 519(define (take lis k) 520 (check-arg integer? k take) 521 (let recur ((lis lis) (k k)) 522 (if (zero? k) '() 523 (cons (car lis) 524 (recur (cdr lis) (- k 1)))))) 525 526(define (drop lis k) 527 (check-arg integer? k drop) 528 (let iter ((lis lis) (k k)) 529 (if (zero? k) lis (iter (cdr lis) (- k 1))))) 530 531(define (take! lis k) 532 (check-arg integer? k take!) 533 (if (zero? k) '() 534 (begin (set-cdr! (drop lis (- k 1)) '()) 535 lis))) 536 537;;; TAKE-RIGHT and DROP-RIGHT work by getting two pointers into the list, 538;;; off by K, then chasing down the list until the lead pointer falls off 539;;; the end. 540 541(define (take-right lis k) 542 (check-arg integer? k take-right) 543 (let lp ((lag lis) (lead (drop lis k))) 544 (if (pair? lead) 545 (lp (cdr lag) (cdr lead)) 546 lag))) 547 548(define (drop-right lis k) 549 (check-arg integer? k drop-right) 550 (let recur ((lag lis) (lead (drop lis k))) 551 (if (pair? lead) 552 (cons (car lag) (recur (cdr lag) (cdr lead))) 553 '()))) 554 555;;; In this function, LEAD is actually K+1 ahead of LAG. This lets 556;;; us stop LAG one step early, in time to smash its cdr to (). 557(define (drop-right! lis k) 558 (check-arg integer? k drop-right!) 559 (let ((lead (drop lis k))) 560 (if (pair? lead) 561 562 (let lp ((lag lis) (lead (cdr lead))) ; Standard case 563 (if (pair? lead) 564 (lp (cdr lag) (cdr lead)) 565 (begin (set-cdr! lag '()) 566 lis))) 567 568 '()))) ; Special case dropping everything -- no cons to side-effect. 569 570;(define (list-ref lis i) (car (drop lis i))) ; R4RS 571 572;;; These use the APL convention, whereby negative indices mean 573;;; "from the right." I liked them, but they didn't win over the 574;;; SRFI reviewers. 575;;; K >= 0: Take and drop K elts from the front of the list. 576;;; K <= 0: Take and drop -K elts from the end of the list. 577 578;(define (take lis k) 579; (check-arg integer? k take) 580; (if (negative? k) 581; (list-tail lis (+ k (length lis))) 582; (let recur ((lis lis) (k k)) 583; (if (zero? k) '() 584; (cons (car lis) 585; (recur (cdr lis) (- k 1))))))) 586; 587;(define (drop lis k) 588; (check-arg integer? k drop) 589; (if (negative? k) 590; (let recur ((lis lis) (nelts (+ k (length lis)))) 591; (if (zero? nelts) '() 592; (cons (car lis) 593; (recur (cdr lis) (- nelts 1))))) 594; (list-tail lis k))) 595; 596; 597;(define (take! lis k) 598; (check-arg integer? k take!) 599; (cond ((zero? k) '()) 600; ((positive? k) 601; (set-cdr! (list-tail lis (- k 1)) '()) 602; lis) 603; (else (list-tail lis (+ k (length lis)))))) 604; 605;(define (drop! lis k) 606; (check-arg integer? k drop!) 607; (if (negative? k) 608; (let ((nelts (+ k (length lis)))) 609; (if (zero? nelts) '() 610; (begin (set-cdr! (list-tail lis (- nelts 1)) '()) 611; lis))) 612; (list-tail lis k))) 613 614(define (split-at x k) 615 (check-arg integer? k split-at) 616 (let recur ((lis x) (k k)) 617 (if (zero? k) (values '() lis) 618 (receive (prefix suffix) (recur (cdr lis) (- k 1)) 619 (values (cons (car lis) prefix) suffix))))) 620 621(define (split-at! x k) 622 (check-arg integer? k split-at!) 623 (if (zero? k) (values '() x) 624 (let* ((prev (drop x (- k 1))) 625 (suffix (cdr prev))) 626 (set-cdr! prev '()) 627 (values x suffix)))) 628 629 630(define (last lis) (car (last-pair lis))) 631 632(define (last-pair lis) 633 (check-arg pair? lis last-pair) 634 (let lp ((lis lis)) 635 (let ((tail (cdr lis))) 636 (if (pair? tail) (lp tail) lis)))) 637 638 639;;; Unzippers -- 1 through 5 640;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 641 642(define (unzip1 lis) (map car lis)) 643 644(define (unzip2 lis) 645 (let recur ((lis lis)) 646 (if (null-list? lis) (values lis lis) ; Use NOT-PAIR? to handle 647 (let ((elt (car lis))) ; dotted lists. 648 (receive (a b) (recur (cdr lis)) 649 (values (cons (car elt) a) 650 (cons (cadr elt) b))))))) 651 652(define (unzip3 lis) 653 (let recur ((lis lis)) 654 (if (null-list? lis) (values lis lis lis) 655 (let ((elt (car lis))) 656 (receive (a b c) (recur (cdr lis)) 657 (values (cons (car elt) a) 658 (cons (cadr elt) b) 659 (cons (caddr elt) c))))))) 660 661(define (unzip4 lis) 662 (let recur ((lis lis)) 663 (if (null-list? lis) (values lis lis lis lis) 664 (let ((elt (car lis))) 665 (receive (a b c d) (recur (cdr lis)) 666 (values (cons (car elt) a) 667 (cons (cadr elt) b) 668 (cons (caddr elt) c) 669 (cons (cadddr elt) d))))))) 670 671(define (unzip5 lis) 672 (let recur ((lis lis)) 673 (if (null-list? lis) (values lis lis lis lis lis) 674 (let ((elt (car lis))) 675 (receive (a b c d e) (recur (cdr lis)) 676 (values (cons (car elt) a) 677 (cons (cadr elt) b) 678 (cons (caddr elt) c) 679 (cons (cadddr elt) d) 680 (cons (car (cddddr elt)) e))))))) 681 682 683;;; append! append-reverse append-reverse! concatenate concatenate! 684;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 685 686(define (append! . lists) 687 ;; First, scan through lists looking for a non-empty one. 688 (let lp ((lists lists) (prev '())) 689 (if (not (pair? lists)) prev 690 (let ((first (car lists)) 691 (rest (cdr lists))) 692 (if (not (pair? first)) (lp rest first) 693 694 ;; Now, do the splicing. 695 (let lp2 ((tail-cons (last-pair first)) 696 (rest rest)) 697 (if (pair? rest) 698 (let ((next (car rest)) 699 (rest (cdr rest))) 700 (set-cdr! tail-cons next) 701 (lp2 (if (pair? next) (last-pair next) tail-cons) 702 rest)) 703 first))))))) 704 705;;; APPEND is R4RS. 706;(define (append . lists) 707; (if (pair? lists) 708; (let recur ((list1 (car lists)) (lists (cdr lists))) 709; (if (pair? lists) 710; (let ((tail (recur (car lists) (cdr lists)))) 711; (fold-right cons tail list1)) ; Append LIST1 & TAIL. 712; list1)) 713; '())) 714 715;(define (append-reverse rev-head tail) (fold cons tail rev-head)) 716 717;(define (append-reverse! rev-head tail) 718; (pair-fold (lambda (pair tail) (set-cdr! pair tail) pair) 719; tail 720; rev-head)) 721 722;;; Hand-inline the FOLD and PAIR-FOLD ops for speed. 723 724(define (append-reverse rev-head tail) 725 (let lp ((rev-head rev-head) (tail tail)) 726 (if (null-list? rev-head) tail 727 (lp (cdr rev-head) (cons (car rev-head) tail))))) 728 729(define (append-reverse! rev-head tail) 730 (let lp ((rev-head rev-head) (tail tail)) 731 (if (null-list? rev-head) tail 732 (let ((next-rev (cdr rev-head))) 733 (set-cdr! rev-head tail) 734 (lp next-rev rev-head))))) 735 736 737(define (concatenate lists) (reduce-right append '() lists)) 738(define (concatenate! lists) (reduce-right append! '() lists)) 739 740;;; Fold/map internal utilities 741;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 742;;; These little internal utilities are used by the general 743;;; fold & mapper funs for the n-ary cases . It'd be nice if they got inlined. 744;;; One the other hand, the n-ary cases are painfully inefficient as it is. 745;;; An aggressive implementation should simply re-write these functions 746;;; for raw efficiency; I have written them for as much clarity, portability, 747;;; and simplicity as can be achieved. 748;;; 749;;; I use the dreaded call/cc to do local aborts. A good compiler could 750;;; handle this with extreme efficiency. An implementation that provides 751;;; a one-shot, non-persistent continuation grabber could help the compiler 752;;; out by using that in place of the call/cc's in these routines. 753;;; 754;;; These functions have funky definitions that are precisely tuned to 755;;; the needs of the fold/map procs -- for example, to minimize the number 756;;; of times the argument lists need to be examined. 757 758;;; Return (map cdr lists). 759;;; However, if any element of LISTS is empty, just abort and return '(). 760(define (%cdrs lists) 761 (call-with-current-continuation 762 (lambda (abort) 763 (let recur ((lists lists)) 764 (if (pair? lists) 765 (let ((lis (car lists))) 766 (if (null-list? lis) (abort '()) 767 (cons (cdr lis) (recur (cdr lists))))) 768 '()))))) 769 770(define (%cars+ lists last-elt) ; (append! (map car lists) (list last-elt)) 771 (let recur ((lists lists)) 772 (if (pair? lists) (cons (caar lists) (recur (cdr lists))) (list last-elt)))) 773 774;;; LISTS is a (not very long) non-empty list of lists. 775;;; Return two lists: the cars & the cdrs of the lists. 776;;; However, if any of the lists is empty, just abort and return [() ()]. 777 778(define (%cars+cdrs lists) 779 (call-with-current-continuation 780 (lambda (abort) 781 (let recur ((lists lists)) 782 (if (pair? lists) 783 (receive (list other-lists) (car+cdr lists) 784 (if (null-list? list) (abort '() '()) ; LIST is empty -- bail out 785 (receive (a d) (car+cdr list) 786 (receive (cars cdrs) (recur other-lists) 787 (values (cons a cars) (cons d cdrs)))))) 788 (values '() '())))))) 789 790;;; Like %CARS+CDRS, but we pass in a final elt tacked onto the end of the 791;;; cars list. What a hack. 792(define (%cars+cdrs+ lists cars-final) 793 (call-with-current-continuation 794 (lambda (abort) 795 (let recur ((lists lists)) 796 (if (pair? lists) 797 (receive (list other-lists) (car+cdr lists) 798 (if (null-list? list) (abort '() '()) ; LIST is empty -- bail out 799 (receive (a d) (car+cdr list) 800 (receive (cars cdrs) (recur other-lists) 801 (values (cons a cars) (cons d cdrs)))))) 802 (values (list cars-final) '())))))) 803 804;;; Like %CARS+CDRS, but blow up if any list is empty. 805(define (%cars+cdrs/no-test lists) 806 (let recur ((lists lists)) 807 (if (pair? lists) 808 (receive (list other-lists) (car+cdr lists) 809 (receive (a d) (car+cdr list) 810 (receive (cars cdrs) (recur other-lists) 811 (values (cons a cars) (cons d cdrs))))) 812 (values '() '())))) 813 814 815;;; count 816;;;;;;;;; 817(define (count pred list1 . lists) 818 (check-arg procedure? pred count) 819 (if (pair? lists) 820 821 ;; N-ary case 822 (let lp ((list1 list1) (lists lists) (i 0)) 823 (if (null-list? list1) i 824 (receive (as ds) (%cars+cdrs lists) 825 (if (null? as) i 826 (lp (cdr list1) ds 827 (if (apply pred (car list1) as) (+ i 1) i)))))) 828 829 ;; Fast path 830 (let lp ((lis list1) (i 0)) 831 (if (null-list? lis) i 832 (lp (cdr lis) (if (pred (car lis)) (+ i 1) i)))))) 833 834 835;;; fold/unfold 836;;;;;;;;;;;;;;; 837 838(define (unfold-right p f g seed . maybe-tail) 839 (check-arg procedure? p unfold-right) 840 (check-arg procedure? f unfold-right) 841 (check-arg procedure? g unfold-right) 842 (let lp ((seed seed) (ans (:optional maybe-tail '()))) 843 (if (p seed) ans 844 (lp (g seed) 845 (cons (f seed) ans))))) 846 847 848(define (unfold p f g seed . maybe-tail-gen) 849 (check-arg procedure? p unfold) 850 (check-arg procedure? f unfold) 851 (check-arg procedure? g unfold) 852 (if (pair? maybe-tail-gen) 853 854 (let ((tail-gen (car maybe-tail-gen))) 855 (if (pair? (cdr maybe-tail-gen)) 856 (apply error "Too many arguments" unfold p f g seed maybe-tail-gen) 857 858 (let recur ((seed seed)) 859 (if (p seed) (tail-gen seed) 860 (cons (f seed) (recur (g seed))))))) 861 862 (let recur ((seed seed)) 863 (if (p seed) '() 864 (cons (f seed) (recur (g seed))))))) 865 866 867(define (fold kons knil lis1 . lists) 868 (check-arg procedure? kons fold) 869 (if (pair? lists) 870 (let lp ((lists (cons lis1 lists)) (ans knil)) ; N-ary case 871 (receive (cars+ans cdrs) (%cars+cdrs+ lists ans) 872 (if (null? cars+ans) ans ; Done. 873 (lp cdrs (apply kons cars+ans))))) 874 875 (let lp ((lis lis1) (ans knil)) ; Fast path 876 (if (null-list? lis) ans 877 (lp (cdr lis) (kons (car lis) ans)))))) 878 879 880(define (fold-right kons knil lis1 . lists) 881 (check-arg procedure? kons fold-right) 882 (if (pair? lists) 883 (let recur ((lists (cons lis1 lists))) ; N-ary case 884 (let ((cdrs (%cdrs lists))) 885 (if (null? cdrs) knil 886 (apply kons (%cars+ lists (recur cdrs)))))) 887 888 (let recur ((lis lis1)) ; Fast path 889 (if (null-list? lis) knil 890 (let ((head (car lis))) 891 (kons head (recur (cdr lis)))))))) 892 893 894(define (pair-fold-right f zero lis1 . lists) 895 (check-arg procedure? f pair-fold-right) 896 (if (pair? lists) 897 (let recur ((lists (cons lis1 lists))) ; N-ary case 898 (let ((cdrs (%cdrs lists))) 899 (if (null? cdrs) zero 900 (apply f (append! lists (list (recur cdrs))))))) 901 902 (let recur ((lis lis1)) ; Fast path 903 (if (null-list? lis) zero (f lis (recur (cdr lis))))))) 904 905(define (pair-fold f zero lis1 . lists) 906 (check-arg procedure? f pair-fold) 907 (if (pair? lists) 908 (let lp ((lists (cons lis1 lists)) (ans zero)) ; N-ary case 909 (let ((tails (%cdrs lists))) 910 (if (null? tails) ans 911 (lp tails (apply f (append! lists (list ans))))))) 912 913 (let lp ((lis lis1) (ans zero)) 914 (if (null-list? lis) ans 915 (let ((tail (cdr lis))) ; Grab the cdr now, 916 (lp tail (f lis ans))))))) ; in case F SET-CDR!s LIS. 917 918 919;;; REDUCE and REDUCE-RIGHT only use RIDENTITY in the empty-list case. 920;;; These cannot meaningfully be n-ary. 921 922(define (reduce f ridentity lis) 923 (check-arg procedure? f reduce) 924 (if (null-list? lis) ridentity 925 (fold f (car lis) (cdr lis)))) 926 927(define (reduce-right f ridentity lis) 928 (check-arg procedure? f reduce-right) 929 (if (null-list? lis) ridentity 930 (let recur ((head (car lis)) (lis (cdr lis))) 931 (if (pair? lis) 932 (f head (recur (car lis) (cdr lis))) 933 head)))) 934 935 936 937;;; Mappers: append-map append-map! pair-for-each map! filter-map map-in-order 938;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 939 940(define (append-map f lis1 . lists) 941 (really-append-map append-map append f lis1 lists)) 942(define (append-map! f lis1 . lists) 943 (really-append-map append-map! append! f lis1 lists)) 944 945(define (really-append-map who appender f lis1 lists) 946 (check-arg procedure? f who) 947 (if (pair? lists) 948 (receive (cars cdrs) (%cars+cdrs (cons lis1 lists)) 949 (if (null? cars) '() 950 (let recur ((cars cars) (cdrs cdrs)) 951 (let ((vals (apply f cars))) 952 (receive (cars2 cdrs2) (%cars+cdrs cdrs) 953 (if (null? cars2) vals 954 (appender vals (recur cars2 cdrs2)))))))) 955 956 ;; Fast path 957 (if (null-list? lis1) '() 958 (let recur ((elt (car lis1)) (rest (cdr lis1))) 959 (let ((vals (f elt))) 960 (if (null-list? rest) vals 961 (appender vals (recur (car rest) (cdr rest))))))))) 962 963 964(define (pair-for-each proc lis1 . lists) 965 (check-arg procedure? proc pair-for-each) 966 (if (pair? lists) 967 968 (let lp ((lists (cons lis1 lists))) 969 (let ((tails (%cdrs lists))) 970 (if (pair? tails) 971 (begin (apply proc lists) 972 (lp tails))))) 973 974 ;; Fast path. 975 (let lp ((lis lis1)) 976 (if (not (null-list? lis)) 977 (let ((tail (cdr lis))) ; Grab the cdr now, 978 (proc lis) ; in case PROC SET-CDR!s LIS. 979 (lp tail)))))) 980 981;;; We stop when LIS1 runs out, not when any list runs out. 982(define (map! f lis1 . lists) 983 (check-arg procedure? f map!) 984 (if (pair? lists) 985 (let lp ((lis1 lis1) (lists lists)) 986 (if (not (null-list? lis1)) 987 (receive (heads tails) (%cars+cdrs/no-test lists) 988 (set-car! lis1 (apply f (car lis1) heads)) 989 (lp (cdr lis1) tails)))) 990 991 ;; Fast path. 992 (pair-for-each (lambda (pair) (set-car! pair (f (car pair)))) lis1)) 993 lis1) 994 995 996;;; Map F across L, and save up all the non-false results. 997(define (filter-map f lis1 . lists) 998 (check-arg procedure? f filter-map) 999 (if (pair? lists) 1000 (let recur ((lists (cons lis1 lists))) 1001 (receive (cars cdrs) (%cars+cdrs lists) 1002 (if (pair? cars) 1003 (cond ((apply f cars) => (lambda (x) (cons x (recur cdrs)))) 1004 (else (recur cdrs))) ; Tail call in this arm. 1005 '()))) 1006 1007 ;; Fast path. 1008 (let recur ((lis lis1)) 1009 (if (null-list? lis) lis 1010 (let ((tail (recur (cdr lis)))) 1011 (cond ((f (car lis)) => (lambda (x) (cons x tail))) 1012 (else tail))))))) 1013 1014 1015;;; Map F across lists, guaranteeing to go left-to-right. 1016;;; NOTE: Some implementations of R5RS MAP are compliant with this spec; 1017;;; in which case this procedure may simply be defined as a synonym for MAP. 1018 1019(define (map-in-order f lis1 . lists) 1020 (check-arg procedure? f map-in-order) 1021 (if (pair? lists) 1022 (let recur ((lists (cons lis1 lists))) 1023 (receive (cars cdrs) (%cars+cdrs lists) 1024 (if (pair? cars) 1025 (let ((x (apply f cars))) ; Do head first, 1026 (cons x (recur cdrs))) ; then tail. 1027 '()))) 1028 1029 ;; Fast path. 1030 (let recur ((lis lis1)) 1031 (if (null-list? lis) lis 1032 (let ((tail (cdr lis)) 1033 (x (f (car lis)))) ; Do head first, 1034 (cons x (recur tail))))))) ; then tail. 1035 1036 1037;;; We extend MAP to handle arguments of unequal length. 1038(define map map-in-order) 1039 1040;; Added by yamaken 2007-06-15 1041(define for-each 1042 (lambda args 1043 (apply map-in-order args) 1044 %srfi-1:undefined)) 1045 1046;;; filter, remove, partition 1047;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1048;;; FILTER, REMOVE, PARTITION and their destructive counterparts do not 1049;;; disorder the elements of their argument. 1050 1051;; This FILTER shares the longest tail of L that has no deleted elements. 1052;; If Scheme had multi-continuation calls, they could be made more efficient. 1053 1054(define (filter pred lis) ; Sleazing with EQ? makes this 1055 (check-arg procedure? pred filter) ; one faster. 1056 (let recur ((lis lis)) 1057 (if (null-list? lis) lis ; Use NOT-PAIR? to handle dotted lists. 1058 (let ((head (car lis)) 1059 (tail (cdr lis))) 1060 (if (pred head) 1061 (let ((new-tail (recur tail))) ; Replicate the RECUR call so 1062 (if (eq? tail new-tail) lis 1063 (cons head new-tail))) 1064 (recur tail)))))) ; this one can be a tail call. 1065 1066 1067;;; Another version that shares longest tail. 1068;(define (filter pred lis) 1069; (receive (ans no-del?) 1070; ;; (recur l) returns L with (pred x) values filtered. 1071; ;; It also returns a flag NO-DEL? if the returned value 1072; ;; is EQ? to L, i.e. if it didn't have to delete anything. 1073; (let recur ((l l)) 1074; (if (null-list? l) (values l #t) 1075; (let ((x (car l)) 1076; (tl (cdr l))) 1077; (if (pred x) 1078; (receive (ans no-del?) (recur tl) 1079; (if no-del? 1080; (values l #t) 1081; (values (cons x ans) #f))) 1082; (receive (ans no-del?) (recur tl) ; Delete X. 1083; (values ans #f)))))) 1084; ans)) 1085 1086 1087 1088;(define (filter! pred lis) ; Things are much simpler 1089; (let recur ((lis lis)) ; if you are willing to 1090; (if (pair? lis) ; push N stack frames & do N 1091; (cond ((pred (car lis)) ; SET-CDR! writes, where N is 1092; (set-cdr! lis (recur (cdr lis))); the length of the answer. 1093; lis) 1094; (else (recur (cdr lis)))) 1095; lis))) 1096 1097 1098;;; This implementation of FILTER! 1099;;; - doesn't cons, and uses no stack; 1100;;; - is careful not to do redundant SET-CDR! writes, as writes to memory are 1101;;; usually expensive on modern machines, and can be extremely expensive on 1102;;; modern Schemes (e.g., ones that have generational GC's). 1103;;; It just zips down contiguous runs of in and out elts in LIS doing the 1104;;; minimal number of SET-CDR!s to splice the tail of one run of ins to the 1105;;; beginning of the next. 1106 1107(define (filter! pred lis) 1108 (check-arg procedure? pred filter!) 1109 (let lp ((ans lis)) 1110 (cond ((null-list? ans) ans) ; Scan looking for 1111 ((not (pred (car ans))) (lp (cdr ans))) ; first cons of result. 1112 1113 ;; ANS is the eventual answer. 1114 ;; SCAN-IN: (CDR PREV) = LIS and (CAR PREV) satisfies PRED. 1115 ;; Scan over a contiguous segment of the list that 1116 ;; satisfies PRED. 1117 ;; SCAN-OUT: (CAR PREV) satisfies PRED. Scan over a contiguous 1118 ;; segment of the list that *doesn't* satisfy PRED. 1119 ;; When the segment ends, patch in a link from PREV 1120 ;; to the start of the next good segment, and jump to 1121 ;; SCAN-IN. 1122 (else (letrec ((scan-in (lambda (prev lis) 1123 (if (pair? lis) 1124 (if (pred (car lis)) 1125 (scan-in lis (cdr lis)) 1126 (scan-out prev (cdr lis)))))) 1127 (scan-out (lambda (prev lis) 1128 (let lp ((lis lis)) 1129 (if (pair? lis) 1130 (if (pred (car lis)) 1131 (begin (set-cdr! prev lis) 1132 (scan-in lis (cdr lis))) 1133 (lp (cdr lis))) 1134 (set-cdr! prev lis)))))) 1135 (scan-in ans (cdr ans)) 1136 ans))))) 1137 1138 1139 1140;;; Answers share common tail with LIS where possible; 1141;;; the technique is slightly subtle. 1142 1143(define (partition pred lis) 1144 (check-arg procedure? pred partition) 1145 (let recur ((lis lis)) 1146 (if (null-list? lis) (values lis lis) ; Use NOT-PAIR? to handle dotted lists. 1147 (let ((elt (car lis)) 1148 (tail (cdr lis))) 1149 (receive (in out) (recur tail) 1150 (if (pred elt) 1151 (values (if (pair? out) (cons elt in) lis) out) 1152 (values in (if (pair? in) (cons elt out) lis)))))))) 1153 1154 1155 1156;(define (partition! pred lis) ; Things are much simpler 1157; (let recur ((lis lis)) ; if you are willing to 1158; (if (null-list? lis) (values lis lis) ; push N stack frames & do N 1159; (let ((elt (car lis))) ; SET-CDR! writes, where N is 1160; (receive (in out) (recur (cdr lis)) ; the length of LIS. 1161; (cond ((pred elt) 1162; (set-cdr! lis in) 1163; (values lis out)) 1164; (else (set-cdr! lis out) 1165; (values in lis)))))))) 1166 1167 1168;;; This implementation of PARTITION! 1169;;; - doesn't cons, and uses no stack; 1170;;; - is careful not to do redundant SET-CDR! writes, as writes to memory are 1171;;; usually expensive on modern machines, and can be extremely expensive on 1172;;; modern Schemes (e.g., ones that have generational GC's). 1173;;; It just zips down contiguous runs of in and out elts in LIS doing the 1174;;; minimal number of SET-CDR!s to splice these runs together into the result 1175;;; lists. 1176 1177(define (partition! pred lis) 1178 (check-arg procedure? pred partition!) 1179 (if (null-list? lis) (values lis lis) 1180 1181 ;; This pair of loops zips down contiguous in & out runs of the 1182 ;; list, splicing the runs together. The invariants are 1183 ;; SCAN-IN: (cdr in-prev) = LIS. 1184 ;; SCAN-OUT: (cdr out-prev) = LIS. 1185 (letrec ((scan-in (lambda (in-prev out-prev lis) 1186 (let lp ((in-prev in-prev) (lis lis)) 1187 (if (pair? lis) 1188 (if (pred (car lis)) 1189 (lp lis (cdr lis)) 1190 (begin (set-cdr! out-prev lis) 1191 (scan-out in-prev lis (cdr lis)))) 1192 (set-cdr! out-prev lis))))) ; Done. 1193 1194 (scan-out (lambda (in-prev out-prev lis) 1195 (let lp ((out-prev out-prev) (lis lis)) 1196 (if (pair? lis) 1197 (if (pred (car lis)) 1198 (begin (set-cdr! in-prev lis) 1199 (scan-in lis out-prev (cdr lis))) 1200 (lp lis (cdr lis))) 1201 (set-cdr! in-prev lis)))))) ; Done. 1202 1203 ;; Crank up the scan&splice loops. 1204 (if (pred (car lis)) 1205 ;; LIS begins in-list. Search for out-list's first pair. 1206 (let lp ((prev-l lis) (l (cdr lis))) 1207 (cond ((not (pair? l)) (values lis l)) 1208 ((pred (car l)) (lp l (cdr l))) 1209 (else (scan-out prev-l l (cdr l)) 1210 (values lis l)))) ; Done. 1211 1212 ;; LIS begins out-list. Search for in-list's first pair. 1213 (let lp ((prev-l lis) (l (cdr lis))) 1214 (cond ((not (pair? l)) (values l lis)) 1215 ((pred (car l)) 1216 (scan-in l prev-l (cdr l)) 1217 (values l lis)) ; Done. 1218 (else (lp l (cdr l))))))))) 1219 1220 1221;;; Inline us, please. 1222(define (remove pred l) (filter (lambda (x) (not (pred x))) l)) 1223(define (remove! pred l) (filter! (lambda (x) (not (pred x))) l)) 1224 1225 1226 1227;;; Here's the taxonomy for the DELETE/ASSOC/MEMBER functions. 1228;;; (I don't actually think these are the world's most important 1229;;; functions -- the procedural FILTER/REMOVE/FIND/FIND-TAIL variants 1230;;; are far more general.) 1231;;; 1232;;; Function Action 1233;;; --------------------------------------------------------------------------- 1234;;; remove pred lis Delete by general predicate 1235;;; delete x lis [=] Delete by element comparison 1236;;; 1237;;; find pred lis Search by general predicate 1238;;; find-tail pred lis Search by general predicate 1239;;; member x lis [=] Search by element comparison 1240;;; 1241;;; assoc key lis [=] Search alist by key comparison 1242;;; alist-delete key alist [=] Alist-delete by key comparison 1243 1244(define (delete x lis . maybe-=) 1245 (let ((= (:optional maybe-= equal?))) 1246 (filter (lambda (y) (not (= x y))) lis))) 1247 1248(define (delete! x lis . maybe-=) 1249 (let ((= (:optional maybe-= equal?))) 1250 (filter! (lambda (y) (not (= x y))) lis))) 1251 1252;;; Extended from R4RS to take an optional comparison argument. 1253(define (member x lis . maybe-=) 1254 (let ((= (:optional maybe-= equal?))) 1255 (find-tail (lambda (y) (= x y)) lis))) 1256 1257;;; R4RS, hence we don't bother to define. 1258;;; The MEMBER and then FIND-TAIL call should definitely 1259;;; be inlined for MEMQ & MEMV. 1260;(define (memq x lis) (member x lis eq?)) 1261;(define (memv x lis) (member x lis eqv?)) 1262 1263 1264;;; right-duplicate deletion 1265;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1266;;; delete-duplicates delete-duplicates! 1267;;; 1268;;; Beware -- these are N^2 algorithms. To efficiently remove duplicates 1269;;; in long lists, sort the list to bring duplicates together, then use a 1270;;; linear-time algorithm to kill the dups. Or use an algorithm based on 1271;;; element-marking. The former gives you O(n lg n), the latter is linear. 1272 1273(define (delete-duplicates lis . maybe-=) 1274 (let ((elt= (:optional maybe-= equal?))) 1275 (check-arg procedure? elt= delete-duplicates) 1276 (let recur ((lis lis)) 1277 (if (null-list? lis) lis 1278 (let* ((x (car lis)) 1279 (tail (cdr lis)) 1280 (new-tail (recur (delete x tail elt=)))) 1281 (if (eq? tail new-tail) lis (cons x new-tail))))))) 1282 1283(define (delete-duplicates! lis . maybe-=) 1284 (let ((elt= (:optional maybe-= equal?))) 1285 (check-arg procedure? elt= delete-duplicates!) 1286 (let recur ((lis lis)) 1287 (if (null-list? lis) lis 1288 (let* ((x (car lis)) 1289 (tail (cdr lis)) 1290 (new-tail (recur (delete! x tail elt=)))) 1291 (if (eq? tail new-tail) lis (cons x new-tail))))))) 1292 1293 1294;;; alist stuff 1295;;;;;;;;;;;;;;; 1296 1297;;; Extended from R4RS to take an optional comparison argument. 1298(define (assoc x lis . maybe-=) 1299 (let ((= (:optional maybe-= equal?))) 1300 (find (lambda (entry) (= x (car entry))) lis))) 1301 1302(define (alist-cons key datum alist) (cons (cons key datum) alist)) 1303 1304(define (alist-copy alist) 1305 (map (lambda (elt) (cons (car elt) (cdr elt))) 1306 alist)) 1307 1308(define (alist-delete key alist . maybe-=) 1309 (let ((= (:optional maybe-= equal?))) 1310 (filter (lambda (elt) (not (= key (car elt)))) alist))) 1311 1312(define (alist-delete! key alist . maybe-=) 1313 (let ((= (:optional maybe-= equal?))) 1314 (filter! (lambda (elt) (not (= key (car elt)))) alist))) 1315 1316 1317;;; find find-tail take-while drop-while span break any every list-index 1318;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; 1319 1320(define (find pred list) 1321 (cond ((find-tail pred list) => car) 1322 (else #f))) 1323 1324(define (find-tail pred list) 1325 (check-arg procedure? pred find-tail) 1326 (let lp ((list list)) 1327 (and (not (null-list? list)) 1328 (if (pred (car list)) list 1329 (lp (cdr list)))))) 1330 1331(define (take-while pred lis) 1332 (check-arg procedure? pred take-while) 1333 (let recur ((lis lis)) 1334 (if (null-list? lis) '() 1335 (let ((x (car lis))) 1336 (if (pred x) 1337 (cons x (recur (cdr lis))) 1338 '()))))) 1339 1340(define (drop-while pred lis) 1341 (check-arg procedure? pred drop-while) 1342 (let lp ((lis lis)) 1343 (if (null-list? lis) '() 1344 (if (pred (car lis)) 1345 (lp (cdr lis)) 1346 lis)))) 1347 1348(define (take-while! pred lis) 1349 (check-arg procedure? pred take-while!) 1350 (if (or (null-list? lis) (not (pred (car lis)))) '() 1351 (begin (let lp ((prev lis) (rest (cdr lis))) 1352 (if (pair? rest) 1353 (let ((x (car rest))) 1354 (if (pred x) (lp rest (cdr rest)) 1355 (set-cdr! prev '()))))) 1356 lis))) 1357 1358(define (span pred lis) 1359 (check-arg procedure? pred span) 1360 (let recur ((lis lis)) 1361 (if (null-list? lis) (values '() '()) 1362 (let ((x (car lis))) 1363 (if (pred x) 1364 (receive (prefix suffix) (recur (cdr lis)) 1365 (values (cons x prefix) suffix)) 1366 (values '() lis)))))) 1367 1368(define (span! pred lis) 1369 (check-arg procedure? pred span!) 1370 (if (or (null-list? lis) (not (pred (car lis)))) (values '() lis) 1371 (let ((suffix (let lp ((prev lis) (rest (cdr lis))) 1372 (if (null-list? rest) rest 1373 (let ((x (car rest))) 1374 (if (pred x) (lp rest (cdr rest)) 1375 (begin (set-cdr! prev '()) 1376 rest))))))) 1377 (values lis suffix)))) 1378 1379 1380(define (break pred lis) (span (lambda (x) (not (pred x))) lis)) 1381(define (break! pred lis) (span! (lambda (x) (not (pred x))) lis)) 1382 1383(define (any pred lis1 . lists) 1384 (check-arg procedure? pred any) 1385 (if (pair? lists) 1386 1387 ;; N-ary case 1388 (receive (heads tails) (%cars+cdrs (cons lis1 lists)) 1389 (and (pair? heads) 1390 (let lp ((heads heads) (tails tails)) 1391 (receive (next-heads next-tails) (%cars+cdrs tails) 1392 (if (pair? next-heads) 1393 (or (apply pred heads) (lp next-heads next-tails)) 1394 (apply pred heads)))))) ; Last PRED app is tail call. 1395 1396 ;; Fast path 1397 (and (not (null-list? lis1)) 1398 (let lp ((head (car lis1)) (tail (cdr lis1))) 1399 (if (null-list? tail) 1400 (pred head) ; Last PRED app is tail call. 1401 (or (pred head) (lp (car tail) (cdr tail)))))))) 1402 1403 1404;(define (every pred list) ; Simple definition. 1405; (let lp ((list list)) ; Doesn't return the last PRED value. 1406; (or (not (pair? list)) 1407; (and (pred (car list)) 1408; (lp (cdr list)))))) 1409 1410(define (every pred lis1 . lists) 1411 (check-arg procedure? pred every) 1412 (if (pair? lists) 1413 1414 ;; N-ary case 1415 (receive (heads tails) (%cars+cdrs (cons lis1 lists)) 1416 (or (not (pair? heads)) 1417 (let lp ((heads heads) (tails tails)) 1418 (receive (next-heads next-tails) (%cars+cdrs tails) 1419 (if (pair? next-heads) 1420 (and (apply pred heads) (lp next-heads next-tails)) 1421 (apply pred heads)))))) ; Last PRED app is tail call. 1422 1423 ;; Fast path 1424 (or (null-list? lis1) 1425 (let lp ((head (car lis1)) (tail (cdr lis1))) 1426 (if (null-list? tail) 1427 (pred head) ; Last PRED app is tail call. 1428 (and (pred head) (lp (car tail) (cdr tail)))))))) 1429 1430(define (list-index pred lis1 . lists) 1431 (check-arg procedure? pred list-index) 1432 (if (pair? lists) 1433 1434 ;; N-ary case 1435 (let lp ((lists (cons lis1 lists)) (n 0)) 1436 (receive (heads tails) (%cars+cdrs lists) 1437 (and (pair? heads) 1438 (if (apply pred heads) n 1439 (lp tails (+ n 1)))))) 1440 1441 ;; Fast path 1442 (let lp ((lis lis1) (n 0)) 1443 (and (not (null-list? lis)) 1444 (if (pred (car lis)) n (lp (cdr lis) (+ n 1))))))) 1445 1446;;; Reverse 1447;;;;;;;;;;; 1448 1449;R4RS, so not defined here. 1450;(define (reverse lis) (fold cons '() lis)) 1451 1452;(define (reverse! lis) 1453; (pair-fold (lambda (pair tail) (set-cdr! pair tail) pair) '() lis)) 1454 1455(define (reverse! lis) 1456 (let lp ((lis lis) (ans '())) 1457 (if (null-list? lis) ans 1458 (let ((tail (cdr lis))) 1459 (set-cdr! lis ans) 1460 (lp tail lis))))) 1461 1462;;; Lists-as-sets 1463;;;;;;;;;;;;;;;;; 1464 1465;;; This is carefully tuned code; do not modify casually. 1466;;; - It is careful to share storage when possible; 1467;;; - Side-effecting code tries not to perform redundant writes. 1468;;; - It tries to avoid linear-time scans in special cases where constant-time 1469;;; computations can be performed. 1470;;; - It relies on similar properties from the other list-lib procs it calls. 1471;;; For example, it uses the fact that the implementations of MEMBER and 1472;;; FILTER in this source code share longest common tails between args 1473;;; and results to get structure sharing in the lset procedures. 1474 1475(define (%lset2<= = lis1 lis2) (every (lambda (x) (member x lis2 =)) lis1)) 1476 1477(define (lset<= = . lists) 1478 (check-arg procedure? = lset<=) 1479 (or (not (pair? lists)) ; 0-ary case 1480 (let lp ((s1 (car lists)) (rest (cdr lists))) 1481 (or (not (pair? rest)) 1482 (let ((s2 (car rest)) (rest (cdr rest))) 1483 (and (or (eq? s2 s1) ; Fast path 1484 (%lset2<= = s1 s2)) ; Real test 1485 (lp s2 rest))))))) 1486 1487(define (lset= = . lists) 1488 (check-arg procedure? = lset=) 1489 (or (not (pair? lists)) ; 0-ary case 1490 (let lp ((s1 (car lists)) (rest (cdr lists))) 1491 (or (not (pair? rest)) 1492 (let ((s2 (car rest)) 1493 (rest (cdr rest))) 1494 (and (or (eq? s1 s2) ; Fast path 1495 (and (%lset2<= = s1 s2) (%lset2<= = s2 s1))) ; Real test 1496 (lp s2 rest))))))) 1497 1498 1499(define (lset-adjoin = lis . elts) 1500 (check-arg procedure? = lset-adjoin) 1501 (fold (lambda (elt ans) (if (member elt ans =) ans (cons elt ans))) 1502 lis elts)) 1503 1504 1505(define (lset-union = . lists) 1506 (check-arg procedure? = lset-union) 1507 (reduce (lambda (lis ans) ; Compute ANS + LIS. 1508 (cond ((null? lis) ans) ; Don't copy any lists 1509 ((null? ans) lis) ; if we don't have to. 1510 ((eq? lis ans) ans) 1511 (else 1512 (fold (lambda (elt ans) (if (any (lambda (x) (= x elt)) ans) 1513 ans 1514 (cons elt ans))) 1515 ans lis)))) 1516 '() lists)) 1517 1518(define (lset-union! = . lists) 1519 (check-arg procedure? = lset-union!) 1520 (reduce (lambda (lis ans) ; Splice new elts of LIS onto the front of ANS. 1521 (cond ((null? lis) ans) ; Don't copy any lists 1522 ((null? ans) lis) ; if we don't have to. 1523 ((eq? lis ans) ans) 1524 (else 1525 (pair-fold (lambda (pair ans) 1526 (let ((elt (car pair))) 1527 (if (any (lambda (x) (= x elt)) ans) 1528 ans 1529 (begin (set-cdr! pair ans) pair)))) 1530 ans lis)))) 1531 '() lists)) 1532 1533 1534(define (lset-intersection = lis1 . lists) 1535 (check-arg procedure? = lset-intersection) 1536 (let ((lists (delete lis1 lists eq?))) ; Throw out any LIS1 vals. 1537 (cond ((any null-list? lists) '()) ; Short cut 1538 ((null? lists) lis1) ; Short cut 1539 (else (filter (lambda (x) 1540 (every (lambda (lis) (member x lis =)) lists)) 1541 lis1))))) 1542 1543(define (lset-intersection! = lis1 . lists) 1544 (check-arg procedure? = lset-intersection!) 1545 (let ((lists (delete lis1 lists eq?))) ; Throw out any LIS1 vals. 1546 (cond ((any null-list? lists) '()) ; Short cut 1547 ((null? lists) lis1) ; Short cut 1548 (else (filter! (lambda (x) 1549 (every (lambda (lis) (member x lis =)) lists)) 1550 lis1))))) 1551 1552 1553(define (lset-difference = lis1 . lists) 1554 (check-arg procedure? = lset-difference) 1555 (let ((lists (filter pair? lists))) ; Throw out empty lists. 1556 (cond ((null? lists) lis1) ; Short cut 1557 ((memq lis1 lists) '()) ; Short cut 1558 (else (filter (lambda (x) 1559 (every (lambda (lis) (not (member x lis =))) 1560 lists)) 1561 lis1))))) 1562 1563(define (lset-difference! = lis1 . lists) 1564 (check-arg procedure? = lset-difference!) 1565 (let ((lists (filter pair? lists))) ; Throw out empty lists. 1566 (cond ((null? lists) lis1) ; Short cut 1567 ((memq lis1 lists) '()) ; Short cut 1568 (else (filter! (lambda (x) 1569 (every (lambda (lis) (not (member x lis =))) 1570 lists)) 1571 lis1))))) 1572 1573 1574(define (lset-xor = . lists) 1575 (check-arg procedure? = lset-xor) 1576 (reduce (lambda (b a) ; Compute A xor B: 1577 ;; Note that this code relies on the constant-time 1578 ;; short-cuts provided by LSET-DIFF+INTERSECTION, 1579 ;; LSET-DIFFERENCE & APPEND to provide constant-time short 1580 ;; cuts for the cases A = (), B = (), and A eq? B. It takes 1581 ;; a careful case analysis to see it, but it's carefully 1582 ;; built in. 1583 1584 ;; Compute a-b and a^b, then compute b-(a^b) and 1585 ;; cons it onto the front of a-b. 1586 (receive (a-b a-int-b) (lset-diff+intersection = a b) 1587 (cond ((null? a-b) (lset-difference = b a)) 1588 ((null? a-int-b) (append b a)) 1589 (else (fold (lambda (xb ans) 1590 (if (member xb a-int-b =) ans (cons xb ans))) 1591 a-b 1592 b))))) 1593 '() lists)) 1594 1595 1596(define (lset-xor! = . lists) 1597 (check-arg procedure? = lset-xor!) 1598 (reduce (lambda (b a) ; Compute A xor B: 1599 ;; Note that this code relies on the constant-time 1600 ;; short-cuts provided by LSET-DIFF+INTERSECTION, 1601 ;; LSET-DIFFERENCE & APPEND to provide constant-time short 1602 ;; cuts for the cases A = (), B = (), and A eq? B. It takes 1603 ;; a careful case analysis to see it, but it's carefully 1604 ;; built in. 1605 1606 ;; Compute a-b and a^b, then compute b-(a^b) and 1607 ;; cons it onto the front of a-b. 1608 (receive (a-b a-int-b) (lset-diff+intersection! = a b) 1609 (cond ((null? a-b) (lset-difference! = b a)) 1610 ((null? a-int-b) (append! b a)) 1611 (else (pair-fold (lambda (b-pair ans) 1612 (if (member (car b-pair) a-int-b =) ans 1613 (begin (set-cdr! b-pair ans) b-pair))) 1614 a-b 1615 b))))) 1616 '() lists)) 1617 1618 1619(define (lset-diff+intersection = lis1 . lists) 1620 (check-arg procedure? = lset-diff+intersection) 1621 (cond ((every null-list? lists) (values lis1 '())) ; Short cut 1622 ((memq lis1 lists) (values '() lis1)) ; Short cut 1623 (else (partition (lambda (elt) 1624 (not (any (lambda (lis) (member elt lis =)) 1625 lists))) 1626 lis1)))) 1627 1628(define (lset-diff+intersection! = lis1 . lists) 1629 (check-arg procedure? = lset-diff+intersection!) 1630 (cond ((every null-list? lists) (values lis1 '())) ; Short cut 1631 ((memq lis1 lists) (values '() lis1)) ; Short cut 1632 (else (partition! (lambda (elt) 1633 (not (any (lambda (lis) (member elt lis =)) 1634 lists))) 1635 lis1)))) 1636