pegutils.lisp

   1 ;;; -*- Mode: Lisp; Syntax: ANSI-Common-Lisp; Base: 10; -*-
   2 ;;;
   3 ;;; pegutils.lisp --- Utility functions for implementing PEG parsers
   4
   5 ;; Copyright (C) 2008 Utz-Uwe Haus <lisp@uuhaus.de>
   6 ;; $Id$
   7 ;;
   8 ;; This code is free software; you can redistribute it and/or modify
   9 ;; it under the terms of the version 2.1 of the GNU Lesser General
  10 ;; Public License as published by the Free Software Foundation, as
  11 ;; clarified by the lisp prequel found in LICENSE.
  12 ;;
  13 ;; This code is distributed in the hope that it will be useful, but
  14 ;; without any warranty; without even the implied warranty of
  15 ;; merchantability or fitness for a particular purpose. See the GNU
  16 ;; Lesser General Public License for more details.
  17 ;;
  18 ;; Version 2.1 of the GNU Lesser General Public License is in the file
  19 ;; LICENSE that was distributed with this file. If it is not
  20 ;; present, you can access it from
  21 ;; http://www.gnu.org/copyleft/lgpl.txt (until superseded by a
  22 ;; newer version) or write to the Free Software Foundation, Inc., 59
  23 ;; Temple Place, Suite 330, Boston, MA 02111-1307 USA
  24 ;;
  25 ;; Commentary:
  26
  27 ;; Some code here is inspired by the metapeg library of John Leuner
  28
  29 ;;; Code:
  30
  31
  32 (eval-when (:load-toplevel :compile-toplevel :execute)
  33   (declaim (optimize (speed 0) (safety 3) (debug 3))))
  34
  35 (in-package #:opossum)
  36 \f
  37 (defparameter *trace* nil "If non-nil, do extensive tracing of the parser functions.")
  38 \f
  39
  40 (defclass context ()
  41   (;; these slots are copied when cloning a context for recursion
  42    (input       :accessor input       :initarg :input       :initform nil
  43                 :type 'string
  44                 :documentation "The input string being parsed.")
  45    (destpkg     :accessor destpkg     :initarg :destpkg     :initform nil
  46                 :type 'package
  47                 :documentation "The package into which symbols generated during the parse are interned.")
  48    ;; these slots are shared by all cloned copies of a context -- use only STORE-ACTION to guarantee consistency
  49    (actions     :accessor actions     :initarg :actions     :initform '(NIL)
  50                 :type 'list
  51                 :documentation "The list of actions accumulated during the parse.")
  52    (action-counter :accessor action-counter :initarg :action-counter :initform '(0)
  53                    :type '(cons (integer 0) null)
  54                    :documentation "The counter of actions.")
  55    ;; these slots are what make a context unique
  56    (parent      :accessor parent      :initarg :parent      :initform nil)
  57    (rule        :accessor rule        :initarg :rule        :initform nil)
  58    (children    :accessor children    :initform nil)
  59    (value       :accessor value       :initarg :value       :initform nil)
  60    (start-index :accessor start-index :initarg :start-index :initform nil)
  61    (end-index   :accessor end-index   :initarg :end-index   :initform nil))
  62   (:documentation "The parser context."))
  63
  64 (defmethod print-object ((obj context) stream)
  65   (print-unreadable-object (obj stream :type T :identity NIL)
  66     (format stream "rule ~A (~S) value ~S (~D:~D)"
  67             (rule obj) (children obj) (value obj) (start-index obj) (end-index obj))))
  68
  69 (defmethod store-action ((ctx context) action)
  70   (let ((a (actions ctx)))
  71     (rplacd a (cons (car a) (cdr a)))
  72     (rplaca a action)))
  73
  74 (defvar *context* nil "The current parser context.")
  75
  76 (defun clone-ctx (ctx rule)
  77   "Create clone context of CTX for rule RULE."
  78   (make-instance 'context
  79                  :input (input ctx)
  80                  :destpkg (destpkg ctx)
  81                  :actions (actions ctx)
  82                  :action-counter (action-counter ctx)
  83                  :parent ctx
  84                  :rule rule
  85                  :start-index (end-index ctx)))
  86
  87 (defun ctx-failed-p (ctx)
  88   (unless ctx
  89     (break "Botched context"))
  90   (null (end-index ctx)))
  91
  92 (defun succeed (ctx value start-index end-index)
  93   "Mark CTX as successful: set VALUE and matched region."
  94   (setf (value ctx) value)
  95   (setf (start-index ctx) start-index)
  96   (setf (end-index ctx) end-index)
  97   ;  (break "generated success context ~A" ctx)
  98   (when *trace*
  99     (format *trace-output* "Matched: ~A (~D:~D)~%"
 100             (rule ctx) (start-index ctx) (end-index ctx)))
 101   ctx)
 102
 103 (defun fail ()
 104   "Return a failure context."
 105   (let ((ctx (make-instance 'context
 106                             :input (input *context*)
 107                             :rule ':fail
 108                             :value (rule *context*)
 109                             :start-index (start-index *context*)
 110                             :end-index (end-index *context*)
 111                             ;; probably some of these copies can be saved
 112                             :destpkg (destpkg *context*)
 113                             :actions (actions *context*)
 114                             :action-counter (action-counter *context*))))
 115     ;    (break "generated failure context ~A" ctx )
 116     (when *trace*
 117       (format *trace-output* "(failed: ~A ~A ~A)~%"
 118               (value ctx) (start-index ctx) (end-index ctx)))
 119     ctx))
 120
 121 \f
 122 ;;
 123 (defun make-name (string)
 124   (intern (concatenate 'string "parse-" string)
 125           (destpkg *context*)))
 126
 127 (defun make-action-name (&key ctx)
 128   "Return a symbol suitable to name the next action."
 129   (incf (car (action-counter *context*)))
 130   (let ((aname (if ctx
 131                    (format nil "opossum-action-~D-srcpos-~D-~D"
 132                            (car (action-counter *context*))
 133                            (start-index ctx)
 134                            (end-index ctx))
 135                    (format nil "opossum-action-~D-~D~D"
 136                            (car (action-counter *context*)))) ))
 137     (intern aname (destpkg *context*))))
 138
 139 (defun char-list-to-string (char-list)
 140   (coerce char-list 'string))
 141
 142 (defmacro build-parser-function (name parser)
 143   `(let* ((*context* (clone-ctx *context* ,name))
 144           (result (funcall ,parser offset)))
 145     (unless result
 146       (break "Yow"))
 147     (if (ctx-failed-p result)
 148         (fail)
 149         (succeed *context* (value result) (start-index result) (end-index result)))))
 150
 151
 152 \f
 153 (defun match-string (string)
 154   #'(lambda (offset)
 155       (let ((input (input *context*))
 156             (len (length string)))
 157         (if (and (>= (length input) (+ offset len))
 158                  (string= string input :start2 offset :end2 (+ offset len)))
 159             (succeed (clone-ctx *context* 'opossum-string) string offset (+ offset (length string)))
 160             (fail)))))
 161
 162 (defun match-char (char-list)
 163   #'(lambda (offset)
 164       (let ((input (input *context*)))
 165         (when *trace*
 166           (format *trace-output* "match-char: looking for one of `~{~A~}'~%" char-list))
 167         (if (and (> (length input) offset)
 168                  (member (char input offset)
 169                          char-list :test #'char=))
 170             (succeed (clone-ctx *context* 'opossum-char) (char input offset) offset (+ offset 1))
 171             (fail)))))
 172
 173 (defun match-octal-char-code (i1 i2 i3)
 174   "Compare the character given by i3 + 8*i2 + 64*i1 to the next input character."
 175   (let ((c (+ i3 (* 8 i2) (* 64 i1))))
 176     #'(lambda (offset)
 177         (let ((input (input *context*)))
 178           (when *trace*
 179             (format *trace-output* "match-octal-char-code: looking for ~D~%" c))
 180           (if (and (> (length input) offset)
 181                    (= (char-int (char input offset)) c))
 182               (succeed (clone-ctx *context* 'opossum-char) (char input offset) offset (+ offset 1))
 183               (fail))))))
 184
 185 (defun match-char-range (lower-char upper-char)
 186   "Match characters in the range between LOWER-CHAR and UPPER-CHAR (inclusive) as decided by CL:CHAR-CODE."
 187   #'(lambda (offset)
 188       (let ((input (input *context*)))
 189         (when *trace*
 190           (format *trace-output* "match-char-range: looking for ~A-~A~%" lower-char upper-char))
 191         (if (and (> (length input) offset)
 192                  (let ((x (char-code (char input offset))))
 193                    (and (>= x (char-code lower-char))
 194                         (<= x (char-code upper-char)))))
 195             (succeed (clone-ctx *context* 'opossum-char-range) (char input offset) offset (+ offset 1))
 196             (fail)))))
 197
 198 (defun match-any-char (&optional ignored)
 199   (declare (ignore ignored))
 200   #'(lambda (offset)
 201       "Match any character at OFFSET, fail only on EOF."
 202       (let ((input (input *context*)))
 203         (when *trace*
 204           (format *trace-output* "match-any-char~%"))
 205         (if (< (1+ offset) (length input))
 206             (succeed (clone-ctx *context* 'opossum-anychar) (char input offset) offset (+ offset 1))
 207             (fail)))))
 208
 209 (defun match-char-class (char-class)
 210   "Regexp matching of next input character against [CHAR-CLASS] using cl-ppcre:scan."
 211   (declare (type string char-class))
 212   ;; FIXME: could use a pre-computed scanner
 213   (let ((cc (format nil "[~A]" char-class)))
 214     #'(lambda (offset)
 215         "Match next character at OFFSET against the characters in CHAR-CLASS."
 216         (let ((input (input *context*)))
 217           (when *trace*
 218             (format *trace-output* "match-char-class on ~A~%"))
 219           (if (and (< (1+ offset) (length input))
 220                    (let ((c (char input offset)))
 221                      (cl-ppcre:scan cc (make-string 1 :initial-element c))))
 222               (succeed (clone-ctx *context* 'opossum-charclass)
 223                        (char input offset) offset (+ offset 1))
 224               (fail))))))
 225
 226 (defun fix-escape-sequences (char-list)
 227   "Iterate over the list CHAR-LIST, glueing adjacent #\\ #\n and #\\ #\t chars into
 228 #\Newline and #\Tab."
 229   (cond
 230     ((null char-list) char-list)
 231     ((null (cdr char-list)) char-list)
 232     (T
 233      (loop :with drop := nil
 234            :for (c1 c2) :on char-list
 235            :if drop
 236            :do (setf drop nil)
 237            :else
 238            :collect (if (char= c1 #\\)
 239                         (case c2
 240                           ((#\n) (setf drop T) #\Newline)
 241                           ((#\t) (setf drop T) #\Tab)
 242                           ((#\r) (setf drop T) #\Linefeed)
 243                           (otherwise c1))
 244                         c1)
 245            :end))))
 246 \f
 247 ;;
 248 ;; parsing combinator functions cribbed from libmetapeg
 249 ;;
 250 (defun either (&rest parsers)
 251   "Produce a function that tries each of the functions in PARSERS sequentially until one succeeds and
 252 returns the result of that function, or a failure context if none succeeded."
 253   #'(lambda (offset)
 254       (let ((*context* (clone-ctx *context* 'opossum-either)))
 255         (when *trace*
 256           (format *trace-output* "either: ~A ~A~%" *context* parsers))
 257         (loop :for p :in parsers
 258               :as result = (funcall p offset)
 259               :when (not (ctx-failed-p result))
 260               :return (succeed *context* (value result) offset (end-index result))
 261               :finally (return (fail))))))
 262
 263
 264 (defun optional (parser)
 265   #'(lambda (offset)
 266       (let ((*context* (clone-ctx *context* 'opossum-optional)))
 267         (when *trace*
 268           (format *trace-output* "optional: ~A ~A~%" *context* parser))
 269         (let ((result (funcall parser offset)))
 270           (if (ctx-failed-p result)
 271               (succeed *context* 'optional offset offset)
 272               (succeed *context* (value result) offset (end-index result)))))))
 273
 274 (defun follow (parser)
 275   #'(lambda (offset)
 276       (let ((*context* (clone-ctx *context* 'opossum-follow)))
 277         (when *trace*
 278           (format *trace-output* "follow: ~A ~A~%" *context* parser))
 279         (let ((result (funcall parser offset)))
 280           (if (ctx-failed-p result)
 281               (fail)
 282               (succeed *context* (value result)
 283                        ;; don't consume input
 284                        offset offset))))))
 285
 286 (defun many (parser)
 287   #'(lambda (offset)
 288       (let ((*context* (clone-ctx *context* 'opossum-many))
 289             (start-offset offset)
 290             children)
 291         (when *trace*
 292           (format *trace-output* "many: ~A ~A~%" *context* parser))
 293         (loop :as result := (funcall parser offset)
 294               :while (not (ctx-failed-p result))
 295               :do (progn (push (value result) children)
 296                          (setf offset (end-index result)))
 297               :finally (return (succeed *context* (nreverse children) start-offset offset))))))
 298
 299
 300 (defun many1 (parser)
 301   #'(lambda (offset)
 302       (let* ((*context* (clone-ctx *context* 'opossum-many1))
 303              (result (funcall parser offset)))
 304         (when *trace*
 305           (format *trace-output* "many1: ~A ~A~%" *context* parser))
 306         (if (not (ctx-failed-p result))
 307             (let ((result2 (funcall (many parser) (end-index result))))
 308               (if (end-index result2)
 309                   (succeed *context* (cons (value result) (value result2)) offset (end-index result2))
 310                   (succeed *context* (value result) offset (end-index result))))
 311             (fail)))))
 312
 313
 314 (defun seq (&rest parsers)
 315   #'(lambda (offset)
 316       (assert (> (length parsers) 0))
 317       (let ((*context* (clone-ctx *context* 'opossum-seq))
 318             (start-offset offset)
 319             child-values
 320             child-nodes)
 321         (when *trace*
 322           (format *trace-output* "seq: ~A ~A~%" *context* parsers))
 323         ;; run the parsers
 324         (loop :for p :in parsers
 325               :do (when *trace* (format *trace-output* " (seq ~A) trying ~A~%" *context* p))
 326               :do (cond
 327                     ((consp p)
 328                      (push (succeed (clone-ctx *context* 'action) nil offset offset)  child-nodes)
 329                      (push p child-values)
 330                      (setf (children *context*) (reverse child-nodes)))
 331                     (T
 332                      (let ((result (funcall p offset)))
 333                        (if (end-index result)
 334                            (progn
 335                              (push result child-nodes)
 336                              (push (value result) child-values)
 337                              (setf offset (end-index result))
 338                              (setf (children *context*) (reverse child-nodes)))
 339                            (return (fail))))))
 340               :finally (return (succeed *context* (reverse child-values) start-offset offset))))))
 341
 342 (defun negate (parser)
 343   #'(lambda (offset)
 344       "Return a successful context at OFFSET if PARSER succeeds, without advancing input position."
 345       (let ((*context* (clone-ctx *context* 'opossum-negate)))
 346         (let ((result (funcall parser offset)))
 347           (if (ctx-failed-p result)
 348               (succeed *context* 'negate offset offset)
 349               (fail))))))
 350
 351 \f
 352 ;;
 353 (defun read-stream (stream)
 354   "Read STREAM and return a string of all its contents."
 355   (let ((s ""))
 356     (loop :as line := (read-line stream nil nil)
 357           :while line
 358           :do (setq s (concatenate 'string s line)))
 359     s))
 360
 361 (defun read-file (file)
 362   "Read FILE and return a string of all its contents."
 363   (with-open-file (f file :direction :input)
 364     (let ((len (file-length f)))
 365       (if len
 366           (let ((s (make-string len)))
 367             (read-sequence s f)
 368             s)
 369           (read-stream f)))))
 370 \f
 371 (defun make-default-dst-package (grammarfile)
 372   (make-package (gensym "opossum-parser")
 373                 :documentation (format T "Opossum parser for grammar ~A" grammarfile)))
 374
 375 (defun generate-parser-package (grammarfile dst-package head)
 376   "Create functions to parse using GRAMMARFILE in DST-PACKAGE, starting ar rule named HEAD.
 377 DST-PACKAGE will contain the 3 functions PARSE-STRING, PARSE-FILE and PARSE-STREAM as exported entrypoints."
 378   (let ((*package* dst-package))
 379     (multiple-value-bind (form actions)
 380         (opossum:parse-file grammarfile)
 381       (format *debug-io* "Injecting parser functions into ~A~%" dst-package)
 382       (loop :for aform :in form
 383             :do (progn
 384                   (format *debug-io* "Skipping aform ~A~%" aform)))
 385       (loop :for (sym code) :in actions
 386             :do (intern (compile sym `(lambda (data) (declare (ignorable data)) ,code)))
 387             :do (format *debug-io* "Compiled ~A~%" sym))
 388       (intern (compile 'parse-string
 389                        #'(lambda (s)
 390                            `,(format nil "Parse S using grammar ~A starting at ~A" grammarfile head)
 391                            (let ((*context* (make-instance 'opossum:context :dstpkg *package* :input s)))
 392                              (funcall (make-name head) 0)))))
 393       (intern (compile 'parse-file
 394                        #'(lambda (f) (parse-string (read-file f)))))
 395       (intern (compile 'parse-stream
 396                        #'(lambda (s) (parse-string (read-stream s)))))
 397       (intern '*trace*)
 398       (setf (documentation '*trace* 'cl:variable)
 399             "When non-nil, the generated parser function log to cl:*trace-output*.")
 400       (export '(parse-string parse-file parse-stream *trace*)))))
 401
 402 (defun get-iso-time ()
 403   "Return a string in ISO format for the current time"
 404   (multiple-value-bind (second minute hour date month year day daylight-p zone)
 405       (get-decoded-time)
 406     (declare (ignore day))
 407     (format nil "~4D-~2,'0D-~2,'0D-~2,'0D:~2,'0D:~2,'0D (UCT~@D)"
 408             year month date
 409             hour minute second
 410             (if daylight-p (1+ (- zone)) (- zone)))))
 411
 412 (defun cleanup-action-code (code)
 413   "Remove trailing newlines  so that the string CODE can be printed nicely."
 414   (subseq code 0
 415           (when (char= #\Newline (char code (1- (length code))))
 416             (1+ (position #\Newline code :from-end T :test #'char/=)))))
 417
 418 (defun generate-parser-file (grammarfile dst-package dst-file &optional start-rule)
 419   "Create lisp code in DST-FILE that can be loaded to yield functions to parse using GRAMMARFILE in DST-PACKAGE.
 420 DST-PACKAGE will contain the 3 functions PARSE-STRING, PARSE-FILE and PARSE-STREAM as exported entrypoints."
 421   (let ((*package* dst-package))
 422     (let ((result (opossum:parse-file grammarfile)))
 423       ;; FIXME: check for complete parse
 424       (let ((*context* result)) ;; routines in pegutils.lisp expect *context* to be bound properly
 425         (let ((forms (transform (value result)))
 426               (actions (actions result)))
 427           (with-open-file (s dst-file :direction :output :if-exists :supersede)
 428             (let ((*print-readably* T)
 429                   (*print-pretty* T)
 430                   (*print-circle* NIL))
 431               (format s ";; This is a Common Lisp peg parser automatically generated by OPOSSUM -*- mode:lisp -*-~%")
 432               (format s ";; generated from ~A on ~A~%" grammarfile (get-iso-time))
 433               (format s "(eval-when (:load-toplevel :compile-toplevel :execute) (declaim (optimize (speed 0) (safety 3) (debug 3))))~%~%")
 434               (format s "(defpackage #:~A~%" (package-name dst-package))
 435               (format s " (:use #:cl #:opossum)~%")
 436               (format s " (:export #:parse-string #:parse-file #:parse-stream))~%")
 437               (format s "~%(in-package #:~A)~%" (package-name dst-package))
 438               ;; First form is taken to be the start rule
 439               (let ((entryrule (or (and start-rule (make-name start-rule))
 440                                    (and forms (cadr (first forms))))))
 441                 (if (not entryrule)
 442                     (format *error-output* "Cannot find entry rule for parser")
 443                     (progn
 444                       (when *trace*
 445                         (format *trace-output* "Inserting definitions for parser entry points through ~A~%"
 446                                 entryrule))
 447                       (terpri s)
 448                       (prin1 `(defun parse-string (,(intern :s dst-package))
 449                                ,(format nil "Parse S using grammar ~A starting at ~A" grammarfile entryrule)
 450                                (let ((*context* (make-instance 'opossum:context
 451                                                                :dstpkg ,(package-name dst-package)
 452                                                                :input ,(intern :s dst-package))))
 453                                  (funcall (,entryrule) 0)))
 454                              s)
 455                       (fresh-line s))))
 456               (loop :for aform :in forms
 457                     :do (when *trace* (format *trace-output* "Inserting form ~A~%" aform))
 458                     :do (terpri s)
 459                     :do (prin1 aform s)
 460                     :do (fresh-line s))
 461               (terpri s)
 462               (prin1
 463                `(defparameter ,(intern "*trace*" dst-package) nil
 464                  "When non-nil, the generated parser function log to cl:*trace-output*.")
 465                s)
 466               (terpri s)
 467
 468               (loop :for (sym code) :in actions
 469                     :when sym ;; the final action is named NIL because we push a
 470                     ;; NIL ahead of us in store-actions
 471                     :do (when *trace* (format *trace-output* "Inserting defun for ~A~%" sym))
 472                     :and :do (format s "~%(defun ~S (data)~% ~A)~%"
 473                                      sym (cleanup-action-code code))))))))))
 474
 475 (defun transform (tree &optional (depth 0))
 476   (if (and tree
 477            (consp tree))
 478       (if (eq (first tree) ':action)
 479           (progn
 480             (when *trace*
 481               (format *trace-output* "~AFound action ~A~%"  (make-string depth :initial-element #\Space) tree))
 482             tree)
 483           (let ((data (mapcar #'(lambda (tr) (transform tr (1+ depth)))
 484                               tree)))
 485             (loop :for el :in data
 486                   :when (and (listp el)
 487                              (eq (first el) ':action)
 488                              (symbolp (third el)))
 489                   :do (let ((*package* (destpkg *context*))
 490                             (action (third el)))
 491                         (when *trace*
 492                           (format *trace-output* "~&Applying action ~A to ~A~%" action data))
 493                         (handler-case
 494                             (return-from transform
 495                               (funcall (symbol-function action) data))
 496                           (undefined-function (x)
 497                             (progn
 498                               (format *error-output* "missing definition for ~A: ~A~%" action x)
 499                               ;                       (break "~A in ~A" action *package*)
 500                                     tree)))))
 501             data))
 502       tree))