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  21 ;; http://www.gnu.org/copyleft/lgpl.txt (until superseded by a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   `(lambda (offset)
 144     (let* ((*context* (clone-ctx *context* ,name))
 145            (result (funcall ,parser offset)))
 146       (unless result
 147         (break "Yow"))
 148       (if (ctx-failed-p result)
 149           (fail)
 150           (succeed *context* (value result) (start-index result) (end-index result))))))
 151
 152
 153 \f
 154 (defun match-string (string)
 155   #'(lambda (offset)
 156       (let ((input (input *context*))
 157             (len (length string)))
 158         (if (and (>= (length input) (+ offset len))
 159                  (string= string input :start2 offset :end2 (+ offset len)))
 160             (succeed (clone-ctx *context* 'opossum-string) string offset (+ offset (length string)))
 161             (fail)))))
 162
 163 (defun match-char (char-list)
 164   #'(lambda (offset)
 165       (let ((input (input *context*)))
 166         (when *trace*
 167           (format *trace-output* "match-char: looking for one of `~{~A~}'~%" char-list))
 168         (if (and (> (length input) offset)
 169                  (member (char input offset)
 170                          char-list :test #'char=))
 171             (succeed (clone-ctx *context* 'opossum-char) (char input offset) offset (+ offset 1))
 172             (fail)))))
 173
 174 (defun match-octal-char-code (i1 i2 i3)
 175   "Compare the character given by i3 + 8*i2 + 64*i1 to the next input character."
 176   (let ((c (+ i3 (* 8 i2) (* 64 i1))))
 177     #'(lambda (offset)
 178         (let ((input (input *context*)))
 179           (when *trace*
 180             (format *trace-output* "match-octal-char-code: looking for ~D~%" c))
 181           (if (and (> (length input) offset)
 182                    (= (char-int (char input offset)) c))
 183               (succeed (clone-ctx *context* 'opossum-char) (char input offset) offset (+ offset 1))
 184               (fail))))))
 185
 186 (defun match-char-range (lower-char upper-char)
 187   "Match characters in the range between LOWER-CHAR and UPPER-CHAR (inclusive) as decided by CL:CHAR-CODE."
 188   #'(lambda (offset)
 189       (let ((input (input *context*)))
 190         (when *trace*
 191           (format *trace-output* "match-char-range: looking for ~A-~A~%" lower-char upper-char))
 192         (if (and (> (length input) offset)
 193                  (let ((x (char-code (char input offset))))
 194                    (and (>= x (char-code lower-char))
 195                         (<= x (char-code upper-char)))))
 196             (succeed (clone-ctx *context* 'opossum-char-range) (char input offset) offset (+ offset 1))
 197             (fail)))))
 198
 199 (defun match-any-char (&optional ignored)
 200   (declare (ignore ignored))
 201   #'(lambda (offset)
 202       "Match any character at OFFSET, fail only on EOF."
 203       (let ((input (input *context*)))
 204         (when *trace*
 205           (format *trace-output* "match-any-char~%"))
 206         (if (< (1+ offset) (length input))
 207             (succeed (clone-ctx *context* 'opossum-anychar) (char input offset) offset (+ offset 1))
 208             (fail)))))
 209
 210 (defun match-char-class (char-class)
 211   "Regexp matching of next input character against [CHAR-CLASS] using cl-ppcre:scan."
 212   (declare (type string char-class))
 213   ;; FIXME: could use a pre-computed scanner
 214   (let ((cc (format nil "[~A]" char-class)))
 215     #'(lambda (offset)
 216         "Match next character at OFFSET against the characters in CHAR-CLASS."
 217         (let ((input (input *context*)))
 218           (when *trace*
 219             (format *trace-output* "match-char-class on ~A~%"))
 220           (if (and (< (1+ offset) (length input))
 221                    (let ((c (char input offset)))
 222                      (cl-ppcre:scan cc (make-string 1 :initial-element c))))
 223               (succeed (clone-ctx *context* 'opossum-charclass)
 224                        (char input offset) offset (+ offset 1))
 225               (fail))))))
 226
 227 (defun fix-escape-sequences (char-list)
 228   "Iterate over the list CHAR-LIST, glueing adjacent #\\ #\n and #\\ #\t chars into
 229 #\Newline and #\Tab."
 230   (cond
 231     ((null char-list) char-list)
 232     ((null (cdr char-list)) char-list)
 233     (T
 234      (loop :with drop := nil
 235            :for (c1 c2) :on char-list
 236            :if drop
 237            :do (setf drop nil)
 238            :else
 239            :collect (if (char= c1 #\\)
 240                         (case c2
 241                           ((#\n) (setf drop T) #\Newline)
 242                           ((#\t) (setf drop T) #\Tab)
 243                           ((#\r) (setf drop T) #\Linefeed)
 244                           (otherwise c1))
 245                         c1)
 246            :end))))
 247 \f
 248 ;;
 249 ;; parsing combinator functions cribbed from libmetapeg
 250 ;;
 251 (defun either (&rest parsers)
 252   "Produce a function that tries each of the functions in PARSERS sequentially until one succeeds and
 253 returns the result of that function, or a failure context if none succeeded."
 254   #'(lambda (offset)
 255       (let ((*context* (clone-ctx *context* 'opossum-either)))
 256         (when *trace*
 257           (format *trace-output* "either: ~A ~A~%" *context* parsers))
 258         (loop :for p :in parsers
 259               :as result = (funcall p offset)
 260               :when (not (ctx-failed-p result))
 261               :return (succeed *context* (value result) offset (end-index result))
 262               :finally (return (fail))))))
 263
 264
 265 (defun optional (parser)
 266   #'(lambda (offset)
 267       (let ((*context* (clone-ctx *context* 'opossum-optional)))
 268         (when *trace*
 269           (format *trace-output* "optional: ~A ~A~%" *context* parser))
 270         (let ((result (funcall parser offset)))
 271           (if (ctx-failed-p result)
 272               (succeed *context* 'optional offset offset)
 273               (succeed *context* (value result) offset (end-index result)))))))
 274
 275 (defun follow (parser)
 276   #'(lambda (offset)
 277       (let ((*context* (clone-ctx *context* 'opossum-follow)))
 278         (when *trace*
 279           (format *trace-output* "follow: ~A ~A~%" *context* parser))
 280         (let ((result (funcall parser offset)))
 281           (if (ctx-failed-p result)
 282               (fail)
 283               (succeed *context* (value result)
 284                        ;; don't consume input
 285                        offset offset))))))
 286
 287 (defun many (parser)
 288   #'(lambda (offset)
 289       (let ((*context* (clone-ctx *context* 'opossum-many))
 290             (start-offset offset)
 291             children)
 292         (when *trace*
 293           (format *trace-output* "many: ~A ~A~%" *context* parser))
 294         (loop :as result := (funcall parser offset)
 295               :while (not (ctx-failed-p result))
 296               :do (progn (push (value result) children)
 297                          (setf offset (end-index result)))
 298               :finally (return (succeed *context* (nreverse children) start-offset offset))))))
 299
 300
 301 (defun many1 (parser)
 302   #'(lambda (offset)
 303       (let* ((*context* (clone-ctx *context* 'opossum-many1))
 304              (result (funcall parser offset)))
 305         (when *trace*
 306           (format *trace-output* "many1: ~A ~A~%" *context* parser))
 307         (if (not (ctx-failed-p result))
 308             (let ((result2 (funcall (many parser) (end-index result))))
 309               (if (end-index result2)
 310                   (succeed *context* (cons (value result) (value result2)) offset (end-index result2))
 311                   (succeed *context* (value result) offset (end-index result))))
 312             (fail)))))
 313
 314
 315 (defun seq (&rest parsers)
 316   #'(lambda (offset)
 317       (assert (> (length parsers) 0))
 318       (let ((*context* (clone-ctx *context* 'opossum-seq))
 319             (start-offset offset)
 320             child-values
 321             child-nodes)
 322         (when *trace*
 323           (format *trace-output* "seq: ~A ~A~%" *context* parsers))
 324         ;; run the parsers
 325         (loop :for p :in parsers
 326               :do (when *trace* (format *trace-output* " (seq ~A) trying ~A~%" *context* p))
 327               :do (cond
 328                     ((consp p)
 329                      (push (succeed (clone-ctx *context* 'action) nil offset offset)  child-nodes)
 330                      (push p child-values)
 331                      (setf (children *context*) (reverse child-nodes)))
 332                     (T
 333                      (let ((result (funcall p offset)))
 334                        (if (end-index result)
 335                            (progn
 336                              (push result child-nodes)
 337                              (push (value result) child-values)
 338                              (setf offset (end-index result))
 339                              (setf (children *context*) (reverse child-nodes)))
 340                            (return (fail))))))
 341               :finally (return (succeed *context* (reverse child-values) start-offset offset))))))
 342
 343 (defun negate (parser)
 344   #'(lambda (offset)
 345       "Return a successful context at OFFSET if PARSER succeeds, without advancing input position."
 346       (let ((*context* (clone-ctx *context* 'opossum-negate)))
 347         (let ((result (funcall parser offset)))
 348           (if (ctx-failed-p result)
 349               (succeed *context* 'negate offset offset)
 350               (fail))))))
 351
 352 \f
 353 ;;
 354 (defun read-stream (stream)
 355   "Read STREAM and return a string of all its contents."
 356   (let ((s ""))
 357     (loop :as line := (read-line stream nil nil)
 358           :while line
 359           :do (setq s (concatenate 'string s line)))
 360     s))
 361
 362 (defun read-file (file)
 363   "Read FILE and return a string of all its contents."
 364   (with-open-file (f file :direction :input)
 365     (let ((len (file-length f)))
 366       (if len
 367           (let ((s (make-string len)))
 368             (read-sequence s f)
 369             s)
 370           (read-stream f)))))
 371 \f
 372 (defun make-default-dst-package (grammarfile)
 373   (make-package (gensym "opossum-parser")
 374                 :documentation (format T "Opossum parser for grammar ~A" grammarfile)))
 375
 376 (defun generate-parser-package (grammarfile dst-package head &key start-rule (parse-file-fun #'opossum:parse-file))
 377   "Create functions to parse using GRAMMARFILE in DST-PACKAGE, starting ar rule named HEAD.
 378 DST-PACKAGE will contain the 3 functions PARSE-STRING, PARSE-FILE and PARSE-STREAM as exported entrypoints."
 379   (declare (ignore start-rule))
 380   (error "This is broken and needs to be merged with g-p-f.")
 381   (let ((*package* dst-package))
 382     (multiple-value-bind (form actions)
 383         (funcall parse-file-fun grammarfile)
 384       (format *debug-io* "Injecting parser functions into ~A~%" dst-package)
 385       (loop :for aform :in form
 386             :do (progn
 387                   (format *debug-io* "Skipping aform ~A~%" aform)))
 388       (loop :for (sym code) :in actions
 389             :do (intern (compile sym `(lambda (data) (declare (ignorable data)) ,code)))
 390             :do (format *debug-io* "Compiled ~A~%" sym))
 391       (intern (compile 'parse-string
 392                        #'(lambda (s)
 393                            `,(format nil "Parse S using grammar ~A starting at ~A" grammarfile head)
 394                            (let ((*context* (make-instance 'opossum:context :dstpkg *package* :input s)))
 395                              (funcall (make-name head) 0)))))
 396       (intern (compile 'parse-file
 397                        #'(lambda (f) (parse-string (read-file f)))))
 398       (intern (compile 'parse-stream
 399                        #'(lambda (s) (parse-string (read-stream s)))))
 400       (intern '*trace*)
 401       (setf (documentation '*trace* 'cl:variable)
 402             "When non-nil, the generated parser function log to cl:*trace-output*.")
 403       (export '(parse-string parse-file parse-stream *trace*)))))
 404
 405 (defun get-iso-time ()
 406   "Return a string in ISO format for the current time"
 407   (multiple-value-bind (second minute hour date month year day daylight-p zone)
 408       (get-decoded-time)
 409     (declare (ignore day))
 410     (format nil "~4D-~2,'0D-~2,'0D-~2,'0D:~2,'0D:~2,'0D (UCT~@D)"
 411             year month date
 412             hour minute second
 413             (if daylight-p (1+ (- zone)) (- zone)))))
 414
 415 (defun cleanup-action-code (code)
 416   "Remove trailing newlines  so that the string CODE can be printed nicely."
 417   (subseq code 0
 418           (when (char= #\Newline (char code (1- (length code))))
 419             (1+ (position #\Newline code :from-end T :test #'char/=)))))
 420
 421 (defun generate-parser-file (grammarfile dst-package dst-file &key start-rule (parse-file-fun #'opossum:parse-file))
 422   "Create lisp code in DST-FILE that can be loaded to yield functions to parse using GRAMMARFILE in DST-PACKAGE.
 423 DST-PACKAGE will contain the 3 functions PARSE-STRING, PARSE-FILE and PARSE-STREAM as exported entrypoints."
 424   (let ((*package* dst-package))
 425     (let ((result (funcall parse-file-fun grammarfile)))
 426       ;; FIXME: check for complete parse
 427       (let ((*context* result)) ;; routines in pegutils.lisp expect *context* to be bound properly
 428         (let ((forms (transform (value result)))
 429               (actions (actions result)))
 430           (with-open-file (s dst-file :direction :output :if-exists :supersede)
 431             (let ((*print-readably* T)
 432                   (*print-pretty* T)
 433                   (*print-circle* NIL)
 434                   (dpkg (intern (package-name dst-package) :keyword)))
 435               (format s ";; This is a Common Lisp peg parser automatically generated by OPOSSUM -*- mode:lisp -*-~%")
 436               (format s ";; generated from ~A on ~A~%" grammarfile (get-iso-time))
 437               (prin1
 438                `(eval-when (:load-toplevel :compile-toplevel :execute)
 439                  (declaim (optimize (speed 0) (safety 3) (debug 3))))
 440                s) (terpri s)
 441               (prin1
 442                `(defpackage ,dpkg
 443                  (:use :cl :opossum)
 444                  (:export :parse-string :parse-file :parse-stream))
 445                s) (terpri s)
 446               (prin1 `(in-package ,dpkg) s) (terpri s)
 447               ;; First form is taken to be the start rule
 448               (let ((entryrule (or (and start-rule (make-name start-rule))
 449                                    (and forms (cadr (first forms))))))
 450                 (if (not entryrule)
 451                     (format *error-output* "Cannot find entry rule for parser")
 452                     (progn
 453                       (when *trace*
 454                         (format *trace-output* "Inserting definitions for parser entry points through ~A~%"
 455                                 entryrule))
 456                       (terpri s)
 457                       (prin1 `(defun parse-string (,(intern :s dst-package))
 458                                ,(format nil "Parse S using grammar ~A starting at ~A" grammarfile entryrule)
 459                                (let ((*context* (make-instance 'opossum:context
 460                                                                :dstpkg ,dpkg
 461                                                                :input ,(intern :s dst-package))))
 462                                  (funcall (,entryrule) 0)))
 463                              s)
 464                       (fresh-line s))))
 465               (loop :for aform :in forms
 466                     :do (when *trace* (format *trace-output* "Inserting form ~A~%" aform))
 467                     :do (terpri s)
 468                     :do (prin1 aform s)
 469                     :do (fresh-line s))
 470               (terpri s)
 471               (prin1
 472                `(defparameter ,(intern "*trace*" dst-package) nil
 473                  "When non-nil, the generated parser function log to cl:*trace-output*.")
 474                s)
 475               (terpri s)
 476
 477               (loop :for (sym code) :in actions
 478                     :when sym ;; the final action is named NIL because we push a
 479                     ;; NIL ahead of us in store-actions
 480                     :do (when *trace* (format *trace-output* "Inserting defun for ~A~%" sym))
 481                     :and :do (format s "~%(defun ~S (data)~% ~A)~%"
 482                                      sym (cleanup-action-code code))))))))))
 483
 484 (defun transform (tree &optional (depth 0))
 485   (if (and tree
 486            (consp tree))
 487       (if (eq (first tree) ':action)
 488           (progn
 489             (when *trace*
 490               (format *trace-output* "~AFound action ~A~%"  (make-string depth :initial-element #\Space) tree))
 491             tree)
 492           (let ((data (mapcar #'(lambda (tr) (transform tr (1+ depth)))
 493                               tree)))
 494             (loop :for el :in data
 495                   :when (and (listp el)
 496                              (eq (first el) ':action)
 497                              (symbolp (third el)))
 498                   :do (let ((*package* (destpkg *context*))
 499                             (action (third el)))
 500                         (when *trace*
 501                           (format *trace-output* "~&Applying action ~A to ~A~%" action data))
 502                         (handler-case
 503                             (return-from transform
 504                               (funcall (symbol-function action) data))
 505                           (undefined-function (x)
 506                             (progn
 507                               (format *error-output* "missing definition for ~A: ~A~%" action x)
 508                               ;                       (break "~A in ~A" action *package*)
 509                                     tree)))))
 510             data))
 511       tree))