xslt.lisp

   1 ;;; -*- show-trailing-whitespace: t; indent-tabs: nil -*-
   2
   3 ;;; Copyright (c) 2007,2008 David Lichteblau, Ivan Shvedunov.
   4 ;;; All rights reserved.
   5
   6 ;;; Redistribution and use in source and binary forms, with or without
   7 ;;; modification, are permitted provided that the following conditions
   8 ;;; are met:
   9 ;;;
  10 ;;;   * Redistributions of source code must retain the above copyright
  11 ;;;     notice, this list of conditions and the following disclaimer.
  12 ;;;
  13 ;;;   * Redistributions in binary form must reproduce the above
  14 ;;;     copyright notice, this list of conditions and the following
  15 ;;;     disclaimer in the documentation and/or other materials
  16 ;;;     provided with the distribution.
  17 ;;;
  18 ;;; THIS SOFTWARE IS PROVIDED BY THE AUTHOR 'AS IS' AND ANY EXPRESSED
  19 ;;; OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  20 ;;; WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  21 ;;; ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
  22 ;;; DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  23 ;;; DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
  24 ;;; GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
  25 ;;; INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
  26 ;;; WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
  27 ;;; NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
  28 ;;; SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  29
  30 (in-package :xuriella)
  31
  32 #+sbcl
  33 (declaim (optimize (debug 2)))
  34
  35
  36 ;;;; XSLT-ERROR
  37
  38 (define-condition xslt-error (simple-error)
  39   ()
  40   (:documentation "The class of all XSLT errors."))
  41
  42 (define-condition recoverable-xslt-error (xslt-error)
  43   ()
  44   (:documentation "The class of recoverable XSLT errors."))
  45
  46 (defun xslt-error (fmt &rest args)
  47   (error 'xslt-error :format-control fmt :format-arguments args))
  48
  49 (defun xslt-cerror (fmt &rest args)
  50   (with-simple-restart (recover "recover")
  51     (error 'recoverable-xslt-error
  52            :format-control fmt
  53            :format-arguments args)))
  54
  55 (defvar *debug* nil)
  56
  57 (defmacro handler-case* (form &rest clauses)
  58   ;; like HANDLER-CASE if *DEBUG* is off.  If it's on, don't establish
  59   ;; a handler at all so that we see the real stack traces.  (We could use
  60   ;; HANDLER-BIND here and check at signalling time, but doesn't seem
  61   ;; important.)
  62   (let ((doit (gensym)))
  63     `(flet ((,doit () ,form))
  64        (if *debug*
  65            (,doit)
  66            (handler-case
  67                (,doit)
  68              ,@clauses)))))
  69
  70 (defun compile-xpath (xpath &optional env)
  71   (handler-case*
  72       (xpath:compile-xpath xpath env)
  73     (xpath:xpath-error (c)
  74       (xslt-error "~A" c))))
  75
  76 (defmacro with-stack-limit ((&optional) &body body)
  77   `(invoke-with-stack-limit (lambda () ,@body)))
  78
  79
  80 ;;;; Helper function and macro
  81
  82 (defun map-pipe-eagerly (fn pipe)
  83   (xpath::enumerate pipe :key fn :result nil))
  84
  85 (defmacro do-pipe ((var pipe &optional result) &body body)
  86   `(block nil
  87      (map-pipe-eagerly #'(lambda (,var) ,@body) ,pipe)
  88      ,result))
  89
  90
  91 ;;;; XSLT-ENVIRONMENT and XSLT-CONTEXT
  92
  93 (defparameter *initial-namespaces*
  94   '((nil . "")
  95     ("xmlns" . #"http://www.w3.org/2000/xmlns/")
  96     ("xml" . #"http://www.w3.org/XML/1998/namespace")))
  97
  98 (defparameter *namespaces* *initial-namespaces*)
  99
 100 (defvar *global-variable-declarations*)
 101 (defvar *lexical-variable-declarations*)
 102
 103 (defvar *global-variable-values*)
 104 (defvar *lexical-variable-values*)
 105
 106 (defclass xslt-environment () ())
 107
 108 (defun split-qname (str)
 109   (handler-case
 110       (multiple-value-bind (prefix local-name)
 111           (cxml::split-qname str)
 112         (unless
 113             ;; FIXME: cxml should really offer a function that does
 114             ;; checks for NCName and QName in a sensible way for user code.
 115             ;; cxml::split-qname is tailored to the needs of the parser.
 116             ;;
 117             ;; For now, let's just check the syntax explicitly.
 118             (and (or (null prefix) (xpath::nc-name-p prefix))
 119                  (xpath::nc-name-p local-name))
 120           (xslt-error "not a qname: ~A" str))
 121         (values prefix local-name))
 122     (cxml:well-formedness-violation ()
 123       (xslt-error "not a qname: ~A" str))))
 124
 125 (defun decode-qname (qname env attributep)
 126   (multiple-value-bind (prefix local-name)
 127       (split-qname qname)
 128     (values local-name
 129             (if (or prefix (not attributep))
 130                 (xpath:environment-find-namespace env prefix)
 131                 "")
 132             prefix)))
 133
 134 (defmethod xpath:environment-find-namespace ((env xslt-environment) prefix)
 135   (cdr (assoc prefix *namespaces* :test 'equal)))
 136
 137 (defun find-variable-index (local-name uri table)
 138   (position (cons local-name uri) table :test 'equal))
 139
 140 (defun intern-global-variable (local-name uri)
 141   (or (find-variable-index local-name uri *global-variable-declarations*)
 142       (push-variable local-name uri *global-variable-declarations*)))
 143
 144 (defun push-variable (local-name uri table)
 145   (prog1
 146       (length table)
 147     (vector-push-extend (cons local-name uri) table)))
 148
 149 (defun lexical-variable-value (index &optional (errorp t))
 150   (let ((result (svref *lexical-variable-values* index)))
 151     (when errorp
 152       (assert (not (eq result 'unbound))))
 153     result))
 154
 155 (defun (setf lexical-variable-value) (newval index)
 156   (assert (not (eq newval 'unbound)))
 157   (setf (svref *lexical-variable-values* index) newval))
 158
 159 (defun global-variable-value (index &optional (errorp t))
 160   (let ((result (svref *global-variable-values* index)))
 161     (when errorp
 162       (assert (not (eq result 'unbound))))
 163     result))
 164
 165 (defun (setf global-variable-value) (newval index)
 166   (assert (not (eq newval 'unbound)))
 167   (setf (svref *global-variable-values* index) newval))
 168
 169 (defmethod xpath:environment-find-function
 170     ((env xslt-environment) lname uri)
 171   (if (string= uri "")
 172       (or (xpath:find-xpath-function lname *xsl*)
 173           (xpath:find-xpath-function lname uri))
 174       (xpath:find-xpath-function lname uri)))
 175
 176 (defmethod xpath:environment-find-variable
 177     ((env xslt-environment) lname uri)
 178   (let ((index
 179          (find-variable-index lname uri *lexical-variable-declarations*)))
 180     (when index
 181       (lambda (ctx)
 182         (declare (ignore ctx))
 183         (svref *lexical-variable-values* index)))))
 184
 185 (defclass lexical-xslt-environment (xslt-environment) ())
 186
 187 (defmethod xpath:environment-find-variable
 188     ((env lexical-xslt-environment) lname uri)
 189   (or (call-next-method)
 190       (let ((index
 191              (find-variable-index lname uri *global-variable-declarations*)))
 192         (when index
 193           (xslt-trace-thunk
 194            (lambda (ctx)
 195              (declare (ignore ctx))
 196              (svref *global-variable-values* index))
 197            "global ~s (uri ~s) = ~s" lname uri :result)))))
 198
 199 (defclass global-variable-environment (xslt-environment)
 200   ((initial-global-variable-thunks
 201     :initarg :initial-global-variable-thunks
 202     :accessor initial-global-variable-thunks)))
 203
 204 (defmethod xpath:environment-find-variable
 205     ((env global-variable-environment) lname uri)
 206   (or (call-next-method)
 207       (gethash (cons lname uri) (initial-global-variable-thunks env))))
 208
 209
 210 ;;;; TEXT-OUTPUT-SINK
 211 ;;;;
 212 ;;;; A sink that serializes only text and will error out on any other
 213 ;;;; SAX event.
 214
 215 (defmacro with-text-output-sink ((var) &body body)
 216   `(invoke-with-text-output-sink (lambda (,var) ,@body)))
 217
 218 (defclass text-output-sink (sax:default-handler)
 219   ((target :initarg :target :accessor text-output-sink-target)))
 220
 221 (defmethod sax:characters ((sink text-output-sink) data)
 222   (write-string data (text-output-sink-target sink)))
 223
 224 (defun invoke-with-text-output-sink (fn)
 225   (with-output-to-string (s)
 226     (funcall fn (make-instance 'text-output-sink :target s))))
 227
 228
 229 ;;;; Names
 230
 231 (eval-when (:compile-toplevel :load-toplevel :execute)
 232   (defvar *xsl* "http://www.w3.org/1999/XSL/Transform")
 233   (defvar *xml* "http://www.w3.org/XML/1998/namespace")
 234   (defvar *html* "http://www.w3.org/1999/xhtml"))
 235
 236 (defun of-name (local-name)
 237   (stp:of-name local-name *xsl*))
 238
 239 (defun namep (node local-name)
 240   (and (typep node '(or stp:element stp:attribute))
 241        (equal (stp:namespace-uri node) *xsl*)
 242        (equal (stp:local-name node) local-name)))
 243
 244
 245 ;;;; PARSE-STYLESHEET
 246
 247 (defstruct stylesheet
 248   (modes (make-hash-table :test 'equal))
 249   (global-variables ())
 250   (output-specification (make-output-specification))
 251   (strip-tests nil)
 252   (named-templates (make-hash-table :test 'equal))
 253   (attribute-sets (make-hash-table :test 'equal)))
 254
 255 (defstruct mode (templates nil))
 256
 257 (defun find-mode (stylesheet local-name &optional uri)
 258   (gethash (cons local-name uri) (stylesheet-modes stylesheet)))
 259
 260 (defun ensure-mode (stylesheet &optional local-name uri)
 261   (or (find-mode stylesheet local-name uri)
 262       (setf (gethash (cons local-name uri) (stylesheet-modes stylesheet))
 263             (make-mode))))
 264
 265 (defun ensure-mode/qname (stylesheet qname env)
 266   (if qname
 267       (multiple-value-bind (local-name uri)
 268           (decode-qname qname env nil)
 269         (ensure-mode stylesheet local-name uri))
 270       (find-mode stylesheet nil)))
 271
 272 (defun acons-namespaces (element &optional (bindings *namespaces*))
 273   (map-namespace-declarations (lambda (prefix uri)
 274                                 (push (cons prefix uri) bindings))
 275                               element)
 276   bindings)
 277
 278 (defvar *excluded-namespaces* (list *xsl*))
 279 (defvar *empty-mode*)
 280
 281 (defvar *xsl-include-stack* nil)
 282
 283 (defun uri-to-pathname (uri)
 284   (cxml::uri-to-pathname (puri:parse-uri uri)))
 285
 286 (defun parse-stylesheet-to-stp (input uri-resolver)
 287   (let* ((d (cxml:parse input (make-text-normalizer (cxml-stp:make-builder))))
 288          (<transform> (stp:document-element d)))
 289     (strip-stylesheet <transform>)
 290     ;; FIXME: handle embedded stylesheets
 291     (unless (and (equal (stp:namespace-uri <transform>) *xsl*)
 292                  (or (equal (stp:local-name <transform>) "transform")
 293                      (equal (stp:local-name <transform>) "stylesheet")))
 294       (xslt-error "not a stylesheet"))
 295     (dolist (include (stp:filter-children (of-name "include") <transform>))
 296       (let* ((uri (puri:merge-uris (stp:attribute-value include "href")
 297                                    (stp:base-uri include)))
 298              (uri (if uri-resolver
 299                       (funcall uri-resolver (puri:render-uri uri nil))
 300                       uri))
 301              (str (puri:render-uri uri nil))
 302              (pathname
 303               (handler-case
 304                   (uri-to-pathname uri)
 305                 (cxml:xml-parse-error (c)
 306                   (xslt-error "cannot find included stylesheet ~A: ~A"
 307                               uri c)))))
 308         (with-open-file
 309             (stream pathname
 310                     :element-type '(unsigned-byte 8)
 311                     :if-does-not-exist nil)
 312           (unless stream
 313             (xslt-error "cannot find included stylesheet ~A at ~A"
 314                         uri pathname))
 315           (when (find str *xsl-include-stack* :test #'equal)
 316             (xslt-error "recursive inclusion of ~A" uri))
 317           (let* ((*xsl-include-stack* (cons str *xsl-include-stack*))
 318                  (<transform>2 (parse-stylesheet-to-stp stream uri-resolver)))
 319             (stp:do-children (child <transform>2)
 320               (stp:insert-child-after <transform>
 321                                       (stp:copy child)
 322                                       include))
 323             (stp:detach include)))))
 324     <transform>))
 325
 326 (defvar *instruction-base-uri*)
 327 (defvar *apply-imports-limit*)
 328 (defvar *import-priority*)
 329
 330 (defun parse-1-stylesheet (env stylesheet designator uri-resolver)
 331   (let* ((<transform> (parse-stylesheet-to-stp designator uri-resolver))
 332          (*instruction-base-uri* (stp:base-uri <transform>))
 333          (*namespaces* (acons-namespaces <transform>))
 334          (*apply-imports-limit* (1+ *import-priority*)))
 335     (dolist (import (stp:filter-children (of-name "import") <transform>))
 336       (let ((uri (puri:merge-uris (stp:attribute-value import "href")
 337                                   (stp:base-uri import))))
 338         (parse-imported-stylesheet env stylesheet uri uri-resolver)))
 339     (incf *import-priority*)
 340     (parse-exclude-result-prefixes! <transform> env)
 341     (parse-global-variables! stylesheet <transform>)
 342     (parse-templates! stylesheet <transform> env)
 343     (parse-output! stylesheet <transform>)
 344     (parse-strip/preserve-space! stylesheet <transform> env)
 345     (parse-attribute-sets! stylesheet <transform> env)))
 346
 347 (defvar *xsl-import-stack* nil)
 348
 349 (defun parse-imported-stylesheet (env stylesheet uri uri-resolver)
 350   (let* ((uri (if uri-resolver
 351                   (funcall uri-resolver (puri:render-uri uri nil))
 352                   uri))
 353          (str (puri:render-uri uri nil))
 354          (pathname
 355           (handler-case
 356               (uri-to-pathname uri)
 357             (cxml:xml-parse-error (c)
 358               (xslt-error "cannot find imported stylesheet ~A: ~A"
 359                           uri c)))))
 360     (with-open-file
 361         (stream pathname
 362                 :element-type '(unsigned-byte 8)
 363                 :if-does-not-exist nil)
 364       (unless stream
 365         (xslt-error "cannot find imported stylesheet ~A at ~A"
 366                     uri pathname))
 367       (when (find str *xsl-import-stack* :test #'equal)
 368         (xslt-error "recursive inclusion of ~A" uri))
 369       (let ((*xsl-import-stack* (cons str *xsl-import-stack*)))
 370         (parse-1-stylesheet env stylesheet stream uri-resolver)))))
 371
 372 (defun parse-stylesheet (designator &key uri-resolver)
 373   (let* ((*import-priority* 0)
 374          (puri:*strict-parse* nil)
 375          (stylesheet (make-stylesheet))
 376          (env (make-instance 'lexical-xslt-environment))
 377          (*excluded-namespaces* *excluded-namespaces*)
 378          (*global-variable-declarations* (make-empty-declaration-array)))
 379     (ensure-mode stylesheet nil)
 380     (parse-1-stylesheet env stylesheet designator uri-resolver)
 381     ;; reverse attribute sets:
 382     (let ((table (stylesheet-attribute-sets stylesheet)))
 383       (maphash (lambda (k v)
 384                  (setf (gethash k table) (nreverse v)))
 385                table))
 386     stylesheet))
 387
 388 (defun parse-attribute-sets! (stylesheet <transform> env)
 389   (dolist (elt (stp:filter-children (of-name "attribute-set") <transform>))
 390     (push (let* ((sets
 391                   (mapcar (lambda (qname)
 392                             (multiple-value-list (decode-qname qname env nil)))
 393                           (words
 394                            (stp:attribute-value elt "use-attribute-sets"))))
 395                  (instructions
 396                   (stp:map-children 'list #'parse-instruction elt))
 397                  (*lexical-variable-declarations*
 398                   (make-empty-declaration-array))
 399                  (thunk
 400                   (compile-instruction `(progn ,@instructions) env))
 401                  (n-variables (length *lexical-variable-declarations*)))
 402             (lambda (ctx)
 403               (with-stack-limit ()
 404                 (loop for (local-name uri nil) in sets do
 405                      (dolist (thunk (find-attribute-set local-name uri))
 406                        (funcall thunk ctx)))
 407                 (let ((*lexical-variable-values*
 408                        (make-variable-value-array n-variables)))
 409                   (funcall thunk ctx)))))
 410           (gethash (multiple-value-bind (local-name uri)
 411                        (decode-qname (stp:attribute-value elt "name") env nil)
 412                      (cons local-name uri))
 413                    (stylesheet-attribute-sets stylesheet)))))
 414
 415 (defun parse-exclude-result-prefixes! (<transform> env)
 416   (stp:with-attributes (exclude-result-prefixes) <transform>
 417       (dolist (prefix (words (or exclude-result-prefixes "")))
 418     (when (equal prefix "#default")
 419       (setf prefix nil))
 420     (push (or (xpath:environment-find-namespace env prefix)
 421               (xslt-error "namespace not found: ~A" prefix))
 422           *excluded-namespaces*))))
 423
 424 (defun parse-strip/preserve-space! (stylesheet <transform> env)
 425   (xpath:with-namespaces ((nil #.*xsl*))
 426     (dolist (elt (stp:filter-children (lambda (x)
 427                                         (or (namep x "strip-space")
 428                                             (namep x "preserve-space")))
 429                                       <transform>))
 430       (let ((*namespaces* (acons-namespaces elt))
 431             (mode
 432              (if (equal (stp:local-name elt) "strip-space")
 433                  :strip
 434                  :preserve)))
 435         (dolist (name-test (words (stp:attribute-value elt "elements")))
 436           (let* ((pos (search ":*" name-test))
 437                  (test-function
 438                   (cond
 439                     ((eql pos (- (length name-test) 2))
 440                      (let* ((prefix (subseq name-test 0 pos))
 441                             (name-test-uri
 442                              (xpath:environment-find-namespace env prefix)))
 443                        (unless (xpath::nc-name-p prefix)
 444                          (xslt-error "not an NCName: ~A" prefix))
 445                        (lambda (local-name uri)
 446                          (declare (ignore local-name))
 447                          (if (equal uri name-test-uri)
 448                              mode
 449                              nil))))
 450                     ((equal name-test "*")
 451                      (lambda (local-name uri)
 452                        (declare (ignore local-name uri))
 453                        mode))
 454                     (t
 455                      (multiple-value-bind (name-test-local-name name-test-uri)
 456                          (decode-qname name-test env nil)
 457                        (lambda (local-name uri)
 458                          (if (and (equal local-name name-test-local-name)
 459                                   (equal uri name-test-uri))
 460                              mode
 461                              nil)))))))
 462             (push test-function (stylesheet-strip-tests stylesheet))))))))
 463
 464 (defstruct (output-specification
 465              (:conc-name "OUTPUT-"))
 466   method
 467   indent
 468   omit-xml-declaration
 469   encoding)
 470
 471 (defun parse-output! (stylesheet <transform>)
 472   (let ((outputs (stp:filter-children (of-name "output") <transform>)))
 473     (when outputs
 474       (when (cdr outputs)
 475         ;; FIXME:
 476         ;;   - concatenate cdata-section-elements
 477         ;;   - the others must not conflict
 478         (error "oops, merging of output elements not supported yet"))
 479       (let ((<output> (car outputs))
 480             (spec (stylesheet-output-specification stylesheet)))
 481         (stp:with-attributes (;; version
 482                               method
 483                               indent
 484                               encoding
 485 ;;;                           media-type
 486 ;;;                           doctype-system
 487 ;;;                           doctype-public
 488                               omit-xml-declaration
 489 ;;;                           standalone
 490 ;;;                           cdata-section-elements
 491                               )
 492             <output>
 493           (setf (output-method spec) method)
 494           (setf (output-indent spec) indent)
 495           (setf (output-encoding spec) encoding)
 496           (setf (output-omit-xml-declaration spec) omit-xml-declaration))))))
 497
 498 (defun make-empty-declaration-array ()
 499   (make-array 1 :fill-pointer 0 :adjustable t))
 500
 501 (defun make-variable-value-array (n-lexical-variables)
 502   (make-array n-lexical-variables :initial-element 'unbound))
 503
 504 (defun compile-global-variable (<variable> env) ;; also for <param>
 505   (stp:with-attributes (name select) <variable>
 506     (when (and select (stp:list-children <variable>))
 507       (xslt-error "variable with select and body"))
 508     (let* ((*lexical-variable-declarations* (make-empty-declaration-array))
 509            (inner (cond
 510                     (select
 511                      (compile-xpath select env))
 512                     ((stp:list-children <variable>)
 513                      (let* ((inner-sexpr `(progn ,@(parse-body <variable>)))
 514                             (inner-thunk (compile-instruction inner-sexpr env)))
 515                        (lambda (ctx)
 516                          (apply-to-result-tree-fragment ctx inner-thunk))))
 517                     (t
 518                      (lambda (ctx)
 519                        (declare (ignore ctx))
 520                        ""))))
 521            (n-lexical-variables (length *lexical-variable-declarations*)))
 522       (xslt-trace-thunk
 523        (lambda (ctx)
 524          (let* ((*lexical-variable-values*
 525                  (make-variable-value-array n-lexical-variables)))
 526            (funcall inner ctx)))
 527        "global ~s (~s) = ~s" name select :result))))
 528
 529 (defstruct (variable-information
 530              (:constructor make-variable)
 531              (:conc-name "VARIABLE-"))
 532   index
 533   thunk
 534   local-name
 535   uri
 536   param-p
 537   thunk-setter)
 538
 539 (defun parse-global-variable! (<variable> global-env) ;; also for <param>
 540   (let ((*namespaces* (acons-namespaces <variable>))
 541         (qname (stp:attribute-value <variable> "name")))
 542     (unless qname
 543       (xslt-error "name missing in ~A" (stp:local-name <variable>)))
 544     (multiple-value-bind (local-name uri)
 545         (decode-qname qname global-env nil)
 546       ;; For the normal compilation environment of templates, install it
 547       ;; into *GLOBAL-VARIABLE-DECLARATIONS*:
 548       (let ((index (intern-global-variable local-name uri)))
 549         ;; For the evaluation of a global variable itself, build a thunk
 550         ;; that lazily resolves other variables, stored into
 551         ;; INITIAL-GLOBAL-VARIABLE-THUNKS:
 552         (let* ((value-thunk :unknown)
 553                (global-variable-thunk
 554                 (lambda (ctx)
 555                   (let ((v (global-variable-value index nil)))
 556                     (when (eq v 'seen)
 557                       (xslt-error "recursive variable definition"))
 558                     (cond
 559                       ((eq v 'unbound)
 560                        ;; (print (list :computing index))
 561                        (setf (global-variable-value index) 'seen)
 562                        (setf (global-variable-value index)
 563                              (funcall value-thunk ctx))
 564                        #+nil (print (list :done-computing index
 565                                     (global-variable-value index)))
 566                        #+nil (global-variable-value index))
 567                       (t
 568                        #+nil(print (list :have
 569                                     index v))
 570                        v)))))
 571                (thunk-setter
 572                 (lambda ()
 573                   (setf value-thunk
 574                         (compile-global-variable <variable> global-env)))))
 575           (setf (gethash (cons local-name uri)
 576                          (initial-global-variable-thunks global-env))
 577                 global-variable-thunk)
 578           (make-variable :index index
 579                          :local-name local-name
 580                          :uri uri
 581                          :thunk global-variable-thunk
 582                          :param-p (namep <variable> "param")
 583                          :thunk-setter thunk-setter))))))
 584
 585 (defun parse-global-variables! (stylesheet <transform>)
 586   (xpath:with-namespaces ((nil #.*xsl*))
 587     (let* ((table (make-hash-table :test 'equal))
 588            (global-env (make-instance 'global-variable-environment
 589                                       :initial-global-variable-thunks table))
 590            (specs '()))
 591       (xpath:do-node-set
 592           (<variable> (xpath:evaluate "variable|param" <transform>))
 593         (let ((var (parse-global-variable! <variable> global-env)))
 594           (xslt-trace "parsing global variable ~s (uri ~s)"
 595                       (variable-local-name var)
 596                       (variable-uri var))
 597           (when (find var
 598                       specs
 599                       :test (lambda (a b)
 600                               (and (equal (variable-local-name a)
 601                                           (variable-local-name b))
 602                                    (equal (variable-uri a)
 603                                           (variable-uri b)))))
 604             (xslt-error "duplicate definition for global variable ~A"
 605                         (variable-local-name var)))
 606           (push var specs)))
 607       ;; now that the global environment knows about all variables, run the
 608       ;; thunk setters to perform their compilation
 609       (setf specs (nreverse specs))
 610       (mapc (lambda (spec) (funcall (variable-thunk-setter spec))) specs)
 611       (setf (stylesheet-global-variables stylesheet) specs))))
 612
 613 (defun parse-templates! (stylesheet <transform> env)
 614   (let ((i 0))
 615     (dolist (<template> (stp:filter-children (of-name "template") <transform>))
 616       (let ((*namespaces* (acons-namespaces <template>)))
 617         (dolist (template (compile-template <template> env i))
 618           (let ((name (template-name template)))
 619             (if name
 620                 (let* ((table (stylesheet-named-templates stylesheet))
 621                        (head (car (gethash name table))))
 622                   (when (and head (eql (template-import-priority head)
 623                                        (template-import-priority template)))
 624                     ;; fixme: is this supposed to be a run-time error?
 625                     (xslt-error "conflicting templates for ~A" name))
 626                   (push template (gethash name table)))
 627                 (let ((mode (ensure-mode/qname stylesheet
 628                                                (template-mode-qname template)
 629                                                env)))
 630                   (setf (template-mode template) mode)
 631                   (push template (mode-templates mode)))))))
 632       (incf i))))
 633
 634
 635 ;;;; APPLY-STYLESHEET
 636
 637 (defvar *stylesheet*)
 638 (defvar *mode*)
 639
 640 (deftype xml-designator () '(or runes:xstream runes:rod array stream pathname))
 641
 642 (defstruct (parameter
 643              (:constructor make-parameter (value local-name &optional uri)))
 644   (uri "")
 645   local-name
 646   value)
 647
 648 (defun find-parameter-value (local-name uri parameters)
 649   (dolist (p parameters)
 650     (when (and (equal (parameter-local-name p) local-name)
 651                (equal (parameter-uri p) uri))
 652       (return (parameter-value p)))))
 653
 654 (defvar *uri-resolver*)
 655
 656 (defun parse-allowing-microsoft-bom (pathname handler)
 657   (with-open-file (s pathname :element-type '(unsigned-byte 8))
 658     (unless (and (eql (read-byte s nil) #xef)
 659                  (eql (read-byte s nil) #xbb)
 660                  (eql (read-byte s nil) #xbf))
 661       (file-position s 0))
 662     (cxml:parse s handler)))
 663
 664 (defun %document (uri-string base-uri)
 665   (let* ((absolute-uri
 666           (puri:merge-uris uri-string base-uri))
 667          (resolved-uri
 668           (if *uri-resolver*
 669               (funcall *uri-resolver* (puri:render-uri absolute-uri nil))
 670               absolute-uri))
 671          (pathname
 672           (handler-case
 673               (uri-to-pathname resolved-uri)
 674             (cxml:xml-parse-error (c)
 675               (xslt-error "cannot find referenced document ~A: ~A"
 676                           resolved-uri c))))
 677          (document
 678           (handler-case
 679               (parse-allowing-microsoft-bom pathname (stp:make-builder))
 680             ((or file-error cxml:xml-parse-error) (c)
 681               (xslt-error "cannot parse referenced document ~A: ~A"
 682                           pathname c))))
 683          (xpath-root-node
 684           (make-whitespace-stripper document
 685                                     (stylesheet-strip-tests *stylesheet*))))
 686     (when (puri:uri-fragment absolute-uri)
 687       (xslt-error "use of fragment identifiers in document() not supported"))
 688     xpath-root-node))
 689
 690 (xpath:define-extension xslt *xsl*)
 691
 692 (xpath:define-xpath-function/lazy
 693     xslt :document
 694     (object &optional node-set)
 695   (let ((instruction-base-uri *instruction-base-uri*))
 696     (lambda (ctx)
 697       (declare (ignore ctx))
 698       (let* ((object (funcall object))
 699              (node-set (and node-set (funcall node-set)))
 700              (uri
 701               (when node-set
 702                 ;; FIXME: should use first node of the node set
 703                 ;; _in document order_
 704                 (xpath-protocol:base-uri (xpath:first-node node-set)))))
 705         (xpath::make-node-set
 706          (if (xpath:node-set-p object)
 707              (xpath:map-node-set->list
 708               (lambda (node)
 709                 (%document (xpath:string-value node)
 710                            (or uri (xpath-protocol:base-uri node))))
 711               object)
 712              (list (%document (xpath:string-value object)
 713                               (or uri instruction-base-uri)))))))))
 714
 715 (defun apply-stylesheet
 716     (stylesheet source-document &key output parameters uri-resolver)
 717   (when (typep stylesheet 'xml-designator)
 718     (setf stylesheet (parse-stylesheet stylesheet)))
 719   (when (typep source-document 'xml-designator)
 720     (setf source-document (cxml:parse source-document (stp:make-builder))))
 721   (invoke-with-output-sink
 722    (lambda ()
 723      (handler-case*
 724          (let* ((puri:*strict-parse* nil)
 725                 (*stylesheet* stylesheet)
 726                 (*mode* (find-mode stylesheet nil))
 727                 (*empty-mode* (make-mode))
 728                 (global-variable-specs
 729                  (stylesheet-global-variables stylesheet))
 730                 (*global-variable-values*
 731                  (make-variable-value-array (length global-variable-specs)))
 732                 (*uri-resolver* uri-resolver)
 733                 (xpath-root-node
 734                  (make-whitespace-stripper
 735                   source-document
 736                   (stylesheet-strip-tests stylesheet)))
 737                 (ctx (xpath:make-context xpath-root-node)))
 738            (mapc (lambda (spec)
 739                    (when (variable-param-p spec)
 740                      (let ((value
 741                             (find-parameter-value (variable-local-name spec)
 742                                                   (variable-uri spec)
 743                                                   parameters)))
 744                        (when value
 745                          (setf (global-variable-value (variable-index spec))
 746                                value)))))
 747                  global-variable-specs)
 748            (mapc (lambda (spec)
 749                    (funcall (variable-thunk spec) ctx))
 750                  global-variable-specs)
 751            #+nil (print global-variable-specs)
 752            #+nil (print *global-variable-values*)
 753            (apply-templates ctx))
 754        (xpath:xpath-error (c)
 755                           (xslt-error "~A" c))))
 756    stylesheet
 757    output))
 758
 759 (defun find-attribute-set (local-name uri)
 760   (or (gethash (cons local-name uri) (stylesheet-attribute-sets *stylesheet*))
 761       (xslt-error "no such attribute set: ~A/~A" local-name uri)))
 762
 763 (defun apply-templates/list (list &optional param-bindings sort-predicate)
 764   (when sort-predicate
 765     (setf list (sort list sort-predicate)))
 766   (let* ((n (length list))
 767          (s/d (lambda () n)))
 768     (loop
 769        for i from 1
 770        for child in list
 771        do
 772          (apply-templates (xpath:make-context child s/d i)
 773            param-bindings))))
 774
 775 (defvar *stack-limit* 200)
 776
 777 (defun invoke-with-stack-limit (fn)
 778   (let ((*stack-limit* (1- *stack-limit*)))
 779     (unless (plusp *stack-limit*)
 780       (xslt-error "*stack-limit* reached; stack overflow"))
 781     (funcall fn)))
 782
 783 (defun invoke-template (ctx template param-bindings)
 784   (let ((*lexical-variable-values*
 785          (make-variable-value-array (template-n-variables template))))
 786     (with-stack-limit ()
 787       (loop
 788          for (name-cons value) in param-bindings
 789          for (nil index nil) = (find name-cons
 790                                      (template-params template)
 791                                      :test #'equal
 792                                      :key #'car)
 793          do
 794          (unless index
 795            (xslt-error "invalid template parameter ~A" name-cons))
 796          (setf (lexical-variable-value index) value))
 797       (funcall (template-body template) ctx))))
 798
 799 (defun apply-default-templates (ctx)
 800   (let ((node (xpath:context-node ctx)))
 801     (cond
 802       ((or (xpath-protocol:node-type-p node :processing-instruction)
 803            (xpath-protocol:node-type-p node :comment)))
 804       ((or (xpath-protocol:node-type-p node :text)
 805            (xpath-protocol:node-type-p node :attribute))
 806        (write-text (xpath-protocol:string-value node)))
 807       (t
 808        (apply-templates/list
 809         (xpath::force
 810          (xpath-protocol:child-pipe node)))))))
 811
 812 (defvar *apply-imports*)
 813
 814 (defun apply-applicable-templates (ctx templates param-bindings finally)
 815   (labels ((apply-imports ()
 816              (if templates
 817                  (let* ((this (pop templates))
 818                         (low (template-apply-imports-limit this))
 819                         (high (template-import-priority this)))
 820                    (setf templates
 821                          (remove-if-not
 822                           (lambda (x)
 823                             (<= low (template-import-priority x) high))
 824                           templates))
 825                    (invoke-template ctx this param-bindings))
 826                  (funcall finally))))
 827     (let ((*apply-imports* #'apply-imports))
 828       (apply-imports))))
 829
 830 (defun apply-templates (ctx &optional param-bindings)
 831   (apply-applicable-templates ctx
 832                               (find-templates ctx)
 833                               param-bindings
 834                               (lambda ()
 835                                 (apply-default-templates ctx))))
 836
 837 (defun call-template (ctx name &optional param-bindings)
 838   (apply-applicable-templates ctx
 839                               (find-named-templates name)
 840                               param-bindings
 841                               (lambda ()
 842                                 (error "cannot find named template: ~s"
 843                                        name))))
 844
 845 (defun find-templates (ctx)
 846   (let* ((matching-candidates
 847           (remove-if-not (lambda (template)
 848                            (template-matches-p template ctx))
 849                          (mode-templates *mode*)))
 850          (npriorities
 851           (if matching-candidates
 852               (1+ (reduce #'max
 853                           matching-candidates
 854                           :key #'template-import-priority))
 855               0))
 856          (priority-groups (make-array npriorities :initial-element nil)))
 857     (dolist (template matching-candidates)
 858       (push template
 859             (elt priority-groups (template-import-priority template))))
 860 ;;;     (print (map 'list #'length priority-groups))
 861 ;;;     (force-output)
 862     (loop
 863        for i from (1- npriorities) downto 0
 864        for group = (elt priority-groups i)
 865        for template = (maximize #'template< group)
 866        when template
 867        collect template)))
 868
 869 (defun find-named-templates (name)
 870   (gethash name (stylesheet-named-templates *stylesheet*)))
 871
 872 (defun template< (a b)                  ;assuming same import priority
 873   (let ((p (template-priority a))
 874         (q (template-priority b)))
 875     (cond
 876       ((< p q) t)
 877       ((> p q) nil)
 878       (t
 879        (xslt-cerror "conflicting templates:~_~A,~_~A"
 880                     (template-match-expression a)
 881                     (template-match-expression b))
 882        (< (template-position a) (template-position b))))))
 883
 884 (defun maximize (< things)
 885   (when things
 886     (let ((max (car things)))
 887       (dolist (other (cdr things))
 888         (when (funcall < max other)
 889           (setf max other)))
 890       max)))
 891
 892 (defun template-matches-p (template ctx)
 893   (find (xpath:context-node ctx)
 894         (xpath:all-nodes (funcall (template-match-thunk template) ctx))))
 895
 896 (defun invoke-with-output-sink (fn stylesheet output)
 897   (etypecase output
 898     (pathname
 899      (with-open-file (s output
 900                         :direction :output
 901                         :element-type '(unsigned-byte 8)
 902                         :if-exists :rename-and-delete)
 903        (invoke-with-output-sink fn stylesheet s)))
 904     ((or stream null)
 905      (invoke-with-output-sink fn
 906                               stylesheet
 907                               (make-output-sink stylesheet output)))
 908     ((or hax:abstract-handler sax:abstract-handler)
 909      (with-xml-output output
 910        (funcall fn)))))
 911
 912 (defun make-output-sink (stylesheet stream)
 913   (let* ((ystream
 914           (if stream
 915               (let ((et (stream-element-type stream)))
 916                 (cond
 917                   ((or (null et) (subtypep et '(unsigned-byte 8)))
 918                    (runes:make-octet-stream-ystream stream))
 919                   ((subtypep et 'character)
 920                    (runes:make-character-stream-ystream stream))))
 921               (runes:make-rod-ystream)))
 922          (output-spec (stylesheet-output-specification stylesheet))
 923          (omit-xml-declaration-p
 924           (equal (output-omit-xml-declaration output-spec) "yes"))
 925          (sax-target
 926           (make-instance 'cxml::sink
 927                          :ystream ystream
 928                          :omit-xml-declaration-p omit-xml-declaration-p)))
 929     (if (equalp (output-method (stylesheet-output-specification stylesheet))
 930                 "HTML")
 931         (make-instance 'combi-sink
 932                        :hax-target (make-instance 'chtml::sink
 933                                                   :ystream ystream)
 934                        :sax-target sax-target
 935                        :encoding (output-encoding output-spec))
 936         sax-target)))
 937
 938 (defstruct template
 939   match-expression
 940   match-thunk
 941   name
 942   import-priority
 943   apply-imports-limit
 944   priority
 945   position
 946   mode
 947   mode-qname
 948   params
 949   body
 950   n-variables)
 951
 952 (defun expression-priority (form)
 953   (let ((step (second form)))
 954     (if (and (null (cddr form))
 955              (listp step)
 956              (eq :child (car step))
 957              (null (cddr step)))
 958         (let ((name (second step)))
 959           (cond
 960             ((or (stringp name)
 961                  (and (consp name)
 962                       (or (eq (car name) :qname)
 963                           (eq (car name) :processing-instruction))))
 964              0.0)
 965             ((and (consp name)
 966                   (or (eq (car name) :namespace)
 967                       (eq (car name) '*)))
 968              -0.25)
 969             (t
 970              -0.5)))
 971         0.5)))
 972
 973 (defun valid-expression-p (expr)
 974   (cond
 975     ((atom expr) t)
 976     ((eq (first expr) :path)
 977      (every (lambda (x)
 978               (let ((filter (third x)))
 979                 (or (null filter) (valid-expression-p filter))))
 980             (cdr expr)))
 981     ((eq (first expr) :variable)        ;(!)
 982      nil)
 983     (t
 984      (every #'valid-expression-p (cdr expr)))))
 985
 986 (defun parse-pattern (str)
 987   ;; zzz check here for anything not allowed as an XSLT pattern
 988   ;; zzz can we hack id() and key() here?
 989   (let ((form (xpath:parse-xpath str)))
 990     (unless (consp form)
 991       (xslt-error "not a valid pattern: ~A" str))
 992     (labels ((process-form (form)
 993                (cond ((eq (car form) :union)
 994                       (alexandria:mappend #'process-form (rest form)))
 995                      ((not (eq (car form) :path)) ;zzz: filter statt path
 996                       (xslt-error "not a valid pattern: ~A" str))
 997                      ((not (valid-expression-p form))
 998                       (xslt-error "invalid filter"))
 999                      (t (list form)))))
1000       (process-form form))))
1001
1002 (defun compile-value-thunk (value env)
1003   (if (and (listp value) (eq (car value) 'progn))
1004       (let ((inner-thunk (compile-instruction value env)))
1005         (lambda (ctx)
1006           (apply-to-result-tree-fragment ctx inner-thunk)))
1007       (compile-xpath value env)))
1008
1009 (defun compile-var-bindings/nointern (forms env)
1010   (loop
1011     for (name value) in forms
1012     collect (multiple-value-bind (local-name uri)
1013                 (decode-qname name env nil)
1014               (list (cons local-name uri)
1015                     (xslt-trace-thunk
1016                      (compile-value-thunk value env)
1017                      "local variable ~s = ~s" name :result)))))
1018
1019 (defun compile-var-bindings (forms env)
1020   (loop
1021      for (cons thunk) in (compile-var-bindings/nointern forms env)
1022      for (local-name . uri) = cons
1023      collect (list cons
1024                    (push-variable local-name
1025                                   uri
1026                                   *lexical-variable-declarations*)
1027                    thunk)))
1028
1029 (defun compile-template (<template> env position)
1030   (stp:with-attributes (match name priority mode) <template>
1031     (unless (or name match)
1032       (xslt-error "missing match in template"))
1033     (multiple-value-bind (params body-pos)
1034         (loop
1035           for i from 0
1036           for child in (stp:list-children <template>)
1037           while (namep child "param")
1038           collect (parse-param child) into params
1039           finally (return (values params i)))
1040       (let* ((*lexical-variable-declarations* (make-empty-declaration-array))
1041              (param-bindings (compile-var-bindings params env))
1042              (body (parse-body <template> body-pos (mapcar #'car params)))
1043              (body-thunk (compile-instruction `(progn ,@body) env))
1044              (outer-body-thunk
1045               (xslt-trace-thunk
1046                #'(lambda (ctx)
1047                    (unwind-protect
1048                        (progn
1049                          ;; set params that weren't initialized by apply-templates
1050                          (loop for (name index param-thunk) in param-bindings
1051                                when (eq (lexical-variable-value index nil) 'unbound)
1052                                  do (setf (lexical-variable-value index)
1053                                           (funcall param-thunk ctx)))
1054                          (funcall body-thunk ctx))))
1055                "template: match = ~s name = ~s" match name))
1056              (n-variables (length *lexical-variable-declarations*)))
1057         (append
1058          (when name
1059            (multiple-value-bind (local-name uri)
1060                (decode-qname name env nil)
1061              (list
1062               (make-template :name (cons local-name uri)
1063                              :import-priority *import-priority*
1064                              :apply-imports-limit *apply-imports-limit*
1065                              :params param-bindings
1066                              :body outer-body-thunk
1067                              :n-variables n-variables))))
1068          (when match
1069            (mapcar (lambda (expression)
1070                      (let ((match-thunk
1071                             (xslt-trace-thunk
1072                              (compile-xpath
1073                               `(xpath:xpath
1074                                 (:path (:ancestor-or-self :node)
1075                                        ,@(cdr expression)))
1076                               env)
1077                              "match-thunk for template (match ~s): ~s --> ~s"
1078                              match expression :result))
1079                            (p (if priority
1080                                   (parse-number:parse-number priority)
1081                                   (expression-priority expression))))
1082                        (make-template :match-expression expression
1083                                       :match-thunk match-thunk
1084                                       :import-priority *import-priority*
1085                                       :apply-imports-limit *apply-imports-limit*
1086                                       :priority p
1087                                       :position position
1088                                       :mode-qname mode
1089                                       :params param-bindings
1090                                       :body outer-body-thunk
1091                                       :n-variables n-variables)))
1092                    (parse-pattern match))))))))
1093 #+(or)
1094 (xuriella::parse-stylesheet #p"/home/david/src/lisp/xuriella/test.xsl")