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-sys:environment-find-namespace env prefix)
 131                 "")
 132             prefix)))
 133
 134 (defmethod xpath-sys: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-sys:environment-find-function
 170     ((env xslt-environment) lname uri)
 171   (if (string= uri "")
 172       (or (xpath-sys:find-xpath-function lname *xsl*)
 173           (xpath-sys:find-xpath-function lname uri))
 174       (xpath-sys:find-xpath-function lname uri)))
 175
 176 (defmethod xpath-sys: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-sys: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-sys: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   (keys (make-hash-table :test 'equal)))
 255
 256 (defstruct mode (templates nil))
 257
 258 (defun find-mode (stylesheet local-name &optional uri)
 259   (gethash (cons local-name uri) (stylesheet-modes stylesheet)))
 260
 261 (defun ensure-mode (stylesheet &optional local-name uri)
 262   (or (find-mode stylesheet local-name uri)
 263       (setf (gethash (cons local-name uri) (stylesheet-modes stylesheet))
 264             (make-mode))))
 265
 266 (defun ensure-mode/qname (stylesheet qname env)
 267   (if qname
 268       (multiple-value-bind (local-name uri)
 269           (decode-qname qname env nil)
 270         (ensure-mode stylesheet local-name uri))
 271       (find-mode stylesheet nil)))
 272
 273 (defun acons-namespaces (element &optional (bindings *namespaces*))
 274   (map-namespace-declarations (lambda (prefix uri)
 275                                 (push (cons prefix uri) bindings))
 276                               element)
 277   bindings)
 278
 279 (defun find-key (name stylesheet)
 280   (or (gethash name (stylesheet-keys stylesheet))
 281       (xslt-error "unknown key: ~a" name)))
 282
 283 (defun make-key (match use) (cons match use))
 284
 285 (defun key-match (key) (car key))
 286
 287 (defun key-use (key) (cdr key))
 288
 289 (defun add-key (stylesheet name match use)
 290   (if (gethash name (stylesheet-keys stylesheet))
 291       (xslt-error "duplicate key: ~a" name)
 292       (setf (gethash name (stylesheet-keys stylesheet))
 293             (make-key match use))))
 294
 295 (defvar *excluded-namespaces* (list *xsl*))
 296 (defvar *empty-mode*)
 297
 298 (defvar *xsl-include-stack* nil)
 299
 300 (defun uri-to-pathname (uri)
 301   (cxml::uri-to-pathname (puri:parse-uri uri)))
 302
 303 (defun parse-stylesheet-to-stp (input uri-resolver)
 304   (let* ((d (cxml:parse input (make-text-normalizer (cxml-stp:make-builder))))
 305          (<transform> (stp:document-element d)))
 306     (strip-stylesheet <transform>)
 307     ;; FIXME: handle embedded stylesheets
 308     (unless (and (equal (stp:namespace-uri <transform>) *xsl*)
 309                  (or (equal (stp:local-name <transform>) "transform")
 310                      (equal (stp:local-name <transform>) "stylesheet")))
 311       (xslt-error "not a stylesheet"))
 312     (dolist (include (stp:filter-children (of-name "include") <transform>))
 313       (let* ((uri (puri:merge-uris (stp:attribute-value include "href")
 314                                    (stp:base-uri include)))
 315              (uri (if uri-resolver
 316                       (funcall uri-resolver (puri:render-uri uri nil))
 317                       uri))
 318              (str (puri:render-uri uri nil))
 319              (pathname
 320               (handler-case
 321                   (uri-to-pathname uri)
 322                 (cxml:xml-parse-error (c)
 323                   (xslt-error "cannot find included stylesheet ~A: ~A"
 324                               uri c)))))
 325         (with-open-file
 326             (stream pathname
 327                     :element-type '(unsigned-byte 8)
 328                     :if-does-not-exist nil)
 329           (unless stream
 330             (xslt-error "cannot find included stylesheet ~A at ~A"
 331                         uri pathname))
 332           (when (find str *xsl-include-stack* :test #'equal)
 333             (xslt-error "recursive inclusion of ~A" uri))
 334           (let* ((*xsl-include-stack* (cons str *xsl-include-stack*))
 335                  (<transform>2 (parse-stylesheet-to-stp stream uri-resolver)))
 336             (stp:do-children (child <transform>2)
 337               (stp:insert-child-after <transform>
 338                                       (stp:copy child)
 339                                       include))
 340             (stp:detach include)))))
 341     <transform>))
 342
 343 (defvar *instruction-base-uri*)
 344 (defvar *apply-imports-limit*)
 345 (defvar *import-priority*)
 346
 347 (defun parse-1-stylesheet (env stylesheet designator uri-resolver)
 348   (let* ((<transform> (parse-stylesheet-to-stp designator uri-resolver))
 349          (*instruction-base-uri* (stp:base-uri <transform>))
 350          (*namespaces* (acons-namespaces <transform>))
 351          (*apply-imports-limit* (1+ *import-priority*)))
 352     (dolist (import (stp:filter-children (of-name "import") <transform>))
 353       (let ((uri (puri:merge-uris (stp:attribute-value import "href")
 354                                   (stp:base-uri import))))
 355         (parse-imported-stylesheet env stylesheet uri uri-resolver)))
 356     (incf *import-priority*)
 357     (parse-exclude-result-prefixes! <transform> env)
 358     (parse-global-variables! stylesheet <transform>)
 359     (parse-keys! stylesheet <transform> env)
 360     (parse-templates! stylesheet <transform> env)
 361     (parse-output! stylesheet <transform>)
 362     (parse-strip/preserve-space! stylesheet <transform> env)
 363     (parse-attribute-sets! stylesheet <transform> env)))
 364
 365 (defvar *xsl-import-stack* nil)
 366
 367 (defun parse-imported-stylesheet (env stylesheet uri uri-resolver)
 368   (let* ((uri (if uri-resolver
 369                   (funcall uri-resolver (puri:render-uri uri nil))
 370                   uri))
 371          (str (puri:render-uri uri nil))
 372          (pathname
 373           (handler-case
 374               (uri-to-pathname uri)
 375             (cxml:xml-parse-error (c)
 376               (xslt-error "cannot find imported stylesheet ~A: ~A"
 377                           uri c)))))
 378     (with-open-file
 379         (stream pathname
 380                 :element-type '(unsigned-byte 8)
 381                 :if-does-not-exist nil)
 382       (unless stream
 383         (xslt-error "cannot find imported stylesheet ~A at ~A"
 384                     uri pathname))
 385       (when (find str *xsl-import-stack* :test #'equal)
 386         (xslt-error "recursive inclusion of ~A" uri))
 387       (let ((*xsl-import-stack* (cons str *xsl-import-stack*)))
 388         (parse-1-stylesheet env stylesheet stream uri-resolver)))))
 389
 390 (defun parse-stylesheet (designator &key uri-resolver)
 391   (let* ((*import-priority* 0)
 392          (puri:*strict-parse* nil)
 393          (stylesheet (make-stylesheet))
 394          (env (make-instance 'lexical-xslt-environment))
 395          (*excluded-namespaces* *excluded-namespaces*)
 396          (*global-variable-declarations* (make-empty-declaration-array)))
 397     (ensure-mode stylesheet nil)
 398     (parse-1-stylesheet env stylesheet designator uri-resolver)
 399     ;; reverse attribute sets:
 400     (let ((table (stylesheet-attribute-sets stylesheet)))
 401       (maphash (lambda (k v)
 402                  (setf (gethash k table) (nreverse v)))
 403                table))
 404     stylesheet))
 405
 406 (defun parse-attribute-sets! (stylesheet <transform> env)
 407   (dolist (elt (stp:filter-children (of-name "attribute-set") <transform>))
 408     (push (let* ((sets
 409                   (mapcar (lambda (qname)
 410                             (multiple-value-list (decode-qname qname env nil)))
 411                           (words
 412                            (stp:attribute-value elt "use-attribute-sets"))))
 413                  (instructions
 414                   (stp:map-children 'list #'parse-instruction elt))
 415                  (*lexical-variable-declarations*
 416                   (make-empty-declaration-array))
 417                  (thunk
 418                   (compile-instruction `(progn ,@instructions) env))
 419                  (n-variables (length *lexical-variable-declarations*)))
 420             (lambda (ctx)
 421               (with-stack-limit ()
 422                 (loop for (local-name uri nil) in sets do
 423                      (dolist (thunk (find-attribute-set local-name uri))
 424                        (funcall thunk ctx)))
 425                 (let ((*lexical-variable-values*
 426                        (make-variable-value-array n-variables)))
 427                   (funcall thunk ctx)))))
 428           (gethash (multiple-value-bind (local-name uri)
 429                        (decode-qname (stp:attribute-value elt "name") env nil)
 430                      (cons local-name uri))
 431                    (stylesheet-attribute-sets stylesheet)))))
 432
 433 (defun parse-exclude-result-prefixes! (<transform> env)
 434   (stp:with-attributes (exclude-result-prefixes) <transform>
 435       (dolist (prefix (words (or exclude-result-prefixes "")))
 436     (when (equal prefix "#default")
 437       (setf prefix nil))
 438     (push (or (xpath-sys:environment-find-namespace env prefix)
 439               (xslt-error "namespace not found: ~A" prefix))
 440           *excluded-namespaces*))))
 441
 442 (defun parse-strip/preserve-space! (stylesheet <transform> env)
 443   (xpath:with-namespaces ((nil #.*xsl*))
 444     (dolist (elt (stp:filter-children (lambda (x)
 445                                         (or (namep x "strip-space")
 446                                             (namep x "preserve-space")))
 447                                       <transform>))
 448       (let ((*namespaces* (acons-namespaces elt))
 449             (mode
 450              (if (equal (stp:local-name elt) "strip-space")
 451                  :strip
 452                  :preserve)))
 453         (dolist (name-test (words (stp:attribute-value elt "elements")))
 454           (let* ((pos (search ":*" name-test))
 455                  (test-function
 456                   (cond
 457                     ((eql pos (- (length name-test) 2))
 458                      (let* ((prefix (subseq name-test 0 pos))
 459                             (name-test-uri
 460                              (xpath-sys:environment-find-namespace env prefix)))
 461                        (unless (xpath::nc-name-p prefix)
 462                          (xslt-error "not an NCName: ~A" prefix))
 463                        (lambda (local-name uri)
 464                          (declare (ignore local-name))
 465                          (if (equal uri name-test-uri)
 466                              mode
 467                              nil))))
 468                     ((equal name-test "*")
 469                      (lambda (local-name uri)
 470                        (declare (ignore local-name uri))
 471                        mode))
 472                     (t
 473                      (multiple-value-bind (name-test-local-name name-test-uri)
 474                          (decode-qname name-test env nil)
 475                        (lambda (local-name uri)
 476                          (if (and (equal local-name name-test-local-name)
 477                                   (equal uri name-test-uri))
 478                              mode
 479                              nil)))))))
 480             (push test-function (stylesheet-strip-tests stylesheet))))))))
 481
 482 (defstruct (output-specification
 483              (:conc-name "OUTPUT-"))
 484   method
 485   indent
 486   omit-xml-declaration
 487   encoding)
 488
 489 (defun parse-output! (stylesheet <transform>)
 490   (let ((outputs (stp:filter-children (of-name "output") <transform>)))
 491     (when outputs
 492       (when (cdr outputs)
 493         ;; FIXME:
 494         ;;   - concatenate cdata-section-elements
 495         ;;   - the others must not conflict
 496         (error "oops, merging of output elements not supported yet"))
 497       (let ((<output> (car outputs))
 498             (spec (stylesheet-output-specification stylesheet)))
 499         (stp:with-attributes (;; version
 500                               method
 501                               indent
 502                               encoding
 503 ;;;                           media-type
 504 ;;;                           doctype-system
 505 ;;;                           doctype-public
 506                               omit-xml-declaration
 507 ;;;                           standalone
 508 ;;;                           cdata-section-elements
 509                               )
 510             <output>
 511           (setf (output-method spec) method)
 512           (setf (output-indent spec) indent)
 513           (setf (output-encoding spec) encoding)
 514           (setf (output-omit-xml-declaration spec) omit-xml-declaration))))))
 515
 516 (defun make-empty-declaration-array ()
 517   (make-array 1 :fill-pointer 0 :adjustable t))
 518
 519 (defun make-variable-value-array (n-lexical-variables)
 520   (make-array n-lexical-variables :initial-element 'unbound))
 521
 522 (defun compile-global-variable (<variable> env) ;; also for <param>
 523   (stp:with-attributes (name select) <variable>
 524     (when (and select (stp:list-children <variable>))
 525       (xslt-error "variable with select and body"))
 526     (let* ((*lexical-variable-declarations* (make-empty-declaration-array))
 527            (inner (cond
 528                     (select
 529                      (compile-xpath select env))
 530                     ((stp:list-children <variable>)
 531                      (let* ((inner-sexpr `(progn ,@(parse-body <variable>)))
 532                             (inner-thunk (compile-instruction inner-sexpr env)))
 533                        (lambda (ctx)
 534                          (apply-to-result-tree-fragment ctx inner-thunk))))
 535                     (t
 536                      (lambda (ctx)
 537                        (declare (ignore ctx))
 538                        ""))))
 539            (n-lexical-variables (length *lexical-variable-declarations*)))
 540       (xslt-trace-thunk
 541        (lambda (ctx)
 542          (let* ((*lexical-variable-values*
 543                  (make-variable-value-array n-lexical-variables)))
 544            (funcall inner ctx)))
 545        "global ~s (~s) = ~s" name select :result))))
 546
 547 (defstruct (variable-information
 548              (:constructor make-variable)
 549              (:conc-name "VARIABLE-"))
 550   index
 551   thunk
 552   local-name
 553   uri
 554   param-p
 555   thunk-setter)
 556
 557 (defun parse-global-variable! (<variable> global-env) ;; also for <param>
 558   (let ((*namespaces* (acons-namespaces <variable>))
 559         (qname (stp:attribute-value <variable> "name")))
 560     (unless qname
 561       (xslt-error "name missing in ~A" (stp:local-name <variable>)))
 562     (multiple-value-bind (local-name uri)
 563         (decode-qname qname global-env nil)
 564       ;; For the normal compilation environment of templates, install it
 565       ;; into *GLOBAL-VARIABLE-DECLARATIONS*:
 566       (let ((index (intern-global-variable local-name uri)))
 567         ;; For the evaluation of a global variable itself, build a thunk
 568         ;; that lazily resolves other variables, stored into
 569         ;; INITIAL-GLOBAL-VARIABLE-THUNKS:
 570         (let* ((value-thunk :unknown)
 571                (global-variable-thunk
 572                 (lambda (ctx)
 573                   (let ((v (global-variable-value index nil)))
 574                     (when (eq v 'seen)
 575                       (xslt-error "recursive variable definition"))
 576                     (cond
 577                       ((eq v 'unbound)
 578                        ;; (print (list :computing index))
 579                        (setf (global-variable-value index) 'seen)
 580                        (setf (global-variable-value index)
 581                              (funcall value-thunk ctx))
 582                        #+nil (print (list :done-computing index
 583                                     (global-variable-value index)))
 584                        #+nil (global-variable-value index))
 585                       (t
 586                        #+nil(print (list :have
 587                                     index v))
 588                        v)))))
 589                (thunk-setter
 590                 (lambda ()
 591                   (setf value-thunk
 592                         (compile-global-variable <variable> global-env)))))
 593           (setf (gethash (cons local-name uri)
 594                          (initial-global-variable-thunks global-env))
 595                 global-variable-thunk)
 596           (make-variable :index index
 597                          :local-name local-name
 598                          :uri uri
 599                          :thunk global-variable-thunk
 600                          :param-p (namep <variable> "param")
 601                          :thunk-setter thunk-setter))))))
 602
 603 (defun parse-keys! (stylesheet <transform> env)
 604   (xpath:with-namespaces ((nil #.*xsl*))
 605     (xpath:do-node-set
 606         (<key> (xpath:evaluate "key" <transform>))
 607       (stp:with-attributes (name match use) <key>
 608         (unless name (xslt-error "key name attribute not specified"))
 609         (unless match (xslt-error "key match attribute not specified"))
 610         (unless use (xslt-error "key use attribute not specified"))
 611         (add-key stylesheet name
 612                  (compile-xpath `(xpath:xpath ,(parse-key-pattern match)) env)
 613                  (compile-xpath use env))))))
 614
 615 (defun parse-global-variables! (stylesheet <transform>)
 616   (xpath:with-namespaces ((nil #.*xsl*))
 617     (let* ((table (make-hash-table :test 'equal))
 618            (global-env (make-instance 'global-variable-environment
 619                                       :initial-global-variable-thunks table))
 620            (specs '()))
 621       (xpath:do-node-set
 622           (<variable> (xpath:evaluate "variable|param" <transform>))
 623         (let ((var (parse-global-variable! <variable> global-env)))
 624           (xslt-trace "parsing global variable ~s (uri ~s)"
 625                       (variable-local-name var)
 626                       (variable-uri var))
 627           (when (find var
 628                       specs
 629                       :test (lambda (a b)
 630                               (and (equal (variable-local-name a)
 631                                           (variable-local-name b))
 632                                    (equal (variable-uri a)
 633                                           (variable-uri b)))))
 634             (xslt-error "duplicate definition for global variable ~A"
 635                         (variable-local-name var)))
 636           (push var specs)))
 637       ;; now that the global environment knows about all variables, run the
 638       ;; thunk setters to perform their compilation
 639       (setf specs (nreverse specs))
 640       (mapc (lambda (spec) (funcall (variable-thunk-setter spec))) specs)
 641       (setf (stylesheet-global-variables stylesheet) specs))))
 642
 643 (defun parse-templates! (stylesheet <transform> env)
 644   (let ((i 0))
 645     (dolist (<template> (stp:filter-children (of-name "template") <transform>))
 646       (let ((*namespaces* (acons-namespaces <template>)))
 647         (dolist (template (compile-template <template> env i))
 648           (let ((name (template-name template)))
 649             (if name
 650                 (let* ((table (stylesheet-named-templates stylesheet))
 651                        (head (car (gethash name table))))
 652                   (when (and head (eql (template-import-priority head)
 653                                        (template-import-priority template)))
 654                     ;; fixme: is this supposed to be a run-time error?
 655                     (xslt-error "conflicting templates for ~A" name))
 656                   (push template (gethash name table)))
 657                 (let ((mode (ensure-mode/qname stylesheet
 658                                                (template-mode-qname template)
 659                                                env)))
 660                   (setf (template-mode template) mode)
 661                   (push template (mode-templates mode)))))))
 662       (incf i))))
 663
 664
 665 ;;;; APPLY-STYLESHEET
 666
 667 (defvar *stylesheet*)
 668 (defvar *mode*)
 669
 670 (deftype xml-designator () '(or runes:xstream runes:rod array stream pathname))
 671
 672 (defstruct (parameter
 673              (:constructor make-parameter (value local-name &optional uri)))
 674   (uri "")
 675   local-name
 676   value)
 677
 678 (defun find-parameter-value (local-name uri parameters)
 679   (dolist (p parameters)
 680     (when (and (equal (parameter-local-name p) local-name)
 681                (equal (parameter-uri p) uri))
 682       (return (parameter-value p)))))
 683
 684 (defvar *uri-resolver*)
 685
 686 (defun parse-allowing-microsoft-bom (pathname handler)
 687   (with-open-file (s pathname :element-type '(unsigned-byte 8))
 688     (unless (and (eql (read-byte s nil) #xef)
 689                  (eql (read-byte s nil) #xbb)
 690                  (eql (read-byte s nil) #xbf))
 691       (file-position s 0))
 692     (cxml:parse s handler)))
 693
 694 (defun %document (uri-string base-uri)
 695   (let* ((absolute-uri
 696           (puri:merge-uris uri-string base-uri))
 697          (resolved-uri
 698           (if *uri-resolver*
 699               (funcall *uri-resolver* (puri:render-uri absolute-uri nil))
 700               absolute-uri))
 701          (pathname
 702           (handler-case
 703               (uri-to-pathname resolved-uri)
 704             (cxml:xml-parse-error (c)
 705               (xslt-error "cannot find referenced document ~A: ~A"
 706                           resolved-uri c))))
 707          (document
 708           (handler-case
 709               (parse-allowing-microsoft-bom pathname (stp:make-builder))
 710             ((or file-error cxml:xml-parse-error) (c)
 711               (xslt-error "cannot parse referenced document ~A: ~A"
 712                           pathname c))))
 713          (xpath-root-node
 714           (make-whitespace-stripper document
 715                                     (stylesheet-strip-tests *stylesheet*))))
 716     (when (puri:uri-fragment absolute-uri)
 717       (xslt-error "use of fragment identifiers in document() not supported"))
 718     xpath-root-node))
 719
 720 (xpath-sys:define-extension xslt *xsl*)
 721
 722 (xpath-sys:define-xpath-function/lazy
 723     xslt :document
 724     (object &optional node-set)
 725   (let ((instruction-base-uri *instruction-base-uri*))
 726     (lambda (ctx)
 727       (let* ((object (funcall object ctx))
 728              (node-set (and node-set (funcall node-set ctx)))
 729              (uri
 730               (when node-set
 731                 ;; FIXME: should use first node of the node set
 732                 ;; _in document order_
 733                 (xpath-protocol:base-uri (xpath:first-node node-set)))))
 734         (xpath-sys:make-node-set
 735          (if (xpath:node-set-p object)
 736              (xpath:map-node-set->list
 737               (lambda (node)
 738                 (%document (xpath:string-value node)
 739                            (or uri (xpath-protocol:base-uri node))))
 740               object)
 741              (list (%document (xpath:string-value object)
 742                               (or uri instruction-base-uri)))))))))
 743
 744 (xpath-sys:define-xpath-function/eager xslt :key (name object)
 745   (let ((key (find-key (xpath:string-value name) *stylesheet*)))
 746     (labels ((get-by-key (value)
 747                (let ((value (xpath:string-value value)))
 748                  (xpath::filter-pipe
 749                   #'(lambda (node)
 750                       (equal value (xpath:string-value
 751                                     (xpath:evaluate-compiled
 752                                      (key-use key) node))))
 753                   (xpath-sys:pipe-of
 754                    (xpath:node-set-value
 755                     (xpath:evaluate-compiled
 756                      (key-match key) xpath:context)))))))
 757       (xpath-sys:make-node-set
 758        (xpath::sort-pipe
 759         (if (xpath:node-set-p object)
 760             (xpath::mappend-pipe #'get-by-key (xpath-sys:pipe-of object))
 761             (get-by-key object)))))))
 762
 763 ;; FIXME: add alias mechanism for XPath extensions in order to avoid duplication
 764
 765 (xpath-sys:define-xpath-function/lazy xslt :current ()
 766   #'(lambda (ctx)
 767       (xpath-sys:make-node-set
 768        (xpath-sys:make-pipe
 769         (xpath:context-starting-node ctx)
 770         nil))))
 771
 772 (xpath-sys:define-xpath-function/lazy xslt :generate-id (&optional node-set-thunk)
 773   (if node-set-thunk
 774       #'(lambda (ctx)
 775           (xpath-sys:get-node-id (xpath:node-set-value (funcall node-set-thunk ctx))))
 776       #'(lambda (ctx)
 777           (xpath-sys:get-node-id (xpath:context-node ctx)))))
 778
 779 (defun apply-stylesheet
 780     (stylesheet source-document
 781      &key output parameters uri-resolver navigator)
 782   (when (typep stylesheet 'xml-designator)
 783     (setf stylesheet (parse-stylesheet stylesheet)))
 784   (when (typep source-document 'xml-designator)
 785     (setf source-document (cxml:parse source-document (stp:make-builder))))
 786   (invoke-with-output-sink
 787    (lambda ()
 788      (handler-case*
 789          (let* ((xpath:*navigator* (or navigator :default-navigator))
 790                 (puri:*strict-parse* nil)
 791                 (*stylesheet* stylesheet)
 792                 (*mode* (find-mode stylesheet nil))
 793                 (*empty-mode* (make-mode))
 794                 (global-variable-specs
 795                  (stylesheet-global-variables stylesheet))
 796                 (*global-variable-values*
 797                  (make-variable-value-array (length global-variable-specs)))
 798                 (*uri-resolver* uri-resolver)
 799                 (xpath-root-node
 800                  (make-whitespace-stripper
 801                   source-document
 802                   (stylesheet-strip-tests stylesheet)))
 803                 (ctx (xpath:make-context xpath-root-node)))
 804            (mapc (lambda (spec)
 805                    (when (variable-param-p spec)
 806                      (let ((value
 807                             (find-parameter-value (variable-local-name spec)
 808                                                   (variable-uri spec)
 809                                                   parameters)))
 810                        (when value
 811                          (setf (global-variable-value (variable-index spec))
 812                                value)))))
 813                  global-variable-specs)
 814            (mapc (lambda (spec)
 815                    (funcall (variable-thunk spec) ctx))
 816                  global-variable-specs)
 817            #+nil (print global-variable-specs)
 818            #+nil (print *global-variable-values*)
 819            (apply-templates ctx))
 820        (xpath:xpath-error (c)
 821                           (xslt-error "~A" c))))
 822    stylesheet
 823    output))
 824
 825 (defun find-attribute-set (local-name uri)
 826   (or (gethash (cons local-name uri) (stylesheet-attribute-sets *stylesheet*))
 827       (xslt-error "no such attribute set: ~A/~A" local-name uri)))
 828
 829 (defun apply-templates/list (list &optional param-bindings sort-predicate)
 830   (when sort-predicate
 831     (setf list (sort list sort-predicate)))
 832   (let* ((n (length list))
 833          (s/d (lambda () n)))
 834     (loop
 835        for i from 1
 836        for child in list
 837        do
 838          (apply-templates (xpath:make-context child s/d i)
 839            param-bindings))))
 840
 841 (defvar *stack-limit* 200)
 842
 843 (defun invoke-with-stack-limit (fn)
 844   (let ((*stack-limit* (1- *stack-limit*)))
 845     (unless (plusp *stack-limit*)
 846       (xslt-error "*stack-limit* reached; stack overflow"))
 847     (funcall fn)))
 848
 849 (defun invoke-template (ctx template param-bindings)
 850   (let ((*lexical-variable-values*
 851          (make-variable-value-array (template-n-variables template))))
 852     (with-stack-limit ()
 853       (loop
 854          for (name-cons value) in param-bindings
 855          for (nil index nil) = (find name-cons
 856                                      (template-params template)
 857                                      :test #'equal
 858                                      :key #'car)
 859          do
 860          (unless index
 861            (xslt-error "invalid template parameter ~A" name-cons))
 862          (setf (lexical-variable-value index) value))
 863       (funcall (template-body template) ctx))))
 864
 865 (defun apply-default-templates (ctx)
 866   (let ((node (xpath:context-node ctx)))
 867     (cond
 868       ((or (xpath-protocol:node-type-p node :processing-instruction)
 869            (xpath-protocol:node-type-p node :comment)))
 870       ((or (xpath-protocol:node-type-p node :text)
 871            (xpath-protocol:node-type-p node :attribute))
 872        (write-text (xpath-protocol:node-text node)))
 873       (t
 874        (apply-templates/list
 875         (xpath::force
 876          (xpath-protocol:child-pipe node)))))))
 877
 878 (defvar *apply-imports*)
 879
 880 (defun apply-applicable-templates (ctx templates param-bindings finally)
 881   (labels ((apply-imports ()
 882              (if templates
 883                  (let* ((this (pop templates))
 884                         (low (template-apply-imports-limit this))
 885                         (high (template-import-priority this)))
 886                    (setf templates
 887                          (remove-if-not
 888                           (lambda (x)
 889                             (<= low (template-import-priority x) high))
 890                           templates))
 891                    (invoke-template ctx this param-bindings))
 892                  (funcall finally))))
 893     (let ((*apply-imports* #'apply-imports))
 894       (apply-imports))))
 895
 896 (defun apply-templates (ctx &optional param-bindings)
 897   (apply-applicable-templates ctx
 898                               (find-templates ctx)
 899                               param-bindings
 900                               (lambda ()
 901                                 (apply-default-templates ctx))))
 902
 903 (defun call-template (ctx name &optional param-bindings)
 904   (apply-applicable-templates ctx
 905                               (find-named-templates name)
 906                               param-bindings
 907                               (lambda ()
 908                                 (error "cannot find named template: ~s"
 909                                        name))))
 910
 911 (defun find-templates (ctx)
 912   (let* ((matching-candidates
 913           (remove-if-not (lambda (template)
 914                            (template-matches-p template ctx))
 915                          (mode-templates *mode*)))
 916          (npriorities
 917           (if matching-candidates
 918               (1+ (reduce #'max
 919                           matching-candidates
 920                           :key #'template-import-priority))
 921               0))
 922          (priority-groups (make-array npriorities :initial-element nil)))
 923     (dolist (template matching-candidates)
 924       (push template
 925             (elt priority-groups (template-import-priority template))))
 926 ;;;     (print (map 'list #'length priority-groups))
 927 ;;;     (force-output)
 928     (loop
 929        for i from (1- npriorities) downto 0
 930        for group = (elt priority-groups i)
 931        for template = (maximize #'template< group)
 932        when template
 933        collect template)))
 934
 935 (defun find-named-templates (name)
 936   (gethash name (stylesheet-named-templates *stylesheet*)))
 937
 938 (defun template< (a b)                  ;assuming same import priority
 939   (let ((p (template-priority a))
 940         (q (template-priority b)))
 941     (cond
 942       ((< p q) t)
 943       ((> p q) nil)
 944       (t
 945        (xslt-cerror "conflicting templates:~_~A,~_~A"
 946                     (template-match-expression a)
 947                     (template-match-expression b))
 948        (< (template-position a) (template-position b))))))
 949
 950 (defun maximize (< things)
 951   (when things
 952     (let ((max (car things)))
 953       (dolist (other (cdr things))
 954         (when (funcall < max other)
 955           (setf max other)))
 956       max)))
 957
 958 (defun template-matches-p (template ctx)
 959   (find (xpath:context-node ctx)
 960         (xpath:all-nodes (funcall (template-match-thunk template) ctx))))
 961
 962 (defun invoke-with-output-sink (fn stylesheet output)
 963   (etypecase output
 964     (pathname
 965      (with-open-file (s output
 966                         :direction :output
 967                         :element-type '(unsigned-byte 8)
 968                         :if-exists :rename-and-delete)
 969        (invoke-with-output-sink fn stylesheet s)))
 970     ((or stream null)
 971      (invoke-with-output-sink fn
 972                               stylesheet
 973                               (make-output-sink stylesheet output)))
 974     ((or hax:abstract-handler sax:abstract-handler)
 975      (with-xml-output output
 976        (funcall fn)))))
 977
 978 (defun make-output-sink (stylesheet stream)
 979   (let* ((ystream
 980           (if stream
 981               (let ((et (stream-element-type stream)))
 982                 (cond
 983                   ((or (null et) (subtypep et '(unsigned-byte 8)))
 984                    (runes:make-octet-stream-ystream stream))
 985                   ((subtypep et 'character)
 986                    (runes:make-character-stream-ystream stream))))
 987               (runes:make-rod-ystream)))
 988          (output-spec (stylesheet-output-specification stylesheet))
 989          (omit-xml-declaration-p
 990           (equal (output-omit-xml-declaration output-spec) "yes"))
 991          (sax-target
 992           (make-instance 'cxml::sink
 993                          :ystream ystream
 994                          :omit-xml-declaration-p omit-xml-declaration-p)))
 995     (if (equalp (output-method (stylesheet-output-specification stylesheet))
 996                 "HTML")
 997         (make-instance 'combi-sink
 998                        :hax-target (make-instance 'chtml::sink
 999                                                   :ystream ystream)
1000                        :sax-target sax-target
1001                        :encoding (output-encoding output-spec))
1002         sax-target)))
1003
1004 (defstruct template
1005   match-expression
1006   match-thunk
1007   name
1008   import-priority
1009   apply-imports-limit
1010   priority
1011   position
1012   mode
1013   mode-qname
1014   params
1015   body
1016   n-variables)
1017
1018 (defun expression-priority (form)
1019   (let ((step (second form)))
1020     (if (and (null (cddr form))
1021              (listp step)
1022              (eq :child (car step))
1023              (null (cddr step)))
1024         (let ((name (second step)))
1025           (cond
1026             ((or (stringp name)
1027                  (and (consp name)
1028                       (or (eq (car name) :qname)
1029                           (eq (car name) :processing-instruction))))
1030              0.0)
1031             ((and (consp name)
1032                   (or (eq (car name) :namespace)
1033                       (eq (car name) '*)))
1034              -0.25)
1035             (t
1036              -0.5)))
1037         0.5)))
1038
1039 (defun valid-expression-p (expr)
1040   (cond
1041     ((atom expr) t)
1042     ((eq (first expr) :path)
1043      (every (lambda (x)
1044               (let ((filter (third x)))
1045                 (or (null filter) (valid-expression-p filter))))
1046             (cdr expr)))
1047     ((eq (first expr) :variable)        ;(!)
1048      nil)
1049     (t
1050      (every #'valid-expression-p (cdr expr)))))
1051
1052 (defun parse-xpath (str)
1053   (handler-case
1054       (xpath:parse-xpath str)
1055     (xpath:xpath-error (c)
1056       (xslt-error "~A" c))))
1057
1058 (defun parse-key-pattern (str)
1059   (let ((parsed
1060          (mapcar #'(lambda (item)
1061                      `(:path (:root :node)
1062                              (:descendant-or-self *)
1063                              ,@(cdr item)))
1064                  (parse-pattern str))))
1065     (if (null (rest parsed))
1066         (first parsed)
1067         `(:union ,@parsed))))
1068
1069 (defun parse-pattern (str)
1070   ;; zzz check here for anything not allowed as an XSLT pattern
1071   ;; zzz can we hack id() and key() here?
1072   (let ((form (parse-xpath str)))
1073     (unless (consp form)
1074       (xslt-error "not a valid pattern: ~A" str))
1075     (labels ((process-form (form)
1076                (cond ((eq (car form) :union)
1077                       (alexandria:mappend #'process-form (rest form)))
1078                      ((not (eq (car form) :path)) ;zzz: filter statt path
1079                       (xslt-error "not a valid pattern: ~A" str))
1080                      ((not (valid-expression-p form))
1081                       (xslt-error "invalid filter"))
1082                      (t (list form)))))
1083       (process-form form))))
1084
1085 (defun compile-value-thunk (value env)
1086   (if (and (listp value) (eq (car value) 'progn))
1087       (let ((inner-thunk (compile-instruction value env)))
1088         (lambda (ctx)
1089           (apply-to-result-tree-fragment ctx inner-thunk)))
1090       (compile-xpath value env)))
1091
1092 (defun compile-var-bindings/nointern (forms env)
1093   (loop
1094     for (name value) in forms
1095     collect (multiple-value-bind (local-name uri)
1096                 (decode-qname name env nil)
1097               (list (cons local-name uri)
1098                     (xslt-trace-thunk
1099                      (compile-value-thunk value env)
1100                      "local variable ~s = ~s" name :result)))))
1101
1102 (defun compile-var-bindings (forms env)
1103   (loop
1104      for (cons thunk) in (compile-var-bindings/nointern forms env)
1105      for (local-name . uri) = cons
1106      collect (list cons
1107                    (push-variable local-name
1108                                   uri
1109                                   *lexical-variable-declarations*)
1110                    thunk)))
1111
1112 (defun compile-template (<template> env position)
1113   (stp:with-attributes (match name priority mode) <template>
1114     (unless (or name match)
1115       (xslt-error "missing match in template"))
1116     (multiple-value-bind (params body-pos)
1117         (loop
1118           for i from 0
1119           for child in (stp:list-children <template>)
1120           while (namep child "param")
1121           collect (parse-param child) into params
1122           finally (return (values params i)))
1123       (let* ((*lexical-variable-declarations* (make-empty-declaration-array))
1124              (param-bindings (compile-var-bindings params env))
1125              (body (parse-body <template> body-pos (mapcar #'car params)))
1126              (body-thunk (compile-instruction `(progn ,@body) env))
1127              (outer-body-thunk
1128               (xslt-trace-thunk
1129                #'(lambda (ctx)
1130                    (unwind-protect
1131                        (progn
1132                          ;; set params that weren't initialized by apply-templates
1133                          (loop for (name index param-thunk) in param-bindings
1134                                when (eq (lexical-variable-value index nil) 'unbound)
1135                                  do (setf (lexical-variable-value index)
1136                                           (funcall param-thunk ctx)))
1137                          (funcall body-thunk ctx))))
1138                "template: match = ~s name = ~s" match name))
1139              (n-variables (length *lexical-variable-declarations*)))
1140         (append
1141          (when name
1142            (multiple-value-bind (local-name uri)
1143                (decode-qname name env nil)
1144              (list
1145               (make-template :name (cons local-name uri)
1146                              :import-priority *import-priority*
1147                              :apply-imports-limit *apply-imports-limit*
1148                              :params param-bindings
1149                              :body outer-body-thunk
1150                              :n-variables n-variables))))
1151          (when match
1152            (mapcar (lambda (expression)
1153                      (let ((match-thunk
1154                             (xslt-trace-thunk
1155                              (compile-xpath
1156                               `(xpath:xpath
1157                                 (:path (:ancestor-or-self :node)
1158                                        ,@(cdr expression)))
1159                               env)
1160                              "match-thunk for template (match ~s): ~s --> ~s"
1161                              match expression :result))
1162                            (p (if priority
1163                                   (parse-number:parse-number priority)
1164                                   (expression-priority expression))))
1165                        (make-template :match-expression expression
1166                                       :match-thunk match-thunk
1167                                       :import-priority *import-priority*
1168                                       :apply-imports-limit *apply-imports-limit*
1169                                       :priority p
1170                                       :position position
1171                                       :mode-qname mode
1172                                       :params param-bindings
1173                                       :body outer-body-thunk
1174                                       :n-variables n-variables)))
1175                    (parse-pattern match))))))))
1176 #+(or)
1177 (xuriella::parse-stylesheet #p"/home/david/src/lisp/xuriella/test.xsl")