test.lisp

   1 ;;; -*- show-trailing-whitespace: t; indent-tabs-mode: 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 (defparameter *tests-directory*
  33   "/home/david/src/XSLT-testsuite-04/testsuite/TESTS/")
  34
  35 (defclass test-case ()
  36   ((id :initarg :id :accessor test-id)
  37    (category :initarg :category :accessor test-category)
  38    (operation :initarg :operation :accessor test-operation)
  39    (data-pathname :initarg :data-pathname :accessor test-data-pathname)
  40    (stylesheet-pathname :initarg :stylesheet-pathname
  41                         :accessor test-stylesheet-pathname)
  42    (data-pathname-2 :initarg :data-pathname-2 :accessor test-data-pathname-2)
  43    (stylesheet-pathname-2 :initarg :stylesheet-pathname-2
  44                           :accessor test-stylesheet-pathname-2)
  45    (output-pathname :initarg :output-pathname
  46                     :accessor test-official-output-pathname)
  47    (output-compare :initarg :output-compare
  48                    :accessor test-output-compare)))
  49
  50 (defmethod print-object ((object test-case) stream)
  51   (print-unreadable-object (object stream :identity nil :type t)
  52     (format stream "~A ~A/~A"
  53             (test-operation object)
  54             (test-category object)
  55             (test-id object))))
  56
  57
  58 ;;;; SIMPLIFY-TESTS
  59
  60 ;;; Translate catalog.xml into an actually usable katalog.xml
  61 ;;; by running the test cases through xsltproc to see what it thinks
  62 ;;; about them.
  63
  64 (defun simplify-tests (&optional (d *tests-directory*))
  65   (with-open-file (stream (merge-pathnames "katalog.xml" d)
  66                           :direction :output
  67                           :if-exists :supersede
  68                           :element-type '(unsigned-byte 8))
  69     (cxml:with-xml-output (cxml:make-octet-stream-sink stream)
  70       (cxml:with-element "simplified-test-suite"
  71         (klacks:with-open-source
  72             (source (klacks:make-tapping-source
  73                      (cxml:make-source (merge-pathnames "catalog.xml" d))))
  74           (let ((*default-pathname-defaults* (merge-pathnames d)))
  75             (map-original-tests #'simplify-test source)))))))
  76
  77 (defun map-original-tests (run-test source &key (test (constantly t)))
  78   (let ((total 0)
  79         (pass 0)
  80         major-path)
  81     (loop
  82        while (klacks:find-event source :start-element)
  83        for lname = (klacks:current-lname source)
  84        do
  85        (cond
  86          ((equal lname "major-path")
  87           (klacks:skip source :start-element)
  88           (setf major-path
  89                 (namestring
  90                  (merge-pathnames (klacks:consume-characters source)))))
  91          ((equal lname "test-case")
  92           (let* ((<test-case>
  93                   (stp:document-element
  94                    (klacks:serialize-element source (stp:make-builder))))
  95                  (test-case (parse-original-test major-path <test-case>)))
  96             (when (funcall test test-case)
  97               (incf total)
  98               (when (funcall run-test test-case)
  99                 (incf pass)))))
 100          (t
 101           (klacks:skip source :start-element))))
 102     (format t "~&Passed ~D/~D tests.~%" pass total)))
 103
 104 (defun parse-original-test (major-path <test-case>)
 105   (let* ((file-path
 106           (stp:string-value
 107            (stp:find-recursively-if (stp:of-name "file-path") <test-case>)))
 108          (base (concatenate 'string major-path "/" file-path))
 109          (out-base (concatenate 'string major-path "/REF_OUT/" file-path))
 110          (scenario
 111           (stp:find-recursively-if (stp:of-name "scenario") <test-case>))
 112          data
 113          stylesheet
 114          supplemental-stylesheet
 115          supplemental-data
 116          output
 117          compare)
 118     (dolist (<input> (stp:filter-recursively (stp:of-name "input-file")
 119                                              <test-case>))
 120       (let ((role (stp:attribute-value <input> "role"))
 121             (path (concatenate 'string base "/" (stp:string-value <input>))))
 122         (cond
 123           ((equal role "principal-data")
 124            (setf data path))
 125           ((equal role "principal-stylesheet")
 126            (setf stylesheet path))
 127           ((equal role "supplemental-stylesheet")
 128            (setf supplemental-stylesheet path))
 129           ((equal role "supplemental-data")
 130            (setf supplemental-data path))
 131           (t
 132            (error "unrecognized role: ~A" role)))))
 133     (dolist (<output> (stp:filter-recursively (stp:of-name "output-file")
 134                                             <test-case>))
 135       (let ((role (stp:attribute-value <output> "role"))
 136             (path (concatenate 'string out-base
 137                                "/"
 138                                (stp:string-value <output>))))
 139         (cond
 140           ((equal role "principal")
 141            (setf output path)
 142            (setf compare (stp:attribute-value <output> "compare")))
 143           (t
 144            (error "unrecognized role: ~A" role)))))
 145     (make-instance 'test-case
 146                    :id (stp:attribute-value <test-case> "id")
 147                    :category (stp:attribute-value <test-case> "category")
 148                    :operation (stp:attribute-value scenario "operation")
 149                    :data-pathname data
 150                    :stylesheet-pathname stylesheet
 151                    :stylesheet-pathname-2 supplemental-stylesheet
 152                    :data-pathname-2 supplemental-data
 153                    :output-pathname output
 154                    :output-compare compare)))
 155
 156 (defun write-simplified-test (test-case operation)
 157   (cxml:with-element "test-case"
 158     (cxml:attribute "id" (test-id test-case))
 159     (cxml:attribute "category" (test-category test-case))
 160     (flet ((p (l p)
 161              (cxml:attribute l (and p (namestring p)))))
 162       (p "data" (test-data-pathname test-case))
 163       (p "stylesheet" (noindent-stylesheet-pathname test-case))
 164       (p "data-2" (test-data-pathname-2 test-case))
 165       (p "stylesheet-2" (test-stylesheet-pathname-2 test-case))
 166       (p "output" (test-official-output-pathname test-case))
 167       (p "compare" (test-output-compare test-case)))
 168     (cxml:attribute "operation" operation)))
 169
 170 (defun test-output-pathname (test type)
 171   (make-pathname :name (test-id test)
 172                  :type type
 173                  :defaults (test-data-pathname test)))
 174
 175 (defun sanitize-stylesheet (in out)
 176   (if (probe-file in)
 177       (handler-case
 178           (let ((d (cxml:parse (pathname in) (stp:make-builder))))
 179             (xpath:with-namespaces ((nil #.*xsl*))
 180               (xpath:do-node-set (output (xpath:evaluate "//output" d))
 181                 (let ((a (stp:find-attribute-named output "indent")))
 182                   (when a
 183                     (stp:detach a)))))
 184             (with-open-file (s out
 185                                :direction :output
 186                                :if-exists :rename-and-delete
 187                                :element-type '(unsigned-byte 8))
 188               (stp:serialize d (cxml:make-octet-stream-sink s))))
 189         (error (c)
 190           (warn "ignoring bogus stylesheet ~A: ~A" in c)
 191           (copy-file in out)))
 192       (warn "oops, ignoring missing stylesheet: ~A" in)))
 193
 194 (defun noindent-stylesheet-pathname (test-case)
 195   (make-pathname :type "noindent-xsl"
 196                  :defaults (test-stylesheet-pathname test-case)))
 197
 198 (defun simplify-test (test-case)
 199   (flet ((report (status &optional (fmt "") &rest args)
 200            (format t "~&~A ~A [~A]~?~%"
 201                    status
 202                    (test-id test-case)
 203                    (test-category test-case)
 204                    fmt
 205                    args)))
 206     (let* ((data (test-data-pathname test-case))
 207            (stylesheet (test-stylesheet-pathname test-case))
 208            (noindent-stylesheet (noindent-stylesheet-pathname test-case))
 209            #+xuriella::xsltproc
 210            (out (test-output-pathname test-case "xsltproc"))
 211            (saxon-out (test-output-pathname test-case "saxon")))
 212       (sanitize-stylesheet stylesheet noindent-stylesheet)
 213       (if (equal (test-operation test-case) "standard")
 214           (handler-case
 215               (progn
 216                 #+xuriella::xsltproc (xsltproc noindent-stylesheet data out)
 217                 (saxon noindent-stylesheet data saxon-out)
 218                 (report "PASS")
 219                 (write-simplified-test test-case "standard")
 220                 t)
 221             (error (c)
 222               (report "FAIL" ": ~A" c)
 223               (write-simplified-test test-case "execution-error")
 224               nil))
 225           (handler-case
 226               (progn
 227                 #+xuriella::xsltproc
 228                 (xsltproc noindent-stylesheet data "/dev/null")
 229                 (saxon noindent-stylesheet data "/dev/null")
 230                 (report "FAIL" ": expected error not signalled")
 231                 ;; let's ignore unexpected successes for now
 232                 nil)
 233             (error (c)
 234               (report "PASS" ": expected error ~A" c)
 235               (write-simplified-test test-case "execution-error")
 236               t))))))
 237
 238 (defun xsltproc (stylesheet input output)
 239   (flet ((full-namestring (x)
 240            (namestring (merge-pathnames x))))
 241     (let* ((asdf::*verbose-out* (make-string-output-stream))
 242            (code (asdf:run-shell-command
 243                   "cd ~S && xsltproc ~S ~S >~S"
 244                   (full-namestring "")
 245                   (full-namestring stylesheet)
 246                   (full-namestring input)
 247                   (full-namestring output))))
 248       (unless (zerop code)
 249         (error "running xsltproc failed with code ~A [~%~A~%]"
 250                code
 251                (get-output-stream-string asdf::*verbose-out*))))))
 252
 253 (defun saxon (stylesheet input output)
 254   (flet ((full-namestring (x)
 255            (namestring (merge-pathnames x))))
 256     (let* ((asdf::*verbose-out* (make-string-output-stream))
 257            (code (asdf:run-shell-command
 258                   "cd ~S && java -jar /usr/share/java/saxon.jar ~S ~S >~S"
 259                   (full-namestring "")
 260                   (full-namestring input)
 261                   (full-namestring stylesheet)
 262                   (full-namestring output))))
 263       (unless (zerop code)
 264         (error "running saxon failed with code ~A [~%~A~%]"
 265                code
 266                (get-output-stream-string asdf::*verbose-out*))))))
 267
 268
 269 ;;;; RUN-TESTS and DRIBBLE-TESTS
 270
 271 ;;; Process katalog.xml
 272
 273 (defun dribble-tests
 274     (&key filter (directory *tests-directory*) (file "TEST"))
 275   (let ((*package* (find-package 'cl-user))
 276         (*print-circle* nil))
 277     (with-open-file (dribble
 278                      (merge-pathnames file
 279                                       (slot-value (asdf:find-system :xuriella)
 280                                                   'asdf::relative-pathname))
 281                      :direction :output
 282                      :if-exists :supersede
 283                      :external-format :utf-8)
 284       (let* ((dribble (make-broadcast-stream dribble *standard-output*))
 285              (*standard-output* dribble)
 286              (*trace-output* dribble)
 287              (*error-output* dribble)
 288              (*terminal-io* (make-two-way-stream *standard-input* dribble)))
 289         (handler-bind ((warning
 290                         (lambda (c)
 291                           (warn "~A" (replace-junk (princ-to-string c)))
 292                           (muffle-warning c))))
 293           (run-tests :filter filter
 294                      :directory directory))))))
 295
 296 (defparameter *bad-tests*
 297   '( ;; INCONSISTENT TESTS: some tests wants us to recover from this
 298     ;; error, yet this one doesn't:
 299     "copy_copy61"
 300     "copy_copy62"
 301
 302     ;; I'M TOO LAZY TO FIX THIS: we signal a run-time error when and if it's
 303     ;; actually used.  The test wants a compilation-time error.
 304     ;; So what?
 305     "AttributeSets_RefToUndefinedAttributeSet"
 306
 307     ;; WE WOULD PASS THIS: we perform recovery, but saxon doesn't.  Recovery
 308     ;; results in non-XML output, which we can't parse for comparison
 309     ;; against the official test case.
 310     "output_output75"
 311     ;;
 312     ;; we pass this test, but saxon doesn't
 313     "AttributeSets__91038"
 314     ;;
 315     ;; we'd pass these tests, but the test authors forgot to declare the
 316     ;; entity they're writing, so we can't parse it for comparison.
 317     "output_output06"
 318     "output_output10"
 319     "output_output61"
 320     ;;
 321     ;; another similar test where the output is unparsable, except that
 322     ;; here an entity declaration wouldn't have helped either:
 323     "Copying_ResultTreeFragmentWithEscapedText"
 324
 325     ;; BROKEN TEST: input document not ns-wf
 326     "Attributes__78387"
 327
 328     ;; FIXME: should re-enable these at some point:
 329     ;;
 330     ;; the following tests take a lot of time due to the problems of current matching algorithm:
 331     "impincl_impincl16"
 332     "match_match13"
 333     ;; stack exhaustion -- matching problem i think
 334     "Keys_PerfRepro3"
 335     ;; probably the same problem (but I haven't checked):
 336     "numbering_numbering03"
 337     "numbering_numbering10"
 338     "numbering_numbering11"
 339     "numbering_numbering80"
 340     "numbering_numbering81"
 341     "numbering_numbering94"
 342     "numbering_numbering95"
 343     "Import__91164"))
 344
 345 ;; Tests where the output isn't a match because of extraneous whitespace.
 346 ;; For these tests, we force space normalization before comparing.
 347 ;;
 348 ;; (SANITIZE-STYLESHEET is supposed to get rid of indent="yes", but it
 349 ;; misses imported stylesheets.)
 350 ;;
 351 ;; FIXME: Perhaps some *bad-tests* could instead be *whitespace-issues*,
 352 ;; at least those where the official output wuold be a match.
 353 ;;
 354 (defparameter *whitespace-issues*
 355   '("BVTs_bvt044"
 356     "Namespace-alias__91782"))
 357
 358 (defun run-tests (&key filter (directory *tests-directory*))
 359   (when (typep filter '(or string cons))
 360     (setf filter (cl-ppcre:create-scanner filter)))
 361   (klacks:with-open-source
 362       (source (klacks:make-tapping-source
 363                (cxml:make-source (merge-pathnames "katalog.xml" directory))))
 364     (let ((*default-pathname-defaults* (merge-pathnames directory)))
 365       (map-tests #'run-test
 366                  source
 367                  :test (lambda (test)
 368                          (and (or (null filter)
 369                                   (cl-ppcre:all-matches
 370                                    filter
 371                                    (format nil "~A/~A"
 372                                            (test-category test)
 373                                            (test-id test))))
 374                               (not (find (test-id test)
 375                                          *bad-tests*
 376                                          :test #'equal))))))))
 377
 378 (defun run-named-test (name &optional (d *tests-directory*))
 379   (let ((*break-on-signals* 'error))
 380     (run-tests :filter (format nil "/~A$" name) :directory d)))
 381
 382 (defun copy-file (p q)
 383   (with-open-file (in p :element-type '(unsigned-byte 8))
 384     (with-open-file (out q
 385                          :element-type '(unsigned-byte 8)
 386                          :direction :output
 387                          :if-exists :rename-and-delete)
 388       (let ((buf (make-array 8192 :element-type '(unsigned-byte 8))))
 389         (loop for pos = (read-sequence buf in)
 390            until (zerop pos)
 391            do (write-sequence buf out :end pos))))))
 392
 393 (defun find-named-test (name &optional (d *tests-directory*))
 394   (klacks:with-open-source
 395       (source (klacks:make-tapping-source
 396                (cxml:make-source (merge-pathnames "katalog.xml" d))))
 397     (block nil
 398       (map-tests (lambda (test)
 399                    (return test))
 400                  source
 401                  :test (lambda (test) (equal (test-id test) name))))))
 402
 403 (defun copy-test-files (name &optional (d *tests-directory*))
 404   (let* ((test (find-named-test name d))
 405          (*default-pathname-defaults* (merge-pathnames d))
 406          (*break-on-signals* 'error)
 407          (target-dir (merge-pathnames "copied-test/"
 408                                       (asdf:component-pathname
 409                                        (asdf:find-system :xuriella))))
 410          (xsl (merge-pathnames "test.xsl" target-dir))
 411          (xml (merge-pathnames "test.xml" target-dir))
 412          (txt (merge-pathnames "official-output.txt" target-dir))
 413          (expected (merge-pathnames "expected.xml" target-dir))
 414          (actual (merge-pathnames "actual.xml" target-dir)))
 415     (ensure-directories-exist target-dir)
 416     (copy-file (test-stylesheet-pathname test) xsl)
 417     (copy-file (test-data-pathname test) xml)
 418     (when (test-official-output-pathname test)
 419       (copy-file (test-official-output-pathname test) txt))
 420     (format t "Test stylesheet copied to:~%  ~A~%~%" xsl)
 421     (format t "Test data copied to:~%  ~A~%~%" xml)
 422     (when (test-official-output-pathname test)
 423       (format t "Official output file:~%  ~A~%~%" txt))
 424     (format t "Run xsltproc like this:~%  cd ~A~%  xsltproc ~A ~A >~A~%~%"
 425             (namestring target-dir)
 426             (enough-namestring xsl target-dir)
 427             (enough-namestring xml target-dir)
 428             (enough-namestring expected target-dir))
 429     (format t "Run saxon like this:~%  cd ~A~%  java -jar /usr/share/java/saxon.jar ~A ~A >~A~%~%"
 430             (namestring target-dir)
 431             (enough-namestring xml target-dir)
 432             (enough-namestring xsl target-dir)
 433             (enough-namestring expected target-dir))
 434     (format t "Run MSXSL like this:~%  cd ~A~%  wine msxsl.exe ~A ~A >~A~%~%"
 435             (namestring target-dir)
 436             (enough-namestring xml target-dir)
 437             (enough-namestring xsl target-dir)
 438             (enough-namestring expected target-dir))
 439     (format t "Run xuriella like this:~%")
 440     `(apply-stylesheet ,xsl ,xml :output ,actual)))
 441
 442 (defun map-tests (run-test source &key (test (constantly t)))
 443   (let ((total 0)
 444         (pass 0))
 445     (loop
 446        while (klacks:find-event source :start-element)
 447        for lname = (klacks:current-lname source)
 448        do
 449        (cond
 450          ((equal lname "test-case")
 451           (let* ((<test-case>
 452                   (stp:document-element
 453                    (klacks:serialize-element source (stp:make-builder))))
 454                  (test-case (parse-test <test-case>)))
 455             (when (funcall test test-case)
 456               (incf total)
 457               (when (funcall run-test test-case)
 458                 (incf pass)))))
 459          (t
 460           (klacks:skip source :start-element))))
 461     (format t "~&Passed ~D/~D tests.~%" pass total)))
 462
 463 (defun parse-test (<test-case>)
 464   (stp:with-attributes (id category operation
 465                            data stylesheet data-2 stylesheet-2
 466                            output compare)
 467       <test-case>
 468     (make-instance 'test-case
 469                    :id id
 470                    :category category
 471                    :operation operation
 472                    :data-pathname data
 473                    :stylesheet-pathname stylesheet
 474                    :data-pathname-2 data-2
 475                    :stylesheet-pathname-2 stylesheet-2
 476                    :output-pathname output
 477                    :output-compare compare)))
 478
 479 ;; read from file P, skipping the XMLDecl or TextDecl and Doctype at the
 480 ;; beginning, if any.
 481 (defun slurp-for-comparison (p)
 482   (with-open-file (s p :element-type '(unsigned-byte 8))
 483     (unless (and (eql (read-byte s nil) #xef)
 484                  (eql (read-byte s nil) #xbb)
 485                  (eql (read-byte s nil) #xbf))
 486       (file-position s 0))
 487     (if (plusp (file-length s))
 488         (slurp-for-comparison-1 p s t)
 489         "<wrapper/>")))
 490
 491 (defun slurp-for-comparison-1 (p s junk-info)
 492   (let ((pos (file-position s))         ;for UTF-8 "BOM"
 493         (xstream (runes:make-xstream s :speed 1))
 494         (prev-pos 0))
 495     (setf (runes:xstream-name xstream)
 496           (cxml::make-stream-name
 497            :entity-name "main document"
 498            :entity-kind :main
 499            :uri (cxml::pathname-to-uri (merge-pathnames p))))
 500     (let ((source
 501            (flet ((er (pub sys)
 502                     pub sys
 503                     (flexi-streams:make-in-memory-input-stream
 504                      #())))
 505              (cxml:make-source xstream
 506                                :pathname p
 507                                :entity-resolver #'er))))
 508       (unless (eq junk-info :nada)
 509         (loop
 510            for key = (progn
 511                        (setf prev-pos (runes:xstream-position xstream))
 512                        (klacks:peek-next source))
 513            until (eq key :start-document))
 514         (cxml::with-source (source cxml::context)
 515           (when (eq (cxml::zstream-token-category
 516                      (cxml::main-zstream cxml::context))
 517                     :NMTOKEN)
 518             ;; oops, doesn't look like XML at all
 519             (file-position s pos)
 520             (return-from slurp-for-comparison-1
 521               (slurp-for-comparison-1 p s :nada)))))
 522       (etypecase junk-info
 523         (integer
 524          (dotimes (x junk-info)
 525            (setf prev-pos (runes:xstream-position xstream))
 526            (klacks:peek-next source)))
 527         ((eql t)
 528          (let ((nskip 0))
 529            (handler-case
 530                (loop
 531                   (case (klacks:peek-next source)
 532                     (:start-element (return))
 533                     (:characters
 534                      (if (whitespacep (klacks:current-characters source))
 535                          (incf nskip)
 536                          (return)))
 537                     (t
 538                      (incf nskip))))
 539              ((or file-error cxml:xml-parse-error) ()
 540                (when (zerop nskip)
 541                  (setf nskip nil))))
 542            ;; retry
 543            (with-open-file (u p :element-type '(unsigned-byte 8))
 544              (file-position u pos)
 545              (return-from slurp-for-comparison-1
 546                (slurp-for-comparison-1 p u nskip)))))
 547         ((member nil :nada)))
 548       (with-output-to-string (r)
 549         (let* ((seen-char
 550                 (cxml::with-source (source cxml::context)
 551                   (ecase (cxml::zstream-token-category
 552                           (cxml::main-zstream cxml::context))
 553                     (:seen-< #\<)
 554                     (:? #\?)
 555                     ((nil :s)
 556                      (setf prev-pos (runes:xstream-position xstream))
 557                      nil))))
 558                (off-by-one-p (or seen-char (eq junk-info :nada)))
 559                (new-pos (- prev-pos (if off-by-one-p 1 0))))
 560           ;; copy doctype over
 561           (with-open-file (u p :element-type '(unsigned-byte 8))
 562             (file-position u pos)
 563             (let ((y (runes:make-xstream u :speed 1)))
 564               (loop
 565                  while (< (runes:xstream-position y) new-pos)
 566                  do (write-char (runes:read-rune y) r))))
 567           (write-line "<wrapper>" r)
 568           (when seen-char
 569             (write-char seen-char r)))
 570         (loop
 571            for char = (runes:read-rune xstream)
 572            until (eq char :eof)
 573            do (write-char char r))
 574         (write-line "</wrapper>" r)))))
 575
 576 (defun parse-for-comparison (p)
 577   (let* ((d (flet ((er (pub sys)
 578                     pub sys
 579                     (flexi-streams:make-in-memory-input-stream
 580                      #())))
 581               (cxml:parse (slurp-for-comparison p)
 582                           (make-text-normalizer (stp:make-builder))
 583                           :entity-resolver #'er)))
 584          (de (stp:document-element d)))
 585     (let ((first (stp:first-child de)))
 586       (when (typep first 'stp:text)
 587         (cond
 588           ((whitespacep (stp:data first))
 589            (stp:delete-child first de))
 590           (t
 591            (setf (stp:data first)
 592                  (cl-ppcre:regex-replace #.(format nil "^[~A]+" *whitespace*)
 593                                          (stp:data first)
 594                                          ""))))))
 595     (let ((last (stp:last-child de)))
 596       (when (typep last 'stp:text)
 597         (cond
 598           ((whitespacep (stp:data last))
 599            (stp:delete-child last de))
 600           (t
 601            (setf (stp:data last)
 602                  (cl-ppcre:regex-replace #.(format nil "[~A]+$" *whitespace*)
 603                                          (stp:data last)
 604                                          ""))))))
 605     d))
 606
 607 (defun output-equal-p (compare p q &key normalize)
 608   (handler-case
 609       (ecase compare
 610         (:xml (xml-output-equal-p p q normalize))
 611         (:html (html-output-equal-p p q))
 612         (:text (text-output-equal-p p q)))
 613     ((or error parse-number::invalid-number) (c)
 614       (warn "comparison failed: ~A" c)
 615       nil)))
 616
 617 ;; Workaround for namespace_namespace23 and other tests:
 618 ;;  - For these tests, saxon and msxsl output a declaration for the XSL
 619 ;;    namespace without using that declaration.
 620 ;;  - I think saxon and msxsl are both wrong.
 621 ;;  - The official test output agrees with my assessment.
 622 ;;    (So does libxslt, but that's not to be trusted. :-))
 623 ;;  - Here's the catch: The official test output is broken in its whitespace
 624 ;;    handling.
 625 ;; So let's normalize spaces in test output that looks like an XSLT
 626 ;; stylesheet, allowing us to pass these tests using the official test output.
 627 (defun maybe-normalize-test-spaces (wrapper force)
 628   (let ((i 0))
 629     (loop while (< i (length (cxml-stp-impl::%children wrapper))) do
 630          (let ((wrapper-child (stp:nth-child i wrapper)))
 631            (cond
 632              ((not (typep wrapper-child 'stp:element))
 633               (if force
 634                   (stp:delete-nth-child i wrapper)
 635                   (incf i)))
 636              ((or (equal (stp:namespace-uri wrapper-child) *xsl*)
 637                   force)
 638               (strip-stylesheet wrapper-child)
 639               (labels ((recurse (e &optional preserve)
 640                          (stp:do-children (child e)
 641                            (typecase child
 642                              (stp:text
 643                               (setf (stp:data child)
 644                                     (normalize-whitespace (stp:data child))))
 645                              (stp:element
 646                                  (stp:with-attributes ((space "space" *xml*))
 647                                      child
 648                                    (let ((new-preserve
 649                                           (cond
 650                                             ((namep child "text") t)
 651                                             ((not space) preserve)
 652                                             ((equal space "preserve") t)
 653                                             (t nil))))
 654                                      (recurse child new-preserve))))))))
 655                 (recurse wrapper-child))
 656               (incf i))
 657              (t
 658               (incf i)))))))
 659
 660 (defun xml-output-equal-p (p q normalize)
 661   (let ((r (parse-for-comparison p))
 662         (s (parse-for-comparison q)))
 663     (maybe-normalize-test-spaces (stp:document-element r) normalize)
 664     (maybe-normalize-test-spaces (stp:document-element s) normalize)
 665     (and (let ((u (stp:document-type r))
 666                (v (stp:document-type s)))
 667            (if u
 668                (and v (node= u v))
 669                (null v)))
 670          (node= (stp:document-element r) (stp:document-element s)))))
 671
 672 ;; FIXME: don't do this in <pre> etc.
 673 (defun normalize-html-whitespace (node)
 674   (when (typep node 'stp:parent-node)
 675     ;; ignore newlines after start tags completely
 676     (let ((first (stp:first-child node)))
 677       (when (and (typep first 'stp:text)
 678                  (alexandria:starts-with #\newline (stp:data first)))
 679         (setf (stp:data first) (subseq (stp:data first) 1))))
 680     ;; ignore newlines before end tags completely
 681     (let ((last (stp:last-child node)))
 682       (when (and (typep last 'stp:text)
 683                  (alexandria:ends-with #\newline (stp:data last)))
 684         (setf (stp:data last)
 685               (subseq (stp:data last) 0 (length (stp:data last))))))
 686     ;; normalize sequences of whitespace
 687     (stp:do-children (child node)
 688       (if (typep child 'stp:text)
 689           (setf (stp:data child)
 690                 (let ((str (normalize-whitespace (stp:data child))))
 691                   (when
 692                       ;; FIXME!  Here we remove whitespace entirely.
 693                       ;; Totally incorrect, but I don't see how we could
 694                       ;; watch Saxon's output otherwise.
 695                       (equal str " ")
 696                     (setf str ""))
 697                   str))
 698           (normalize-html-whitespace child)))
 699     ;; just to be sure, join adjacent nodes
 700     (cxml-stp-impl::normalize-text-nodes! node)))
 701
 702 ;; FIXME: this check is too lenient, because chtml is an error-correcting
 703 ;; parser.
 704 (defun html-output-equal-p (p q)
 705   (let ((r (chtml:parse (pathname p) (stp:make-builder)))
 706         (s (chtml:parse (pathname q) (stp:make-builder))))
 707     (normalize-html-whitespace r)
 708     (normalize-html-whitespace s)
 709     (node= (stp:document-element r) (stp:document-element s))))
 710
 711 (defun text-output-equal-p (p q)
 712   (with-open-file (a p :element-type '(unsigned-byte 8))
 713     (with-open-file (b q :element-type '(unsigned-byte 8))
 714       (let ((len (file-length a)))
 715         (and (eql len (file-length b))
 716              (let ((d (make-array len :element-type '(unsigned-byte 8)))
 717                    (e (make-array len :element-type '(unsigned-byte 8))))
 718                (read-sequence d a)
 719                (read-sequence e b)
 720                (equalp d e)))))))
 721
 722 (defun strip-addresses (str)
 723   (cl-ppcre:regex-replace-all "{[0-9a-fA-F]+}\\>" str "{xxxxxxxx}>"))
 724
 725 (defun slurp-output-method (p)
 726   (xpath:with-namespaces ((nil #.*xsl*))
 727     (let* ((d (handler-bind
 728                   ((warning #'muffle-warning))
 729                 (cxml:parse (pathname p) (stp:make-builder))))
 730            (output (xpath:first-node (xpath:evaluate "//output" d))))
 731       (if output
 732           (let ((method (stp:attribute-value output "method")))
 733             (if method
 734                 (intern (string-upcase method) :keyword)
 735                 :xml))
 736           :xml))))
 737
 738 (defun replace-junk (str)
 739   (cl-ppcre:regex-replace-all
 740    `(:group ,(namestring *tests-directory*))
 741    (map 'string
 742         (lambda (c)
 743           (if (or (eql c #\newline) (<= 32 (char-code c) 126))
 744               c
 745               #\?))
 746         str)
 747    "..."))
 748
 749 (defun run-test (test)
 750   (let ((expected-saxon (test-output-pathname test "saxon"))
 751         #+xuriella::xsltproc
 752         (expected-xsltproc (test-output-pathname test "xsltproc"))
 753         (actual (test-output-pathname test "xuriella"))
 754         (official (test-official-output-pathname test))
 755         (force-normalization
 756          (find (test-id test) *whitespace-issues* :test #'equal))
 757         (output-method nil))
 758     (handler-bind ((|hey test suite, this is an HTML document|
 759                     (lambda (c)
 760                       (declare (ignore c))
 761                       (setf output-method :html))))
 762       (labels ((uri-resolver (uri)
 763                  (if (search "%5c%5c%5c%5cwebxtest%5c%5cmanagedshadow%5c%5cmanaged_b2%5c%5ctestdata%5c%5cxslt%5c%5celement%5c%5cxslt_element_NSShared.xml"
 764                              uri)
 765                      (cxml::pathname-to-uri
 766                       (merge-pathnames
 767                        "MSFT_Conformance_Tests/Elements/xslt_element_NSShared.xml"
 768                        *tests-directory*))
 769                      uri))
 770                (doit ()
 771                  (with-open-file (s actual
 772                                     :if-exists :rename-and-delete
 773                                     :direction :output
 774                                     :element-type '(unsigned-byte 8))
 775                    (handler-bind ((xslt-error
 776                                    (lambda (c)
 777                                      (declare (ignore c))
 778                                      (when (find-restart 'recover)
 779                                        (invoke-restart 'recover)))))
 780                      (apply-stylesheet (pathname (test-stylesheet-pathname test))
 781                                        (pathname (test-data-pathname test))
 782                                        :output s
 783                                        :uri-resolver #'uri-resolver))))
 784                (pp (label pathname)
 785                  (when pathname
 786                    (format t "  ~A: ~A~%"
 787                            label
 788                            (enough-namestring pathname *tests-directory*))))
 789                (report (ok &optional (fmt "") &rest args)
 790                  (write-string
 791                   (replace-junk
 792                    (strip-addresses
 793                     (format nil "~&~:[FAIL~;PASS~] ~A [~A]~?~%"
 794                             ok
 795                             (test-id test)
 796                             (test-category test)
 797                             fmt
 798                             args))))
 799                  (pp "Stylesheet" (test-stylesheet-pathname test))
 800                  (pp "Data" (test-data-pathname test))
 801                  (pp "Supplemental stylesheet"
 802                      (test-stylesheet-pathname-2 test))
 803                  (pp "Supplemental data" (test-data-pathname-2 test))
 804                  (pp "Expected output (1)" expected-saxon)
 805                  #+xuriella::xsltproc
 806                  (pp "Expected output (2)" expected-xsltproc)
 807                  (pp "Actual output" actual)
 808                  (terpri)
 809                  ok))
 810         (cond
 811           ((equal (test-operation test) "standard")
 812            (handler-case
 813                (progn
 814                  (when (find (test-id test)
 815                              nil ;;'("axes_axes47" "attribset_attribset20")
 816                              :test #'equal)
 817                    (error "skipping problematic test"))
 818                  (doit)
 819                  (let* ((output-method
 820                          (or output-method
 821                              (slurp-output-method
 822                               (test-stylesheet-pathname test))))
 823                         (saxon-matches-p
 824                          (output-equal-p output-method
 825                                          expected-saxon
 826                                          actual
 827                                          :normalize force-normalization))
 828                         #+xuriella::xsltproc
 829                         (xsltproc-matches-p
 830                          (output-equal-p output-method
 831                                          expected-xsltproc
 832                                          actual))
 833                         (official-matches-p
 834                          (output-equal-p output-method
 835                                          official
 836                                          actual
 837                                          :normalize force-normalization)))
 838                    (cond
 839                      ((or saxon-matches-p
 840                           #+xuriella::xsltproc xsltproc-matches-p
 841                           official-matches-p)
 842                       (report t)
 843                       #+xuriella::xsltproc
 844                       (report t ": saxon ~A, xsltproc ~A~:[~; (MISMATCH)~]"
 845                               saxon-matches-p
 846                               xsltproc-matches-p
 847                               (if saxon-matches-p
 848                                   (not xsltproc-matches-p)
 849                                   xsltproc-matches-p)))
 850                      (t
 851                       (report nil ": output doesn't match")))))
 852              ((or error parse-number::invalid-number) (c)
 853                (report nil ": ~A" c))))
 854           (t
 855            (handler-case
 856                (doit)
 857              (xslt-error (c)
 858                (report t ": raised an xslt-error as expected" c))
 859              ((or error parse-number::invalid-number) (c)
 860                (report nil ": condition of incorrect type: ~%~A" c))
 861              (:no-error (result)
 862                (cond
 863                  ((not (and official (probe-file official)))
 864                   (report nil ": expected error not signalled: " result))
 865                  ((output-equal-p
 866                    (or output-method
 867                        (slurp-output-method (test-stylesheet-pathname test)))
 868                    official
 869                    actual
 870                    :normalize force-normalization)
 871                   (report t))
 872                  (t
 873                   (report nil ": saxon error not signalled and official output not a match")))))))))))
 874
 875 (defun run-xpath-tests ()
 876   (run-tests "XPath-Expression/|XSLT-Data-Model/"))
 877
 878
 879 ;;;; from cxml-stp-test
 880
 881 (defun assert-node= (a b)
 882   (unless (node= a b)
 883     (error "assertion failed: ~S and ~S are not NODE=" a b)))
 884
 885 (defun child-count (node)
 886   (stp:count-children-if (constantly t) node))
 887
 888 (defun named-node-= (a b)
 889   (and (equal (stp:namespace-uri a) (stp:namespace-uri b))
 890        ;; (equal (stp:namespace-prefix a) (stp:namespace-prefix b))
 891        (equal (stp:local-name a) (stp:local-name b))))
 892
 893 (defun parent-node-= (e f)
 894   (and (eql (child-count e)
 895             (child-count f))
 896        (every #'node= (stp:list-children e) (stp:list-children f))))
 897
 898 (defmethod node= ((e stp:element) (f stp:element))
 899   (and (named-node-= e f)
 900        (parent-node-= e f)
 901        (null
 902         (set-exclusive-or (stp:list-attributes e) (stp:list-attributes f)
 903                           :test #'node=))
 904        (block nil
 905          (flet ((check-namespaces (a b)
 906                   (let ((result ()))
 907                     (stp:map-extra-namespaces
 908                      (lambda (k v)
 909                        (unless (equal v (stp:find-namespace k b))
 910                          (return nil)))
 911                      a)
 912                     result)))
 913            (check-namespaces e f)
 914            (check-namespaces f e))
 915          t)))
 916
 917 (defmethod node= ((a stp:node) (b stp:node))
 918   nil)
 919
 920 (defmethod node= ((e stp:document) (f stp:document))
 921   (parent-node-= e f))
 922
 923 (defmethod node= ((a stp:attribute) (b stp:attribute))
 924   (and (named-node-= a b)
 925        (equal (stp:value a) (stp:value b))))
 926
 927 (defmethod node= ((a stp:comment) (b stp:comment))
 928   (equal (stp:data a) (stp:data b)))
 929
 930 (defmethod node= ((a stp:text) (b stp:text))
 931   (equal (stp:data a) (stp:data b)))
 932
 933 (defmethod node= ((a stp:processing-instruction)
 934                   (b stp:processing-instruction))
 935   (and (equal (stp:data a) (stp:data b))
 936        (equal (stp:target a) (stp:target b))))
 937
 938 (defmethod node= ((a stp:document-type) (b stp:document-type))
 939   (and (equal (stp:root-element-name a) (stp:root-element-name b))
 940        (equal (stp:public-id a) (stp:public-id b))
 941        (equal (stp:system-id a) (stp:system-id b))
 942        (equal (stp:internal-subset a) (stp:internal-subset b))))
 943
 944 (xpath-sys:define-xpath-function/eager
 945     xslt :print
 946     (thing)
 947   (if (xpath:node-set-p thing)
 948       (loop
 949          initially (format t ";;; node set:~%")
 950          for i from 0
 951          for node in (xpath:all-nodes thing)
 952          do
 953            (format t ";;;   ~D: ~A~%" i (type-of node)))
 954       (format t ";;; ~A~%" thing))
 955   thing)