tests/utils/tap-driver.sh

   1 #! /bin/sh
   2 # Copyright (C) 2011-2013 Free Software Foundation, Inc.
   3 #
   4 # This program is free software; you can redistribute it and/or modify
   5 # it under the terms of the GNU General Public License as published by
   6 # the Free Software Foundation; either version 2, or (at your option)
   7 # any later version.
   8 #
   9 # This program is distributed in the hope that it will be useful,
  10 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  11 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12 # GNU General Public License for more details.
  13 #
  14 # You should have received a copy of the GNU General Public License
  15 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
  16
  17 # As a special exception to the GNU General Public License, if you
  18 # distribute this file as part of a program that contains a
  19 # configuration script generated by Autoconf, you may include it under
  20 # the same distribution terms that you use for the rest of that program.
  21
  22 # This file is maintained in Automake, please report
  23 # bugs to <bug-automake@gnu.org> or send patches to
  24 # <automake-patches@gnu.org>.
  25
  26 scriptversion=2013-12-23.17; # UTC
  27
  28 # Make unconditional expansion of undefined variables an error.  This
  29 # helps a lot in preventing typo-related bugs.
  30 set -u
  31
  32 me=tap-driver.sh
  33
  34 fatal ()
  35 {
  36   echo "$me: fatal: $*" >&2
  37   exit 1
  38 }
  39
  40 usage_error ()
  41 {
  42   echo "$me: $*" >&2
  43   print_usage >&2
  44   exit 2
  45 }
  46
  47 print_usage ()
  48 {
  49   cat <<END
  50 Usage:
  51   tap-driver.sh --test-name=NAME --log-file=PATH --trs-file=PATH
  52                 [--expect-failure={yes|no}] [--color-tests={yes|no}]
  53                 [--enable-hard-errors={yes|no}] [--ignore-exit]
  54                 [--diagnostic-string=STRING] [--merge|--no-merge]
  55                 [--comments|--no-comments] [--post-script] [--] TEST-COMMAND
  56 The '--test-name', '-log-file' and '--trs-file' options are mandatory.
  57 END
  58 }
  59
  60 # TODO: better error handling in option parsing (in particular, ensure
  61 # TODO: $log_file, $trs_file and $test_name are defined).
  62 test_name= # Used for reporting.
  63 log_file=  # Where to save the result and output of the test script.
  64 trs_file=  # Where to save the metadata of the test run.
  65 post_script= # Script to be run after the test.
  66 expect_failure=0
  67 color_tests=0
  68 merge=0
  69 ignore_exit=0
  70 comments=0
  71 diag_string='#'
  72 while test $# -gt 0; do
  73   case $1 in
  74   --help) print_usage; exit $?;;
  75   --version) echo "$me $scriptversion"; exit $?;;
  76   --test-name) test_name=$2; shift;;
  77   --log-file) log_file=$2; shift;;
  78   --trs-file) trs_file=$2; shift;;
  79   --color-tests) color_tests=$2; shift;;
  80   --expect-failure) expect_failure=$2; shift;;
  81   --enable-hard-errors) shift;; # No-op.
  82   --merge) merge=1;;
  83   --no-merge) merge=0;;
  84   --ignore-exit) ignore_exit=1;;
  85   --comments) comments=1;;
  86   --no-comments) comments=0;;
  87   --diagnostic-string) diag_string=$2; shift;;
  88   --post-script) post_script=$2; shift;;
  89   --) shift; break;;
  90   -*) usage_error "invalid option: '$1'";;
  91   esac
  92   shift
  93 done
  94
  95 test $# -gt 0 || usage_error "missing test command"
  96
  97 case $expect_failure in
  98   yes) expect_failure=1;;
  99     *) expect_failure=0;;
 100 esac
 101
 102 if test $color_tests = yes; then
 103   init_colors='
 104     color_map["red"]="\e[0;31m" # Red.
 105     color_map["grn"]="\e[0;32m" # Green.
 106     color_map["lgn"]="\e[1;32m" # Light green.
 107     color_map["blu"]="\e[1;34m" # Blue.
 108     color_map["mgn"]="\e[0;35m" # Magenta.
 109     color_map["std"]="\e[m"     # No color.
 110     color_for_result["ERROR"] = "mgn"
 111     color_for_result["PASS"]  = "grn"
 112     color_for_result["XPASS"] = "red"
 113     color_for_result["FAIL"]  = "red"
 114     color_for_result["XFAIL"] = "lgn"
 115     color_for_result["SKIP"]  = "blu"'
 116 else
 117   init_colors=''
 118 fi
 119
 120 # :; is there to work around a bug in bash 3.2 (and earlier) which
 121 # does not always set '$?' properly on redirection failure.
 122 # See the Autoconf manual for more details.
 123 :;{
 124   (
 125     # Ignore common signals (in this subshell only!), to avoid potential
 126     # problems with Korn shells.  Some Korn shells are known to propagate
 127     # to themselves signals that have killed a child process they were
 128     # waiting for; this is done at least for SIGINT (and usually only for
 129     # it, in truth).  Without the `trap' below, such a behaviour could
 130     # cause a premature exit in the current subshell, e.g., in case the
 131     # test command it runs gets terminated by a SIGINT.  Thus, the awk
 132     # script we are piping into would never seen the exit status it
 133     # expects on its last input line (which is displayed below by the
 134     # last `echo $?' statement), and would thus die reporting an internal
 135     # error.
 136     # For more information, see the Autoconf manual and the threads:
 137     # <http://lists.gnu.org/archive/html/bug-autoconf/2011-09/msg00004.html>
 138     # <http://mail.opensolaris.org/pipermail/ksh93-integration-discuss/2009-February/004121.html>
 139     trap : 1 3 2 13 15
 140     if test $merge -gt 0; then
 141       exec 2>&1
 142     else
 143       exec 2>&3
 144     fi
 145     "$@"
 146     echo $?
 147   ) | LC_ALL=C ${AM_TAP_AWK-awk} \
 148         -v me="$me" \
 149         -v test_script_name="$test_name" \
 150         -v log_file="$log_file" \
 151         -v trs_file="$trs_file" \
 152         -v expect_failure="$expect_failure" \
 153         -v merge="$merge" \
 154         -v ignore_exit="$ignore_exit" \
 155         -v comments="$comments" \
 156         -v diag_string="$diag_string" \
 157 '
 158 # TODO: the usages of "cat >&3" below could be optimized when using
 159 #       GNU awk, and/on on systems that supports /dev/fd/.
 160
 161 # Implementation note: in what follows, `result_obj` will be an
 162 # associative array that (partly) simulates a TAP result object
 163 # from the `TAP::Parser` perl module.
 164
 165 ## ----------- ##
 166 ##  FUNCTIONS  ##
 167 ## ----------- ##
 168
 169 function fatal(msg)
 170 {
 171   print me ": " msg | "cat >&2"
 172   exit 1
 173 }
 174
 175 function abort(where)
 176 {
 177   fatal("internal error " where)
 178 }
 179
 180 # Convert a boolean to a "yes"/"no" string.
 181 function yn(bool)
 182 {
 183   return bool ? "yes" : "no";
 184 }
 185
 186 function add_test_result(result)
 187 {
 188   if (!test_results_index)
 189     test_results_index = 0
 190   test_results_list[test_results_index] = result
 191   test_results_index += 1
 192   test_results_seen[result] = 1;
 193 }
 194
 195 # Whether the test script should be re-run by "make recheck".
 196 function must_recheck()
 197 {
 198   for (k in test_results_seen)
 199     if (k != "XFAIL" && k != "PASS" && k != "SKIP")
 200       return 1
 201   return 0
 202 }
 203
 204 # Whether the content of the log file associated to this test should
 205 # be copied into the "global" test-suite.log.
 206 function copy_in_global_log()
 207 {
 208   for (k in test_results_seen)
 209     if (k != "PASS")
 210       return 1
 211   return 0
 212 }
 213
 214 function gettime_ns(    r,    cmd)
 215 {
 216    cmd = "date +%s.%N"
 217    r = ""
 218    if (cmd | getline r) {
 219      r = r
 220    }
 221    close(cmd)
 222    return sprintf("%f", r)
 223 }
 224
 225 function get_global_test_result()
 226 {
 227     if ("ERROR" in test_results_seen)
 228       return "ERROR"
 229     if ("FAIL" in test_results_seen || "XPASS" in test_results_seen)
 230       return "FAIL"
 231     all_skipped = 1
 232     for (k in test_results_seen)
 233       if (k != "SKIP")
 234         all_skipped = 0
 235     if (all_skipped)
 236       return "SKIP"
 237     return "PASS";
 238 }
 239
 240 function stringify_result_obj(result_obj)
 241 {
 242   if (result_obj["is_unplanned"] || result_obj["number"] != testno)
 243     return "ERROR"
 244
 245   if (plan_seen == LATE_PLAN)
 246     return "ERROR"
 247
 248   if (result_obj["directive"] == "TODO")
 249     return result_obj["is_ok"] ? "XPASS" : "XFAIL"
 250
 251   if (result_obj["directive"] == "SKIP")
 252     return result_obj["is_ok"] ? "SKIP" : COOKED_FAIL;
 253
 254   if (length(result_obj["directive"]))
 255       abort("in function stringify_result_obj()")
 256
 257   return result_obj["is_ok"] ? COOKED_PASS : COOKED_FAIL
 258 }
 259
 260 function decorate_result(result)
 261 {
 262   color_name = color_for_result[result]
 263   if (color_name)
 264     return color_map[color_name] "" result "" color_map["std"]
 265   # If we are not using colorized output, or if we do not know how
 266   # to colorize the given result, we should return it unchanged.
 267   return result
 268 }
 269
 270 function report(result, details)
 271 {
 272   if (result ~ /^(X?(PASS|FAIL)|SKIP|ERROR)/)
 273     {
 274       msg = ": " test_script_name
 275       add_test_result(result)
 276     }
 277   else if (result == "#")
 278     {
 279       msg = " " test_script_name ":"
 280     }
 281   else
 282     {
 283       abort("in function report()")
 284     }
 285   if (length(details))
 286     msg = msg " " details
 287   # Output on console might be colorized.
 288   print decorate_result(result) msg
 289   # Flush stdout after each test result, this is useful when stdout
 290   # is buffered, for example in a CI system.
 291   fflush()
 292   # Log the result in the log file too, to help debugging (this is
 293   # especially true when said result is a TAP error or "Bail out!").
 294   print result msg | "cat >&3";
 295 }
 296
 297 function testsuite_error(error_message)
 298 {
 299   report("ERROR", "- " error_message)
 300 }
 301
 302 function handle_tap_result()
 303 {
 304   details = result_obj["number"];
 305   if (length(result_obj["description"]))
 306     details = details " " result_obj["description"]
 307
 308   if (plan_seen == LATE_PLAN)
 309     {
 310       details = details " # AFTER LATE PLAN";
 311     }
 312   else if (result_obj["is_unplanned"])
 313     {
 314        details = details " # UNPLANNED";
 315     }
 316   else if (result_obj["number"] != testno)
 317     {
 318        details = sprintf("%s # OUT-OF-ORDER (expecting %d)",
 319                          details, testno);
 320     }
 321   else if (result_obj["directive"])
 322     {
 323       details = details " # " result_obj["directive"];
 324       if (length(result_obj["explanation"]))
 325         details = details " " result_obj["explanation"]
 326     }
 327
 328   report(stringify_result_obj(result_obj), details)
 329 }
 330
 331 # `skip_reason` should be empty whenever planned > 0.
 332 function handle_tap_plan(planned, skip_reason)
 333 {
 334   planned += 0 # Avoid getting confused if, say, `planned` is "00"
 335   if (length(skip_reason) && planned > 0)
 336     abort("in function handle_tap_plan()")
 337   if (plan_seen)
 338     {
 339       # Error, only one plan per stream is acceptable.
 340       testsuite_error("multiple test plans")
 341       return;
 342     }
 343   planned_tests = planned
 344   # The TAP plan can come before or after *all* the TAP results; we speak
 345   # respectively of an "early" or a "late" plan.  If we see the plan line
 346   # after at least one TAP result has been seen, assume we have a late
 347   # plan; in this case, any further test result seen after the plan will
 348   # be flagged as an error.
 349   plan_seen = (testno >= 1 ? LATE_PLAN : EARLY_PLAN)
 350   # If testno > 0, we have an error ("too many tests run") that will be
 351   # automatically dealt with later, so do not worry about it here.  If
 352   # $plan_seen is true, we have an error due to a repeated plan, and that
 353   # has already been dealt with above.  Otherwise, we have a valid "plan
 354   # with SKIP" specification, and should report it as a particular kind
 355   # of SKIP result.
 356   if (planned == 0 && testno == 0)
 357     {
 358       if (length(skip_reason))
 359         skip_reason = "- "  skip_reason;
 360       report("SKIP", skip_reason);
 361     }
 362 }
 363
 364 function extract_tap_comment(line)
 365 {
 366   if (index(line, diag_string) == 1)
 367     {
 368       # Strip leading `diag_string` from `line`.
 369       line = substr(line, length(diag_string) + 1)
 370       # And strip any leading and trailing whitespace left.
 371       sub("^[ \t]*", "", line)
 372       sub("[ \t]*$", "", line)
 373       # Return what is left (if any).
 374       return line;
 375     }
 376   return "";
 377 }
 378
 379 # When this function is called, we know that line is a TAP result line,
 380 # so that it matches the (perl) RE "^(not )?ok\b".
 381 function setup_result_obj(line)
 382 {
 383   # Get the result, and remove it from the line.
 384   result_obj["is_ok"] = (substr(line, 1, 2) == "ok" ? 1 : 0)
 385   sub("^(not )?ok[ \t]*", "", line)
 386
 387   # If the result has an explicit number, get it and strip it; otherwise,
 388   # automatically assing the next progresive number to it.
 389   if (line ~ /^[0-9]+$/ || line ~ /^[0-9]+[^a-zA-Z0-9_]/)
 390     {
 391       match(line, "^[0-9]+")
 392       # The final `+ 0` is to normalize numbers with leading zeros.
 393       result_obj["number"] = substr(line, 1, RLENGTH) + 0
 394       line = substr(line, RLENGTH + 1)
 395     }
 396   else
 397     {
 398       result_obj["number"] = testno
 399     }
 400
 401   if (plan_seen == LATE_PLAN)
 402     # No further test results are acceptable after a "late" TAP plan
 403     # has been seen.
 404     result_obj["is_unplanned"] = 1
 405   else if (plan_seen && testno > planned_tests)
 406     result_obj["is_unplanned"] = 1
 407   else
 408     result_obj["is_unplanned"] = 0
 409
 410   # Strip trailing and leading whitespace.
 411   sub("^[ \t]*", "", line)
 412   sub("[ \t]*$", "", line)
 413
 414   # This will have to be corrected if we have a "TODO"/"SKIP" directive.
 415   result_obj["description"] = line
 416   result_obj["directive"] = ""
 417   result_obj["explanation"] = ""
 418
 419   if (index(line, "#") == 0)
 420     return # No possible directive, nothing more to do.
 421
 422   # Directives are case-insensitive.
 423   rx = "[ \t]*#[ \t]*([tT][oO][dD][oO]|[sS][kK][iI][pP])[ \t]*"
 424
 425   # See whether we have the directive, and if yes, where.
 426   pos = match(line, rx "$")
 427   if (!pos)
 428     pos = match(line, rx "[^a-zA-Z0-9_]")
 429
 430   # If there was no TAP directive, we have nothing more to do.
 431   if (!pos)
 432     return
 433
 434   # Let`s now see if the TAP directive has been escaped.  For example:
 435   #  escaped:     ok \# SKIP
 436   #  not escaped: ok \\# SKIP
 437   #  escaped:     ok \\\\\# SKIP
 438   #  not escaped: ok \ # SKIP
 439   if (substr(line, pos, 1) == "#")
 440     {
 441       bslash_count = 0
 442       for (i = pos; i > 1 && substr(line, i - 1, 1) == "\\"; i--)
 443         bslash_count += 1
 444       if (bslash_count % 2)
 445         return # Directive was escaped.
 446     }
 447
 448   # Strip the directive and its explanation (if any) from the test
 449   # description.
 450   result_obj["description"] = substr(line, 1, pos - 1)
 451   # Now remove the test description from the line, that has been dealt
 452   # with already.
 453   line = substr(line, pos)
 454   # Strip the directive, and save its value (normalized to upper case).
 455   sub("^[ \t]*#[ \t]*", "", line)
 456   result_obj["directive"] = toupper(substr(line, 1, 4))
 457   line = substr(line, 5)
 458   # Now get the explanation for the directive (if any), with leading
 459   # and trailing whitespace removed.
 460   sub("^[ \t]*", "", line)
 461   sub("[ \t]*$", "", line)
 462   result_obj["explanation"] = line
 463 }
 464
 465 function get_test_exit_message(status)
 466 {
 467   if (status == 0)
 468     return ""
 469   if (status !~ /^[1-9][0-9]*$/)
 470     abort("getting exit status")
 471   if (status < 127)
 472     exit_details = ""
 473   else if (status == 127)
 474     exit_details = " (command not found?)"
 475   else if (status >= 128 && status <= 255)
 476     exit_details = sprintf(" (terminated by signal %d?)", status - 128)
 477   else if (status > 256 && status <= 384)
 478     # We used to report an "abnormal termination" here, but some Korn
 479     # shells, when a child process die due to signal number n, can leave
 480     # in $? an exit status of 256+n instead of the more standard 128+n.
 481     # Apparently, both behaviours are allowed by POSIX (2008), so be
 482     # prepared to handle them both.  See also Austing Group report ID
 483     # 0000051 <http://www.austingroupbugs.net/view.php?id=51>
 484     exit_details = sprintf(" (terminated by signal %d?)", status - 256)
 485   else
 486     # Never seen in practice.
 487     exit_details = " (abnormal termination)"
 488   return sprintf("exited with status %d%s", status, exit_details)
 489 }
 490
 491 function write_test_results()
 492 {
 493   print ":global-test-result: " get_global_test_result() > trs_file
 494   print ":recheck: "  yn(must_recheck()) > trs_file
 495   print ":copy-in-global-log: " yn(copy_in_global_log()) > trs_file
 496   print ":time-taken: " sprintf("%f", gettime_ns() - TIME_START) > trs_file
 497   for (i = 0; i < test_results_index; i += 1)
 498     print ":test-result: " test_results_list[i] > trs_file
 499   close(trs_file);
 500 }
 501
 502 BEGIN {
 503
 504 ## ------- ##
 505 ##  SETUP  ##
 506 ## ------- ##
 507
 508 '"$init_colors"'
 509
 510 # Properly initialized once the TAP plan is seen.
 511 planned_tests = 0
 512
 513 COOKED_PASS = expect_failure ? "XPASS": "PASS";
 514 COOKED_FAIL = expect_failure ? "XFAIL": "FAIL";
 515
 516 # Enumeration-like constants to remember which kind of plan (if any)
 517 # has been seen.  It is important that NO_PLAN evaluates "false" as
 518 # a boolean.
 519 NO_PLAN = 0
 520 EARLY_PLAN = 1
 521 LATE_PLAN = 2
 522
 523 testno = 0     # Number of test results seen so far.
 524 bailed_out = 0 # Whether a "Bail out!" directive has been seen.
 525
 526 # Whether the TAP plan has been seen or not, and if yes, which kind
 527 # it is ("early" is seen before any test result, "late" otherwise).
 528 plan_seen = NO_PLAN
 529
 530 TIME_START = gettime_ns();
 531 ## --------- ##
 532 ##  PARSING  ##
 533 ## --------- ##
 534
 535 is_first_read = 1
 536
 537 while (1)
 538   {
 539     # Involutions required so that we are able to read the exit status
 540     # from the last input line.
 541     st = getline
 542     if (st < 0) # I/O error.
 543       fatal("I/O error while reading from input stream")
 544     else if (st == 0) # End-of-input
 545       {
 546         if (is_first_read)
 547           abort("in input loop: only one input line")
 548         break
 549       }
 550     if (is_first_read)
 551       {
 552         is_first_read = 0
 553         nextline = $0
 554         continue
 555       }
 556     else
 557       {
 558         curline = nextline
 559         nextline = $0
 560         $0 = curline
 561       }
 562     # Copy any input line verbatim into the log file.
 563     print | "cat >&3"
 564     # Parsing of TAP input should stop after a "Bail out!" directive.
 565     if (bailed_out)
 566       continue
 567
 568     # TAP test result.
 569     if ($0 ~ /^(not )?ok$/ || $0 ~ /^(not )?ok[^a-zA-Z0-9_]/)
 570       {
 571         testno += 1
 572         setup_result_obj($0)
 573         handle_tap_result()
 574       }
 575     # TAP plan (normal or "SKIP" without explanation).
 576     else if ($0 ~ /^1\.\.[0-9]+[ \t]*$/)
 577       {
 578         # The next two lines will put the number of planned tests in $0.
 579         sub("^1\\.\\.", "")
 580         sub("[^0-9]*$", "")
 581         handle_tap_plan($0, "")
 582         continue
 583       }
 584     # TAP "SKIP" plan, with an explanation.
 585     else if ($0 ~ /^1\.\.0+[ \t]*#/)
 586       {
 587         # The next lines will put the skip explanation in $0, stripping
 588         # any leading and trailing whitespace.  This is a little more
 589         # tricky in truth, since we want to also strip a potential leading
 590         # "SKIP" string from the message.
 591         sub("^[^#]*#[ \t]*(SKIP[: \t][ \t]*)?", "")
 592         sub("[ \t]*$", "");
 593         handle_tap_plan(0, $0)
 594       }
 595     # "Bail out!" magic.
 596     # Older versions of prove and TAP::Harness (e.g., 3.17) did not
 597     # recognize a "Bail out!" directive when preceded by leading
 598     # whitespace, but more modern versions (e.g., 3.23) do.  So we
 599     # emulate the latter, "more modern" behaviour.
 600     else if ($0 ~ /^[ \t]*Bail out!/)
 601       {
 602         bailed_out = 1
 603         # Get the bailout message (if any), with leading and trailing
 604         # whitespace stripped.  The message remains stored in `$0`.
 605         sub("^[ \t]*Bail out![ \t]*", "");
 606         sub("[ \t]*$", "");
 607         # Format the error message for the
 608         bailout_message = "Bail out!"
 609         if (length($0))
 610           bailout_message = bailout_message " " $0
 611         testsuite_error(bailout_message)
 612       }
 613     # Maybe we have too look for dianogtic comments too.
 614     else if (comments != 0)
 615       {
 616         comment = extract_tap_comment($0);
 617         if (length(comment))
 618           report("#", comment);
 619       }
 620   }
 621
 622 ## -------- ##
 623 ##  FINISH  ##
 624 ## -------- ##
 625
 626 # A "Bail out!" directive should cause us to ignore any following TAP
 627 # error, as well as a non-zero exit status from the TAP producer.
 628 if (!bailed_out)
 629   {
 630     if (!plan_seen)
 631       {
 632         testsuite_error("missing test plan")
 633       }
 634     else if (planned_tests != testno)
 635       {
 636         bad_amount = testno > planned_tests ? "many" : "few"
 637         testsuite_error(sprintf("too %s tests run (expected %d, got %d)",
 638                                 bad_amount, planned_tests, testno))
 639       }
 640     if (!ignore_exit)
 641       {
 642         # Fetch exit status from the last line.
 643         exit_message = get_test_exit_message(nextline)
 644         if (exit_message)
 645           testsuite_error(exit_message)
 646       }
 647   }
 648
 649 write_test_results()
 650
 651 exit 0
 652
 653 } # End of "BEGIN" block.
 654 '
 655
 656 # TODO: document that we consume the file descriptor 3 :-(
 657 } 3>"$log_file"
 658
 659 test $? -eq 0 || fatal "I/O or internal error"
 660
 661 if test ! -z $post_script ; then
 662   $post_script
 663   test $? -eq 0 || fatal "Post script returned an error. See $log_file"
 664 fi
 665
 666 # Local Variables:
 667 # mode: shell-script
 668 # sh-indentation: 2
 669 # eval: (add-hook 'write-file-hooks 'time-stamp)
 670 # time-stamp-start: "scriptversion="
 671 # time-stamp-format: "%:y-%02m-%02d.%02H"
 672 # time-stamp-time-zone: "UTC"
 673 # time-stamp-end: "; # UTC"
 674 # End: