source: other-projects/nightly-tasks/diffcol/trunk/task.pl@ 28620

Last change on this file since 28620 was 28620, checked in by sjm84, 10 years ago

Do over of previous 2 commits: the new commandline arguments are more general.

  • Property svn:executable set to *
File size: 37.4 KB
Line 
1#!/usr/bin/perl -w
2
3# This program is meant to run the nightly diffcol
4# It is meant to be an equivalent for the existing task bash script
5# But it is intended to be expanded to work for Windows and GS3
6# For windows, need to REMEMBER to set the correct shebangs at the top
7
8
9# TODO:
10# Have a caveat mode and a stable mode (as in snapshot/task.pl)
11#
12#} elsif ( $ENV{'TASK_NAME'} =~ "gs2-diffcol-(caveat|stable)" ) {
13# $major_version = 2;
14# $prefix="2t";
15# $rk="tk2"; # test kit
16#} elsif ( $ENV{'TASK_NAME'} =~ "gs3-diffcol-(caveat|stable)" ) {
17# $major_version = 3;
18# $prefix="3t";
19# $rk="tk3"; # test kit
20
21package diffcoltask;
22
23use Cwd;
24use Switch; # for switch(val) { case: ; ...}
25use File::Path; # for rmdir and mkdir type functions
26use File::Copy; # for recursive copying of folders but skipping .svn
27use File::Basename;
28
29use strict;
30no strict 'subs'; # allow barewords (eg STDERR) as function arguments
31
32
33my $isWin = ($^O =~ m/mswin/i) ? 1 : 0;
34my $isMac = ($^O =~ m/macos|darwin/i) ? 1 : 0;
35
36my $osversion="";
37# Need to get the correct gnome-lib-minimal for the OS
38# darwin11* Lion, darwin12* Mountain Lion, darwin9* and darwin10* are Leopard and Snow Leopard
39if ($^O eq "darwin") {
40 $osversion=`uname -r`; # e.g. 12.x.x
41 #$osversion =~ s@\..*$@@; # e.g.12
42 $osversion = ($osversion =~ m@^1[1-9](\.)?@i) ? "Lion-" : "";
43}
44
45
46my $sep = $isWin ? "\\" : "/";
47my $pathsep = $isWin ? ";" : ":";
48#my $script_ext = $isWin ? ".bat" : ".bash";
49my $setup_script = "setup"; # needs to become gs3-setup for GS3
50my $use_blat = 0; # if we ever get blat to send mail/attachments on Windows working, set this to 1
51
52my $install_type = "svn";
53my $install_version = "2";
54
55my $use_local_rebuild = 0; # set to 1 (true) if just diffing and so we needn't copy model-collection over to the test collection again nor rebuild it (This is useful when having built the collection locally once before)
56my $use_static_model = 0; # set to 1 (true) if working with a non-svn model-collection. Defaults to 1 if $use_local_rebuild is turned on
57
58# if use_local_rebuild is on, use_static_model should be on
59if ($use_local_rebuild && !$use_static_model) {
60 $use_static_model = 1;
61}
62
63my $test_os = $isWin ? "windows" : ($isMac ? "darwin" : "linux");
64my $model_os = "linux"; # default
65
66# TASK_HOME should be the toplevel diffcol folder
67$ENV{'TASK_HOME'} = getcwd unless defined $ENV{'TASK_HOME'};
68if($isWin) {
69 $ENV{'TASK_HOME'} =~ s@\/@\\@g;
70 # need to convert TASK_HOME path name to resolve very subtle bug when running task.pl via
71 # run-gs2-diffcol.bat which uses environment.pl's TASK_HOME setting via envi
72 # At that point TASK_HOME is already defined but ends up lowercase, so that entries in archiveinf-doc
73 # end up sorted differently when db2txt -sort is applied compared to if TASK_HOME had kept its case.
74 require Win32; # for working out Windows Long Filenames from Win 8.3 short filenames
75 $ENV{'TASK_HOME'} = &Win32::GetLongPathName($ENV{'TASK_HOME'});
76}
77## print STDERR "@@@ TASK_HOME: ".$ENV{'TASK_HOME'}."\n";
78
79
80$ENV{'BIN_DIR'} = &filename_concat($ENV{'TASK_HOME'}, "bin");
81
82# we'll be using BLAT to send mail attachments on Windows
83my $blat = $use_blat ? &filename_concat($ENV{'BIN_DIR'}, "blat", "full", "blat.exe") : 0;
84if($isWin && $use_blat && ! -e $blat) {
85 print STDERR "\n***********************************\n";
86 print STDERR "No blat.exe found in $blat.\n";
87 print STDERR "Blat needed to send mail with attachments on Windows.\n";
88 print STDERR "Extract the blat zip file found in $ENV{'BIN_DIR'}\n";
89 print STDERR "for your bit architecture and name the folder 'blat'\n";
90 print STDERR "***********************************\n\n";
91 $blat = 0;
92}
93
94
95$ENV{'DATA_DIR'} = &filename_concat($ENV{'TASK_HOME'}, "diffcol-data");
96$ENV{'UPLOAD_DIR'} = &filename_concat($ENV{'TASK_HOME'}, "diffcol-reports");
97$ENV{'MONITOR_EMAIL'} = "greenstone_team\@cs.waikato.ac.nz"; # need to escape @ sign
98$ENV{'GSDL_SMTP'} = ""; #"smtp.gmail.com";
99##print STDERR "@@@ email: ".$ENV{'MONITOR_EMAIL'}."\n";
100
101# control if an existing compiled greenstone is used
102# or, if one should be checked out, which revision to checkout from svn
103$ENV{'SVN_OPT_REV'} = "-r head";
104#$ENV{'GSDLHOME'}=
105#$ENV{'GSDL3SRCHOME'}=
106
107
108# if the first arg is a digit, it's the new envi verbosity param. Take it off the array
109my $envi_verbose = shift(@ARGV) if(exists $ARGV[0] && $ARGV[0] =~ m/^\d+$/);
110
111#parse arguments
112my $action = "all";
113my $subaction = ""; # run_test can take subactions: --just_diff and --no_svn
114my @collections = (); # list of collections that run_test should process
115
116if(scalar(@ARGV) == 0) {
117 $action="all";
118}
119
120# process any arguments that are --gs2|--gs3 and --bin|--svn, and delete them from the array
121# if none provided, it's gs2 and svn by default.
122for (my $i = $#ARGV; $i >= 0; --$i) {
123 if($ARGV[$i] =~ m/--(bin|svn)/) {
124 $install_type = $1;
125 splice @ARGV, $i, 1; # remove the element from the argument array
126 } elsif($ARGV[$i] =~ m/--gs(2|3)/) {
127 $install_version = $1;
128 splice @ARGV, $i, 1; # remove the element from the argument array
129 }
130}
131
132
133# run_test can take any number of args
134if(scalar(@ARGV) > 1 && $ARGV[0] ne "run_test") {
135 print STDERR "**** Wrong number of arguments\n";
136 &printusage();
137 exit -1;
138}
139
140if(scalar(@ARGV) > 0) {
141 switch ($ARGV[0]) {
142 case qr/^(-h|--?help|help)$/i { &printusage; exit 0; }
143 case qr/^(setup_greenstone|run_test|summarise|upload|all)$/ { $action=$ARGV[0]; }
144 else {
145 print STDERR "**** Bad subcommand.\n";
146 &printusage;
147 exit -1;
148 }
149 }
150
151 # run_test action can take a subaction: nosvn|justdiff. It can also take --modelOS (windows|linux|darwin)
152 # nosvn: uses the model-collect as static and copies it over to collect, rebuilding what's currently in model-collect instead of copying
153 # it out from the svn model-collect again.
154 # justdiff: same as nosvn, but doesn't copy over model-collection to collect, and doesn't rebuild either of them. Just does the diff part.
155
156 if($action eq "run_test" && scalar(@ARGV) >= 2) {
157 push(@collections, @ARGV);
158 shift @collections; # remove action from array
159
160 for (my $i=0; $i < scalar(@ARGV); $i++) {
161 if($ARGV[$i] =~ m@^--@) {
162 shift @collections; # remove subaction/flag from array
163
164 $subaction = $ARGV[$i];
165 if($subaction eq "--justdiff") {
166 $use_local_rebuild = $use_static_model = 1;
167 } elsif ($subaction eq "--nosvn") {
168 $use_static_model = 1;
169 #} elsif ($subaction =~ m/\-\-testOS/i && defined $ARGV[$i+1]) {
170 # $test_os = $ARGV[$i+1];
171 # $i++;
172 # shift @collections; # remove test_os value from array
173 } elsif ($subaction =~ m/\-\-modelOS/i && defined $ARGV[$i+1] && $ARGV[$i+1] =~ m/windows|linux|darwin/i) {
174 $model_os = $ARGV[$i+1];
175 $i++;
176 shift @collections; # remove model_os value from array
177 #print STDERR "Model_os specified: $model_os\n";
178 } else {
179 print STDERR "**** Bad subaction/value: ".$ARGV[$i]."\n";
180 &printusage;
181 exit -1;
182 }
183 }
184 }
185
186# foreach my $col (@collections) {
187# print STDERR "Collection: $col\n";
188# }
189 }
190}
191
192#print STDERR "Install type $install_type\n";
193#print STDERR "Install version $install_version\n";
194
195
196#check key environment vars are set
197if(!defined $ENV{'UPLOAD_DIR'}) {
198 print STDERR "Please set a UPLOAD_DIR for the test in an environment.sh file\n";
199 #return 1;
200}
201if(!defined $ENV{'DATA_DIR'}) {
202 print STDERR "Please set a DATA_DIR for the test in an environment.sh file\n";
203 #return 1;
204}
205if(!defined $ENV{'MONITOR_EMAIL'}) {
206 print STDERR "Please set a MONITOR_EMAIL for the test in an environment.sh file\n";
207 #return 1;
208}
209
210if($ENV{'DATA_DIR'} eq "/") {
211 print STDERR "DATA_DIR should not be the fs root\n";
212 #return 1;
213}
214
215print STDERR "DATA_DIR: ".$ENV{'DATA_DIR'}."\n";
216print STDERR "UPLOAD_DIR: ".$ENV{'UPLOAD_DIR'}."\n";
217
218#create an id for this test
219my ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = localtime(time);
220$year += 1900;
221$mon += 1;
222$mon = "0$mon" if ($mon < 10);
223$mday = "0$mday" if ($mday < 10);
224my $dateid="$year.$mon.$mday"; #my $dateid=($year+1900)."-".($mon+1)."-$mday";
225
226print STDERR "Starting test '$dateid'\n";
227
228
229# http://stackoverflow.com/questions/2149368/how-can-i-loop-through-files-in-a-directory-in-perl
230$ENV{'CLASSPATH'} = "";
231my $jar_lib_path = $ENV{'TASK_HOME'}.$sep."lib";
232my @files = <$jar_lib_path/*.jar>; # /full/path/to/diffcol/lib/*jar
233foreach my $file (@files) {
234 $file =~ s@\/@\\@g if $isWin;
235 $ENV{'CLASSPATH'}=$file.$pathsep.$ENV{'CLASSPATH'};
236}
237##print STDERR "**** classpath: ".$ENV{'CLASSPATH'}."\n";
238
239
240#set the location of the full report
241my $xmlout=filename_concat($ENV{'DATA_DIR'}, "full-report-$dateid.xml");
242##print STDERR "XML: $xmlout\n";
243
244# the toplevel folder of the greenstone installation being used
245my $greenstone_home="";
246# gsdl is the checkout folder and can be greenstone2 or greenstone3
247my $gsdl="greenstone2";
248
249
250# Check if using existing compiled-up greenstone installation
251# and set the greenstone_home location accordingly
252
253if(defined $ENV{'GSDL3SRCHOME'} || defined $ENV{'GSDLHOME'}) {
254 print STDERR "Found existing Greenstone home, will use that instead\n";
255 $greenstone_home=$ENV{'GSDLHOME'};
256} else {
257 $greenstone_home=filename_concat($ENV{'DATA_DIR'}, $gsdl);
258}
259##print STDERR "GSHOME: $greenstone_home\n";
260
261#do the requested action
262if($action eq "setup_greenstone") {
263 &setup_greenstone;
264}
265elsif ($action eq "run_test") {
266 &run_test;
267}
268elsif ($action eq "summarise") {
269 &summarise;
270}
271elsif ($action eq "upload") {
272 &upload;
273 &mail_with_report_attached;
274}
275elsif ($action eq "all") {
276 &setup_greenstone;
277 &run_test;
278 &summarise;
279 &upload;
280 &mail_with_report_attached;
281}
282
283##********************************
284
285sub printusage
286{
287# print STDERR "Run as: $0 (help|setup_greenstone|run_test <--modelOS windows|darwin|linux> <--justdiff|--nosvn> <col1 col2 ...> |summarise|upload|all)\n";
288 print STDERR "Run as: $0 (help|setup_greenstone|run_test|summarise|upload|all)\n";
289 print STDERR "where run_test can further take the following optional parameters:\n";
290 print STDERR "\t--modelOS (windows|darwin|linux)\n";
291 print STDERR "\t--justdiff|--nosvn\n";
292 print STDERR "\t<col1 col2 ...>\n";
293 print STDERR "where setup_greenstone can further take the following optional parameters:\n";
294 print STDERR "\t--gs2|--gs3\n";
295 print STDERR "\t--svn|--bin\n";
296}
297
298#http://stackoverflow.com/questions/7427262/read-a-file-and-save-it-in-variable-using-shell-script
299
300sub setup_greenstone
301{
302 #clean up from previous tests
303 print STDERR "about to clean up any old tests (Ctrl-C to cancel)"; # no newline
304 for my $i ( 1..5 ) {
305 sleep 1; # 1 second
306 print STDERR ".";
307 }
308 print STDERR "\n";
309
310 # http://perldoc.perl.org/File/Path.html
311 print STDERR "cleaning up previous tests\n";
312 &File::Path::remove_tree($ENV{'DATA_DIR'});
313
314 print STDERR "creating the data dir\n";
315 &File::Path::make_path($ENV{'DATA_DIR'}); # works like mkdir -p
316
317 chdir($ENV{'DATA_DIR'});
318
319 # use existing compiled-up greenstone installation, if a GSDLHOME set
320 if(defined $ENV{'GSDL3SRCHOME'} || defined $ENV{'GSDLHOME'}) {
321 print STDERR "Found existing Greenstone home, will use that instead\n";
322 return;
323 }
324
325 # Else checkout a GS from svn into DATA_DIR
326
327 #svn checkout of main gsdl directory
328 print STDERR "checkout $gsdl:\n";
329 my $cmd = "svn co ".$ENV{'SVN_OPT_REV'}." http://svn.greenstone.org/main/trunk/greenstone2 $gsdl";
330 ##print STDERR "Checkout CMD: $cmd\n";
331
332 # # unlike backticks operator, system() will print the output of the command to the screen as it executes
333 # http://stackoverflow.com/questions/758611/how-to-flush-output-in-backticks-in-perl?rq=1
334 my $status = system "$cmd"; #my $status = `$cmd`;
335 if($status != 0) {
336 print STDERR "@@@ SVN checkout of $gsdl failed\n";
337 exit -1;
338 }
339 print STDERR "done\n";
340
341 ##print STDERR "$ENV{'DATA_DIR'}$sep$gsdl\n";
342
343 chdir("$ENV{'DATA_DIR'}$sep$gsdl");
344
345 ##print STDERR "@@@ OS: $^O.|".$Config{'archname64'}."|\n";
346
347 if($isWin) {
348 print STDERR "Compiling $gsdl using makegs2.bat running in auto (silent) mode\n";
349
350 # we're now in the GS2 folder, call makegs2 with silent param
351 $cmd = "makegs2.bat silent 2>> $ENV{'DATA_DIR'}/compilation-errors"; # STDERR is sent to compilation-errors file
352 $status = system $cmd;
353 if($status != 0) {
354 print STDERR "Greenstone compilation on Windows failed\n";
355 exit -1;
356 }
357
358 } else { # if we're on linux/darwin, need gnome-lib for the correct architecture. And need imagemagick to build imgs in collections
359
360 my $bit_arch=`uname -m`;
361
362 # imagmagick binary
363 print STDERR "Getting imagemagick binary\n";
364
365 my $os = $isMac ? "darwin" : "linux";
366 my $imagickzip = "imagemagick-$os";
367
368 if($isMac) {
369# $imagickzip .= "-10.5.tar.gz";
370 # at present, only the Imagemagick binaries created by Max for darwin work on the Macs
371 &File::Path::make_path($ENV{'DATA_DIR'}."$sep$gsdl$sep$os"); # need to ensure gsdl/bin/darwin exists
372 $cmd = "svn export http://svn.greenstone.org/main/trunk/binaries/mac/intel/imagemagick bin/darwin/imagemagick";
373 $status = system($cmd);
374 if($status != 0) {
375 print STDERR "@@@ Unable to get imagemagick for darwin\n";
376 }
377
378 # need ghostscript mac binary too for pdf to img conversions on mac
379 $cmd = "svn export http://svn.greenstone.org/main/trunk/binaries/mac/intel/ghostscript bin/darwin/ghostscript";
380 $status = system($cmd);
381 if($status != 0) {
382 print STDERR "@@@ Unable to get ghostscript for darwin\n";
383 }
384
385 # the imagemagick and ghostscript binaries have been set to executable on svn trac now
386# system("chmod -R u+x $ENV{'DATA_DIR'}/$gsdl/bin/darwin/imagemagick/bin/*");
387# system("chmod -R u+x $ENV{'DATA_DIR'}/$gsdl/bin/darwin/ghostscript/bin/*");
388 } else { # linux
389 my $extension64 = ($bit_arch =~ m/64$/) ? "-x64" : "";
390 $imagickzip .= "$extension64.tar.gz";
391
392 # now these next imagemagick steps (and those near the end of this sub) are just for linux, no longer also for mac
393 $cmd = "svn export http://svn.greenstone.org/gs2-extensions/imagemagick/trunk/$imagickzip ext/$imagickzip";
394 $status = system ($cmd);
395 system("cd ext && tar -xvzf $imagickzip");
396 }
397
398 # gnomelib binary
399 print STDERR "setting up gnome-lib-minimal for compilation\n";
400
401 # To get gnome-lib, need to determine bit architecture of the linux/darwin
402 # http://stackoverflow.com/questions/8963400/the-correct-way-to-read-a-data-file-into-an-array
403 # $Config{'archname64'} doesn't work on the Ubuntu and the Sys::Info package seems to not be supported
404 # well on ActivePerl.
405 # But since we know we're on a Linux/Darwin machine at this point, wecan just run `uname -m` and other linux cmds
406
407 # osversion will be "Lion" or ""
408 # and assuming all darwin is intel, not ppc!!
409 my $gnome_lib_file = $isMac ? "darwin-".$osversion."intel" : "linux";
410
411 $gnome_lib_file .= "-x64" if($bit_arch =~ m/64$/ && !$isMac); # linux only case
412
413 #svn checkout gnome-lib for this linux/darwin
414 chdir("$ENV{'DATA_DIR'}$sep$gsdl$sep"."ext"); #cd $DATA_DIR/$gsdl/ext
415
416 ##print STDERR "**** gnomelib: $gnome_lib_file\n";
417
418 # checkout and unpack gnome-lib-minimal
419
420 #svn export http://svn.greenstone.org/gs2-extensions/gnome-lib/trunk/gnome-lib-minimal-linux-x64.tar.gz gl.tar.gz
421 $cmd = "svn export http://svn.greenstone.org/gs2-extensions/gnome-lib/trunk/gnome-lib-minimal-".$gnome_lib_file.".tar.gz gl.tar.gz";
422 system $cmd;
423 system ("tar -xvzf gl.tar.gz");
424
425 chdir("gnome-lib-minimal");
426 ##print STDERR "*** ARCH: $bit_arch\n";
427
428 # need to run source devel.bash on gnome-lib followed by configure, make, make install
429 # in one go, in order to preserve the compile environment set up by sourcing devel.bash
430
431 # http://stackoverflow.com/questions/7369145/activating-a-virtualenv-using-a-shell-script-doesnt-seem-to-work
432 # http://ubuntuforums.org/showthread.php?t=1932504 linking /bin/sh to bash instead of dash
433
434# $cmd = "bash -c \"source ./devel.bash && cd ../.. && ./configure --enable-apache-httpd && make && make install\"";
435 $cmd = "bash -c \"";
436
437 $cmd .= "source ./devel.bash";
438 $cmd .= " && cd ../..";
439
440 #configure
441 # $cmd .= " && ./configure";
442 $cmd .= " && echo 'configure $gsdl: ' ";
443 $cmd .= " && echo '<configure>' >> $xmlout";
444 $cmd .= " && ./configure 2>> $ENV{'DATA_DIR'}/compilation-errors"; # configure
445 $cmd .= " && echo '</configure>' >> $xmlout";
446 $cmd .= " && echo 'done'";
447
448 #make
449 $cmd .= " && echo 'make $gsdl: '";
450 $cmd .= " && echo '<make>' >> $xmlout";
451 $cmd .= " && make 2>> $ENV{'DATA_DIR'}/compilation-errors"; # make
452 $cmd .= " && echo '</make>' >> $xmlout";
453 $cmd .= " && echo 'done'";
454
455 #make install
456 $cmd .= " && echo 'make install $gsdl: '";
457 $cmd .= " && echo '<make-install>' >> $xmlout";
458 $cmd .= " && make install 2>> $ENV{'DATA_DIR'}/compilation-errors"; # make install
459 $cmd .= " && echo '</make-install>' >> $xmlout";
460 $cmd .= " && echo 'done'";
461
462 $cmd .= "\""; # close off cmd to bash and run it
463 $status = system $cmd;
464
465 if(!$isMac) { # Linux
466 # Moving imagemagick after instead of before compilation, since bin/darwin gets overwritten during compilation
467 move("$ENV{'DATA_DIR'}$sep$gsdl$sep"."ext/imagemagick/$os", "$ENV{'DATA_DIR'}$sep$gsdl$sep"."bin/$os/imagemagick"); # http://www.perlmonks.org/?node_id=586537
468 unlink "$ENV{'DATA_DIR'}/$gsdl"."/ext/$imagickzip" or warn "Could not unlink ext/$imagickzip: $!";
469 &File::Path::remove_tree("$ENV{'DATA_DIR'}$sep$gsdl$sep"."ext/imagemagick"); # the untarred parent folder
470 }
471
472 }
473
474 if($status != 0) {
475 print STDERR "@@@ Compilation of Greenstone on Linux/Mac failed\n";
476 exit -1;
477 }
478
479 &getIsisGdl("$ENV{'DATA_DIR'}/$gsdl");
480
481 # set the path to the greenstone_home variable
482 $greenstone_home="$ENV{'DATA_DIR'}$sep$gsdl";
483
484}
485
486sub getPDFBox
487{
488 # current revision is 27763, but using "head" works
489 my $PDFBOX_TRAC_URL="http://trac.greenstone.org/export/head/gs2-extensions/pdf-box/trunk/pdf-box-java"; # both for .zip and .tar.gz extension
490 #"http://trac.greenstone.org/export/".$ENV{'SVN_OPT_REV'}."/gs2-extensions/pdf-box/trunk/pdf-box-java";
491
492 # now get the PDFBox extension for PDFBox tutorial
493 print STDERR "Getting pdfbox from $PDFBOX_TRAC_URL:\n";
494
495 chdir($greenstone_home);
496 my $cmd = "";
497 if ($isWin) {
498 $cmd = "setup.bat && cd ext && wget $PDFBOX_TRAC_URL.zip && unzip pdf-box-java.zip";
499
500 } elsif ($isMac) { # need to use curl not wget
501 $cmd = "cd ext && curl $PDFBOX_TRAC_URL.tar.gz > pdf-box-java.tar.gz && tar -xzf pdf-box-java.tar.gz";
502 }
503 else { # linux
504 $cmd = "bash -c \"export GSDLHOME=&& source setup.bash && cd ext && wget $PDFBOX_TRAC_URL.tar.gz && tar -xzf pdf-box-java.tar.gz\"";
505 }
506 my $status = system $cmd;
507 if($status != 0) {
508 print STDERR "@@@ Failed to set up PDFBox\n";
509 exit -1; # or proceed to testing other tutorials?
510 }
511}
512
513sub getOpenOfficeExt
514{
515 # current revision is 27763, but using "head" works
516 my $OOEXT_TRAC_URL="http://trac.greenstone.org/export/head/gs2-extensions/open-office/trunk/open-office-java"; # both for .zip and .tar.gz extension
517 #"http://trac.greenstone.org/export/".$ENV{'SVN_OPT_REV'}."/gs2-extensions/open-office/trunk/open-office-java";
518
519 # now get the OpenOffice extension for the AllDocTypes collection
520 print STDERR "Getting open office ext from $OOEXT_TRAC_URL:\n";
521
522 chdir($greenstone_home);
523 my $cmd = "";
524 if ($isWin) {
525 $cmd = "setup.bat && cd ext && wget $OOEXT_TRAC_URL.zip && unzip open-office-java.zip";
526
527 } elsif ($isMac) { # need to use curl not wget
528 $cmd = "cd ext && curl $OOEXT_TRAC_URL.tar.gz > open-office-java.tar.gz && tar -xzf open-office-java.tar.gz";
529 }
530 else { # linux
531 $cmd = "bash -c \"export GSDLHOME=&& source setup.bash && cd ext && wget $OOEXT_TRAC_URL.tar.gz && tar -xzf open-office-java.tar.gz\"";
532 }
533 my $status = system $cmd;
534 if($status != 0) {
535 print STDERR "@@@ Failed to set up the Open Office Extension\n";
536 exit -1; # or proceed to testing other tutorials?
537 }
538}
539
540sub getIsisGdl {
541 my $gsfolder = shift(@_);
542
543 if(!$isWin) {
544 my $bit_arch=`uname -m`;
545 if ($bit_arch =~ m/64$/) {
546 my $cmd = "";
547 if($isMac) {
548 $cmd = "cd $gsfolder/bin/darwin && curl http://www.greenstone.org/caveat-emptor/IsisGdl.macleopard > IsisGdl && chmod u+x IsisGdl";
549 } else { # linux
550 $cmd = "cd $gsfolder/bin/linux && wget http://www.greenstone.org/caveat-emptor/IsisGdl.bin32 && mv IsisGdl.bin32 IsisGdl && chmod u+x IsisGdl";
551 }
552 my $isis_status = system $cmd;
553 if($isis_status != 0) {
554 print STDERR "Unable to get IsisGdl from caveat page\n";
555 }
556 }
557 }
558}
559
560# http://stackoverflow.com/questions/3377879/how-do-i-receive-command-output-immediately
561sub run_test
562{
563 my $num_cols = scalar(@collections); # remember the empty case
564
565 if($num_cols == 0) { # deal with all collections
566 push (@collections, "");
567 # putting the empty string in the array so that the "all collections" case
568 # can be handled similar to how the case of user-specified collections is handled
569
570 } else { # deal with user specified set of collections
571 # prefix the directory separator to each collection name
572 @collections = map { $sep.$_ } @collections;
573 }
574
575 my $pdfbox = &filename_concat($greenstone_home, "ext", "pdf-box");
576 if(!-d $pdfbox) {
577 &getPDFBox();
578 }
579
580 my $openofficeext = &filename_concat($greenstone_home, "ext", "open-office");
581 if(!-d $openofficeext) {
582 &getOpenOfficeExt();
583 }
584
585 #&getIsisGdl("$greenstone_home");
586
587 open (my $xml_fh, '>'.$xmlout) || die "Could not open xml file $xmlout for appending: $!\n";
588
589 # perform the requested subcommands, outputting xml information
590 print $xml_fh "<test time=\"$dateid\" id=\"$dateid\">\n";
591
592 my ($cmd, $status);
593 # make sure that diffcol/model-collect is up to date before copying it over to greenstone-home
594
595 if(!$use_local_rebuild) {
596 print $xml_fh "Updating $ENV{'TASK_HOME'}/model-collect:\n";
597 for my $col (@collections) {
598 $cmd = "svn up $ENV{'TASK_HOME'}/model-collect$col"; #chdir("$ENV{'TASK_HOME'}/model-collect");
599 $status = system "$cmd";
600 }
601 }
602
603 # go to whichever greenstone_home we're using
604 chdir($greenstone_home);
605
606 # get svn info
607 print STDERR "getting svn info: $xmlout\n";
608 print $xml_fh "<svn-info>\n";
609 &run_and_print_cmd("svn info", $xml_fh);
610 print $xml_fh "</svn-info>\n";
611 print STDERR "done\n";
612
613 if(!$use_local_rebuild) {
614
615 #make two copies of the model-collect directory in gsdl
616 #one to be rebuilt and one as the basis for comparison
617 #strip both of all .svn directories
618
619 #copy the model collections to the collect folder to be rebuilt
620 print STDERR "installing test collections and model collections to new $gsdl installation... ";
621
622 #clean up
623 if(-d "collect") {
624 for my $col (@collections) {
625 if(-d "collect$col") {
626 &File::Path::remove_tree("collect$col") || die "Error could not delete collect: $!";
627 }
628 }
629 }
630
631 if($use_static_model) {
632 for my $col (@collections) {
633 #copy to collect and strip .svn subfolders
634 &File::Path::make_path("collect$col"); # create the collect folder and copy contents from static model-collection across
635 &copy_recursively("model-collect$col", "collect$col", ".svn");
636 }
637
638 } else { # the default situation: where we check out the model-collect from svn
639 for my $col (@collections) {
640 &File::Path::remove_tree("model-collect$col");
641
642 #copy to collect and strip .svn subfolders
643 &File::Path::make_path("collect$col"); # create the folder and copy contents across
644 &copy_recursively(&filename_concat("$ENV{'TASK_HOME'}","model-collect$col"), "collect$col", ".svn");
645
646 #make the model copy
647 &File::Path::make_path("model-collect$col");
648 &copy_recursively("collect$col", "model-collect$col"); # copy contents across
649 }
650 }
651
652 print STDERR "done\n";
653 }
654
655 #for each collection, import, build and diff with its model counterpart
656
657 # if working with all collections, read the list of collections from the folders in collect
658 if($num_cols == 0) {
659 @collections = (); # get rid of the empty string put in the array to represent "all collections"
660
661 opendir my($collect_handle), "collect" or die "Could not open dir $greenstone_home/collect: $!";
662
663 for my $collection (readdir $collect_handle) {
664 next if ($collection eq "." || $collection eq "..");
665 next if ($collection eq "modelcol");
666 push(@collections, $collection);
667 }
668 closedir $collect_handle; # close handle to collect dir
669 }
670
671 for my $collection (@collections) {
672
673 # next if ($collection ne "Demo-Lucene"); ## TEMPORARY, FOR TESTING THIS SCRIPT
674 # next if ($collection !~ m/OAI|METS|DSpace|MGPP|Lucene/); ## TEMPORARY, FOR TESTING THIS SCRIPT
675
676 #escape the filename (in case of space)
677 $collection =~ s@ @\\ @g;
678 #getting just the basename of the collection would have been necessary had we not cd-ed into $gsdl
679
680 $collection =~ s@^[\\/]@@g; # take the dir-sep prefix away again for user-specified collection names
681
682 if (! -d "collect$sep$collection") {
683 print STDERR "Collection $collection does not exist\n";
684 next;
685 }
686
687 print STDERR "*** Found collection $collection\n";
688 print $xml_fh "<collection-test name=\"$collection\">\n";
689
690 if(!$use_local_rebuild) {
691 #import
692 # Ensure the OIDtype for importing is hash_on_full_filename
693 # "to make document identifiers more stable across upgrades of the software,
694 # although it means that duplicate documents contained in the collection are
695 # no longer detected automatically."
696 print STDERR "$collection - Importing:\n";
697 print $xml_fh "<import>\n";
698 &run_build_script("import.pl -removeold $collection"); #-OIDtype hash_on_full_filename
699 print $xml_fh "</import>\n";
700 print STDERR "done\n";
701
702 #build
703 print STDERR "$collection - Building:\n";
704 print $xml_fh "<build>\n";
705 &run_build_script("buildcol.pl -removeold $collection");
706 print $xml_fh "</build>\n";
707 print STDERR "done\n";
708
709 #rename the intermediate 'building' directory 'index'
710 print STDERR "$collection - Move \"building\" to \"index\"... ";
711 my $index = &filename_concat("collect", $collection, "index");
712 my $building = &filename_concat("collect", $collection, "building");
713 &File::Path::remove_tree($index);
714 # Renaming Directories, http://www.perlmonks.org/?node_id=177421
715 move($building, $index) or die "copy failed: $!"; # File::Copy::move
716 print STDERR "done\n";
717 }
718 #diffcol
719 print STDERR "$collection - Diffing:\n";
720 my $diffcol_dir = &filename_concat($ENV{'TASK_HOME'},"diffcol");
721
722 # help diffcol to know on what os the model cols were generated
723 # and what os this test machine is (on which the test cols will be generated)
724 $cmd = "diffcol.pl -testos $test_os -modelos $model_os -output xml -verbosity 10 $collection"; # need to run with ./diffcol.pl if bash script
725 &run_diff_script($cmd, $xml_fh, $diffcol_dir);
726
727 chdir($greenstone_home); # this is actually where we are
728 print STDERR "done\n";
729 print $xml_fh "</collection-test>\n";
730 }
731
732 print $xml_fh "</test>\n";
733 close($xml_fh);
734
735 print STDERR "done\n";
736}
737
738##***************************************************************
739# runs setup in greenstone_home before running the diff command
740sub run_diff_script {
741 my ($cmd, $fh, $diffcol_dir) = @_;
742
743 # we're in greenstone_home now
744 if(!$isWin) {
745 $cmd = "bash -c \"export GSDLHOME=&& source $setup_script.bash && cd $diffcol_dir && ./$cmd\"";
746
747 } else { # Need to prefix cmd -c/-k as necessary
748 $cmd = "cmd /c \"set GSDLHOME=&& $setup_script.bat && cd $diffcol_dir && perl -S $cmd\"";
749## print STDERR "@@@@ Going to call command: $cmd\n";
750 }
751
752 return &run_and_print_cmd($cmd, $fh);
753}
754
755# runs setup in greenstone_home before running the given build command
756sub run_build_script {
757 my ($cmd, $fh) = @_;
758
759# chdir($greenstone_home);
760 # we are in $greenstone_home already, can directly run the build cmd on the collection
761 if(!$isWin) {
762 $cmd = "bash -c \"export GSDLHOME=&& source $setup_script.bash && $cmd\"";
763
764 } else { # Need to prefix cmd -c/-k as necessary
765 $cmd = "cmd /c \"set GSDLHOME=&& $setup_script.bat && perl -S $cmd\"";
766 }
767## print STDERR "@@@@ Going to call command: $cmd\n";
768
769 return system($cmd);
770 #return &run_and_print_cmd($cmd, $fh); # doesn't work on cmds chained with bash -c
771}
772
773
774# http://stackoverflow.com/questions/758611/how-to-flush-output-in-backticks-in-perl?rq=1activeperl%20sys::info
775# http://stackoverflow.com/questions/1477500/how-do-i-get-the-output-of-an-external-command-in-perl
776sub run_and_print_cmd {
777 my ($cmd, $fh) = @_;
778
779 open my $pin, "$cmd|" or die "unable to run cmd $cmd: $!"; # open(my $fh, '-|', 'powercfg -l') or die $!;
780
781 if(defined $fh) { # print cmd output both to the filehandle and to stdout
782 while (my $line = <$pin>) {
783 print $fh $line;
784# print STDOUT $line; # if also printing cmd output to STDOUT
785 }
786 }
787 else { # no filehandle, so just need to print to stdout
788
789 # unlike backticks operator, system() will print the output of the command to the screen as it executes
790 # http://stackoverflow.com/questions/758611/how-to-flush-output-in-backticks-in-perl?rq=1
791
792 my $status = system $cmd;
793 if($status != 0) {
794 print STDERR "ERROR ($status) running $cmd: $!\n";
795 }
796 }
797 close($pin);
798}
799
800sub filename_concat {
801 my $first_file = shift(@_);
802 my (@filenames) = @_;
803
804 # If first_file is not null or empty, then add it back into the list
805 if (defined $first_file && $first_file =~ /\S/)
806 {
807 unshift(@filenames, $first_file);
808 }
809
810 my $filename = join($sep, @filenames);
811 $filename =~ s/[\\\/]$//; # remove trailing slashes if any
812 return $filename;
813}
814
815
816# The following code is from
817# http://stackoverflow.com/questions/227613/how-can-i-copy-a-directory-recursively-and-filter-filenames-in-perl
818# It also states that "Perl's File::Copy is a bit broken (it doesn't copy permissions on Unix systems, for example)"
819sub copy_recursively {
820 my ($from_dir, $to_dir, $regex) = @_;
821 opendir my($dh), $from_dir or die "Could not open dir '$from_dir': $!";
822
823# if(-d !$to_dir) {
824# mkdir $to_dir or die "mkdir '$to_dir' failed: $!" if not -e $to_dir;
825# }
826
827 for my $entry (readdir $dh) {
828 next if ($entry eq "." || $entry eq "..");
829 next if (defined $regex && $entry =~ /$regex/);
830 my $source = "$from_dir/$entry";
831 my $destination = "$to_dir/$entry";
832 if (-d $source) {
833 mkdir $destination or die "mkdir '$destination' failed: $!" if not -e $destination;
834 copy_recursively($source, $destination, $regex);
835 } else {
836 copy($source, $destination) or die "copy failed: $!";
837 }
838 }
839 closedir $dh;
840 return;
841}
842
843sub summarise {
844
845 # make a summarised Xml report
846 print STDERR "Summarizing the xml report... ";
847 my $cmd = "java org.apache.xalan.xslt.Process -IN $xmlout -XSL $ENV{'TASK_HOME'}/xsl/xml-report.xsl -OUT $ENV{'DATA_DIR'}/report-$dateid.xml";
848 my $status = system($cmd);
849 print STDERR "done\n";
850
851 # make a summarised HTMl report
852 print STDERR "Creating an html summary report... ";
853 $cmd = "java org.apache.xalan.xslt.Process -IN $ENV{'DATA_DIR'}/report-$dateid.xml -XSL $ENV{'TASK_HOME'}/xsl/html-report.xsl -OUT $ENV{'DATA_DIR'}/report-$dateid.html";
854 $status = system($cmd);
855 print STDERR "done\n";
856
857 # Print whether the tests passed or failed
858 print STDERR "*******************************************\n";
859 print STDERR "Checking if successful... \n";
860 $cmd = "java org.apache.xalan.xslt.Process -IN $xmlout -XSL $ENV{'TASK_HOME'}/xsl/passed-or-not.xsl";
861 $status = `$cmd`; #$status = system($cmd);
862 print STDERR "result: $status\n";
863 print STDERR "*******************************************\n";
864}
865
866sub upload {
867 # if the upload dir already existed, clear it of contents
868 if (-d $ENV{'UPLOAD_DIR'}) { #else rm $UPLOAD_DIR/*
869 # don't want to keep previous days reports
870 # else we will have to manually clear them at some point
871 # just generate the set of reports for this run of task.pl upload
872 # and
873 &File::Path::remove_tree($ENV{'UPLOAD_DIR'});
874 }
875 # recreate the upload directory
876 &File::Path::make_path($ENV{'UPLOAD_DIR'});
877
878 # copy all *.xml and *.html files across to UPLOAD_DIR
879 opendir my($dh), $ENV{'DATA_DIR'} or die "Could not open DATA_DIR: $!";
880 for my $entry (readdir $dh) {
881 next if ($entry !~ m/(\.xml|\.html?)$/);
882
883 # copy the reports across with different names: with OS prefixed to them. And for the HTML file on Win, rename to HTM
884 # html files uploaded from windows to nzdl are empty for no reason. Uploading as htm seems to work
885 my $os_entry = $entry;
886 $os_entry =~ s@\[email protected]@ if $isWin;
887 if($isMac) {
888 $os_entry = "diffcol-".$^O."-".$osversion."$os_entry"; # darwin-Lion for Lion/Mountain Lion
889 } else {
890 $os_entry = "diffcol-".$^O."-$os_entry";
891 }
892
893 # get the absolute path to the original files before copying them over
894 $entry = &filename_concat($ENV{'DATA_DIR'}, $entry);
895
896 # copy them over with their new names
897## print STDERR "@@@@ copying across $entry to $ENV{'UPLOAD_DIR'} as $os_entry\n";
898 copy($entry, "$ENV{'UPLOAD_DIR'}$sep$os_entry"); #copy($entry, "$ENV{'UPLOAD_DIR'}");
899 }
900 closedir $dh;
901
902
903 # Upload the html file to puka
904 #default identity dir
905 if ( ! exists $ENV{'IDENTITY_DIR'} ) {
906 $ENV{'IDENTITY_DIR'} = "$ENV{'HOME'}${sep}.ssh"; # "C:\\Research\\Nightly\\tools\\keys" on windows, see environment.pl
907 }
908 if (! exists $ENV{'SNAPSHOT_MODE'} ) {
909 $ENV{'SNAPSHOT_MODE'} = "caveat";
910 }
911
912 #use the correct key for uploading
913 $ENV{'IDENTITY_FILE'} = "$ENV{'IDENTITY_DIR'}${sep}upload-" . $ENV{'SNAPSHOT_MODE'} . ($^O eq "MSWin32" ? ".ppk" : "");
914 if(-f $ENV{'IDENTITY_FILE'}) {
915 # if you need to touch the file on windows: http://stackoverflow.com/questions/51435/windows-version-of-the-unix-touch-command
916
917 # the report we want to upload is actually just os-diffcol-report-$dateid.html
918 my $command = "cd \"$ENV{'UPLOAD_DIR'}\" && tar -c *.htm* | "; #&& cat *.html | "; # && tar -c * |
919 $command .= ($^O eq "MSWin32" ? "plink" : "ssh");
920 $command .= " -T -i \"$ENV{'IDENTITY_FILE'}\" nzdl\@puka.cs.waikato.ac.nz";
921 #print "$command\n";
922 my $status = system("$command");
923 if($status != 0) {
924 print STDERR "*** Failed to upload test report to nzdl $status\n";
925 }
926 } else {
927 print STDERR "*** Cannot upload the test report to nzdl from this machine\n";
928 }
929
930 print STDERR "Finished uploading\n";
931}
932
933# Sending emails with perl: http://learn.perl.org/examples/email.html
934# Sending email attachments with perl: http://www.perlmonks.org/?node_id=19430
935# Sadly none of the packages are installed by default and use of MIME::Lite is discouraged
936sub mail_with_report_attached
937{
938 # email out with report attached, if the tests failed
939 print STDERR "Checking if successful... \n";
940 my $cmd = "java org.apache.xalan.xslt.Process -IN $xmlout -XSL $ENV{'TASK_HOME'}/xsl/passed-or-not.xsl";
941 #my $result = system($cmd);
942 my $result = `$cmd`;
943
944 print STDERR "result: $result\n";
945
946 if($result ne "yes") {
947 my $msg = "$gsdl regression test for $dateid failed";
948 my $subject = "Regression Test Failed"; #"$gsdl regression test for $dateid failed\n";
949 my $attach_file = &filename_concat($ENV{'DATA_DIR'}, "report-$dateid.html");
950
951 if($isWin) {
952 if($use_blat && $blat && $ENV{'GSDL_SMTP'}) {
953 # http://stackoverflow.com/questions/709635/sending-mail-from-batch-file
954 #blat -to [email protected] -server smtp.example.com -f [email protected] -subject "subject" -body "body"
955
956 # need to install blat on windows
957 $cmd = "$blat -to $ENV{'MONITOR_EMAIL'} -server $ENV{'GSDL_SMTP'} -f $ENV{'MONITOR_EMAIL'} -attach $attach_file -subject \"$subject\" -body \"$msg\"";
958 $result = system($cmd);
959 }
960 else {
961 $result = 1; # status from running mail command is 0 if success, 1 if fail
962 print STDERR "********************************************\n";
963 if ($use_blat) {
964 print STDERR "Need blat and SMTP set to send mail attachment\n" ;
965 } else {
966 print STDERR "Not set up to send mail on Windows\n";
967 }
968 print STDERR "Inspect report at: $attach_file\n";
969 print STDERR "********************************************\n";
970 }
971 } else { # linux
972 my $status = system("command -v mutt > /dev/null 2>&1;"); #better way of doing "which mutt"
973
974 if($status != 0) { # mutt doesn't exist, can't send attachments, so send simple email
975 $cmd="echo '$gsdl regression test for $dateid failed.' | mail -s 'Regression Test Failed' $ENV{'MONITOR_EMAIL'}";
976
977 print STDERR "********************************************\n";
978 print STDERR "No mutt installed, unable to mail attachment\n";
979 print STDERR "Inspect report at: $attach_file\n";
980 print STDERR "********************************************\n";
981 } else {
982 #$cmd = "bash -c \"echo '$gsdl regression test for $dateid failed' | mutt -a $attach_file -s 'Regression Test Failed' -- $ENV{'MONITOR_EMAIL'}\"";
983 $cmd = "echo '$gsdl regression test for $dateid failed' | mutt -a $attach_file -s 'Regression Test Failed' -- $ENV{'MONITOR_EMAIL'}";
984 }
985
986 # run the mail command
987 $result = system($cmd); #&run_and_print_cmd($cmd);
988 }
989
990
991 if($result != 0) {
992 print STDERR "*** Unable to send email: $?\n";
993 }
994 else {
995 print STDERR "Sent mail with report attached.\n";
996 }
997 } else {
998 print STDERR "********************************************\n";
999 print STDERR "Tests were successful. Not sending mail.\n";
1000 print STDERR "********************************************\n";
1001 }
1002}
1003
1004# The old version of this program contained the following, consisting of 1 line of active code:
1005
1006 # Invoke as: sjmc@br:/research/sjm84/envi/bin$ ./envi diffcol summarise
1007 # Doing so will call this pl file and pass in "summarise" in ARGV
1008 # This pl file will in turn call the task executable in this folder
1009 # passing in "summarise" as a parameter.
1010#system("/bin/bash -c \"../etc/tasks/diffcol/task @ARGV\"");
1011
1012 ##system("/bin/bash -c \"./task @ARGV\"");
1013 ##print STDERR "/bin/bash -c ../etc/tasks/diffcol/task @ARGV"
1014
Note: See TracBrowser for help on using the repository browser.