source: other-projects/nightly-tasks/diffcol/trunk/task.pl@ 28616

Last change on this file since 28616 was 28616, checked in by sjm84, 10 years ago

Beginnings of adding further cmdline args (svn/bin, gs2/gs3).

  • 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 = ();
115
116# run_test can take any number of args
117if(scalar(@ARGV) > 1 && $ARGV[0] !~ m/run_test|setup_greenstone/) {
118 print STDERR "**** Wrong number of arguments\n";
119 &printusage();
120 exit -1;
121}
122
123if(scalar(@ARGV) == 0) {
124 $action="all";
125}
126else {
127 switch ($ARGV[0]) {
128 case qr/^(-h|--?help|help)$/i { &printusage; exit 0; }
129 case qr/^(setup_greenstone|run_test|summarise|upload|all)$/ { $action=$ARGV[0]; }
130 else {
131 print STDERR "**** Bad subcommand.\n";
132 &printusage;
133 exit -1;
134 }
135 }
136
137 # setup_greenstone can take gs2|gs3 and bin|svn. gs2 and svn by default
138
139 # run_test action can take a subaction: nosvn|justdiff. It can also take --modelOS (windows|linux|darwin)
140 # nosvn: uses the model-collect as static and copies it over to collect, rebuilding what's currently in model-collect instead of copying
141 # it out from the svn model-collect again.
142 # 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.
143 if($action eq "setup_greenstone" && scalar(@ARGV) >= 2) {
144
145 for (my $i=0; $i < scalar(@ARGV); $i++) {
146 if($ARGV[$i] =~ m@^--@) {
147 $subaction = $ARGV[$i];
148 if($subaction =~ m/--gs(2|3)/) {
149 $install_version = $1;
150 } elsif ($subaction =~ m/--(bin|svn)/) {
151 $install_type = $1;
152 } else {
153 print STDERR "**** Bad subaction: ".$ARGV[$i]."\n";
154 &printusage;
155 exit -1;
156 }
157 }
158 }
159
160 #print STDERR "Install type $install_type\n";
161 #print STDERR "Install version $install_version\n";
162
163 } elsif($action eq "run_test" && scalar(@ARGV) >= 2) {
164 push(@collections, @ARGV);
165 shift @collections; # remove action from array
166
167 for (my $i=0; $i < scalar(@ARGV); $i++) {
168 if($ARGV[$i] =~ m@^--@) {
169 shift @collections; # remove subaction/flag from array
170
171 $subaction = $ARGV[$i];
172 if($subaction eq "--justdiff") {
173 $use_local_rebuild = $use_static_model = 1;
174 } elsif ($subaction eq "--nosvn") {
175 $use_static_model = 1;
176 #} elsif ($subaction =~ m/\-\-testOS/i && defined $ARGV[$i+1]) {
177 # $test_os = $ARGV[$i+1];
178 # $i++;
179 # shift @collections; # remove test_os value from array
180 } elsif ($subaction =~ m/\-\-modelOS/i && defined $ARGV[$i+1] && $ARGV[$i+1] =~ m/windows|linux|darwin/i) {
181 $model_os = $ARGV[$i+1];
182 $i++;
183 shift @collections; # remove model_os value from array
184 #print STDERR "Model_os specified: $model_os\n";
185 } else {
186 print STDERR "**** Bad subaction/value: ".$ARGV[$i]."\n";
187 &printusage;
188 exit -1;
189 }
190 }
191 }
192
193# foreach my $col (@collections) {
194# print STDERR "Collection: $col\n";
195# }
196 }
197}
198
199
200#check key environment vars are set
201if(!defined $ENV{'UPLOAD_DIR'}) {
202 print STDERR "Please set a UPLOAD_DIR for the test in an environment.sh file\n";
203 #return 1;
204}
205if(!defined $ENV{'DATA_DIR'}) {
206 print STDERR "Please set a DATA_DIR for the test in an environment.sh file\n";
207 #return 1;
208}
209if(!defined $ENV{'MONITOR_EMAIL'}) {
210 print STDERR "Please set a MONITOR_EMAIL for the test in an environment.sh file\n";
211 #return 1;
212}
213
214if($ENV{'DATA_DIR'} eq "/") {
215 print STDERR "DATA_DIR should not be the fs root\n";
216 #return 1;
217}
218
219print STDERR "DATA_DIR: ".$ENV{'DATA_DIR'}."\n";
220print STDERR "UPLOAD_DIR: ".$ENV{'UPLOAD_DIR'}."\n";
221
222#create an id for this test
223my ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = localtime(time);
224$year += 1900;
225$mon += 1;
226$mon = "0$mon" if ($mon < 10);
227$mday = "0$mday" if ($mday < 10);
228my $dateid="$year.$mon.$mday"; #my $dateid=($year+1900)."-".($mon+1)."-$mday";
229
230print STDERR "Starting test '$dateid'\n";
231
232
233# http://stackoverflow.com/questions/2149368/how-can-i-loop-through-files-in-a-directory-in-perl
234$ENV{'CLASSPATH'} = "";
235my $jar_lib_path = $ENV{'TASK_HOME'}.$sep."lib";
236my @files = <$jar_lib_path/*.jar>; # /full/path/to/diffcol/lib/*jar
237foreach my $file (@files) {
238 $file =~ s@\/@\\@g if $isWin;
239 $ENV{'CLASSPATH'}=$file.$pathsep.$ENV{'CLASSPATH'};
240}
241##print STDERR "**** classpath: ".$ENV{'CLASSPATH'}."\n";
242
243
244#set the location of the full report
245my $xmlout=filename_concat($ENV{'DATA_DIR'}, "full-report-$dateid.xml");
246##print STDERR "XML: $xmlout\n";
247
248# the toplevel folder of the greenstone installation being used
249my $greenstone_home="";
250# gsdl is the checkout folder and can be greenstone2 or greenstone3
251my $gsdl="greenstone2";
252
253
254# Check if using existing compiled-up greenstone installation
255# and set the greenstone_home location accordingly
256
257if(defined $ENV{'GSDL3SRCHOME'} || defined $ENV{'GSDLHOME'}) {
258 print STDERR "Found existing Greenstone home, will use that instead\n";
259 $greenstone_home=$ENV{'GSDLHOME'};
260} else {
261 $greenstone_home=filename_concat($ENV{'DATA_DIR'}, $gsdl);
262}
263##print STDERR "GSHOME: $greenstone_home\n";
264
265#do the requested action
266if($action eq "setup_greenstone") {
267 &setup_greenstone;
268}
269elsif ($action eq "run_test") {
270 &run_test;
271}
272elsif ($action eq "summarise") {
273 &summarise;
274}
275elsif ($action eq "upload") {
276 &upload;
277 &mail_with_report_attached;
278}
279elsif ($action eq "all") {
280 &setup_greenstone;
281 &run_test;
282 &summarise;
283 &upload;
284 &mail_with_report_attached;
285}
286
287##********************************
288
289sub printusage
290{
291# print STDERR "Run as: $0 (help|setup_greenstone|run_test <--modelOS windows|darwin|linux> <--justdiff|--nosvn> <col1 col2 ...> |summarise|upload|all)\n";
292 print STDERR "Run as: $0 (help|setup_greenstone|run_test|summarise|upload|all)\n";
293 print STDERR "where run_test can further take the following optional parameters:\n";
294 print STDERR "\t--modelOS (windows|darwin|linux)\n";
295 print STDERR "\t--justdiff|--nosvn\n";
296 print STDERR "\t<col1 col2 ...>\n";
297 print STDERR "where setup_greenstone can further take the following optional parameters:\n";
298 print STDERR "\t--gs2|--gs3\n";
299 print STDERR "\t--svn|--bin\n";
300}
301
302#http://stackoverflow.com/questions/7427262/read-a-file-and-save-it-in-variable-using-shell-script
303
304sub setup_greenstone
305{
306 #clean up from previous tests
307 print STDERR "about to clean up any old tests (Ctrl-C to cancel)"; # no newline
308 for my $i ( 1..5 ) {
309 sleep 1; # 1 second
310 print STDERR ".";
311 }
312 print STDERR "\n";
313
314 # http://perldoc.perl.org/File/Path.html
315 print STDERR "cleaning up previous tests\n";
316 &File::Path::remove_tree($ENV{'DATA_DIR'});
317
318 print STDERR "creating the data dir\n";
319 &File::Path::make_path($ENV{'DATA_DIR'}); # works like mkdir -p
320
321 chdir($ENV{'DATA_DIR'});
322
323 # use existing compiled-up greenstone installation, if a GSDLHOME set
324 if(defined $ENV{'GSDL3SRCHOME'} || defined $ENV{'GSDLHOME'}) {
325 print STDERR "Found existing Greenstone home, will use that instead\n";
326 return;
327 }
328
329 # Else checkout a GS from svn into DATA_DIR
330
331 #svn checkout of main gsdl directory
332 print STDERR "checkout $gsdl:\n";
333 my $cmd = "svn co ".$ENV{'SVN_OPT_REV'}." http://svn.greenstone.org/main/trunk/greenstone2 $gsdl";
334 ##print STDERR "Checkout CMD: $cmd\n";
335
336 # # unlike backticks operator, system() will print the output of the command to the screen as it executes
337 # http://stackoverflow.com/questions/758611/how-to-flush-output-in-backticks-in-perl?rq=1
338 my $status = system "$cmd"; #my $status = `$cmd`;
339 if($status != 0) {
340 print STDERR "@@@ SVN checkout of $gsdl failed\n";
341 exit -1;
342 }
343 print STDERR "done\n";
344
345 ##print STDERR "$ENV{'DATA_DIR'}$sep$gsdl\n";
346
347 chdir("$ENV{'DATA_DIR'}$sep$gsdl");
348
349 ##print STDERR "@@@ OS: $^O.|".$Config{'archname64'}."|\n";
350
351 if($isWin) {
352 print STDERR "Compiling $gsdl using makegs2.bat running in auto (silent) mode\n";
353
354 # we're now in the GS2 folder, call makegs2 with silent param
355 $cmd = "makegs2.bat silent 2>> $ENV{'DATA_DIR'}/compilation-errors"; # STDERR is sent to compilation-errors file
356 $status = system $cmd;
357 if($status != 0) {
358 print STDERR "Greenstone compilation on Windows failed\n";
359 exit -1;
360 }
361
362 } else { # if we're on linux/darwin, need gnome-lib for the correct architecture. And need imagemagick to build imgs in collections
363
364 my $bit_arch=`uname -m`;
365
366 # imagmagick binary
367 print STDERR "Getting imagemagick binary\n";
368
369 my $os = $isMac ? "darwin" : "linux";
370 my $imagickzip = "imagemagick-$os";
371
372 if($isMac) {
373# $imagickzip .= "-10.5.tar.gz";
374 # at present, only the Imagemagick binaries created by Max for darwin work on the Macs
375 &File::Path::make_path($ENV{'DATA_DIR'}."$sep$gsdl$sep$os"); # need to ensure gsdl/bin/darwin exists
376 $cmd = "svn export http://svn.greenstone.org/main/trunk/binaries/mac/intel/imagemagick bin/darwin/imagemagick";
377 $status = system($cmd);
378 if($status != 0) {
379 print STDERR "@@@ Unable to get imagemagick for darwin\n";
380 }
381
382 # need ghostscript mac binary too for pdf to img conversions on mac
383 $cmd = "svn export http://svn.greenstone.org/main/trunk/binaries/mac/intel/ghostscript bin/darwin/ghostscript";
384 $status = system($cmd);
385 if($status != 0) {
386 print STDERR "@@@ Unable to get ghostscript for darwin\n";
387 }
388
389 # the imagemagick and ghostscript binaries have been set to executable on svn trac now
390# system("chmod -R u+x $ENV{'DATA_DIR'}/$gsdl/bin/darwin/imagemagick/bin/*");
391# system("chmod -R u+x $ENV{'DATA_DIR'}/$gsdl/bin/darwin/ghostscript/bin/*");
392 } else { # linux
393 my $extension64 = ($bit_arch =~ m/64$/) ? "-x64" : "";
394 $imagickzip .= "$extension64.tar.gz";
395
396 # now these next imagemagick steps (and those near the end of this sub) are just for linux, no longer also for mac
397 $cmd = "svn export http://svn.greenstone.org/gs2-extensions/imagemagick/trunk/$imagickzip ext/$imagickzip";
398 $status = system ($cmd);
399 system("cd ext && tar -xvzf $imagickzip");
400 }
401
402 # gnomelib binary
403 print STDERR "setting up gnome-lib-minimal for compilation\n";
404
405 # To get gnome-lib, need to determine bit architecture of the linux/darwin
406 # http://stackoverflow.com/questions/8963400/the-correct-way-to-read-a-data-file-into-an-array
407 # $Config{'archname64'} doesn't work on the Ubuntu and the Sys::Info package seems to not be supported
408 # well on ActivePerl.
409 # But since we know we're on a Linux/Darwin machine at this point, wecan just run `uname -m` and other linux cmds
410
411 # osversion will be "Lion" or ""
412 # and assuming all darwin is intel, not ppc!!
413 my $gnome_lib_file = $isMac ? "darwin-".$osversion."intel" : "linux";
414
415 $gnome_lib_file .= "-x64" if($bit_arch =~ m/64$/ && !$isMac); # linux only case
416
417 #svn checkout gnome-lib for this linux/darwin
418 chdir("$ENV{'DATA_DIR'}$sep$gsdl$sep"."ext"); #cd $DATA_DIR/$gsdl/ext
419
420 ##print STDERR "**** gnomelib: $gnome_lib_file\n";
421
422 # checkout and unpack gnome-lib-minimal
423
424 #svn export http://svn.greenstone.org/gs2-extensions/gnome-lib/trunk/gnome-lib-minimal-linux-x64.tar.gz gl.tar.gz
425 $cmd = "svn export http://svn.greenstone.org/gs2-extensions/gnome-lib/trunk/gnome-lib-minimal-".$gnome_lib_file.".tar.gz gl.tar.gz";
426 system $cmd;
427 system ("tar -xvzf gl.tar.gz");
428
429 chdir("gnome-lib-minimal");
430 ##print STDERR "*** ARCH: $bit_arch\n";
431
432 # need to run source devel.bash on gnome-lib followed by configure, make, make install
433 # in one go, in order to preserve the compile environment set up by sourcing devel.bash
434
435 # http://stackoverflow.com/questions/7369145/activating-a-virtualenv-using-a-shell-script-doesnt-seem-to-work
436 # http://ubuntuforums.org/showthread.php?t=1932504 linking /bin/sh to bash instead of dash
437
438# $cmd = "bash -c \"source ./devel.bash && cd ../.. && ./configure --enable-apache-httpd && make && make install\"";
439 $cmd = "bash -c \"";
440
441 $cmd .= "source ./devel.bash";
442 $cmd .= " && cd ../..";
443
444 #configure
445 # $cmd .= " && ./configure";
446 $cmd .= " && echo 'configure $gsdl: ' ";
447 $cmd .= " && echo '<configure>' >> $xmlout";
448 $cmd .= " && ./configure 2>> $ENV{'DATA_DIR'}/compilation-errors"; # configure
449 $cmd .= " && echo '</configure>' >> $xmlout";
450 $cmd .= " && echo 'done'";
451
452 #make
453 $cmd .= " && echo 'make $gsdl: '";
454 $cmd .= " && echo '<make>' >> $xmlout";
455 $cmd .= " && make 2>> $ENV{'DATA_DIR'}/compilation-errors"; # make
456 $cmd .= " && echo '</make>' >> $xmlout";
457 $cmd .= " && echo 'done'";
458
459 #make install
460 $cmd .= " && echo 'make install $gsdl: '";
461 $cmd .= " && echo '<make-install>' >> $xmlout";
462 $cmd .= " && make install 2>> $ENV{'DATA_DIR'}/compilation-errors"; # make install
463 $cmd .= " && echo '</make-install>' >> $xmlout";
464 $cmd .= " && echo 'done'";
465
466 $cmd .= "\""; # close off cmd to bash and run it
467 $status = system $cmd;
468
469 if(!$isMac) { # Linux
470 # Moving imagemagick after instead of before compilation, since bin/darwin gets overwritten during compilation
471 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
472 unlink "$ENV{'DATA_DIR'}/$gsdl"."/ext/$imagickzip" or warn "Could not unlink ext/$imagickzip: $!";
473 &File::Path::remove_tree("$ENV{'DATA_DIR'}$sep$gsdl$sep"."ext/imagemagick"); # the untarred parent folder
474 }
475
476 }
477
478 if($status != 0) {
479 print STDERR "@@@ Compilation of Greenstone on Linux/Mac failed\n";
480 exit -1;
481 }
482
483 &getIsisGdl("$ENV{'DATA_DIR'}/$gsdl");
484
485 # set the path to the greenstone_home variable
486 $greenstone_home="$ENV{'DATA_DIR'}$sep$gsdl";
487
488}
489
490sub getPDFBox
491{
492 # current revision is 27763, but using "head" works
493 my $PDFBOX_TRAC_URL="http://trac.greenstone.org/export/head/gs2-extensions/pdf-box/trunk/pdf-box-java"; # both for .zip and .tar.gz extension
494 #"http://trac.greenstone.org/export/".$ENV{'SVN_OPT_REV'}."/gs2-extensions/pdf-box/trunk/pdf-box-java";
495
496 # now get the PDFBox extension for PDFBox tutorial
497 print STDERR "Getting pdfbox from $PDFBOX_TRAC_URL:\n";
498
499 chdir($greenstone_home);
500 my $cmd = "";
501 if ($isWin) {
502 $cmd = "setup.bat && cd ext && wget $PDFBOX_TRAC_URL.zip && unzip pdf-box-java.zip";
503
504 } elsif ($isMac) { # need to use curl not wget
505 $cmd = "cd ext && curl $PDFBOX_TRAC_URL.tar.gz > pdf-box-java.tar.gz && tar -xzf pdf-box-java.tar.gz";
506 }
507 else { # linux
508 $cmd = "bash -c \"export GSDLHOME=&& source setup.bash && cd ext && wget $PDFBOX_TRAC_URL.tar.gz && tar -xzf pdf-box-java.tar.gz\"";
509 }
510 my $status = system $cmd;
511 if($status != 0) {
512 print STDERR "@@@ Failed to set up PDFBox\n";
513 exit -1; # or proceed to testing other tutorials?
514 }
515}
516
517sub getOpenOfficeExt
518{
519 # current revision is 27763, but using "head" works
520 my $OOEXT_TRAC_URL="http://trac.greenstone.org/export/head/gs2-extensions/open-office/trunk/open-office-java"; # both for .zip and .tar.gz extension
521 #"http://trac.greenstone.org/export/".$ENV{'SVN_OPT_REV'}."/gs2-extensions/open-office/trunk/open-office-java";
522
523 # now get the OpenOffice extension for the AllDocTypes collection
524 print STDERR "Getting open office ext from $OOEXT_TRAC_URL:\n";
525
526 chdir($greenstone_home);
527 my $cmd = "";
528 if ($isWin) {
529 $cmd = "setup.bat && cd ext && wget $OOEXT_TRAC_URL.zip && unzip open-office-java.zip";
530
531 } elsif ($isMac) { # need to use curl not wget
532 $cmd = "cd ext && curl $OOEXT_TRAC_URL.tar.gz > open-office-java.tar.gz && tar -xzf open-office-java.tar.gz";
533 }
534 else { # linux
535 $cmd = "bash -c \"export GSDLHOME=&& source setup.bash && cd ext && wget $OOEXT_TRAC_URL.tar.gz && tar -xzf open-office-java.tar.gz\"";
536 }
537 my $status = system $cmd;
538 if($status != 0) {
539 print STDERR "@@@ Failed to set up the Open Office Extension\n";
540 exit -1; # or proceed to testing other tutorials?
541 }
542}
543
544sub getIsisGdl {
545 my $gsfolder = shift(@_);
546
547 if(!$isWin) {
548 my $bit_arch=`uname -m`;
549 if ($bit_arch =~ m/64$/) {
550 my $cmd = "";
551 if($isMac) {
552 $cmd = "cd $gsfolder/bin/darwin && curl http://www.greenstone.org/caveat-emptor/IsisGdl.macleopard > IsisGdl && chmod u+x IsisGdl";
553 } else { # linux
554 $cmd = "cd $gsfolder/bin/linux && wget http://www.greenstone.org/caveat-emptor/IsisGdl.bin32 && mv IsisGdl.bin32 IsisGdl && chmod u+x IsisGdl";
555 }
556 my $isis_status = system $cmd;
557 if($isis_status != 0) {
558 print STDERR "Unable to get IsisGdl from caveat page\n";
559 }
560 }
561 }
562}
563
564# http://stackoverflow.com/questions/3377879/how-do-i-receive-command-output-immediately
565sub run_test
566{
567 my $num_cols = scalar(@collections); # remember the empty case
568
569 if($num_cols == 0) { # deal with all collections
570 push (@collections, "");
571 # putting the empty string in the array so that the "all collections" case
572 # can be handled similar to how the case of user-specified collections is handled
573
574 } else { # deal with user specified set of collections
575 # prefix the directory separator to each collection name
576 @collections = map { $sep.$_ } @collections;
577 }
578
579 my $pdfbox = &filename_concat($greenstone_home, "ext", "pdf-box");
580 if(!-d $pdfbox) {
581 &getPDFBox();
582 }
583
584 my $openofficeext = &filename_concat($greenstone_home, "ext", "open-office");
585 if(!-d $openofficeext) {
586 &getOpenOfficeExt();
587 }
588
589 #&getIsisGdl("$greenstone_home");
590
591 open (my $xml_fh, '>'.$xmlout) || die "Could not open xml file $xmlout for appending: $!\n";
592
593 # perform the requested subcommands, outputting xml information
594 print $xml_fh "<test time=\"$dateid\" id=\"$dateid\">\n";
595
596 my ($cmd, $status);
597 # make sure that diffcol/model-collect is up to date before copying it over to greenstone-home
598
599 if(!$use_local_rebuild) {
600 print $xml_fh "Updating $ENV{'TASK_HOME'}/model-collect:\n";
601 for my $col (@collections) {
602 $cmd = "svn up $ENV{'TASK_HOME'}/model-collect$col"; #chdir("$ENV{'TASK_HOME'}/model-collect");
603 $status = system "$cmd";
604 }
605 }
606
607 # go to whichever greenstone_home we're using
608 chdir($greenstone_home);
609
610 # get svn info
611 print STDERR "getting svn info: $xmlout\n";
612 print $xml_fh "<svn-info>\n";
613 &run_and_print_cmd("svn info", $xml_fh);
614 print $xml_fh "</svn-info>\n";
615 print STDERR "done\n";
616
617 if(!$use_local_rebuild) {
618
619 #make two copies of the model-collect directory in gsdl
620 #one to be rebuilt and one as the basis for comparison
621 #strip both of all .svn directories
622
623 #copy the model collections to the collect folder to be rebuilt
624 print STDERR "installing test collections and model collections to new $gsdl installation... ";
625
626 #clean up
627 if(-d "collect") {
628 for my $col (@collections) {
629 if(-d "collect$col") {
630 &File::Path::remove_tree("collect$col") || die "Error could not delete collect: $!";
631 }
632 }
633 }
634
635 if($use_static_model) {
636 for my $col (@collections) {
637 #copy to collect and strip .svn subfolders
638 &File::Path::make_path("collect$col"); # create the collect folder and copy contents from static model-collection across
639 &copy_recursively("model-collect$col", "collect$col", ".svn");
640 }
641
642 } else { # the default situation: where we check out the model-collect from svn
643 for my $col (@collections) {
644 &File::Path::remove_tree("model-collect$col");
645
646 #copy to collect and strip .svn subfolders
647 &File::Path::make_path("collect$col"); # create the folder and copy contents across
648 &copy_recursively(&filename_concat("$ENV{'TASK_HOME'}","model-collect$col"), "collect$col", ".svn");
649
650 #make the model copy
651 &File::Path::make_path("model-collect$col");
652 &copy_recursively("collect$col", "model-collect$col"); # copy contents across
653 }
654 }
655
656 print STDERR "done\n";
657 }
658
659 #for each collection, import, build and diff with its model counterpart
660
661 # if working with all collections, read the list of collections from the folders in collect
662 if($num_cols == 0) {
663 @collections = (); # get rid of the empty string put in the array to represent "all collections"
664
665 opendir my($collect_handle), "collect" or die "Could not open dir $greenstone_home/collect: $!";
666
667 for my $collection (readdir $collect_handle) {
668 next if ($collection eq "." || $collection eq "..");
669 next if ($collection eq "modelcol");
670 push(@collections, $collection);
671 }
672 closedir $collect_handle; # close handle to collect dir
673 }
674
675 for my $collection (@collections) {
676
677 # next if ($collection ne "Demo-Lucene"); ## TEMPORARY, FOR TESTING THIS SCRIPT
678 # next if ($collection !~ m/OAI|METS|DSpace|MGPP|Lucene/); ## TEMPORARY, FOR TESTING THIS SCRIPT
679
680 #escape the filename (in case of space)
681 $collection =~ s@ @\\ @g;
682 #getting just the basename of the collection would have been necessary had we not cd-ed into $gsdl
683
684 $collection =~ s@^[\\/]@@g; # take the dir-sep prefix away again for user-specified collection names
685
686 if (! -d "collect$sep$collection") {
687 print STDERR "Collection $collection does not exist\n";
688 next;
689 }
690
691 print STDERR "*** Found collection $collection\n";
692 print $xml_fh "<collection-test name=\"$collection\">\n";
693
694 if(!$use_local_rebuild) {
695 #import
696 # Ensure the OIDtype for importing is hash_on_full_filename
697 # "to make document identifiers more stable across upgrades of the software,
698 # although it means that duplicate documents contained in the collection are
699 # no longer detected automatically."
700 print STDERR "$collection - Importing:\n";
701 print $xml_fh "<import>\n";
702 &run_build_script("import.pl -removeold $collection"); #-OIDtype hash_on_full_filename
703 print $xml_fh "</import>\n";
704 print STDERR "done\n";
705
706 #build
707 print STDERR "$collection - Building:\n";
708 print $xml_fh "<build>\n";
709 &run_build_script("buildcol.pl -removeold $collection");
710 print $xml_fh "</build>\n";
711 print STDERR "done\n";
712
713 #rename the intermediate 'building' directory 'index'
714 print STDERR "$collection - Move \"building\" to \"index\"... ";
715 my $index = &filename_concat("collect", $collection, "index");
716 my $building = &filename_concat("collect", $collection, "building");
717 &File::Path::remove_tree($index);
718 # Renaming Directories, http://www.perlmonks.org/?node_id=177421
719 move($building, $index) or die "copy failed: $!"; # File::Copy::move
720 print STDERR "done\n";
721 }
722 #diffcol
723 print STDERR "$collection - Diffing:\n";
724 my $diffcol_dir = &filename_concat($ENV{'TASK_HOME'},"diffcol");
725
726 # help diffcol to know on what os the model cols were generated
727 # and what os this test machine is (on which the test cols will be generated)
728 $cmd = "diffcol.pl -testos $test_os -modelos $model_os -output xml -verbosity 10 $collection"; # need to run with ./diffcol.pl if bash script
729 &run_diff_script($cmd, $xml_fh, $diffcol_dir);
730
731 chdir($greenstone_home); # this is actually where we are
732 print STDERR "done\n";
733 print $xml_fh "</collection-test>\n";
734 }
735
736 print $xml_fh "</test>\n";
737 close($xml_fh);
738
739 print STDERR "done\n";
740}
741
742##***************************************************************
743# runs setup in greenstone_home before running the diff command
744sub run_diff_script {
745 my ($cmd, $fh, $diffcol_dir) = @_;
746
747 # we're in greenstone_home now
748 if(!$isWin) {
749 $cmd = "bash -c \"export GSDLHOME=&& source $setup_script.bash && cd $diffcol_dir && ./$cmd\"";
750
751 } else { # Need to prefix cmd -c/-k as necessary
752 $cmd = "cmd /c \"set GSDLHOME=&& $setup_script.bat && cd $diffcol_dir && perl -S $cmd\"";
753## print STDERR "@@@@ Going to call command: $cmd\n";
754 }
755
756 return &run_and_print_cmd($cmd, $fh);
757}
758
759# runs setup in greenstone_home before running the given build command
760sub run_build_script {
761 my ($cmd, $fh) = @_;
762
763# chdir($greenstone_home);
764 # we are in $greenstone_home already, can directly run the build cmd on the collection
765 if(!$isWin) {
766 $cmd = "bash -c \"export GSDLHOME=&& source $setup_script.bash && $cmd\"";
767
768 } else { # Need to prefix cmd -c/-k as necessary
769 $cmd = "cmd /c \"set GSDLHOME=&& $setup_script.bat && perl -S $cmd\"";
770 }
771## print STDERR "@@@@ Going to call command: $cmd\n";
772
773 return system($cmd);
774 #return &run_and_print_cmd($cmd, $fh); # doesn't work on cmds chained with bash -c
775}
776
777
778# http://stackoverflow.com/questions/758611/how-to-flush-output-in-backticks-in-perl?rq=1activeperl%20sys::info
779# http://stackoverflow.com/questions/1477500/how-do-i-get-the-output-of-an-external-command-in-perl
780sub run_and_print_cmd {
781 my ($cmd, $fh) = @_;
782
783 open my $pin, "$cmd|" or die "unable to run cmd $cmd: $!"; # open(my $fh, '-|', 'powercfg -l') or die $!;
784
785 if(defined $fh) { # print cmd output both to the filehandle and to stdout
786 while (my $line = <$pin>) {
787 print $fh $line;
788# print STDOUT $line; # if also printing cmd output to STDOUT
789 }
790 }
791 else { # no filehandle, so just need to print to stdout
792
793 # unlike backticks operator, system() will print the output of the command to the screen as it executes
794 # http://stackoverflow.com/questions/758611/how-to-flush-output-in-backticks-in-perl?rq=1
795
796 my $status = system $cmd;
797 if($status != 0) {
798 print STDERR "ERROR ($status) running $cmd: $!\n";
799 }
800 }
801 close($pin);
802}
803
804sub filename_concat {
805 my $first_file = shift(@_);
806 my (@filenames) = @_;
807
808 # If first_file is not null or empty, then add it back into the list
809 if (defined $first_file && $first_file =~ /\S/)
810 {
811 unshift(@filenames, $first_file);
812 }
813
814 my $filename = join($sep, @filenames);
815 $filename =~ s/[\\\/]$//; # remove trailing slashes if any
816 return $filename;
817}
818
819
820# The following code is from
821# http://stackoverflow.com/questions/227613/how-can-i-copy-a-directory-recursively-and-filter-filenames-in-perl
822# It also states that "Perl's File::Copy is a bit broken (it doesn't copy permissions on Unix systems, for example)"
823sub copy_recursively {
824 my ($from_dir, $to_dir, $regex) = @_;
825 opendir my($dh), $from_dir or die "Could not open dir '$from_dir': $!";
826
827# if(-d !$to_dir) {
828# mkdir $to_dir or die "mkdir '$to_dir' failed: $!" if not -e $to_dir;
829# }
830
831 for my $entry (readdir $dh) {
832 next if ($entry eq "." || $entry eq "..");
833 next if (defined $regex && $entry =~ /$regex/);
834 my $source = "$from_dir/$entry";
835 my $destination = "$to_dir/$entry";
836 if (-d $source) {
837 mkdir $destination or die "mkdir '$destination' failed: $!" if not -e $destination;
838 copy_recursively($source, $destination, $regex);
839 } else {
840 copy($source, $destination) or die "copy failed: $!";
841 }
842 }
843 closedir $dh;
844 return;
845}
846
847sub summarise {
848
849 # make a summarised Xml report
850 print STDERR "Summarizing the xml report... ";
851 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";
852 my $status = system($cmd);
853 print STDERR "done\n";
854
855 # make a summarised HTMl report
856 print STDERR "Creating an html summary report... ";
857 $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";
858 $status = system($cmd);
859 print STDERR "done\n";
860
861 # Print whether the tests passed or failed
862 print STDERR "*******************************************\n";
863 print STDERR "Checking if successful... \n";
864 $cmd = "java org.apache.xalan.xslt.Process -IN $xmlout -XSL $ENV{'TASK_HOME'}/xsl/passed-or-not.xsl";
865 $status = `$cmd`; #$status = system($cmd);
866 print STDERR "result: $status\n";
867 print STDERR "*******************************************\n";
868}
869
870sub upload {
871 # if the upload dir already existed, clear it of contents
872 if (-d $ENV{'UPLOAD_DIR'}) { #else rm $UPLOAD_DIR/*
873 # don't want to keep previous days reports
874 # else we will have to manually clear them at some point
875 # just generate the set of reports for this run of task.pl upload
876 # and
877 &File::Path::remove_tree($ENV{'UPLOAD_DIR'});
878 }
879 # recreate the upload directory
880 &File::Path::make_path($ENV{'UPLOAD_DIR'});
881
882 # copy all *.xml and *.html files across to UPLOAD_DIR
883 opendir my($dh), $ENV{'DATA_DIR'} or die "Could not open DATA_DIR: $!";
884 for my $entry (readdir $dh) {
885 next if ($entry !~ m/(\.xml|\.html?)$/);
886
887 # copy the reports across with different names: with OS prefixed to them. And for the HTML file on Win, rename to HTM
888 # html files uploaded from windows to nzdl are empty for no reason. Uploading as htm seems to work
889 my $os_entry = $entry;
890 $os_entry =~ s@\[email protected]@ if $isWin;
891 if($isMac) {
892 $os_entry = "diffcol-".$^O."-".$osversion."$os_entry"; # darwin-Lion for Lion/Mountain Lion
893 } else {
894 $os_entry = "diffcol-".$^O."-$os_entry";
895 }
896
897 # get the absolute path to the original files before copying them over
898 $entry = &filename_concat($ENV{'DATA_DIR'}, $entry);
899
900 # copy them over with their new names
901## print STDERR "@@@@ copying across $entry to $ENV{'UPLOAD_DIR'} as $os_entry\n";
902 copy($entry, "$ENV{'UPLOAD_DIR'}$sep$os_entry"); #copy($entry, "$ENV{'UPLOAD_DIR'}");
903 }
904 closedir $dh;
905
906
907 # Upload the html file to puka
908 #default identity dir
909 if ( ! exists $ENV{'IDENTITY_DIR'} ) {
910 $ENV{'IDENTITY_DIR'} = "$ENV{'HOME'}${sep}.ssh"; # "C:\\Research\\Nightly\\tools\\keys" on windows, see environment.pl
911 }
912 if (! exists $ENV{'SNAPSHOT_MODE'} ) {
913 $ENV{'SNAPSHOT_MODE'} = "caveat";
914 }
915
916 #use the correct key for uploading
917 $ENV{'IDENTITY_FILE'} = "$ENV{'IDENTITY_DIR'}${sep}upload-" . $ENV{'SNAPSHOT_MODE'} . ($^O eq "MSWin32" ? ".ppk" : "");
918 if(-f $ENV{'IDENTITY_FILE'}) {
919 # if you need to touch the file on windows: http://stackoverflow.com/questions/51435/windows-version-of-the-unix-touch-command
920
921 # the report we want to upload is actually just os-diffcol-report-$dateid.html
922 my $command = "cd \"$ENV{'UPLOAD_DIR'}\" && tar -c *.htm* | "; #&& cat *.html | "; # && tar -c * |
923 $command .= ($^O eq "MSWin32" ? "plink" : "ssh");
924 $command .= " -T -i \"$ENV{'IDENTITY_FILE'}\" nzdl\@puka.cs.waikato.ac.nz";
925 #print "$command\n";
926 my $status = system("$command");
927 if($status != 0) {
928 print STDERR "*** Failed to upload test report to nzdl $status\n";
929 }
930 } else {
931 print STDERR "*** Cannot upload the test report to nzdl from this machine\n";
932 }
933
934 print STDERR "Finished uploading\n";
935}
936
937# Sending emails with perl: http://learn.perl.org/examples/email.html
938# Sending email attachments with perl: http://www.perlmonks.org/?node_id=19430
939# Sadly none of the packages are installed by default and use of MIME::Lite is discouraged
940sub mail_with_report_attached
941{
942 # email out with report attached, if the tests failed
943 print STDERR "Checking if successful... \n";
944 my $cmd = "java org.apache.xalan.xslt.Process -IN $xmlout -XSL $ENV{'TASK_HOME'}/xsl/passed-or-not.xsl";
945 #my $result = system($cmd);
946 my $result = `$cmd`;
947
948 print STDERR "result: $result\n";
949
950 if($result ne "yes") {
951 my $msg = "$gsdl regression test for $dateid failed";
952 my $subject = "Regression Test Failed"; #"$gsdl regression test for $dateid failed\n";
953 my $attach_file = &filename_concat($ENV{'DATA_DIR'}, "report-$dateid.html");
954
955 if($isWin) {
956 if($use_blat && $blat && $ENV{'GSDL_SMTP'}) {
957 # http://stackoverflow.com/questions/709635/sending-mail-from-batch-file
958 #blat -to [email protected] -server smtp.example.com -f [email protected] -subject "subject" -body "body"
959
960 # need to install blat on windows
961 $cmd = "$blat -to $ENV{'MONITOR_EMAIL'} -server $ENV{'GSDL_SMTP'} -f $ENV{'MONITOR_EMAIL'} -attach $attach_file -subject \"$subject\" -body \"$msg\"";
962 $result = system($cmd);
963 }
964 else {
965 $result = 1; # status from running mail command is 0 if success, 1 if fail
966 print STDERR "********************************************\n";
967 if ($use_blat) {
968 print STDERR "Need blat and SMTP set to send mail attachment\n" ;
969 } else {
970 print STDERR "Not set up to send mail on Windows\n";
971 }
972 print STDERR "Inspect report at: $attach_file\n";
973 print STDERR "********************************************\n";
974 }
975 } else { # linux
976 my $status = system("command -v mutt > /dev/null 2>&1;"); #better way of doing "which mutt"
977
978 if($status != 0) { # mutt doesn't exist, can't send attachments, so send simple email
979 $cmd="echo '$gsdl regression test for $dateid failed.' | mail -s 'Regression Test Failed' $ENV{'MONITOR_EMAIL'}";
980
981 print STDERR "********************************************\n";
982 print STDERR "No mutt installed, unable to mail attachment\n";
983 print STDERR "Inspect report at: $attach_file\n";
984 print STDERR "********************************************\n";
985 } else {
986 #$cmd = "bash -c \"echo '$gsdl regression test for $dateid failed' | mutt -a $attach_file -s 'Regression Test Failed' -- $ENV{'MONITOR_EMAIL'}\"";
987 $cmd = "echo '$gsdl regression test for $dateid failed' | mutt -a $attach_file -s 'Regression Test Failed' -- $ENV{'MONITOR_EMAIL'}";
988 }
989
990 # run the mail command
991 $result = system($cmd); #&run_and_print_cmd($cmd);
992 }
993
994
995 if($result != 0) {
996 print STDERR "*** Unable to send email: $?\n";
997 }
998 else {
999 print STDERR "Sent mail with report attached.\n";
1000 }
1001 } else {
1002 print STDERR "********************************************\n";
1003 print STDERR "Tests were successful. Not sending mail.\n";
1004 print STDERR "********************************************\n";
1005 }
1006}
1007
1008# The old version of this program contained the following, consisting of 1 line of active code:
1009
1010 # Invoke as: sjmc@br:/research/sjm84/envi/bin$ ./envi diffcol summarise
1011 # Doing so will call this pl file and pass in "summarise" in ARGV
1012 # This pl file will in turn call the task executable in this folder
1013 # passing in "summarise" as a parameter.
1014#system("/bin/bash -c \"../etc/tasks/diffcol/task @ARGV\"");
1015
1016 ##system("/bin/bash -c \"./task @ARGV\"");
1017 ##print STDERR "/bin/bash -c ../etc/tasks/diffcol/task @ARGV"
1018
Note: See TracBrowser for help on using the repository browser.