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 |
|
---|
21 | package diffcoltask;
|
---|
22 |
|
---|
23 | use Cwd;
|
---|
24 | use Switch; # for switch(val) { case: ; ...}
|
---|
25 | use File::Path; # for rmdir and mkdir type functions
|
---|
26 | use File::Copy; # for recursive copying of folders but skipping .svn
|
---|
27 | use File::Basename;
|
---|
28 |
|
---|
29 | use strict;
|
---|
30 | no strict 'subs'; # allow barewords (eg STDERR) as function arguments
|
---|
31 |
|
---|
32 |
|
---|
33 | my $isWin = ($^O =~ m/mswin/i) ? 1 : 0;
|
---|
34 | my $isMac = ($^O =~ m/macos|darwin/i) ? 1 : 0;
|
---|
35 | my $sep = $isWin ? "\\" : "/";
|
---|
36 | my $pathsep = $isWin ? ";" : ":";
|
---|
37 | #my $script_ext = $isWin ? ".bat" : ".bash";
|
---|
38 | my $setup_script = "setup"; # needs to become gs3-setup for GS3
|
---|
39 | my $use_blat = 0; # if we ever get blat to send mail/attachments on Windows working, set this to 1
|
---|
40 |
|
---|
41 |
|
---|
42 | # TASK_HOME should be the toplevel diffcol folder
|
---|
43 | $ENV{'TASK_HOME'} = getcwd unless defined $ENV{'TASK_HOME'};
|
---|
44 | if($isWin) {
|
---|
45 | $ENV{'TASK_HOME'} =~ s@\/@\\@g;
|
---|
46 | # need to convert TASK_HOME path name to resolve very subtle bug when running task.pl via
|
---|
47 | # run-gs2-diffcol.bat which uses environment.pl's TASK_HOME setting via envi
|
---|
48 | # At that point TASK_HOME is already defined but ends up lowercase, so that entries in archiveinf-doc
|
---|
49 | # end up sorted differently when db2txt -sort is applied compared to if TASK_HOME had kept its case.
|
---|
50 | require Win32; # for working out Windows Long Filenames from Win 8.3 short filenames
|
---|
51 | $ENV{'TASK_HOME'} = &Win32::GetLongPathName($ENV{'TASK_HOME'});
|
---|
52 | }
|
---|
53 | ## print STDERR "@@@ TASK_HOME: ".$ENV{'TASK_HOME'}."\n";
|
---|
54 |
|
---|
55 |
|
---|
56 | $ENV{'BIN_DIR'} = &filename_concat($ENV{'TASK_HOME'}, "bin");
|
---|
57 |
|
---|
58 | # we'll be using BLAT to send mail attachments on Windows
|
---|
59 | my $blat = $use_blat ? &filename_concat($ENV{'BIN_DIR'}, "blat", "full", "blat.exe") : 0;
|
---|
60 | if($isWin && $use_blat && ! -e $blat) {
|
---|
61 | print STDERR "\n***********************************\n";
|
---|
62 | print STDERR "No blat.exe found in $blat.\n";
|
---|
63 | print STDERR "Blat needed to send mail with attachments on Windows.\n";
|
---|
64 | print STDERR "Extract the blat zip file found in $ENV{'BIN_DIR'}\n";
|
---|
65 | print STDERR "for your bit architecture and name the folder 'blat'\n";
|
---|
66 | print STDERR "***********************************\n\n";
|
---|
67 | $blat = 0;
|
---|
68 | }
|
---|
69 |
|
---|
70 |
|
---|
71 | $ENV{'DATA_DIR'} = &filename_concat($ENV{'TASK_HOME'}, "diffcol-data");
|
---|
72 | $ENV{'UPLOAD_DIR'} = &filename_concat($ENV{'TASK_HOME'}, "diffcol-reports");
|
---|
73 | $ENV{'MONITOR_EMAIL'} = "greenstone_team\@cs.waikato.ac.nz"; # need to escape @ sign
|
---|
74 | $ENV{'GSDL_SMTP'} = ""; #"smtp.gmail.com";
|
---|
75 | ##print STDERR "@@@ email: ".$ENV{'MONITOR_EMAIL'}."\n";
|
---|
76 |
|
---|
77 | # control if an existing compiled greenstone is used
|
---|
78 | # or, if one should be checked out, which revision to checkout from svn
|
---|
79 | $ENV{'SVN_OPT_REV'} = "-r head";
|
---|
80 | #$ENV{'GSDLHOME'}=
|
---|
81 | #$ENV{'GSDL3SRCHOME'}=
|
---|
82 |
|
---|
83 |
|
---|
84 | # if the first arg is a digit, it's the new envi verbosity param. Take it off the array
|
---|
85 | my $envi_verbose = shift(@ARGV) if(exists $ARGV[0] && $ARGV[0] =~ m/^\d+$/);
|
---|
86 |
|
---|
87 | #parse arguments
|
---|
88 | my $action = "all";
|
---|
89 | if(scalar(@ARGV) > 1) {
|
---|
90 | &printusage();
|
---|
91 | exit 0;
|
---|
92 | }
|
---|
93 |
|
---|
94 | if(scalar(@ARGV) == 0) {
|
---|
95 | $action="all";
|
---|
96 | }
|
---|
97 | else {
|
---|
98 | switch ($ARGV[0]) {
|
---|
99 | case qr/^(-h|-help|help)$/i { &printusage; exit 0; }
|
---|
100 | case qr/^(setup_greenstone|run_test|summarise|upload|all)$/ { $action=$ARGV[0]; }
|
---|
101 | else {
|
---|
102 | print STDERR "Bad subcommand.\n";
|
---|
103 | &printusage;
|
---|
104 | exit -1;
|
---|
105 | }
|
---|
106 | }
|
---|
107 | }
|
---|
108 |
|
---|
109 | #check key environment vars are set
|
---|
110 | if(!defined $ENV{'UPLOAD_DIR'}) {
|
---|
111 | print STDERR "Please set a UPLOAD_DIR for the test in an environment.sh file\n";
|
---|
112 | #return 1;
|
---|
113 | }
|
---|
114 | if(!defined $ENV{'DATA_DIR'}) {
|
---|
115 | print STDERR "Please set a DATA_DIR for the test in an environment.sh file\n";
|
---|
116 | #return 1;
|
---|
117 | }
|
---|
118 | if(!defined $ENV{'MONITOR_EMAIL'}) {
|
---|
119 | print STDERR "Please set a MONITOR_EMAIL for the test in an environment.sh file\n";
|
---|
120 | #return 1;
|
---|
121 | }
|
---|
122 |
|
---|
123 | if($ENV{'DATA_DIR'} eq "/") {
|
---|
124 | print STDERR "DATA_DIR should not be the fs root\n";
|
---|
125 | #return 1;
|
---|
126 | }
|
---|
127 |
|
---|
128 | print STDERR "DATA_DIR: ".$ENV{'DATA_DIR'}."\n";
|
---|
129 | print STDERR "UPLOAD_DIR: ".$ENV{'UPLOAD_DIR'}."\n";
|
---|
130 |
|
---|
131 | #create an id for this test
|
---|
132 | my ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = localtime(time);
|
---|
133 | $year += 1900;
|
---|
134 | $mon += 1;
|
---|
135 | $mon = "0$mon" if ($mon < 10);
|
---|
136 | $mday = "0$mday" if ($mday < 10);
|
---|
137 | my $dateid="$year.$mon.$mday"; #my $dateid=($year+1900)."-".($mon+1)."-$mday";
|
---|
138 |
|
---|
139 | print STDERR "Starting test '$dateid'\n";
|
---|
140 |
|
---|
141 |
|
---|
142 | # http://stackoverflow.com/questions/2149368/how-can-i-loop-through-files-in-a-directory-in-perl
|
---|
143 | $ENV{'CLASSPATH'} = "";
|
---|
144 | my $jar_lib_path = $ENV{'TASK_HOME'}.$sep."lib";
|
---|
145 | my @files = <$jar_lib_path/*.jar>; # /full/path/to/diffcol/lib/*jar
|
---|
146 | foreach my $file (@files) {
|
---|
147 | $file =~ s@\/@\\@g if $isWin;
|
---|
148 | $ENV{'CLASSPATH'}=$file.$pathsep.$ENV{'CLASSPATH'};
|
---|
149 | }
|
---|
150 | ##print STDERR "**** classpath: ".$ENV{'CLASSPATH'}."\n";
|
---|
151 |
|
---|
152 |
|
---|
153 | #set the location of the full report
|
---|
154 | my $xmlout=filename_concat($ENV{'DATA_DIR'}, "full-report-$dateid.xml");
|
---|
155 | ##print STDERR "XML: $xmlout\n";
|
---|
156 |
|
---|
157 | # the toplevel folder of the greenstone installation being used
|
---|
158 | my $greenstone_home="";
|
---|
159 | # gsdl is the checkout folder and can be greenstone2 or greenstone3
|
---|
160 | my $gsdl="greenstone2";
|
---|
161 |
|
---|
162 |
|
---|
163 | # Check if using existing compiled-up greenstone installation
|
---|
164 | # and set the greenstone_home location accordingly
|
---|
165 |
|
---|
166 | if(defined $ENV{'GSDL3SRCHOME'} || defined $ENV{'GSDLHOME'}) {
|
---|
167 | print STDERR "Found existing Greenstone home, will use that instead\n";
|
---|
168 | $greenstone_home=$ENV{'GSDLHOME'};
|
---|
169 | } else {
|
---|
170 | $greenstone_home=filename_concat($ENV{'DATA_DIR'}, $gsdl);
|
---|
171 | }
|
---|
172 | ##print STDERR "GSHOME: $greenstone_home\n";
|
---|
173 |
|
---|
174 | #do the requested action
|
---|
175 | if($action eq "setup_greenstone") {
|
---|
176 | &setup_greenstone;
|
---|
177 | }
|
---|
178 | elsif ($action eq "run_test") {
|
---|
179 | &run_test;
|
---|
180 | }
|
---|
181 | elsif ($action eq "summarise") {
|
---|
182 | &summarise;
|
---|
183 | }
|
---|
184 | elsif ($action eq "upload") {
|
---|
185 | &upload;
|
---|
186 | &mail_with_report_attached;
|
---|
187 | }
|
---|
188 | elsif ($action eq "all") {
|
---|
189 | &setup_greenstone;
|
---|
190 | &run_test;
|
---|
191 | &summarise;
|
---|
192 | &upload;
|
---|
193 | &mail_with_report_attached;
|
---|
194 | }
|
---|
195 |
|
---|
196 | ##********************************
|
---|
197 |
|
---|
198 | sub printusage
|
---|
199 | {
|
---|
200 | print STDERR "Run as: $0 (help|setup_greenstone|run_test|summarise|upload|all)\n";
|
---|
201 | }
|
---|
202 |
|
---|
203 | #http://stackoverflow.com/questions/7427262/read-a-file-and-save-it-in-variable-using-shell-script
|
---|
204 |
|
---|
205 | sub setup_greenstone
|
---|
206 | {
|
---|
207 | #clean up from previous tests
|
---|
208 | print STDERR "about to clean up any old tests (Ctrl-C to cancel)"; # no newline
|
---|
209 | for my $i ( 1..5 ) {
|
---|
210 | sleep 1; # 1 second
|
---|
211 | print STDERR ".";
|
---|
212 | }
|
---|
213 | print STDERR "\n";
|
---|
214 |
|
---|
215 | # http://perldoc.perl.org/File/Path.html
|
---|
216 | print STDERR "cleaning up previous tests\n";
|
---|
217 | &File::Path::remove_tree($ENV{'DATA_DIR'});
|
---|
218 |
|
---|
219 | print STDERR "creating the data dir\n";
|
---|
220 | &File::Path::make_path($ENV{'DATA_DIR'}); # works like mkdir -p
|
---|
221 |
|
---|
222 | chdir($ENV{'DATA_DIR'});
|
---|
223 |
|
---|
224 | # use existing compiled-up greenstone installation, if a GSDLHOME set
|
---|
225 | if(defined $ENV{'GSDL3SRCHOME'} || defined $ENV{'GSDLHOME'}) {
|
---|
226 | print STDERR "Found existing Greenstone home, will use that instead\n";
|
---|
227 | return;
|
---|
228 | }
|
---|
229 |
|
---|
230 | # Else checkout a GS from svn into DATA_DIR
|
---|
231 |
|
---|
232 | #svn checkout of main gsdl directory
|
---|
233 | print STDERR "checkout $gsdl:\n";
|
---|
234 | my $cmd = "svn co ".$ENV{'SVN_OPT_REV'}." http://svn.greenstone.org/main/trunk/greenstone2 $gsdl";
|
---|
235 | ##print STDERR "Checkout CMD: $cmd\n";
|
---|
236 |
|
---|
237 | # # unlike backticks operator, system() will print the output of the command to the screen as it executes
|
---|
238 | # http://stackoverflow.com/questions/758611/how-to-flush-output-in-backticks-in-perl?rq=1
|
---|
239 | my $status = system "$cmd"; #my $status = `$cmd`;
|
---|
240 | if($status != 0) {
|
---|
241 | print STDERR "@@@ SVN checkout of $gsdl failed\n";
|
---|
242 | exit -1;
|
---|
243 | }
|
---|
244 | print STDERR "done\n";
|
---|
245 |
|
---|
246 | ##print STDERR "$ENV{'DATA_DIR'}$sep$gsdl\n";
|
---|
247 |
|
---|
248 | chdir("$ENV{'DATA_DIR'}$sep$gsdl");
|
---|
249 |
|
---|
250 | ##print STDERR "@@@ OS: $^O.|".$Config{'archname64'}."|\n";
|
---|
251 |
|
---|
252 | if($isWin) {
|
---|
253 | print STDERR "Compiling $gsdl using makegs2.bat running in auto (silent) mode\n";
|
---|
254 |
|
---|
255 | # we're now in the GS2 folder, call makegs2 with silent param
|
---|
256 | $cmd = "makegs2.bat silent 2>> $ENV{'DATA_DIR'}/compilation-errors"; # STDERR is sent to compilation-errors file
|
---|
257 | $status = system $cmd;
|
---|
258 |
|
---|
259 | } else { # if we're on linux/darwin, need gnome-lib for the correct architecture. And need imagemagick to build imgs in collections
|
---|
260 |
|
---|
261 | my $bit_arch=`uname -m`;
|
---|
262 |
|
---|
263 | # imagmagick binary
|
---|
264 | print STDERR "Getting imagemagick binary\n";
|
---|
265 |
|
---|
266 | my $os = $isMac ? "darwin" : "linux";
|
---|
267 | my $imagickzip = "imagemagick-$os";
|
---|
268 |
|
---|
269 | if($isMac) {
|
---|
270 | # $imagickzip .= "-10.5.tar.gz";
|
---|
271 | # at present, only the Imagemagick binaries created by Max for darwin work on the Macs
|
---|
272 | &File::Path::make_path($ENV{'DATA_DIR'}."$sep$gsdl$sep$os"); # need to ensure gsdl/bin/darwin exists
|
---|
273 | $cmd = "svn export http://svn.greenstone.org/main/trunk/binaries/mac/intel/imagemagick bin/darwin/imagemagick";
|
---|
274 | $status = system($cmd);
|
---|
275 | if($status != 0) {
|
---|
276 | print STDERR "@@@ Unable to get imagemagick for darwin\n";
|
---|
277 | }
|
---|
278 |
|
---|
279 | # need ghostscript mac binary too for pdf to img conversions on mac
|
---|
280 | $cmd = "svn export http://svn.greenstone.org/main/trunk/binaries/mac/intel/ghostscript bin/darwin/ghostscript";
|
---|
281 | $status = system($cmd);
|
---|
282 | if($status != 0) {
|
---|
283 | print STDERR "@@@ Unable to get ghostscript for darwin\n";
|
---|
284 | }
|
---|
285 |
|
---|
286 | # need to set imagemagick and ghostscript binaries to executable
|
---|
287 | # system("chmod -R u+x $ENV{'DATA_DIR'}/$gsdl/bin/darwin/imagemagick/bin/*");
|
---|
288 | # system("chmod -R u+x $ENV{'DATA_DIR'}/$gsdl/bin/darwin/ghostscript/bin/*");
|
---|
289 | } else { # linux
|
---|
290 | my $extension64 = ($bit_arch =~ m/64$/) ? "-x64" : "";
|
---|
291 | $imagickzip .= "$extension64.tar.gz";
|
---|
292 |
|
---|
293 | # now these next imagemagick steps (and those near the end of this sub) are just for linux, no longer also for mac
|
---|
294 | $cmd = "svn export http://svn.greenstone.org/gs2-extensions/imagemagick/trunk/$imagickzip ext/$imagickzip";
|
---|
295 | $status = system ($cmd);
|
---|
296 | system("cd ext && tar -xvzf $imagickzip");
|
---|
297 | }
|
---|
298 |
|
---|
299 | # gnomelib binary
|
---|
300 | print STDERR "setting up gnome-lib-minimal for compilation\n";
|
---|
301 |
|
---|
302 | # To get gnome-lib, need to determine bit architecture of the linux/darwin
|
---|
303 | # http://stackoverflow.com/questions/8963400/the-correct-way-to-read-a-data-file-into-an-array
|
---|
304 | # $Config{'archname64'} doesn't work on the Ubuntu and the Sys::Info package seems to not be supported
|
---|
305 | # well on ActivePerl.
|
---|
306 | # But since we know we're on a Linux/Darwin machine at this point, wecan just run `uname -m` and other linux cmds
|
---|
307 |
|
---|
308 | my $gnome_lib_file = $isMac ? "darwin-intel" : "linux"; # assuming all darwin is intel, not ppc!!
|
---|
309 |
|
---|
310 | $gnome_lib_file .= "-x64" if($bit_arch =~ m/64$/);
|
---|
311 |
|
---|
312 | #svn checkout gnome-lib for this linux/darwin
|
---|
313 | chdir("$ENV{'DATA_DIR'}$sep$gsdl$sep"."ext"); #cd $DATA_DIR/$gsdl/ext
|
---|
314 |
|
---|
315 | ##print STDERR "**** gnomelib: $gnome_lib_file\n";
|
---|
316 |
|
---|
317 | # checkout and unpack gnome-lib-minimal
|
---|
318 |
|
---|
319 | #svn export http://svn.greenstone.org/gs2-extensions/gnome-lib/trunk/gnome-lib-minimal-linux-x64.tar.gz gl.tar.gz
|
---|
320 | $cmd = "svn export http://svn.greenstone.org/gs2-extensions/gnome-lib/trunk/gnome-lib-minimal-".$gnome_lib_file.".tar.gz gl.tar.gz";
|
---|
321 | system $cmd;
|
---|
322 | system ("tar -xvzf gl.tar.gz");
|
---|
323 |
|
---|
324 | chdir("gnome-lib-minimal");
|
---|
325 | ##print STDERR "*** ARCH: $bit_arch\n";
|
---|
326 |
|
---|
327 | # need to run source devel.bash on gnome-lib followed by configure, make, make install
|
---|
328 | # in one go, in order to preserve the compile environment set up by sourcing devel.bash
|
---|
329 |
|
---|
330 | # http://stackoverflow.com/questions/7369145/activating-a-virtualenv-using-a-shell-script-doesnt-seem-to-work
|
---|
331 | # http://ubuntuforums.org/showthread.php?t=1932504 linking /bin/sh to bash instead of dash
|
---|
332 |
|
---|
333 | # $cmd = "bash -c \"source ./devel.bash && cd ../.. && ./configure --enable-apache-httpd && make && make install\"";
|
---|
334 | $cmd = "bash -c \"";
|
---|
335 |
|
---|
336 | $cmd .= "source ./devel.bash";
|
---|
337 | $cmd .= " && cd ../..";
|
---|
338 |
|
---|
339 | #configure
|
---|
340 | # $cmd .= " && ./configure";
|
---|
341 | $cmd .= " && echo 'configure $gsdl: ' ";
|
---|
342 | $cmd .= " && echo '<configure>' >> $xmlout";
|
---|
343 | $cmd .= " && ./configure 2>> $ENV{'DATA_DIR'}/compilation-errors"; # configure
|
---|
344 | $cmd .= " && echo '</configure>' >> $xmlout";
|
---|
345 | $cmd .= " && echo 'done'";
|
---|
346 |
|
---|
347 | #make
|
---|
348 | $cmd .= " && echo 'make $gsdl: '";
|
---|
349 | $cmd .= " && echo '<make>' >> $xmlout";
|
---|
350 | $cmd .= " && make 2>> $ENV{'DATA_DIR'}/compilation-errors"; # make
|
---|
351 | $cmd .= " && echo '</make>' >> $xmlout";
|
---|
352 | $cmd .= " && echo 'done'";
|
---|
353 |
|
---|
354 | #make install
|
---|
355 | $cmd .= " && echo 'make install $gsdl: '";
|
---|
356 | $cmd .= " && echo '<make-install>' >> $xmlout";
|
---|
357 | $cmd .= " && make install 2>> $ENV{'DATA_DIR'}/compilation-errors"; # make install
|
---|
358 | $cmd .= " && echo '</make-install>' >> $xmlout";
|
---|
359 | $cmd .= " && echo 'done'";
|
---|
360 |
|
---|
361 | $cmd .= "\""; # close off cmd to bash and run it
|
---|
362 | $status = system $cmd;
|
---|
363 |
|
---|
364 | if(!$isMac) { # Linux
|
---|
365 | # Moving imagemagick after instead of before compilation, since bin/darwin gets overwritten during compilation
|
---|
366 | 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
|
---|
367 | unlink "$ENV{'DATA_DIR'}/$gsdl"."/ext/$imagickzip" or warn "Could not unlink ext/$imagickzip: $!";
|
---|
368 | &File::Path::remove_tree("$ENV{'DATA_DIR'}$sep$gsdl$sep"."ext/imagemagick"); # the untarred parent folder
|
---|
369 | }
|
---|
370 |
|
---|
371 | }
|
---|
372 |
|
---|
373 | if($status != 0) {
|
---|
374 | print STDERR "@@@ Compile failed\n";
|
---|
375 | exit -1;
|
---|
376 | }
|
---|
377 |
|
---|
378 | # set the path to the greenstone_home variable
|
---|
379 | $greenstone_home="$ENV{'DATA_DIR'}$sep$gsdl";
|
---|
380 |
|
---|
381 | }
|
---|
382 |
|
---|
383 | sub getPDFBox
|
---|
384 | {
|
---|
385 | # current revision is 27763, but using "head" works
|
---|
386 | my $PDFBOX_TRAC_URL="http://trac.greenstone.org/export/head/gs2-extensions/pdf-box/trunk/pdf-box-java"; # both for .zip and .tar.gz extension
|
---|
387 | #"http://trac.greenstone.org/export/".$ENV{'SVN_OPT_REV'}."/gs2-extensions/pdf-box/trunk/pdf-box-java";
|
---|
388 |
|
---|
389 | # now get the PDFBox extension for PDFBox tutorial
|
---|
390 | print STDERR "Getting pdfbox from $PDFBOX_TRAC_URL:\n";
|
---|
391 |
|
---|
392 | chdir($greenstone_home);
|
---|
393 | my $cmd = "";
|
---|
394 | if ($isWin) {
|
---|
395 | $cmd = "setup.bat && cd ext && wget $PDFBOX_TRAC_URL.zip && unzip pdf-box-java.zip";
|
---|
396 |
|
---|
397 | } elsif ($isMac) { # need to use curl not wget
|
---|
398 | $cmd = "cd ext && curl $PDFBOX_TRAC_URL.tar.gz > pdf-box-java.tar.gz && tar -xzf pdf-box-java.tar.gz";
|
---|
399 | }
|
---|
400 | else { # linux
|
---|
401 | $cmd = "bash -c \"export GSDLHOME=&& source setup.bash && cd ext && wget $PDFBOX_TRAC_URL.tar.gz && tar -xzf pdf-box-java.tar.gz\"";
|
---|
402 | }
|
---|
403 | my $status = system $cmd;
|
---|
404 | if($status != 0) {
|
---|
405 | print STDERR "@@@ Failed to set up PDFBox\n";
|
---|
406 | exit -1; # or proceed to testing other tutorials?
|
---|
407 | }
|
---|
408 | }
|
---|
409 |
|
---|
410 | # http://stackoverflow.com/questions/3377879/how-do-i-receive-command-output-immediately
|
---|
411 | sub run_test
|
---|
412 | {
|
---|
413 | my $pdfbox = &filename_concat($greenstone_home, "ext", "pdf-box");
|
---|
414 | if(!-d $pdfbox) {
|
---|
415 | &getPDFBox();
|
---|
416 | }
|
---|
417 |
|
---|
418 | open (my $xml_fh, '>'.$xmlout) || die "Could not open xml file $xmlout for appending: $!\n";
|
---|
419 |
|
---|
420 | # perform the requested subcommands, outputting xml information
|
---|
421 | print $xml_fh "<test time=\"$dateid\" id=\"$dateid\">\n";
|
---|
422 |
|
---|
423 | # make sure that diffcol/model-collect is up to date before copying it over to greenstone-home
|
---|
424 |
|
---|
425 | print $xml_fh "Updating $ENV{'TASK_HOME'}/model-collect:\n";
|
---|
426 | my $cmd = "svn up $ENV{'TASK_HOME'}/model-collect"; #chdir("$ENV{'TASK_HOME'}/model-collect");
|
---|
427 | my $status = system "$cmd";
|
---|
428 |
|
---|
429 | # go to whichever greenstone_home we're using
|
---|
430 | chdir($greenstone_home);
|
---|
431 |
|
---|
432 | # get svn info
|
---|
433 | print STDERR "getting svn info: $xmlout\n";
|
---|
434 | print $xml_fh "<svn-info>\n";
|
---|
435 | &run_and_print_cmd("svn info", $xml_fh);
|
---|
436 | print $xml_fh "</svn-info>\n";
|
---|
437 | print STDERR "done\n";
|
---|
438 |
|
---|
439 | #make two copies of the model-collect directory in gsdl
|
---|
440 | #one to be rebuilt and one as the basis for comparison
|
---|
441 | #strip both of all .svn directories
|
---|
442 |
|
---|
443 | #copy the model collections to the collect folder to be rebuilt
|
---|
444 | print STDERR "installing test collections and model collections to new $gsdl installation... ";
|
---|
445 |
|
---|
446 | #clean up
|
---|
447 | if(-d "collect") {
|
---|
448 | &File::Path::remove_tree("collect") || die "Error could not delete collect: $!";
|
---|
449 | }
|
---|
450 | &File::Path::remove_tree("model-collect");
|
---|
451 |
|
---|
452 | #copy to collect and strip .svn subfolders
|
---|
453 | &File::Path::make_path("collect"); # create the folder and copy contents across
|
---|
454 | ©_recursively(&filename_concat("$ENV{'TASK_HOME'}","model-collect"), "collect", ".svn");
|
---|
455 |
|
---|
456 | #make the model copy
|
---|
457 | &File::Path::make_path("model-collect");
|
---|
458 | ©_recursively("collect", "model-collect"); # copy contents across
|
---|
459 |
|
---|
460 | print STDERR "done\n";
|
---|
461 |
|
---|
462 |
|
---|
463 | #for each collection, import, build and diff with its model counterpart
|
---|
464 | opendir my($collect_handle), "collect" or die "Could not open dir $greenstone_home/collect: $!";
|
---|
465 | for my $collection (readdir $collect_handle) {
|
---|
466 | next if ($collection eq "." || $collection eq "..");
|
---|
467 | # next if ($collection ne "Word-PDF-Basic"); ## TEMPORARY, FOR TESTING THIS SCRIPT
|
---|
468 |
|
---|
469 | #escape the filename (in case of space)
|
---|
470 | $collection =~ s@ @\\ @g;
|
---|
471 | #getting just the basename of the collection would have been necessary had we not cd-ed into $gsdl
|
---|
472 |
|
---|
473 | print STDERR "*** Found collection $collection\n";
|
---|
474 | print $xml_fh "<collection-test name=\"$collection\">\n";
|
---|
475 |
|
---|
476 | #import
|
---|
477 | # Ensure the OIDtype for importing is hash_on_full_filename
|
---|
478 | # "to make document identifiers more stable across upgrades of the software,
|
---|
479 | # although it means that duplicate documents contained in the collection are
|
---|
480 | # no longer detected automatically."
|
---|
481 | print STDERR "$collection - Importing:\n";
|
---|
482 | print $xml_fh "<import>\n";
|
---|
483 | &run_build_script("import.pl -removeold $collection"); #-OIDtype hash_on_full_filename
|
---|
484 | print $xml_fh "</import>\n";
|
---|
485 | print STDERR "done\n";
|
---|
486 |
|
---|
487 | #build
|
---|
488 | print STDERR "$collection - Building:\n";
|
---|
489 | print $xml_fh "<build>\n";
|
---|
490 | &run_build_script("buildcol.pl -removeold $collection");
|
---|
491 | print $xml_fh "</build>\n";
|
---|
492 | print STDERR "done\n";
|
---|
493 |
|
---|
494 | #rename the intermediate 'building' directory 'index'
|
---|
495 | print STDERR "$collection - Move \"building\" to \"index\"... ";
|
---|
496 | my $index = &filename_concat("collect", $collection, "index");
|
---|
497 | my $building = &filename_concat("collect", $collection, "building");
|
---|
498 | &File::Path::remove_tree($index);
|
---|
499 | # Renaming Directories, http://www.perlmonks.org/?node_id=177421
|
---|
500 | move($building, $index) or die "copy failed: $!"; # File::Copy::move
|
---|
501 | print STDERR "done\n";
|
---|
502 |
|
---|
503 | #diffcol
|
---|
504 | print STDERR "$collection - Diffing:\n";
|
---|
505 | my $diffcol_dir = &filename_concat($ENV{'TASK_HOME'},"diffcol");
|
---|
506 | $cmd = "diffcol.pl -output xml -verbosity 10 $collection"; # need to run with ./diffcol.pl if bash script
|
---|
507 | &run_diff_script($cmd, $xml_fh, $diffcol_dir);
|
---|
508 |
|
---|
509 | chdir($greenstone_home); # this is actually where we are
|
---|
510 | print STDERR "done\n";
|
---|
511 | print $xml_fh "</collection-test>\n";
|
---|
512 | }
|
---|
513 | closedir $collect_handle; # close handle to collect dir
|
---|
514 |
|
---|
515 | print $xml_fh "</test>\n";
|
---|
516 | close($xml_fh);
|
---|
517 |
|
---|
518 | print STDERR "done\n";
|
---|
519 | }
|
---|
520 |
|
---|
521 | ##***************************************************************
|
---|
522 | # runs setup in greenstone_home before running the diff command
|
---|
523 | sub run_diff_script {
|
---|
524 | my ($cmd, $fh, $diffcol_dir) = @_;
|
---|
525 |
|
---|
526 | # we're in greenstone_home now
|
---|
527 | if(!$isWin) {
|
---|
528 | $cmd = "bash -c \"export GSDLHOME=&& source $setup_script.bash && cd $diffcol_dir && ./$cmd\"";
|
---|
529 |
|
---|
530 | } else { # Need to prefix cmd -c/-k as necessary
|
---|
531 | $cmd = "cmd /c \"set GSDLHOME=&& $setup_script.bat && cd $diffcol_dir && perl -S $cmd\"";
|
---|
532 | ## print STDERR "@@@@ Going to call command: $cmd\n";
|
---|
533 | }
|
---|
534 |
|
---|
535 | return &run_and_print_cmd($cmd, $fh);
|
---|
536 | }
|
---|
537 |
|
---|
538 | # runs setup in greenstone_home before running the given build command
|
---|
539 | sub run_build_script {
|
---|
540 | my ($cmd, $fh) = @_;
|
---|
541 |
|
---|
542 | # chdir($greenstone_home);
|
---|
543 | # we are in $greenstone_home already, can directly run the build cmd on the collection
|
---|
544 | if(!$isWin) {
|
---|
545 | $cmd = "bash -c \"export GSDLHOME=&& source $setup_script.bash && $cmd\"";
|
---|
546 |
|
---|
547 | } else { # Need to prefix cmd -c/-k as necessary
|
---|
548 | $cmd = "cmd /c \"set GSDLHOME=&& $setup_script.bat && perl -S $cmd\"";
|
---|
549 | }
|
---|
550 | ## print STDERR "@@@@ Going to call command: $cmd\n";
|
---|
551 |
|
---|
552 | return system($cmd);
|
---|
553 | #return &run_and_print_cmd($cmd, $fh); # doesn't work on cmds chained with bash -c
|
---|
554 | }
|
---|
555 |
|
---|
556 |
|
---|
557 | # http://stackoverflow.com/questions/758611/how-to-flush-output-in-backticks-in-perl?rq=1activeperl%20sys::info
|
---|
558 | # http://stackoverflow.com/questions/1477500/how-do-i-get-the-output-of-an-external-command-in-perl
|
---|
559 | sub run_and_print_cmd {
|
---|
560 | my ($cmd, $fh) = @_;
|
---|
561 |
|
---|
562 | open my $pin, "$cmd|" or die "unable to run cmd $cmd: $!"; # open(my $fh, '-|', 'powercfg -l') or die $!;
|
---|
563 |
|
---|
564 | if(defined $fh) { # print cmd output both to the filehandle and to stdout
|
---|
565 | while (my $line = <$pin>) {
|
---|
566 | print $fh $line;
|
---|
567 | # print STDOUT $line; # if also printing cmd output to STDOUT
|
---|
568 | }
|
---|
569 | }
|
---|
570 | else { # no filehandle, so just need to print to stdout
|
---|
571 |
|
---|
572 | # unlike backticks operator, system() will print the output of the command to the screen as it executes
|
---|
573 | # http://stackoverflow.com/questions/758611/how-to-flush-output-in-backticks-in-perl?rq=1
|
---|
574 |
|
---|
575 | my $status = system $cmd;
|
---|
576 | if($status != 0) {
|
---|
577 | print STDERR "ERROR ($status) running $cmd: $!\n";
|
---|
578 | }
|
---|
579 | }
|
---|
580 | close($pin);
|
---|
581 | }
|
---|
582 |
|
---|
583 | sub filename_concat {
|
---|
584 | my $first_file = shift(@_);
|
---|
585 | my (@filenames) = @_;
|
---|
586 |
|
---|
587 | # If first_file is not null or empty, then add it back into the list
|
---|
588 | if (defined $first_file && $first_file =~ /\S/)
|
---|
589 | {
|
---|
590 | unshift(@filenames, $first_file);
|
---|
591 | }
|
---|
592 |
|
---|
593 | my $filename = join($sep, @filenames);
|
---|
594 | $filename =~ s/[\\\/]$//; # remove trailing slashes if any
|
---|
595 | return $filename;
|
---|
596 | }
|
---|
597 |
|
---|
598 |
|
---|
599 | # The following code is from
|
---|
600 | # http://stackoverflow.com/questions/227613/how-can-i-copy-a-directory-recursively-and-filter-filenames-in-perl
|
---|
601 | # It also states that "Perl's File::Copy is a bit broken (it doesn't copy permissions on Unix systems, for example)"
|
---|
602 | sub copy_recursively {
|
---|
603 | my ($from_dir, $to_dir, $regex) = @_;
|
---|
604 | opendir my($dh), $from_dir or die "Could not open dir '$from_dir': $!";
|
---|
605 |
|
---|
606 | # if(-d !$to_dir) {
|
---|
607 | # mkdir $to_dir or die "mkdir '$to_dir' failed: $!" if not -e $to_dir;
|
---|
608 | # }
|
---|
609 |
|
---|
610 | for my $entry (readdir $dh) {
|
---|
611 | next if ($entry eq "." || $entry eq "..");
|
---|
612 | next if (defined $regex && $entry =~ /$regex/);
|
---|
613 | my $source = "$from_dir/$entry";
|
---|
614 | my $destination = "$to_dir/$entry";
|
---|
615 | if (-d $source) {
|
---|
616 | mkdir $destination or die "mkdir '$destination' failed: $!" if not -e $destination;
|
---|
617 | copy_recursively($source, $destination, $regex);
|
---|
618 | } else {
|
---|
619 | copy($source, $destination) or die "copy failed: $!";
|
---|
620 | }
|
---|
621 | }
|
---|
622 | closedir $dh;
|
---|
623 | return;
|
---|
624 | }
|
---|
625 |
|
---|
626 | sub summarise {
|
---|
627 |
|
---|
628 | # make a summarised Xml report
|
---|
629 | print STDERR "Summarizing the xml report... ";
|
---|
630 | 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";
|
---|
631 | my $status = system($cmd);
|
---|
632 | print STDERR "done\n";
|
---|
633 |
|
---|
634 | # make a summarised HTMl report
|
---|
635 | print STDERR "Creating an html summary report... ";
|
---|
636 | $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";
|
---|
637 | $status = system($cmd);
|
---|
638 | print STDERR "done\n";
|
---|
639 | }
|
---|
640 |
|
---|
641 | sub upload {
|
---|
642 | # if the upload dir already existed, clear it of contents
|
---|
643 | if (-d $ENV{'UPLOAD_DIR'}) { #else rm $UPLOAD_DIR/*
|
---|
644 | # don't want to keep previous days reports
|
---|
645 | # else we will have to manually clear them at some point
|
---|
646 | # just generate the set of reports for this run of task.pl upload
|
---|
647 | # and
|
---|
648 | &File::Path::remove_tree($ENV{'UPLOAD_DIR'});
|
---|
649 | }
|
---|
650 | # recreate the upload directory
|
---|
651 | &File::Path::make_path($ENV{'UPLOAD_DIR'});
|
---|
652 |
|
---|
653 | # copy all *.xml and *.html files across to UPLOAD_DIR
|
---|
654 | opendir my($dh), $ENV{'DATA_DIR'} or die "Could not open DATA_DIR: $!";
|
---|
655 | for my $entry (readdir $dh) {
|
---|
656 | next if ($entry !~ m/(\.xml|\.html?)$/);
|
---|
657 |
|
---|
658 | # copy the reports across with different names: with OS prefixed to them. And for the HTML file on Win, rename to HTM
|
---|
659 | # html files uploaded from windows to nzdl are empty for no reason. Uploading as htm seems to work
|
---|
660 | my $os_entry = $entry;
|
---|
661 | $os_entry =~ s@\.html$@.htm@ if $isWin;
|
---|
662 | $os_entry = $^O."-diffcol-$os_entry";
|
---|
663 |
|
---|
664 | # get the absolute path to the original files before copying them over
|
---|
665 | $entry = &filename_concat($ENV{'DATA_DIR'}, $entry);
|
---|
666 |
|
---|
667 | # copy them over with their new names
|
---|
668 | ## print STDERR "@@@@ copying across $entry to $ENV{'UPLOAD_DIR'} as $os_entry\n";
|
---|
669 | copy($entry, "$ENV{'UPLOAD_DIR'}$sep$os_entry"); #copy($entry, "$ENV{'UPLOAD_DIR'}");
|
---|
670 | }
|
---|
671 | closedir $dh;
|
---|
672 |
|
---|
673 |
|
---|
674 | # Upload the html file to puka
|
---|
675 | #default identity dir
|
---|
676 | if ( ! exists $ENV{'IDENTITY_DIR'} ) {
|
---|
677 | $ENV{'IDENTITY_DIR'} = "$ENV{'HOME'}${sep}.ssh"; # "C:\\Research\\Nightly\\tools\\keys" on windows, see environment.pl
|
---|
678 | }
|
---|
679 | if (! exists $ENV{'SNAPSHOT_MODE'} ) {
|
---|
680 | $ENV{'SNAPSHOT_MODE'} = "caveat";
|
---|
681 | }
|
---|
682 |
|
---|
683 | #use the correct key for uploading
|
---|
684 | $ENV{'IDENTITY_FILE'} = "$ENV{'IDENTITY_DIR'}${sep}upload-" . $ENV{'SNAPSHOT_MODE'} . ($^O eq "MSWin32" ? ".ppk" : "");
|
---|
685 | if(-f $ENV{'IDENTITY_FILE'}) {
|
---|
686 | # if you need to touch the file on windows: http://stackoverflow.com/questions/51435/windows-version-of-the-unix-touch-command
|
---|
687 |
|
---|
688 | # the report we want to upload is actually just os-diffcol-report-$dateid.html
|
---|
689 | my $command = "cd \"$ENV{'UPLOAD_DIR'}\" && tar -c *.htm* | "; #&& cat *.html | "; # && tar -c * |
|
---|
690 | $command .= ($^O eq "MSWin32" ? "plink" : "ssh");
|
---|
691 | $command .= " -T -i \"$ENV{'IDENTITY_FILE'}\" nzdl\@puka.cs.waikato.ac.nz";
|
---|
692 | #print "$command\n";
|
---|
693 | my $status = system("$command");
|
---|
694 | if($status != 0) {
|
---|
695 | print STDERR "*** Failed to upload test report to nzdl $status\n";
|
---|
696 | }
|
---|
697 | } else {
|
---|
698 | print STDERR "*** Cannot upload the test report to nzdl from this machine\n";
|
---|
699 | }
|
---|
700 |
|
---|
701 | print STDERR "Finished uploading\n";
|
---|
702 | }
|
---|
703 |
|
---|
704 | # Sending emails with perl: http://learn.perl.org/examples/email.html
|
---|
705 | # Sending email attachments with perl: http://www.perlmonks.org/?node_id=19430
|
---|
706 | # Sadly none of the packages are installed by default and use of MIME::Lite is discouraged
|
---|
707 | sub mail_with_report_attached
|
---|
708 | {
|
---|
709 | # email out with report attached, if the tests failed
|
---|
710 | print STDERR "Checking if successful... \n";
|
---|
711 | my $cmd = "java org.apache.xalan.xslt.Process -IN $xmlout -XSL $ENV{'TASK_HOME'}/xsl/passed-or-not.xsl";
|
---|
712 | #my $result = system($cmd);
|
---|
713 | my $result = `$cmd`;
|
---|
714 |
|
---|
715 | print STDERR "result: $result\n";
|
---|
716 |
|
---|
717 | if($result ne "yes") {
|
---|
718 | my $msg = "$gsdl regression test for $dateid failed";
|
---|
719 | my $subject = "Regression Test Failed"; #"$gsdl regression test for $dateid failed\n";
|
---|
720 | my $attach_file = &filename_concat($ENV{'DATA_DIR'}, "report-$dateid.html");
|
---|
721 |
|
---|
722 | if($isWin) {
|
---|
723 | if($use_blat && $blat && $ENV{'GSDL_SMTP'}) {
|
---|
724 | # http://stackoverflow.com/questions/709635/sending-mail-from-batch-file
|
---|
725 | #blat -to user@example.com -server smtp.example.com -f batch_script@example.com -subject "subject" -body "body"
|
---|
726 |
|
---|
727 | # need to install blat on windows
|
---|
728 | $cmd = "$blat -to $ENV{'MONITOR_EMAIL'} -server $ENV{'GSDL_SMTP'} -f $ENV{'MONITOR_EMAIL'} -attach $attach_file -subject \"$subject\" -body \"$msg\"";
|
---|
729 | $result = system($cmd);
|
---|
730 | }
|
---|
731 | else {
|
---|
732 | $result = 1; # status from running mail command is 0 if success, 1 if fail
|
---|
733 | print STDERR "********************************************\n";
|
---|
734 | if ($use_blat) {
|
---|
735 | print STDERR "Need blat and SMTP set to send mail attachment\n" ;
|
---|
736 | } else {
|
---|
737 | print STDERR "Not set up to send mail on Windows\n";
|
---|
738 | }
|
---|
739 | print STDERR "Inspect report at: $attach_file\n";
|
---|
740 | print STDERR "********************************************\n";
|
---|
741 | }
|
---|
742 | } else { # linux
|
---|
743 | my $status = system("command -v mutt > /dev/null 2>&1;"); #better way of doing "which mutt"
|
---|
744 |
|
---|
745 | if($status != 0) { # mutt doesn't exist, can't send attachments, so send simple email
|
---|
746 | $cmd="echo '$gsdl regression test for $dateid failed.' | mail -s 'Regression Test Failed' $ENV{'MONITOR_EMAIL'}";
|
---|
747 |
|
---|
748 | print STDERR "********************************************\n";
|
---|
749 | print STDERR "No mutt installed, unable to mail attachment\n";
|
---|
750 | print STDERR "Inspect report at: $attach_file\n";
|
---|
751 | print STDERR "********************************************\n";
|
---|
752 | } else {
|
---|
753 | #$cmd = "bash -c \"echo '$gsdl regression test for $dateid failed' | mutt -a $attach_file -s 'Regression Test Failed' -- $ENV{'MONITOR_EMAIL'}\"";
|
---|
754 | $cmd = "echo '$gsdl regression test for $dateid failed' | mutt -a $attach_file -s 'Regression Test Failed' -- $ENV{'MONITOR_EMAIL'}";
|
---|
755 | }
|
---|
756 |
|
---|
757 | # run the mail command
|
---|
758 | $result = system($cmd); #&run_and_print_cmd($cmd);
|
---|
759 | }
|
---|
760 |
|
---|
761 |
|
---|
762 | if($result != 0) {
|
---|
763 | print STDERR "*** Unable to send email: $?\n";
|
---|
764 | }
|
---|
765 | else {
|
---|
766 | print STDERR "Sent mail with report attached.\n";
|
---|
767 | }
|
---|
768 | } else {
|
---|
769 | print STDERR "********************************************\n";
|
---|
770 | print STDERR "Tests were successful. Not sending mail.\n";
|
---|
771 | print STDERR "********************************************\n";
|
---|
772 | }
|
---|
773 | }
|
---|
774 |
|
---|
775 | # The old version of this program contained the following, consisting of 1 line of active code:
|
---|
776 |
|
---|
777 | # Invoke as: sjmc@br:/research/sjm84/envi/bin$ ./envi diffcol summarise
|
---|
778 | # Doing so will call this pl file and pass in "summarise" in ARGV
|
---|
779 | # This pl file will in turn call the task executable in this folder
|
---|
780 | # passing in "summarise" as a parameter.
|
---|
781 | #system("/bin/bash -c \"../etc/tasks/diffcol/task @ARGV\"");
|
---|
782 |
|
---|
783 | ##system("/bin/bash -c \"./task @ARGV\"");
|
---|
784 | ##print STDERR "/bin/bash -c ../etc/tasks/diffcol/task @ARGV"
|
---|
785 |
|
---|