1 | #!/usr/bin/perl
|
---|
2 |
|
---|
3 | # This perl script may be called directly or by running build.bat on
|
---|
4 | # windows (build.bat is in bin\windows)
|
---|
5 |
|
---|
6 | use FileHandle;
|
---|
7 |
|
---|
8 | BEGIN {
|
---|
9 |
|
---|
10 | die "GSDLHOME not set - did you remember to source setup.bash (unix) or " .
|
---|
11 | "run setup.bat (windows)?\n" unless defined $ENV{'GSDLHOME'};
|
---|
12 | die "GSDLOS not set - did you remember to source setup.bash (unix) or " .
|
---|
13 | "run setup.bat (windows)?\n" unless defined $ENV{'GSDLOS'};
|
---|
14 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib");
|
---|
15 |
|
---|
16 | STDOUT->autoflush(1);
|
---|
17 | STDERR->autoflush(1);
|
---|
18 | }
|
---|
19 |
|
---|
20 | use parsargv;
|
---|
21 | use util;
|
---|
22 |
|
---|
23 | if (!parsargv::parse(\@ARGV,
|
---|
24 | 'append', \$append,
|
---|
25 | 'remove_archives', \$remove_archives,
|
---|
26 | 'remove_import', \$remove_import,
|
---|
27 | 'buildtype/^(build|import)$/import', \$buildtype,
|
---|
28 | 'maxdocs/^\-?\d+/-1', \$maxdocs,
|
---|
29 | 'download/.+', \@download,
|
---|
30 | 'out/.*/STDERR', \$out)) {
|
---|
31 |
|
---|
32 | &print_usage();
|
---|
33 | die "\n";
|
---|
34 | }
|
---|
35 |
|
---|
36 | my ($collection) = @ARGV;
|
---|
37 |
|
---|
38 | if (!defined $collection || $collection !~ /\w/) {
|
---|
39 | print STDERR "You must specify a collection to build\n";
|
---|
40 | &print_usage();
|
---|
41 | die "\n";
|
---|
42 | }
|
---|
43 |
|
---|
44 | if ($maxdocs == -1) {
|
---|
45 | $maxdocs = "";
|
---|
46 | } else {
|
---|
47 | $maxdocs = "-maxdocs $maxdocs";
|
---|
48 | }
|
---|
49 |
|
---|
50 | my $collectdir = &util::filename_cat ($ENV{'GSDLHOME'}, "collect", $collection);
|
---|
51 | my $importdir = &util::filename_cat ($collectdir, "import");
|
---|
52 | my $archivedir = &util::filename_cat ($collectdir, "archives");
|
---|
53 | my $buildingdir = &util::filename_cat ($collectdir, "building");
|
---|
54 | my $indexdir = &util::filename_cat ($collectdir, "index");
|
---|
55 | my $bindir = &util::filename_cat ($ENV{'GSDLHOME'}, "bin");
|
---|
56 |
|
---|
57 | my $use_out = 0;
|
---|
58 | my $outfile = $out;
|
---|
59 | if ($out !~ /^(STDERR|STDOUT)$/i) {
|
---|
60 | open (OUT, ">$out") || die "Couldn't open output file $out\n";
|
---|
61 | $out = "OUT";
|
---|
62 |
|
---|
63 | # delete any existing .final file
|
---|
64 | &util::rm ("$outfile.final") if -e "$outfile.final";
|
---|
65 |
|
---|
66 | $use_out = 1;
|
---|
67 | }
|
---|
68 | $out->autoflush(1);
|
---|
69 |
|
---|
70 | &main();
|
---|
71 |
|
---|
72 | close OUT if $use_out;
|
---|
73 |
|
---|
74 | sub print_usage {
|
---|
75 | print STDERR "\n usage: $0 [options] collection-name\n\n";
|
---|
76 | print STDERR " options:\n";
|
---|
77 | print STDERR " -append Add new files to existing collection\n";
|
---|
78 | print STDERR " -remove_archives Remove archives directory after successfully\n";
|
---|
79 | print STDERR " building the collection.\n";
|
---|
80 | print STDERR " -remove_import Remove import directory after successfully\n";
|
---|
81 | print STDERR " importing the collection.\n";
|
---|
82 | print STDERR " -buildtype build|import If 'build' attempt to build directly\n";
|
---|
83 | print STDERR " from archives directory (bypassing import\n";
|
---|
84 | print STDERR " stage). Defaults to 'import'\n";
|
---|
85 | print STDERR " -maxdocs number Maximum number of documents to build\n";
|
---|
86 | print STDERR " -download directory Directory (or file) to get import documents from.\n";
|
---|
87 | print STDERR " There may be multiple download directories and they\n";
|
---|
88 | print STDERR " may be of type http://, ftp://, or file://\n.";
|
---|
89 | print STDERR " Note that any existing import directory will be\n";
|
---|
90 | print STDERR " deleted to make way for the downloaded data if\n";
|
---|
91 | print STDERR " a -download option is supplied\n";
|
---|
92 | print STDERR " -out Filename or handle to print output status to.\n";
|
---|
93 | print STDERR " The default is STDERR\n\n";
|
---|
94 | }
|
---|
95 |
|
---|
96 | sub main {
|
---|
97 |
|
---|
98 | # do the download thing if we have any -download options
|
---|
99 | if (scalar (@download)) {
|
---|
100 | # remove any existing import data
|
---|
101 | if (&has_content ($importdir)) {
|
---|
102 | print $out "build: WARNING: removing contents of $importdir\n";
|
---|
103 | &util::rm_r ($importdir);
|
---|
104 | }
|
---|
105 |
|
---|
106 | foreach $download_dir (@download) {
|
---|
107 |
|
---|
108 | if ($download_dir =~ /^http:\/\//) {
|
---|
109 | # http download
|
---|
110 |
|
---|
111 | } elsif ($download_dir =~ /^ftp:\/\//) {
|
---|
112 | # ftp download
|
---|
113 |
|
---|
114 | } else {
|
---|
115 | # we assume anything not beginning with http:// or ftp://
|
---|
116 | # is a file or directory on the local file system.
|
---|
117 | $download_dir =~ s/^file:\/\///;
|
---|
118 |
|
---|
119 | if (-e $download_dir) {
|
---|
120 | # copy download_dir and all it contains to the import directory
|
---|
121 | my $download_cmd = "perl " . &util::filename_cat ($bindir, "script", "filecopy.pl");
|
---|
122 | $download_cmd .= " -out \"$outfile.download\"" if $use_out;
|
---|
123 | $download_cmd .= " \"" . $download_dir . "\" " . $collection;
|
---|
124 | system ($download_cmd);
|
---|
125 | # if using output directory append the file download output to it
|
---|
126 | &append_file ($out, "$outfile.download");
|
---|
127 | } else {
|
---|
128 | print $out "WARNING: $download_dir does not exist\n";
|
---|
129 | }
|
---|
130 | }
|
---|
131 | }
|
---|
132 | }
|
---|
133 |
|
---|
134 | if (-e &util::filename_cat ($archivedir, "archives.inf")) {
|
---|
135 | if (&has_content ($importdir)) {
|
---|
136 | if ($buildtype eq "build") {
|
---|
137 | &gsdl_build();
|
---|
138 | } else {
|
---|
139 | &gsdl_import();
|
---|
140 | &gsdl_build();
|
---|
141 | }
|
---|
142 | } else {
|
---|
143 | # there are archives but no import, build directly from archives
|
---|
144 | print $out "build: no import material was found, building directly\n";
|
---|
145 | print $out " from archives\n";
|
---|
146 | &gsdl_build();
|
---|
147 | }
|
---|
148 | } else {
|
---|
149 | if (&has_content ($importdir)) {
|
---|
150 | if ($buildtype eq "build") {
|
---|
151 | print $out "build: can't build directly from archives as no\n";
|
---|
152 | print $out " imported archives exist (did you forget to\n";
|
---|
153 | print $out " move the contents of $collection/import to\n";
|
---|
154 | print $out " collection/archives?)\n";
|
---|
155 | }
|
---|
156 | &gsdl_import();
|
---|
157 | &gsdl_build();
|
---|
158 | } else {
|
---|
159 | # no import or archives
|
---|
160 | print $out "build: ERROR: The $collection collection has no import\n";
|
---|
161 | print $out " or archives data. Try downloading an unbuilt version\n";
|
---|
162 | print $out " of the collection from www.nzdl.org\n";
|
---|
163 | &final_out (1) if $use_out;
|
---|
164 | die "\n";
|
---|
165 | }
|
---|
166 | }
|
---|
167 | &final_out (0) if $use_out;
|
---|
168 | }
|
---|
169 |
|
---|
170 | sub gsdl_import {
|
---|
171 |
|
---|
172 | print $out "importing the $collection collection\n\n";
|
---|
173 |
|
---|
174 | my $import_cmd = "perl " . &util::filename_cat ($bindir, "script", "import.pl");
|
---|
175 | $import_cmd .= " -out \"$outfile.import\"" if $use_out;
|
---|
176 | $import_cmd .= " -removeold" unless $append;
|
---|
177 | $import_cmd .= " $maxdocs $collection";
|
---|
178 | system ($import_cmd);
|
---|
179 | # if using output directory append the import output to it
|
---|
180 | &append_file ($out, "$outfile.import");
|
---|
181 |
|
---|
182 | if (-e &util::filename_cat ($archivedir, "archives.inf")) {
|
---|
183 | print $out "$collection collection imported successfully\n\n";
|
---|
184 | if ($remove_import) {
|
---|
185 | print $out "removing import directory ($importdir)\n";
|
---|
186 | &util::rm_r ($importdir);
|
---|
187 | }
|
---|
188 | } else {
|
---|
189 | &final_out (2) if $use_out;
|
---|
190 | die "\nimport.pl failed\n";
|
---|
191 | }
|
---|
192 | }
|
---|
193 |
|
---|
194 | sub gsdl_build {
|
---|
195 |
|
---|
196 | print $out "building the $collection collection\n\n";
|
---|
197 |
|
---|
198 | my $build_cmd = "perl " . &util::filename_cat ($bindir, "script", "buildcol.pl");
|
---|
199 | $build_cmd .= " -out \"$outfile.build\"" if $use_out;
|
---|
200 | $build_cmd .= " $maxdocs $collection";
|
---|
201 | system ($build_cmd);
|
---|
202 | # if using output directory append the buildcol output to it
|
---|
203 | &append_file ($out, "$outfile.build");
|
---|
204 |
|
---|
205 | if (-e &util::filename_cat ($buildingdir, "text", "$collection.ldb") ||
|
---|
206 | -e &util::filename_cat ($buildingdir, "text", "$collection.bdb")) {
|
---|
207 | print $out "$collection collection built successfully\n\n";
|
---|
208 | if ($remove_archives) {
|
---|
209 | print $out "removing archives directory ($archivedir)\n";
|
---|
210 | &util::rm_r ($archivedir);
|
---|
211 | }
|
---|
212 | } else {
|
---|
213 | &final_out (2) if $use_out;
|
---|
214 | die "\nbuildcol.pl failed\n";
|
---|
215 | }
|
---|
216 |
|
---|
217 | # replace old indexes with new ones
|
---|
218 | if (&has_content ($indexdir)) {
|
---|
219 | print $out "removing old indexes\n";
|
---|
220 | &util::rm_r ($indexdir);
|
---|
221 | }
|
---|
222 | rmdir ($indexdir) if -d $indexdir;
|
---|
223 | rename ($buildingdir, $indexdir);
|
---|
224 | }
|
---|
225 |
|
---|
226 | sub has_content {
|
---|
227 | my ($dir) = @_;
|
---|
228 |
|
---|
229 | if (!-d $dir) {return 0;}
|
---|
230 |
|
---|
231 | opendir (DIR, $dir) || return 0;
|
---|
232 | my @files = readdir DIR;
|
---|
233 | close DIR;
|
---|
234 |
|
---|
235 | foreach my $file (@files) {
|
---|
236 | if ($file !~ /^\.{1,2}$/) {
|
---|
237 | return 1;
|
---|
238 | }
|
---|
239 | }
|
---|
240 | return 0;
|
---|
241 | }
|
---|
242 |
|
---|
243 | sub append_file {
|
---|
244 | my ($handle, $file) = @_;
|
---|
245 |
|
---|
246 | open (FILE, $file) || return;
|
---|
247 | undef $/;
|
---|
248 | print $handle <FILE>;
|
---|
249 | $/ = "\n";
|
---|
250 | close FILE;
|
---|
251 | &util::rm ($file);
|
---|
252 | }
|
---|
253 |
|
---|
254 | # creates a file called $outfile.final (should only be called if -out option
|
---|
255 | # is used and isn't STDERR or STDOUT) and writes an output code to it.
|
---|
256 | # An output code of 0 specifies that there was no error
|
---|
257 | sub final_out {
|
---|
258 | my ($exit_code) = @_;
|
---|
259 |
|
---|
260 | if (open (FINAL, ">$outfile.final")) {
|
---|
261 | print FINAL $exit_code;
|
---|
262 | close FINAL;
|
---|
263 | }
|
---|
264 | }
|
---|