source: trunk/gsdl/bin/script/build@ 1678

#!/usr/bin/perl

# This perl script may be called directly or by running build.bat on
# windows (build.bat is in bin\windows)

package build;

use FileHandle;
use File::Copy;

BEGIN {

    die "GSDLHOME not set - did you remember to source setup.bash (unix) or " .
    "run setup.bat (windows)?\n" unless defined $ENV{'GSDLHOME'};
    die "GSDLOS not set - did you remember to source setup.bash (unix) or " .
    "run setup.bat (windows)?\n" unless defined $ENV{'GSDLOS'};
    unshift (@INC, "$ENV{'GSDLHOME'}/perllib");

    STDOUT->autoflush(1);
    STDERR->autoflush(1);
}

use parsargv;
use util;
use cfgread;

# set up path - this allows for paths not to be supplied to system calls
# and overcomes problems when GSDLHOME contains spaces (double quoting
# the call doesn't work on win2k and probably other variants of winnt)
my $path_separator = ":";
$path_separator = ";" if $ENV{'GSDLOS'} =~ /^windows$/;
$ENV{'PATH'} = &util::filename_cat($ENV{'GSDLHOME'}, "bin", $ENV{'GSDLOS'}) .
    $path_separator . &util::filename_cat($ENV{'GSDLHOME'}, "bin", "script") .
    $path_separator . $ENV{'PATH'};

&parse_args (\@ARGV);

my ($collection) = @ARGV;

if (!defined $collection || $collection !~ /\w/) {
    print STDERR "You must specify a collection to build\n";
    &print_usage();
    die "\n";
}

if ($optionfile =~ /\w/) {
    open (OPTIONS, $optionfile) || die "Couldn't open $optionfile\n";
    my $line = [];
    my $options = [];
    while (defined ($line = &cfgread::read_cfg_line ('build::OPTIONS'))) {
    push (@$options, @$line);
    }
    close OPTIONS;
    &parse_args ($options);
}

if ($maxdocs == -1) {
    $maxdocs = "";
} else {
    $maxdocs = "-maxdocs $maxdocs";
}

my $cdir = $collectdir;
$cdir = &util::filename_cat ($ENV{'GSDLHOME'}, "collect") unless $collectdir =~ /\w/;
my $importdir = &util::filename_cat ($cdir, $collection, "import");
my $archivedir = &util::filename_cat ($cdir, $collection, "archives");
my $buildingdir = &util::filename_cat ($cdir, $collection, "building");
my $indexdir = &util::filename_cat ($cdir, $collection, "index");
my $bindir = &util::filename_cat ($ENV{'GSDLHOME'}, "bin");

my $use_out = 0;
my $outfile = $out;
if ($out !~ /^(STDERR|STDOUT)$/i) {
    open (OUT, ">$out") || die "Couldn't open output file $out\n";
    $out = "OUT";
    
    # delete any existing .final file
    &util::rm ("$outfile.final") if -e "$outfile.final";

    $use_out = 1;
}
$out->autoflush(1);

# delete any .kill file left laying around from a previously aborted build
if (-e &util::filename_cat ($cdir, $collection, ".kill")) {
    &util::rm (&util::filename_cat ($cdir, $collection, ".kill"));
}

&main();

close OUT if $use_out;

sub print_usage {
    print STDERR "\n  usage: $0 [options] collection-name\n\n";
    print STDERR "  options:\n";
    print STDERR "   -optionfile file        Get options from file, useful on systems where\n";
    print STDERR "                           long command lines may cause problems\n";
    print STDERR "   -append                 Add new files to existing collection\n";
    print STDERR "   -remove_archives        Remove archives directory after successfully\n";
    print STDERR "                           building the collection.\n";
    print STDERR "   -remove_import          Remove import directory after successfully\n";
    print STDERR "                           importing the collection.\n";
    print STDERR "   -buildtype build|import If 'build' attempt to build directly\n";
    print STDERR "                           from archives directory (bypassing import\n";
    print STDERR "                           stage). Defaults to 'import'\n";
    print STDERR "   -maxdocs number         Maximum number of documents to build\n";
    print STDERR "   -download directory     Directory (or file) to get import documents from.\n";
    print STDERR "                           There may be multiple download directories and they\n";
    print STDERR "                           may be of type http://, ftp://, or file://\n";
    print STDERR "                           Note that any existing import directory will be\n";
    print STDERR "                           deleted to make way for the downloaded data if\n";
    print STDERR "                           a -download option is supplied\n";
    print STDERR "   -collectdir directory   Collection directory (defaults to " .
    &util::filename_cat ($ENV{'GSDLHOME'}, "collect") . ")\n";
    print STDERR "   -dontinstall            Only applicable if -collectdir is set to something\n";
    print STDERR "                           other than the default. -dontinstall will suppress the\n";
    print STDERR "                           default behaviour which is to install the collection to\n";
    print STDERR "                           the gsdl/collect directory once it has been built.\n";
    print STDERR "   -save_archives          Create a copy of the existing archives directory called\n";
    print STDERR "                           archives.org\n";
    print STDERR "   -out                    Filename or handle to print output status to.\n";
    print STDERR "                           The default is STDERR\n\n";
}

sub main {

    if ($save_archives && -d $archivedir) {
    print $out "caching original archives to ${archivedir}.org\n";
    &util::cp_r ($archivedir, "${archivedir}.org");
    }

    # do the download thing if we have any -download options
    if (scalar (@download)) {
    # remove any existing import data
    if (&has_content ($importdir)) {
        print $out "build: WARNING: removing contents of $importdir\n";
        &util::rm_r ($importdir);
    }
    
    foreach $download_dir (@download) {

        # remove any leading or trailing whitespace from filenames (just in case)
        $download_dir =~ s/^\s+//;
        $download_dir =~ s/\s+$//;
        
        if ($download_dir =~ /^http:\/\//) {
        # http download

        } elsif ($download_dir =~ /^ftp:\/\//) {
        # ftp download

        } else {
        # we assume anything not beginning with http:// or ftp://
        # is a file or directory on the local file system.
        $download_dir =~ s/^file:(\/\/)?//;
        $download_dir =~ s/^\s+//; # may be whitespace between "file://" and the rest
        
        if (-e $download_dir) {
            # copy download_dir and all it contains to the import directory
            my $download_cmd = "perl -S filecopy.pl";
            $download_cmd .= " -collectdir \"$collectdir\"" if $collectdir =~ /\w/;
            $download_cmd .= " -out \"$outfile.download\"" if $use_out;
            $download_cmd .= " \"" . $download_dir . "\" " . $collection;
            system ($download_cmd);
            # if using output directory append the file download output to it
            &append_file ($out, "$outfile.download");
        } else {
            print $out "WARNING: '$download_dir' does not exist\n";
        }
        }
    }
    }

    if (-e &util::filename_cat ($archivedir, "archives.inf")) {
    if (&has_content ($importdir)) {
        if ($buildtype eq "build") {
        &gsdl_build();
        } else {
        &gsdl_import();
        &gsdl_build();
        }
    } else {
        # there are archives but no import, build directly from archives
        print $out "build: no import material was found, building directly\n";
        print $out "       from archives\n";
        &gsdl_build();
    }
    } else {
    if (&has_content ($importdir)) {
        if ($buildtype eq "build") {
        print $out "build: can't build directly from archives as no\n";
        print $out "       imported archives exist (did you forget to\n";
        print $out "       move the contents of $collection/import to\n";
        print $out "       collection/archives?)\n";
        }
        &gsdl_import();
        &gsdl_build();
    } else {
        # no import or archives
        print $out "build: ERROR: The $collection collection has no import or archives data.\n";
        &final_out (1) if $use_out;
        die "\n";
    }
    }

    if ($collectdir ne "" && !$dontinstall) {
    my $install_collectdir = &util::filename_cat ($ENV{'GSDLHOME'}, "collect");
    if (!&util::filenames_equal ($collectdir, $install_collectdir)) {
        
        # install collection to gsdl/collect
        print $out "installing the $collection collection\n";
        my $newdir = &util::filename_cat ($install_collectdir, $collection);
        my $olddir = &util::filename_cat ($collectdir, $collection);
        if (-d $newdir) {
        print $out "build: Could not install collection as $newdir\n";
        print $out "       already exists. Collection will remain at\n";
        print $out "       $olddir\n";
        &final_out (4) if $use_out;
        die "\n";
        }
        if (!&File::Copy::move ($olddir, $newdir)) {
        print $out "build: Failed to install collection to $newdir\n";
        print $out "       Collection will remain at $olddir\n";
        &final_out (5) if $use_out;
        die "\n";
        }
    }
    }

    &final_out (0) if $use_out;
}

sub gsdl_import {

    print $out "importing the $collection collection\n\n";

    my $import_cmd = "perl -S import.pl";
    $import_cmd .= " -out \"$outfile.import\"" if $use_out;
    $import_cmd .= " -removeold" unless $append;
    $import_cmd .= " -collectdir \"$collectdir\"" if $collectdir =~ /\w/;
    $import_cmd .= " $maxdocs $collection";
    system ($import_cmd);
    # if using output directory append the import output to it
    &append_file ($out, "$outfile.import");

    if (-e &util::filename_cat ($archivedir, "archives.inf")) {
    print $out "$collection collection imported successfully\n\n";
    if ($remove_import) {
        print $out "removing import directory ($importdir)\n";
        &util::rm_r ($importdir);
    }
    } else {
    &final_out (2) if $use_out;
    print $out "\nimport.pl failed\n";
    die "\n";
    }
}

sub gsdl_build {

    print $out "building the $collection collection\n\n";

    my $build_cmd = "perl -S buildcol.pl";
    $build_cmd .= " -out \"$outfile.build\"" if $use_out;
    $build_cmd .= " -collectdir \"$collectdir\"" if $collectdir =~ /\w/;
    $build_cmd .= " $maxdocs $collection";
    system ($build_cmd);
    # if using output directory append the buildcol output to it
    &append_file ($out, "$outfile.build");

    if (-e &util::filename_cat ($buildingdir, "text", "$collection.ldb") ||
    -e &util::filename_cat ($buildingdir, "text", "$collection.bdb")) {
    print $out "$collection collection built successfully\n\n";
    if ($remove_archives) {
        print $out "removing archives directory ($archivedir)\n";
        &util::rm_r ($archivedir);
    }
    } else {
    &final_out (3) if $use_out;
    print $out "\nbuildcol.pl failed\n";
    die "\n";
    }

    # replace old indexes with new ones
    if (&has_content ($indexdir)) {
    print $out "removing old indexes\n";
    &util::rm_r ($indexdir);
    }
    rmdir ($indexdir) if -d $indexdir;
    &File::Copy::move ($buildingdir, $indexdir);

    # remove the cached arhives
    if ($save_archives && -d "${archivedir}.org") {
    &util::rm_r ("${archivedir}.org");
    }
}

sub has_content {
    my ($dir) = @_;

    if (!-d $dir) {return 0;}

    opendir (DIR, $dir) || return 0;
    my @files = readdir DIR;
    close DIR;
    
    foreach my $file (@files) {
    if ($file !~ /^\.{1,2}$/) {
        return 1;
    }
    }
    return 0;
}

sub append_file {
    my ($handle, $file) = @_;

    open (FILE, $file) || return;
    undef $/;
    print $handle <FILE>;
    $/ = "\n";
    close FILE;
    &util::rm ($file);
}

# creates a file called $outfile.final (should only be called if -out option
# is used and isn't STDERR or STDOUT) and writes an output code to it.
# An output code of 0 specifies that there was no error
sub final_out {
    my ($exit_code) = @_;

    if (open (FINAL, ">$outfile.final")) {
    print FINAL $exit_code;
    close FINAL;
    }
}

sub parse_args {
    my ($argref) = @_;

    if (!parsargv::parse($argref, 
             'optionfile/.*/', \$optionfile,
             'append', \$append,
             'remove_archives', \$remove_archives,
             'remove_import', \$remove_import,
             'buildtype/^(build|import)$/import', \$buildtype,
             'maxdocs/^\-?\d+/-1', \$maxdocs,
             'download/.+', \@download,
             'collectdir/.*/', \$collectdir,
             'dontinstall', \$dontinstall,
             'save_archives', \$save_archives,
             'out/.*/STDERR', \$out)) {
    
    &print_usage();
    die "\n";
    }
}