# $Id: Index.pm 84 2020-05-31 06:29:34Z stro $

package CPAN::SQLite::Index;
use strict;
use warnings;

our $VERSION = '0.219';

use English qw/-no_match_vars/;

use CPAN::SQLite::Info;
use CPAN::SQLite::State;
use CPAN::SQLite::Populate;
use CPAN::SQLite::DBI qw($tables);
use File::Spec::Functions qw(catfile);
use File::Basename;
use File::Path;
use HTTP::Tiny;

use Scalar::Util 'weaken';


our ($oldout);
my $log_file = 'cpan_sqlite_log.' . time;

# This is usually already defined in real life, but tests need it to be set
$CPAN::FrontEnd ||= "CPAN::Shell";

sub new {
  my ($class, %args) = @_;
  if ($args{setup} and $args{reindex}) {
    die "Reindexing must be done on an exisiting database";

  my $self = { index => undef, state => undef, %args };

  return bless $self, $class;

sub download_index {
  my $self = shift;

    $ENV{'CPAN_SQLITE_DOWNLOAD_URL'} = 'http://cpansqlite.trouchelle.com/' unless $ENV{'CPAN_SQLITE_DOWNLOAD_URL'};

  return 0 unless $ENV{'CPAN_SQLITE_DOWNLOAD_URL'};

  $CPAN::FrontEnd->myprint("Downloading the compiled index db ... ");

  if (my $response =
    HTTP::Tiny->new->mirror($ENV{'CPAN_SQLITE_DOWNLOAD_URL'} => catfile($self->{'db_dir'}, $self->{'db_name'})))
    if ($response->{'success'} and $response->{'status'} and $response->{'status'} eq '200') {
      if (my $type = $response->{'headers'}->{'content-type'}) {
        if ($type eq 'application/x-sqlite3') {
          return 1;

  $CPAN::FrontEnd->mywarn('Cannot download the compiled index db');
  return 0;

sub index {
  my $self  = shift;
  my $setup = $self->{'setup'};

  if ($setup) {
    my $db_name = catfile($self->{'db_dir'}, $self->{db_name});
    if (-f $db_name) {
      $CPAN::FrontEnd->myprint("Removing existing $db_name ... ");
      if (unlink $db_name) {
      } else {
        $CPAN::FrontEnd->mywarn("Failed: $!\n");

  my $log = catfile($self->{'log_dir'}, $log_file);

    $oldout = error_fh($log);

  my $log_cleanup = $ENV{'CPAN_SQLITE_LOG_FILES_CLEANUP'};
  $log_cleanup = 30 unless defined $log_cleanup;
  if ($log_cleanup and $log_cleanup =~ /^\d+$/) {
    if (opendir(my $DIR, $self->{'log_dir'})) {
      my @files = grep { /cpan_sqlite_log\./ } readdir $DIR;
      closedir $DIR;

      @files = grep { -C $_ > $log_cleanup } map { catfile($self->{'log_dir'}, $_) } @files;

      if (@files) {
        $CPAN::FrontEnd->myprint('Cleaning old log files ... ');
        unlink @files;

  if ($self->download_index()) {
    return 1;

  if ($self->{'update_indices'}) {
    $CPAN::FrontEnd->myprint('Fetching index files ... ');
    if ($self->fetch_cpan_indices()) {
    } else {

  $CPAN::FrontEnd->myprint('Gathering information from index files ... ');
  if ($self->fetch_info()) {
  } else {

  unless ($setup) {
    $CPAN::FrontEnd->myprint('Obtaining current state of database ... ');
    if ($self->state()) {
    } else {

  $CPAN::FrontEnd->myprint('Populating database tables ... ');
  if ($self->populate()) {
  } else {

  return 1;

sub fetch_cpan_indices {
  my $self = shift;

  my $CPAN    = $self->{CPAN};
  my $indices = {
    '01mailrc.txt.gz'           => 'authors',
    '02packages.details.txt.gz' => 'modules',

  foreach my $index (keys %$indices) {
    my $file = catfile($CPAN, $indices->{$index}, $index);
    next if (-e $file and -M $file < 1);
    my $dir = dirname($file);
    unless (-d $dir) {
      mkpath($dir, 0, oct(755)) or die "Cannot mkpath $dir: $!";
    my @urllist = @{ $self->{urllist} };
    foreach my $cpan (@urllist) {
      my $from = join '/', ($cpan, $indices->{$index}, $index);
      if (my $response = HTTP::Tiny->new->get($from)) {
        if ($response->{'success'}) {
          if (open(my $FILE, '>', $file)) {
            binmode $FILE;
            print $FILE $response->{'content'};
            if (close($FILE)) {
    unless (-f $file) {
      $CPAN::FrontEnd->mywarn("Cannot retrieve '$file'");
  return 1;

sub fetch_info {
  my $self   = shift;
  my %wanted = map { $_ => $self->{$_} } qw(CPAN ignore keep_source_where);
  my $info   = CPAN::SQLite::Info->new(%wanted);
  $info->fetch_info() or return;
  my @tables = qw(dists mods auths info);
  my $index;
  foreach my $table (@tables) {
    my $class = __PACKAGE__ . '::' . $table;
    my $this = { info => $info->{$table} };
    $index->{$table} = bless $this, $class;
  $self->{index} = $index;
  return 1;

sub state {
  my $self = shift;

  my %wanted = map { $_ => $self->{$_} } qw(db_name index setup reindex db_dir);
  my $state = CPAN::SQLite::State->new(%wanted);
  $state->state() or return;
  $self->{state} = $state;
  return 1;

sub populate {
  my $self   = shift;
  my %wanted = map { $_ => $self->{$_} } qw(db_name index setup state db_dir);
  my $db     = CPAN::SQLite::Populate->new(%wanted);
  $db->populate() or return;
  return 1;

sub error_fh {
  my $file = shift;
  open(my $tmp, '>', $file) or die "Cannot open $file: $!";
  close $tmp;

  # Should be open(my $oldout, '>&', \*STDOUT); but it fails on 5.6.2
  open(my $oldout, '>&STDOUT');
  open(STDOUT, '>', $file) or die "Cannot tie STDOUT to $file: $!";
  select STDOUT;
  $| = 1;
  return $oldout;

    close STDOUT;
    open(STDOUT, '>&', $oldout) if $oldout;


=head1 NAME

CPAN::SQLite::Index - set up or update database tables.

=head1 VERSION

version 0.219


 my $index = CPAN::SQLite::Index->new(setup => 1);


This is the main module used to set up or update the
database tables used to store information from the
CPAN and ppm indices. The creation of the object

 my $index = CPAN::SQLite::Index->new(%args);

accepts two possible arguments:

=over 3

=item * setup =E<gt> 1

This (optional) argument specifies that the database is being set up.
Any existing tables will be dropped.

=item * reindex =E<gt> value

This (optional) argument specifies distribution names that
one would like to reindex in an existing database. These may
be specified as either a scalar, for a single distribution,
or as an array reference for a list of distributions.


=head1 DETAILS



will start the indexing procedure. Various messages
detailing the progress will written to I<STDOUT>,
which by default will be captured into a file
F<cpan_sqlite_log.dddddddddd>, where the extension
is the C<time> that the method was invoked. Error messages
are not captured, and will appear in I<STDERR>.

The steps of the indexing procedure are as follows.

=over 4

=item * download existing pre-compiled index (optional)

already existing and up-to-date cpandb.sql file will be downloaded from
either specified URL or http://cpansqlite.trouchelle.com/ where it's
updated every hour. This greatly increases performance and decreases CPU
and memory consumption during the indexing process but if your CPAN
mirror is out-of-sync or you're using DarkPAN, it obviously wouldn't
work. It also wouldn't work without an internet connection.

See L<WWW::CPAN::SQLite> if you want to setup your own service for
pre-compiling the database.

If neither variable is set, this step is skipped.

=item * fetch index data

The necessary CPAN index files
F<$CPAN/authors/01mailrc.txt.gz> and
F<$CPAN/modules/02packages.details.txt.gz> will be fetched
from the CPAN mirror specified by the C<$cpan> variable
at the beginning of L<CPAN::SQLite::Index>. If you are
using this option, it is recommended to use the
same CPAN mirror with subsequent updates, to ensure consistency
of the database. As well, the information on the locations
of the CPAN mirrors used for Template-Toolkit and GeoIP
is written.

=item * get index information

Information from the CPAN indices is extracted through

=item * get state information

Unless the C<setup> argument within the C<new>
method of L<CPAN::SQLite::Index> is specified,
this will get information on the state of the database
through L<CPAN::SQLite::State>.
A comparison is then made between this information
and that gathered from the CPAN indices, and if there's
a discrepancy in some items, those items are marked
for either insertion, updating, or deletion, as appropriate.

=item * populate the database

At this stage the gathered information is used to populate
the database, through L<CPAN::SQLite::Populate>,
either inserting new items, updating
existing ones, or deleting obsolete items.


=head1 SEE ALSO

L<CPAN::SQLite::Info>, L<CPAN::SQLite::State>,
and L<CPAN::SQLite::Util>.
Development takes place on the CPAN-SQLite project
at L<http://sourceforge.net/projects/cpan-search/>.

=head1 AUTHORS

Randy Kobes (passed away on September 18, 2010)

Serguei Trouchelle E<lt>stro@cpan.orgE<gt>


Copyright 2006 by Randy Kobes E<lt>r.kobes@uwinnipeg.caE<gt>.

Copyright 2011 by Serguei Trouchelle E<lt>stro@cpan.orgE<gt>.

Use and redistribution are under the same terms as Perl itself.