path: root/Bugzilla/Install/Util.pm

# -*- Mode: perl; indent-tabs-mode: nil -*-
#
# The contents of this file are subject to the Mozilla Public
# License Version 1.1 (the "License"); you may not use this file
# except in compliance with the License. You may obtain a copy of
# the License at http://www.mozilla.org/MPL/
#
# Software distributed under the License is distributed on an "AS
# IS" basis, WITHOUT WARRANTY OF ANY KIND, either express or
# implied. See the License for the specific language governing
# rights and limitations under the License.
#
# The Original Code is the Bugzilla Bug Tracking System.
#
# The Initial Developer of the Original Code is Everything Solved.
# Portions created by Everything Solved are Copyright (C) 2006
# Everything Solved. All Rights Reserved.
#
# Contributor(s): Max Kanat-Alexander <mkanat@bugzilla.org>

package Bugzilla::Install::Util;

# The difference between this module and Bugzilla::Util is that this
# module may require *only* Bugzilla::Constants and built-in
# perl modules.

use strict;

use Bugzilla::Constants;

use File::Basename;
use POSIX qw(setlocale LC_CTYPE);
use Safe;

use base qw(Exporter);
our @EXPORT_OK = qw(
    bin_loc
    get_version_and_os
    indicate_progress
    install_string
    include_languages
    template_include_path
    vers_cmp
    get_console_locale
);

sub bin_loc {
    my ($bin) = @_;
    return '' if ON_WINDOWS;
    # Don't print any errors from "which"
    open(my $saveerr, ">&STDERR");
    open(STDERR, '>/dev/null');
    my $loc = `which $bin`;
    close(STDERR);
    open(STDERR, ">&", $saveerr);
    my $exit_code = $? >> 8; # See the perlvar manpage.
    return '' if $exit_code > 0;
    chomp($loc);
    return $loc;
}

sub get_version_and_os {
    # Display version information
    my @os_details = POSIX::uname;
    # 0 is the name of the OS, 2 is the major version,
    my $os_name = $os_details[0] . ' ' . $os_details[2];
    if (ON_WINDOWS) {
        require Win32;
        $os_name = Win32::GetOSName();
    }
    # $os_details[3] is the minor version.
    return { bz_ver   => BUGZILLA_VERSION,
             perl_ver => sprintf('%vd', $^V),
             os_name  => $os_name,
             os_ver   => $os_details[3] };
}

sub indicate_progress {
    my ($params) = @_;
    my $current = $params->{current};
    my $total   = $params->{total};
    my $every   = $params->{every} || 1;

    print "." if !($current % $every);
    if ($current == $total || $current % ($every * 60) == 0) {
        print "$current/$total (" . int($current * 100 / $total) . "%)\n";
    }
}

sub install_string {
    my ($string_id, $vars) = @_;
    _cache()->{template_include_path} ||= template_include_path();
    my $path = _cache()->{template_include_path};
    
    my $string_template;
    # Find the first template that defines this string.
    foreach my $dir (@$path) {
        my $base = "$dir/setup/strings";
        $string_template = _get_string_from_file($string_id, "$base.txt.pl")
            if !defined $string_template;
        last if defined $string_template;
    }
    
    die "No language defines the string '$string_id'"
        if !defined $string_template;

    $vars ||= {};
    my @replace_keys = keys %$vars;
    foreach my $key (@replace_keys) {
        my $replacement = $vars->{$key};
        die "'$key' in '$string_id' is tainted: '$replacement'"
            if is_tainted($replacement);
        # We don't want people to start getting clever and inserting
        # ##variable## into their values. So we check if any other
        # key is listed in the *replacement* string, before doing
        # the replacement. This is mostly to protect programmers from
        # making mistakes.
        if (grep($replacement =~ /##$key##/, @replace_keys)) {
            die "Unsafe replacement for '$key' in '$string_id': '$replacement'";
        }
        $string_template =~ s/\Q##$key##\E/$replacement/g;
    }
    
    return $string_template;
}

sub include_languages {
    my ($params) = @_;
    $params ||= {};

    # Basically, the way this works is that we have a list of languages
    # that we *want*, and a list of languages that Bugzilla actually
    # supports. The caller tells us what languages they want, by setting
    # $ENV{HTTP_ACCEPT_LANGUAGE} or $params->{only_language}. The languages
    # we support are those specified in $params->{use_languages}. Otherwise
    # we support every language installed in the template/ directory.
    
    my @wanted;
    if ($params->{only_language}) {
        @wanted = ($params->{only_language});
    }
    else {
        @wanted = _sort_accept_language($ENV{'HTTP_ACCEPT_LANGUAGE'} || '');
    }
    
    my @supported;
    if (defined $params->{use_languages}) {
        @supported = @{$params->{use_languages}};
    }
    else {
        my @dirs = glob(bz_locations()->{'templatedir'} . "/*");
        @dirs = map(basename($_), @dirs);
        @supported = grep($_ ne 'CVS', @dirs);
    }
    
    my @usedlanguages;
    foreach my $wanted (@wanted) {
        # If we support the language we want, or *any version* of
        # the language we want, it gets pushed into @usedlanguages.
        #
        # Per RFC 1766 and RFC 2616, things like 'en' match 'en-us' and
        # 'en-uk', but not the other way around. (This is unfortunately
        # not very clearly stated in those RFC; see comment just over 14.5
        # in http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.4)
        if(my @found = grep /^\Q$wanted\E(-.+)?$/i, @supported) {
            push (@usedlanguages, @found);
        }
    }

    # We always include English at the bottom if it's not there, even if
    # somebody removed it from use_languages.
    if (!grep($_ eq 'en', @usedlanguages)) {
        push(@usedlanguages, 'en');
    }

    return @usedlanguages;
}
    
sub template_include_path {
    my @usedlanguages = include_languages(@_);
    # Now, we add template directories in the order they will be searched:
    
    # First, we add extension template directories, because extension templates
    # override standard templates. Extensions may be localized in the same way
    # that Bugzilla templates are localized.
    my @include_path;
    my @extensions = glob(bz_locations()->{'extensionsdir'} . "/*");
    foreach my $extension (@extensions) {
        foreach my $lang (@usedlanguages) {
            _add_language_set(\@include_path, $lang, "$extension/template");
        }
    }
    
    # Then, we add normal template directories, sorted by language.
    foreach my $lang (@usedlanguages) {
        _add_language_set(\@include_path, $lang);
    }
    
    return \@include_path;
}

# This is taken straight from Sort::Versions 1.5, which is not included
# with perl by default.
sub vers_cmp {
    my ($a, $b) = @_;

    # Remove leading zeroes - Bug 344661
    $a =~ s/^0*(\d.+)/$1/;
    $b =~ s/^0*(\d.+)/$1/;

    my @A = ($a =~ /([-.]|\d+|[^-.\d]+)/g);
    my @B = ($b =~ /([-.]|\d+|[^-.\d]+)/g);

    my ($A, $B);
    while (@A and @B) {
        $A = shift @A;
        $B = shift @B;
        if ($A eq '-' and $B eq '-') {
            next;
        } elsif ( $A eq '-' ) {
            return -1;
        } elsif ( $B eq '-') {
            return 1;
        } elsif ($A eq '.' and $B eq '.') {
            next;
        } elsif ( $A eq '.' ) {
            return -1;
        } elsif ( $B eq '.' ) {
            return 1;
        } elsif ($A =~ /^\d+$/ and $B =~ /^\d+$/) {
            if ($A =~ /^0/ || $B =~ /^0/) {
                return $A cmp $B if $A cmp $B;
            } else {
                return $A <=> $B if $A <=> $B;
            }
        } else {
            $A = uc $A;
            $B = uc $B;
            return $A cmp $B if $A cmp $B;
        }
    }
    @A <=> @B;
}

######################
# Helper Subroutines #
######################

# Used by install_string
sub _get_string_from_file {
    my ($string_id, $file) = @_;
    
    return undef if !-e $file;
    my $safe = new Safe;
    $safe->rdo($file);
    my %strings = %{$safe->varglob('strings')};
    return $strings{$string_id};
}

# Used by template_include_path.
sub _add_language_set {
    my ($array, $lang, $templatedir) = @_;
    
    $templatedir ||= bz_locations()->{'templatedir'};
    my @add = ("$templatedir/$lang/custom", "$templatedir/$lang/default");
    
    my $project = bz_locations->{'project'};
    unshift(@add, "$templatedir/$lang/$project") if $project;
    
    foreach my $dir (@add) {
        if (-d $dir) {
            trick_taint($dir);
            push(@$array, $dir);
        }
    }
}

# Make an ordered list out of a HTTP Accept-Language header (see RFC 2616, 14.4)
# We ignore '*' and <language-range>;q=0
# For languages with the same priority q the order remains unchanged.
sub _sort_accept_language {
    sub sortQvalue { $b->{'qvalue'} <=> $a->{'qvalue'} }
    my $accept_language = $_[0];

    # clean up string.
    $accept_language =~ s/[^A-Za-z;q=0-9\.\-,]//g;
    my @qlanguages;
    my @languages;
    foreach(split /,/, $accept_language) {
        if (m/([A-Za-z\-]+)(?:;q=(\d(?:\.\d+)))?/) {
            my $lang   = $1;
            my $qvalue = $2;
            $qvalue = 1 if not defined $qvalue;
            next if $qvalue == 0;
            $qvalue = 1 if $qvalue > 1;
            push(@qlanguages, {'qvalue' => $qvalue, 'language' => $lang});
        }
    }

    return map($_->{'language'}, (sort sortQvalue @qlanguages));
}

sub get_console_locale {
    require Locale::Language;
    my $locale = setlocale(LC_CTYPE);
    my $language;
    # Some distros set e.g. LC_CTYPE = fr_CH.UTF-8. We clean it up.
    if ($locale =~ /^([^\.]+)/) {
        $locale = $1;
    }
    $locale =~ s/_/-/;
    # It's pretty sure that there is no language pack of the form fr-CH
    # installed, so we also include fr as a wanted language.
    if ($locale =~ /^(\S+)\-/) {
        $language = $1;
        $locale .= ",$language";
    }
    else {
        $language = $locale;
    }

    # Some OSs or distributions may have setlocale return a string of the form
    # German_Germany.1252 (this example taken from a Windows XP system), which
    # is unsuitable for our needs because Bugzilla works on language codes.
    # We try and convert them here.
    if ($language = Locale::Language::language2code($language)) {
        $locale .= ",$language";
    }

    return $locale;
}


# This is like request_cache, but it's used only by installation code
# for setup.cgi and things like that.
our $_cache = {};
sub _cache {
    if ($ENV{MOD_PERL}) {
        require Apache2::RequestUtil;
        return Apache2::RequestUtil->request->pnotes();
    }
    return $_cache;
}

###############################
# Copied from Bugzilla::Util #
##############################

sub trick_taint {
    require Carp;
    Carp::confess("Undef to trick_taint") unless defined $_[0];
    my $match = $_[0] =~ /^(.*)$/s;
    $_[0] = $match ? $1 : undef;
    return (defined($_[0]));
}

sub is_tainted {
    return not eval { my $foo = join('',@_), kill 0; 1; };
}

__END__

=head1 NAME

Bugzilla::Install::Util - Utility functions that are useful both during
installation and afterwards.

=head1 DESCRIPTION

This module contains various subroutines that are used primarily
during installation. However, these subroutines can also be useful to
non-installation code, so they have been split out into this module.

The difference between this module and L<Bugzilla::Util> is that this
module is safe to C<use> anywhere in Bugzilla, even during installation,
because it depends only on L<Bugzilla::Constants> and built-in perl modules.

None of the subroutines are exported by default--you must explicitly
export them.

=head1 SUBROUTINES

=over

=item C<bin_loc>

On *nix systems, given the name of a binary, returns the path to that
binary, if the binary is in the C<PATH>.

=item C<get_version_and_os>

Returns a hash containing information about what version of Bugzilla we're
running, what perl version we're using, and what OS we're running on.

=item C<get_console_locale>

Returns the language to use based on the LC_CTYPE value returned by the OS.
If LC_CTYPE is of the form fr-CH, then fr is appended to the list.

=item C<indicate_progress>

=over

=item B<Description>

This prints out lines of dots as a long update is going on, to let the user
know where we are and that we're not frozen. A new line of dots will start
every 60 dots.

Sample usage: C<indicate_progress({ total =E<gt> $total, current =E<gt>
$count, every =E<gt> 1 })>

=item B<Sample Output>

Here's some sample output with C<total = 1000> and C<every = 10>:

 ............................................................600/1000 (60%)
 ........................................

=item B<Params>

=over

=item C<total> - The total number of items we're processing.

=item C<current> - The number of the current item we're processing.

=item C<every> - How often the function should print out a dot.
For example, if this is 10, the function will print out a dot every
ten items. Defaults to 1 if not specified.

=back

=item B<Returns>: nothing

=back

=item C<install_string>

=over

=item B<Description>

This is a very simple method of templating strings for installation.
It should only be used by code that has to run before the Template Toolkit
can be used. (See the comments at the top of the various L<Bugzilla::Install>
modules to find out when it's safe to use Template Toolkit.)

It pulls strings out of the F<strings.txt.pl> "template" and replaces
any variable surrounded by double-hashes (##) with a value you specify.

This allows for localization of strings used during installation.

=item B<Example>

Let's say your template string looks like this:

 The ##animal## jumped over the ##plant##.

Let's say that string is called 'animal_jump_plant'. So you call the function
like this:

 install_string('animal_jump_plant', { animal => 'fox', plant => 'tree' });

That will output this:

 The fox jumped over the tree.

=item B<Params>

=over

=item C<$string_id> - The name of the string from F<strings.txt.pl>.

=item C<$vars> - A hashref containing the replacement values for variables
inside of the string.

=back

=item B<Returns>: The appropriate string, with variables replaced.

=back

=item C<template_include_path>

Used by L<Bugzilla::Template> and L</install_string> to determine the
directories where templates are installed. Templates can be installed
in many places. They're listed here in the basic order that they're
searched:

=over

=item extensions/C<$extension>/template/C<$language>/C<$project>

=item extensions/C<$extension>/template/C<$language>/custom

=item extensions/C<$extension>/template/C<$language>/default

=item template/C<$language>/C<$project>

=item template/C<$language>/custom

=item template/C<$language>/default

=back

C<$project> has to do with installations that are using the C<$ENV{PROJECT}>
variable to have different "views" on a single Bugzilla.

The F<default> directory includes templates shipped with Bugzilla.

The F<custom> directory is a directory for local installations to override
the F<default> templates. Any individual template in F<custom> will
override a template of the same name and path in F<default>.

C<$language> is a language code, C<en> being the default language shipped
with Bugzilla. Localizers ship other languages.

C<$extension> is the name of any directory in the F<extensions/> directory.
Each extension has its own directory.

Note that languages are sorted by the user's preference (as specified
in their browser, usually), and extensions are sorted alphabetically.

=item C<include_languages>

Used by L<Bugzilla::Template> to determine the languages' list which 
are compiled with the browser's I<Accept-Language> and the languages 
of installed templates.

=item C<vers_cmp>

=over

=item B<Description>

This is a comparison function, like you would use in C<sort>, except that
it compares two version numbers. So, for example, 2.10 would be greater
than 2.2.

It's based on versioncmp from L<Sort::Versions>, with some Bugzilla-specific
fixes.

=item B<Params>: C<$a> and C<$b> - The versions you want to compare.

=item B<Returns>

C<-1> if C<$a> is less than C<$b>, C<0> if they are equal, or C<1> if C<$a>
is greater than C<$b>.

=back

=back