mirror of
https://github.com/webmin/webmin.git
synced 2025-07-21 23:40:34 +00:00
Add second step in supporting "help/" transcoding/translation
This commit is contained in:
Ilia Rostovtsev
@ -3,7 +3,7 @@
|
||||
|
||||
use strict;
|
||||
use warnings;
|
||||
use 5.010;
|
||||
use 5.014;
|
||||
|
||||
use File::Spec;
|
||||
use File::Basename;
|
||||
@ -87,7 +87,7 @@ sub main
|
||||
$opt{'translate-format'} = 'text';
|
||||
|
||||
# Force HTML format for "module" type translates
|
||||
if (defined($opt{'type'}) && $opt{'type'} eq 'module') {
|
||||
if (defined($opt{'type'}) && $opt{'type'} =~ /module|help/) {
|
||||
$opt{'translate-format'} = 'html';
|
||||
}
|
||||
|
||||
@ -181,8 +181,7 @@ sub main
|
||||
},
|
||||
},
|
||||
$path);
|
||||
@modules = uniq(@modules);
|
||||
@modules = sort @modules;
|
||||
@modules = sort(uniq(@modules));
|
||||
}
|
||||
$data{'modules'} = \@modules;
|
||||
|
||||
@ -360,58 +359,14 @@ sub language_source_file
|
||||
|
||||
# Force encoding based on map
|
||||
if ($opt->{'mode'} eq 'full' && $opt->{'language-source-encoding'} eq 'map') {
|
||||
my $map_auto;
|
||||
my $code = $data{'language-code'};
|
||||
|
||||
if ($code eq 'ja') {
|
||||
$language_source_file_encoding = "euc-jp";
|
||||
} elsif ($code eq 'ko') {
|
||||
$language_source_file_encoding = "euc-kr";
|
||||
} elsif ($code eq 'ru' || $code eq 'bg' || $code eq 'uk') {
|
||||
$language_source_file_encoding = "cp1251";
|
||||
} elsif ($code eq 'ca' ||
|
||||
$code eq 'fr' ||
|
||||
$code eq 'hr' ||
|
||||
$code eq 'lt' ||
|
||||
$code eq 'no')
|
||||
{
|
||||
$language_source_file_encoding = "cp1252";
|
||||
} elsif ($code eq 'cs' ||
|
||||
$code eq 'sk' ||
|
||||
$code eq 'pl' ||
|
||||
$code eq 'sl' ||
|
||||
$code eq 'hu')
|
||||
{
|
||||
$language_source_file_encoding = "iso-8859-2";
|
||||
} elsif ($code eq 'tr') {
|
||||
$language_source_file_encoding = "iso-8859-9";
|
||||
} elsif ($code eq 'he') {
|
||||
$language_source_file_encoding = "cp1255";
|
||||
} elsif ($code eq 'th') {
|
||||
$language_source_file_encoding = "tis-620";
|
||||
} elsif ($code eq 'zh') {
|
||||
$language_source_file_encoding = "gb2312";
|
||||
} elsif ($code eq 'zh_TW') {
|
||||
$language_source_file_encoding = "big5";
|
||||
} else {
|
||||
my $language_source_file_data = read_file_contents($language_source_file_target);
|
||||
my $detected_encoding = Encode::Detect::Detector::detect($language_source_file_data);
|
||||
if ($detected_encoding) {
|
||||
$language_source_file_encoding = $detected_encoding;
|
||||
$map_auto = " (auto)";
|
||||
} else {
|
||||
$language_source_file_encoding = 'utf-8';
|
||||
$map_auto = " (auto enforced)";
|
||||
}
|
||||
}
|
||||
|
||||
my ($language_source_file_encoding, $map_auto) = language_file_encoding($code, $language_source_file_target);
|
||||
talk_log(
|
||||
("" . CYAN . " .. Force file encoding to \`$language_source_file_encoding\`" .
|
||||
($map_auto // '') . " as derived from language map" . RESET . ""
|
||||
),
|
||||
$data,
|
||||
1);
|
||||
|
||||
}
|
||||
|
||||
# Figure out encoding automatically
|
||||
@ -447,6 +402,56 @@ sub language_source_file
|
||||
$language_source_file_target);
|
||||
}
|
||||
|
||||
sub language_file_encoding
|
||||
{
|
||||
my ($code, $file) = @_;
|
||||
my $encoding = 'utf-8';
|
||||
my $auto;
|
||||
if ($code eq 'ja') {
|
||||
$encoding = "euc-jp";
|
||||
} elsif ($code eq 'ko') {
|
||||
$encoding = "euc-kr";
|
||||
} elsif ($code eq 'ru' || $code eq 'bg' || $code eq 'uk') {
|
||||
$encoding = "cp1251";
|
||||
} elsif ($code eq 'ca' ||
|
||||
$code eq 'fr' ||
|
||||
$code eq 'hr' ||
|
||||
$code eq 'lt' ||
|
||||
$code eq 'no')
|
||||
{
|
||||
$encoding = "cp1252";
|
||||
} elsif ($code eq 'cs' ||
|
||||
$code eq 'sk' ||
|
||||
$code eq 'pl' ||
|
||||
$code eq 'sl' ||
|
||||
$code eq 'hu')
|
||||
{
|
||||
$encoding = "iso-8859-2";
|
||||
} elsif ($code eq 'tr') {
|
||||
$encoding = "iso-8859-9";
|
||||
} elsif ($code eq 'he') {
|
||||
$encoding = "cp1255";
|
||||
} elsif ($code eq 'th') {
|
||||
$encoding = "tis-620";
|
||||
} elsif ($code eq 'zh') {
|
||||
$encoding = "gb2312";
|
||||
} elsif ($code eq 'zh_TW') {
|
||||
$encoding = "big5";
|
||||
} else {
|
||||
my $file_data = read_file_contents($file);
|
||||
my $detected = Encode::Detect::Detector::detect($file);
|
||||
if ($detected) {
|
||||
$encoding = $detected;
|
||||
$auto = " (auto)";
|
||||
} else {
|
||||
$encoding = 'utf-8';
|
||||
$auto = " (auto enforced)";
|
||||
}
|
||||
}
|
||||
|
||||
return ($encoding, $auto);
|
||||
}
|
||||
|
||||
sub language_transcode
|
||||
{
|
||||
my ($string, $encoding, $opt) = @_;
|
||||
@ -804,7 +809,157 @@ sub go
|
||||
my $mode_sync = $opt->{'mode'} ne 'full';
|
||||
my $verbose_silent_mode = $mode_sync && $verbose != 2;
|
||||
|
||||
if ($type ne 'help') {
|
||||
if ($type eq 'help') {
|
||||
foreach $module (@{$modules}) {
|
||||
my @module_help;
|
||||
my @module_help_info;
|
||||
my $ext = '.html';
|
||||
my $utf8 = '.UTF-8';
|
||||
my $old_map = $language_source_encoding eq 'map';
|
||||
|
||||
my ($exists, $help_path) = source_data($module, $data, $opt);
|
||||
|
||||
# Build targets first
|
||||
talk_log(("Transcoding/translating " . CYAN BOLD, $module, RESET . " module's help .."), $data, 1);
|
||||
talk_log(("" . CYAN . " .. Building list of help files to process" . RESET . ""), $data, 1);
|
||||
find(
|
||||
{
|
||||
wanted => sub {
|
||||
my $found = $File::Find::name;
|
||||
my $found_nonutf8 = $found =~ s/$utf8//r;
|
||||
my $found_nonutf8_big5 = $found_nonutf8 =~ s/$ext/.Big5$ext/r;
|
||||
my $found_nonutf8_euc = $found_nonutf8 =~ s/$ext/.euc$ext/r;
|
||||
my $found_relative_name = $found =~ s/$data->{'path'}\/$module\/$type\///r;
|
||||
|
||||
# Check if file exists in both UTF-8 and original encoding, if so, keep original only
|
||||
my @found_nonutf8 = ($found_nonutf8, $found_nonutf8_big5, $found_nonutf8_euc);
|
||||
foreach my $found_nonutf8 (@found_nonutf8) {
|
||||
if (-r $found_nonutf8 && $found ne $found_nonutf8) {
|
||||
talk_log(
|
||||
("" . RED .
|
||||
" .. Deleting duplicate file in UTF-8 encoding..\n - $found_relative_name" .
|
||||
RESET . ""
|
||||
),
|
||||
$data,
|
||||
1);
|
||||
unlink($found);
|
||||
} elsif (-f $found) {
|
||||
push(@module_help, $found);
|
||||
push(@module_help_info, $found_relative_name);
|
||||
}
|
||||
}
|
||||
},
|
||||
},
|
||||
$help_path);
|
||||
@module_help = sort(uniq(@module_help));
|
||||
@module_help_info = sort(uniq(@module_help_info));
|
||||
|
||||
if (scalar(@module_help)) {
|
||||
talk_log(
|
||||
("" . GREEN . " .. Found help files to process" .
|
||||
RESET . " \n - @{[join(\"\n - \", @module_help_info)]}"
|
||||
),
|
||||
$data,
|
||||
1);
|
||||
}
|
||||
|
||||
# Store template files
|
||||
my @templates = ();
|
||||
|
||||
# Store which human translated files do we have
|
||||
my @help_translated = ();
|
||||
|
||||
foreach $language (@{ $data->{'languages_source_list'} }) {
|
||||
|
||||
# Get target language code and other attributes
|
||||
my $code = $language->{'lang'};
|
||||
my $code_ = language_map($code);
|
||||
my $code__ = $code_ =~ s/\.(euc|Big5)//r;
|
||||
my $rtl = $language->{'rtl'};
|
||||
|
||||
# Skip translating source, base language
|
||||
next if ($code eq $language_source);
|
||||
|
||||
# Process only user defined languages or do all
|
||||
if (@{$language_target}) {
|
||||
next if (!any {$_ =~ /^$code$/} @{$language_target});
|
||||
}
|
||||
|
||||
# Do not process excluded languages
|
||||
if ($language_target_exclude) {
|
||||
my @languages_excluded = split(',', $language_target_exclude);
|
||||
if (any {$_ =~ /^$code$/} @languages_excluded) {
|
||||
next;
|
||||
}
|
||||
}
|
||||
|
||||
# Transcode each help file first
|
||||
foreach my $help_file (@module_help) {
|
||||
my $source;
|
||||
|
||||
if ($help_file =~
|
||||
/(?|\.($code__|$code).(UTF-8)\.html|\.($code__|$code).(euc)\.html|\.($code__|$code).(Big5)\.html|\.($code__|$code)\.html)/)
|
||||
{
|
||||
my $e_lang = $1;
|
||||
my $e_attr = $2 || '';
|
||||
|
||||
$e_lang = "$e_lang.$e_attr" if ($e_attr);
|
||||
|
||||
# Final file name to be writtten
|
||||
my $help_file_write = $help_file =~ s/$e_lang/$code/r;
|
||||
my $help_file_write_short = $help_file_write =~ s/.*\/(.+)$/$1/r;
|
||||
my $help_file_short = $help_file =~ s/.*\/(.+)$/$1/r;
|
||||
|
||||
# Rename, if old style format
|
||||
if ($code ne $e_lang || $e_attr) {
|
||||
talk_log(
|
||||
("" . YELLOW . " .. Renaming help file to new format - " . RED . "$help_file_short" .
|
||||
RESET . " --> " . GREEN . "$help_file_write_short" . RESET . "" . RESET . ""
|
||||
),
|
||||
$data,
|
||||
1);
|
||||
rename_file($help_file, $help_file_write);
|
||||
|
||||
}
|
||||
|
||||
# Transcode if needed
|
||||
if ($old_map && $e_attr ne "UTF-8") {
|
||||
talk_log(
|
||||
("" . BRIGHT_MAGENTA . " .. Transcoding help file $help_file_write_short" . RESET . ""),
|
||||
$data, 1);
|
||||
|
||||
my $data_original = read_file_contents($help_file_write);
|
||||
my ($data_encoding) = language_file_encoding($code, $help_file_write_short);
|
||||
my $data_converted = Encode::encode('utf-8', Encode::decode($data_encoding, $data_original));
|
||||
if ($data_encoding eq 'utf-8') {
|
||||
$data_converted = $data_original;
|
||||
}
|
||||
write_file_contents("$help_file_write", $data_converted);
|
||||
}
|
||||
push(@help_translated, $code);
|
||||
} elsif ($help_file =~ /\/\w+\.html$/) {
|
||||
push(@templates, $help_file);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@templates = sort(uniq(@templates));
|
||||
@help_translated = sort(uniq(@help_translated));
|
||||
|
||||
# Translate help files that don't have human translations
|
||||
if (!$opt->{'only-transcode'}) {
|
||||
my @help_untranslated = grep {"@help_translated" !~ /\b$_\b/} @{ $data->{'languages_source_list_codes'} };
|
||||
talk_log(
|
||||
("" . GREEN .
|
||||
" .. Found help files to translate to @{[scalar(@help_untranslated)]} more languages" .
|
||||
RESET . " \n - @{[join(\"\n - \", @templates)]}"
|
||||
),
|
||||
$data,
|
||||
1);
|
||||
}
|
||||
}
|
||||
|
||||
} else {
|
||||
foreach $module (@{$modules}) {
|
||||
my (%template);
|
||||
my ($exists, $mpath, $mfile) = source_data($module, $data, $opt);
|
||||
@ -1193,14 +1348,6 @@ sub go
|
||||
|
||||
say GREEN, ".. done ", RESET;
|
||||
}
|
||||
} else {
|
||||
foreach $module (@{$modules}) {
|
||||
my ($exists, $help_path) = source_data($module, $data, $opt);
|
||||
|
||||
talk_log(("Transcoding/translating " . CYAN BOLD, $module, RESET . " module's help .."), $data, 1);
|
||||
foreach $language (@{ $data->{'languages_source_list'} }) {
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user