po4a: Do not pass empty PO files to msgmerge

Fixes #442

MANIFEST

@@ -234,6 +234,13 @@ t/cfg/split-yaml/i18n.pot
 t/cfg/split-yaml/i18n.vi.po
 t/cfg/split-yaml/po4a.cfg
 t/cfg/split-yaml/vi.yaml
+t/cfg/single-emptypo/_fr.po
+t/cfg/single-emptypo/_output
+t/cfg/single-emptypo/_single.man.fr.1
+t/cfg/single-emptypo/fr.po
+t/cfg/single-emptypo/po4a.conf
+t/cfg/single-emptypo/single.man.1
+t/cfg/single-emptypo/single.pot
 t/cfg/single-nopotpo/single.man.1
 t/cfg/single-nopotpo/po4a.conf
 t/cfg/single-nopotpo/_single.pot

NEWS

@@ -68,6 +68,8 @@ General:
 po4a main script (the deprecated po4a-* scripts are still buggy):
  * Deal properly with CRLF files coming from windows (GitHub's #409).
  * Allow to escape spaces in file names, or to quote them (GitHub's #412).
+ * Do not pass empty PO files to msgmerge, as it stumbles on such things when
+   the POT file has UTF chars in the msgids. (GitHub's #442) [Mt]
 
 po4a-translate:
  * Remove the wrap-po option that was not used internally.

lib/Locale/Po4a/Po.pm

@@ -317,23 +317,28 @@ sub read {
     my $self     = shift;
     my $filename = shift
       or croak wrap_mod( "po4a::po", dgettext( "po4a", "Please provide a non-null filename" ) );
+
     my $charset = shift // 'UTF-8';
     $charset = 'UTF-8' if $charset eq "CHARSET";
     warn "Read $filename with encoding: $charset" if $debug{'encoding'};
 
+    my $checkvalidity = shift // 1;
+
     my $lang = basename($filename);
     $lang =~ s/\.po$//;
     $self->{lang} = $lang;
 
-    my $cmd = "msgfmt" . $Config{_exe} . " --check-format --check-domain -o /dev/null \"" . $filename . '"';
+    if ($checkvalidity) {   # We sometimes need to read a file even if it may be invalid (eg to test whether it's empty)
+        my $cmd = "msgfmt" . $Config{_exe} . " --check-format --check-domain -o /dev/null \"" . $filename . '"';
 
-    my $locale = $ENV{'LC_ALL'};
-    $ENV{'LC_ALL'} = "C";
-    my $out = qx/$cmd 2>&1/;
-    $ENV{'LC_ALL'} = $locale;
+        my $locale = $ENV{'LC_ALL'};
+        $ENV{'LC_ALL'} = "C";
+        my $out = qx/$cmd 2>&1/;
+        $ENV{'LC_ALL'} = $locale;
 
-    die wrap_msg( dgettext( "po4a", "Invalid po file %s:\n%s" ), $filename, $out )
-      unless ( $? == 0 );
+        die wrap_msg( dgettext( "po4a", "Invalid po file %s:\n%s" ), $filename, $out )
+          unless ( $? == 0 );
+    }
 
     my $fh;
     if ( $filename eq '-' ) {
@@ -348,6 +353,7 @@ sub read {
     while ( defined( my $textline = <$fh> ) ) {
         $pofile .= $textline;
     }
+    $pofile =~ s/\r\n/\n/sg;    # Reading a DOS-encoded file from Linux (native files are handled in all cases)
 
     # If we did not get the charset right, reload the file with the right one
     if ( $pofile =~ /charset=(.*?)[\s\\]/ ) {
@@ -357,7 +363,7 @@ sub read {
         {
             warn "Reloading the PO file, changing the charset from '$charset' to '$detected_charset'"
               if $debug{'encoding'};
-            $self->read( $filename, $detected_charset );
+            $self->read( $filename, $detected_charset, $checkvalidity );
             return;
         }
     }

po4a

@@ -1756,7 +1756,16 @@ if ( not $po4a_opts{"no-update"} ) {
               ( find_input_file( $po_filename{$lang} ), find_output_file( $po_filename{$lang} ) );
             my $updated_potfile = find_input_file($pot_filename);
 
+            my $usable_pofile = 0;
             if ( -e $infile ) {
+
+                # Check that the po file is not empty, as msgmerge has issues in this case (see Debian's #1022216 and GitHub's #442)
+
+                my $pofile = Locale::Po4a::Po->new();
+                $pofile->read( $infile, undef, 0 );    # Guess the charset ; Don't fail if msgfmt spits an error
+                $usable_pofile = 1 if ( $pofile->count_entries() > 0 );
+            }
+            if ($usable_pofile) {
                 my $dir = dirname($outfile);
                 if ( not -d $dir ) {
                     mkdir $dir or die wrap_msg( gettext("Cannot create directory '%s': %s"), $dir, $! );
@@ -1765,7 +1774,7 @@ if ( not $po4a_opts{"no-update"} ) {
                 my $msgmerge_opt = $po4a_opts{"msgmerge-opt"};
                 $msgmerge_opt =~ s/\$lang\b/$lang/g if scalar @langs;
                 my $cmd = "msgmerge" . $Config{_exe} . " \"$infile\" \"$updated_potfile\" " . $msgmerge_opt;
-                if ( $infile eq $outfile ) {    # in place
+                if ( $infile eq $outfile ) {           # in place
                     $cmd .= " --backup=none --update";
                 } else {
                     $cmd .= " --output \"$outfile\"";
@@ -1785,6 +1794,14 @@ if ( not $po4a_opts{"no-update"} ) {
                 }
 
             } else {
+                if ( $po4a_opts{"verbose"} ) {
+                    if ( $po4a_opts{'split'} ) {
+                        printf( gettext("Creating an empty PO file for the language %s.\n"), $lang );
+                    } else {
+                        printf( gettext("Creating an empty PO file in %s.\n"), $po_filename{$lang} );
+                    }
+                }
+
                 my $read_pot_filename = find_input_file($pot_filename);
                 my $cmd =
                   "msginit$Config{_exe} -i \"$read_pot_filename\" --locale $lang -o \"$outfile\" --no-translator >$devnull";

po4a-updatepo

@@ -304,7 +304,7 @@ while ( my $po_filename = shift @pofiles ) {
     my $usable_pofile = 0;
     if ( -e $po_filename ) {
 
-        # Check that the po file is not empty, as msgmerge has issues in this case (see Debian's #1022216)
+        # Check that the po file is not empty, as msgmerge has issues in this case (see Debian's #1022216 and GitHub's #442)
         my $pofile = Locale::Po4a::Po->new();
         $pofile->read($po_filename);
         $usable_pofile = 1 if ( $pofile->count_entries() > 0 );

t/cfg-single.t

@@ -101,12 +101,12 @@ push @tests, {
     'po4a.conf'      => 'cfg/single-podirectory-emptypot-emptypo/po4a.conf',
     'closed_path'    => 'cfg/*/',
     'expected_files' => 'fr.po  single.pot',
-        'tests'          => [
-            'PODIFF -I#: $path/_fr.po.expected $tmppath/fr.po', #
-            'PODIFF -I#: $path/_single.pot.expected $tmppath/single.pot',
-            'rm $tmppath/single.pot && touch $tmppath/single.pot',    # This file is empty on purpose
-            'rm $tmppath/fr.po && touch $tmppath/fr.po', # This file is empty on purpose
-        ]
+    'tests'          => [
+        'PODIFF -I#: $path/_fr.po.expected $tmppath/fr.po',       #
+        'PODIFF -I#: $path/_single.pot.expected $tmppath/single.pot',
+        'rm $tmppath/single.pot && touch $tmppath/single.pot',    # This file is empty on purpose
+        'rm $tmppath/fr.po && touch $tmppath/fr.po',              # This file is empty on purpose
+    ]
 
   },
   {
@@ -115,12 +115,17 @@ push @tests, {
     'closed_path'    => 'cfg/*/',
     'expected_files' => 'po single.man.fr.1',
     'tests'          => [
-        'PODIFF -I#: $path/po/fr.po $tmppath/po/fr.po',    #
+        'PODIFF -I#: $path/po/fr.po $tmppath/po/fr.po',           #
         'PODIFF -I#: $path/_single.pot $tmppath/po/single.pot',
         'rm $tmppath/po/single.pot && touch $tmppath/po/single.pot'
-        ,                                                  # The $path/po/single.pot exists, but it's empty (on purpose)
+        ,    # The $path/po/single.pot exists, but it's empty (on purpose)
     ]
-
+  },
+  {
+    'doc'            => 'Single language, empty PO file and UTF in msgids (see Debian\'s #1022216)',
+    'po4a.conf'      => 'cfg/single-emptypo/po4a.conf',
+    'closed_path'    => 'cfg/*/',
+    'expected_files' => 'fr.po single.pot single.man.fr.1',
   };
 
 run_all_tests(@tests);

t/cfg/multiple-nopo/_output

@@ -1,4 +1,8 @@
 Updating multiple.pot: (4 entries)
+Creating an empty PO file in multiple.de.po.
+Creating an empty PO file in multiple.es.po.
+Creating an empty PO file in multiple.fr.po.
+Creating an empty PO file in multiple.it.po.
 Discard multiple.man.de.1 (0 of 4 strings; only 0% translated; need 80%).
 Discard multiple.man.es.1 (0 of 4 strings; only 0% translated; need 80%).
 Discard multiple.man.fr.1 (0 of 4 strings; only 0% translated; need 80%).

t/cfg/multiple-nopotpo/_output

@@ -1,4 +1,8 @@
 Creating multiple.pot: (4 entries)
+Creating an empty PO file in multiple.de.po.
+Creating an empty PO file in multiple.es.po.
+Creating an empty PO file in multiple.fr.po.
+Creating an empty PO file in multiple.it.po.
 Discard multiple.man.de.1 (0 of 4 strings; only 0% translated; need 80%).
 Discard multiple.man.es.1 (0 of 4 strings; only 0% translated; need 80%).
 Discard multiple.man.fr.1 (0 of 4 strings; only 0% translated; need 80%).

t/cfg/single-emptypo/_fr.po

@@ -0,0 +1,39 @@
+# French translations for po package
+# Copyright (C) 2020 Free Software Foundation, Inc.
+# This file is distributed under the same license as the po package.
+# Automatically generated, 2020.
+#
+msgid ""
+msgstr ""
+"Project-Id-Version: po 4a\n"
+"POT-Creation-Date: 2005-04-24  2:54+0200\n"
+"PO-Revision-Date: 2005-04-24  2:54+0200\n"
+"Last-Translator: Automatically generated\n"
+"Language-Team: none\n"
+"Language: fr\n"
+"MIME-Version: 1.0\n"
+"Content-Type: text/plain; charset=UTF-8\n"
+"Content-Transfer-Encoding: 8bit\n"
+"Plural-Forms: nplurals=2; plural=(n > 1);\n"
+
+#. type: TH
+#: single.man.1:1
+#, no-wrap
+msgid "test1"
+msgstr ""
+
+#. type: SH
+#: single.man.1:2
+#, no-wrap
+msgid "NAME"
+msgstr ""
+
+#. type: Plain text
+#: single.man.1:4
+msgid "test - just a test"
+msgstr ""
+
+#. type: Plain text
+#: single.man.1:5
+msgid "UTF in msgid is sometimes difficult, but mandatory for some nämes."
+msgstr ""

t/cfg/single-emptypo/_output

@@ -0,0 +1,3 @@
+Updating single.pot: (4 entries)
+Creating an empty PO file in fr.po.
+single.man.fr.1 is 0% translated (0 of 4 strings).

t/cfg/single-emptypo/_single.man.fr.1

@@ -0,0 +1,10 @@
+.\"*******************************************************************
+.\"
+.\" This file was generated with po4a. Translate the source file.
+.\"
+.\"*******************************************************************
+.TH test1 1   
+.SH NAME
+test \- just a test
+
+UTF in msgid is sometimes difficult, but mandatory for some nämes.

t/cfg/single-emptypo/po4a.conf

@@ -0,0 +1,5 @@
+[po4a_paths] single.pot fr:fr.po
+
+[options] opt:"--verbose --keep 0 "
+
+[type:man] single.man.1 fr:single.man.fr.1

t/cfg/single-emptypo/single.man.1

@@ -0,0 +1,5 @@
+.TH test1 1
+.SH NAME
+test \- just a test
+
+UTF in msgid is sometimes difficult, but mandatory for some nämes.

t/cfg/single-emptypo/single.pot

@@ -0,0 +1,39 @@
+# SOME DESCRIPTIVE TITLE
+# Copyright (C) YEAR Free Software Foundation, Inc.
+# This file is distributed under the same license as the PACKAGE package.
+# FIRST AUTHOR <EMAIL@ADDRESS>, YEAR.
+#
+#, fuzzy
+msgid ""
+msgstr ""
+"Project-Id-Version: PACKAGE VERSION\n"
+"POT-Creation-Date: 2005-04-24  2:54+0200\n"
+"PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n"
+"Last-Translator: FULL NAME <EMAIL@ADDRESS>\n"
+"Language-Team: LANGUAGE <LL@li.org>\n"
+"Language: \n"
+"MIME-Version: 1.0\n"
+"Content-Type: text/plain; charset=UTF-8\n"
+"Content-Transfer-Encoding: 8bit\n"
+
+#. type: TH
+#: single.man.1:1
+#, no-wrap
+msgid "test1"
+msgstr ""
+
+#. type: SH
+#: single.man.1:2
+#, no-wrap
+msgid "NAME"
+msgstr ""
+
+#. type: Plain text
+#: single.man.1:4
+msgid "test - just a test"
+msgstr ""
+
+#. type: Plain text
+#: single.man.1:5
+msgid "UTF in msgid is sometimes difficult, but mandatory for some nämes."
+msgstr ""

t/cfg/single-nopo/_output

@@ -1,2 +1,3 @@
 Updating single.pot: (4 entries)
+Creating an empty PO file in single.fr.po.
 Discard single.man.fr.1 (0 of 4 strings; only 0% translated; need 80%).

t/cfg/single-nopotpo/_output

@@ -1,2 +1,3 @@
 Creating single.pot: (4 entries)
+Creating an empty PO file in single.fr.po.
 Discard single.man.fr.1 (0 of 4 strings; only 0% translated; need 80%).

t/cfg/single-podirectory-emptypot-emptypo/_fr.po.expected

@@ -1,3 +1,21 @@
+# French translations for po package
+# Copyright (C) 2024 Free Software Foundation, Inc.
+# This file is distributed under the same license as the po package.
+# Automatically generated, 2024.
+#
+msgid ""
+msgstr ""
+"Project-Id-Version: po 4a\n"
+"POT-Creation-Date: 2024-01-04 22:50+0100\n"
+"PO-Revision-Date: 2024-01-04 22:50+0100\n"
+"Last-Translator: Automatically generated\n"
+"Language-Team: none\n"
+"Language: fr\n"
+"MIME-Version: 1.0\n"
+"Content-Type: text/plain; charset=UTF-8\n"
+"Content-Transfer-Encoding: 8bit\n"
+"Plural-Forms: nplurals=2; plural=(n > 1);\n"
+
 #. type: TH
 #: single.man.1:1
 #, no-wrap

t/cfg/single-podirectory-emptypot-emptypo/_output

@@ -2,5 +2,5 @@ Using '.' as a po_directory.
 Found language 'fr' in the provided po_directory: ./fr.po
 Found POT file './single.pot' in the provided po_directory.
 Updating ./single.pot: (4 entries)
-Updating ./fr.po: 0 translated messages, 4 untranslated messages.
+Creating an empty PO file in ./fr.po.
 Discard single.man.fr.1 (0 of 4 strings; only 0% translated; need 80%).

t/cfg/split-nopo/_output

@@ -3,6 +3,8 @@ The translation of master file 'first.man' in language 'de' is missing (file: fi
 The translation of master file 'second.man' in language 'de' is missing (file: second.man.de.po) -- skipping.
 The translation of master file 'first.man' in language 'fr' is missing (file: first.man.fr.po) -- skipping.
 The translation of master file 'second.man' in language 'fr' is missing (file: second.man.fr.po) -- skipping.
+Creating an empty PO file for the language de.
+Creating an empty PO file for the language fr.
 Discard first.man.de (0 of 5 strings; only 0% translated; need 80%).
 Discard second.man.de (0 of 5 strings; only 0% translated; need 80%).
 Discard first.man.fr (0 of 5 strings; only 0% translated; need 80%).

t/cfg/split-nopotpo/_output

@@ -3,6 +3,8 @@ The translation of master file 'first.man' in language 'de' is missing (file: fi
 The translation of master file 'second.man' in language 'de' is missing (file: second.man.de.po) -- skipping.
 The translation of master file 'first.man' in language 'fr' is missing (file: first.man.fr.po) -- skipping.
 The translation of master file 'second.man' in language 'fr' is missing (file: second.man.fr.po) -- skipping.
+Creating an empty PO file for the language de.
+Creating an empty PO file for the language fr.
 Discard first.man.de (0 of 5 strings; only 0% translated; need 80%).
 Discard second.man.de (0 of 5 strings; only 0% translated; need 80%).
 Discard first.man.fr (0 of 5 strings; only 0% translated; need 80%).

t/core/porefs/_output

@@ -1 +1,2 @@
 Creating list.pot: (2 entries)
+Creating an empty PO file in up.po.