Sync with (CVS snapshot of) po4a 0.39.1.

1 files changed, 117 insertions, 4 deletions

diff --git a/perl/Locale/Po4a/Text.pm b/perl/Locale/Po4a/Text.pm
index d9fcc88..676363d 100644
--- a/perl/Locale/Po4a/Text.pm
+++ b/perl/Locale/Po4a/Text.pm
@@ -21,6 +21,8 @@
 #
 ########################################################################
 
+=encoding UTF-8
+
 =head1 NAME
 
 Locale::Po4a::Text - Convert text documents from/to PO files
@@ -34,7 +36,7 @@ areas where they were not expected like documentation.
 Locale::Po4a::Text is a module to help the translation of text documents into
 other [human] languages.
 
-Paragraphs are splitted on empty lines (or lines containing only spaces or
+Paragraphs are split on empty lines (or lines containing only spaces or
 tabulations).
 
 If a paragraph contains a line starting by a space (or tabulation), this
@@ -62,7 +64,7 @@ These are this module's particular options:
 
 =over
 
-=item B<nobullet>
+=item B<nobullets>
 
 Deactivate detection of bullets.
 
@@ -74,6 +76,37 @@ rewrap this paragraph in the generated PO file and in the translation.
 
 my $bullets = 1;
 
+=item B<tabs>=I<mode>
+
+Specify how tabulations shall be handled. The I<mode> can be any of:
+
+=over
+
+=item B<split>
+
+Lines with tabulations introduce breaks in the current paragraph.
+
+=item B<verbatim>
+
+Paragraph containing tabulations will not be re-wrapped.
+
+=back
+
+By default, tabulations are considered as spaces.
+
+=cut
+
+my $tabs = "";
+
+=item B<breaks>=I<regex>
+
+A regular expression matching lines which introduce breaks.
+The regular expression will be anchored so that the whole line must match.
+
+=cut
+
+my $breaks;
+
 =item B<debianchangelog>
 
 Handle the header and footer of
@@ -116,12 +149,35 @@ sub initialize {
     my $self = shift;
     my %options = @_;
 
-    $self->{options}{'nobullets'}='';
+    $self->{options}{'asciidoc'} = 1;
+    $self->{options}{'breaks'} = 1;
+    $self->{options}{'debianchangelog'} = 1;
+    $self->{options}{'debug'} = 1;
+    $self->{options}{'fortunes'} = 1;
+    $self->{options}{'markdown'} = 1;
+    $self->{options}{'nobullets'} = 1;
+    $self->{options}{'tabs'} = 1;
+    $self->{options}{'verbose'} = 1;
+
+    foreach my $opt (keys %options) {
+        die wrap_mod("po4a::text",
+                     dgettext("po4a", "Unknown option: %s"), $opt)
+            unless exists $self->{options}{$opt};
+        $self->{options}{$opt} = $options{$opt};
+    }
 
     if (defined $options{'nobullets'}) {
         $bullets = 0;
     }
 
+    if (defined $options{'tabs'}) {
+        $tabs = $options{'tabs'};
+    }
+
+    if (defined $options{'breaks'}) {
+        $breaks = $options{'breaks'};
+    }
+
     if (defined $options{'debianchangelog'}) {
         $debianchangelog=1;
     }
@@ -194,7 +250,9 @@ sub parse {
                 undef $self->{type};
                 $wrapped_mode = 1;
             }
-        } elsif ($line =~ /^\s*$/) {
+        } elsif (   ($line =~ /^\s*$/)
+                 or (    defined $breaks
+                     and $line =~ m/^$breaks$/)) {
             # Break paragraphs on lines containing only spaces
             do_paragraph($self,$paragraph,$wrapped_mode);
             $paragraph="";
@@ -467,6 +525,50 @@ sub parse {
                 $self->{indent} = $indent;
                 $self->{bullet} = "";
             }
+        } elsif ($markdown and
+                 (not defined($self->{verbatim})) and
+                 ($line =~ m/^(={4,}|-{4,})$/) and
+                 (defined($paragraph) )and
+                 ($paragraph =~ m/^[^\n]*\n$/s) and
+                 (length($paragraph) == (length($line)+1))) {
+            # XXX: There can be any number of underlining according
+            #      to the documentation. This detection, which avoid
+            #      translating the formatting, is only supported if
+            #      the underlining has the same size as the herder text.
+            # Found title
+            $wrapped_mode = 0;
+            my $level = $line;
+            $level =~ s/^(.).*$/$1/;
+            my $t = $self->translate($paragraph,
+                                     $self->{ref},
+                                     "Title $level",
+                                     "wrap" => 0);
+            $self->pushline($t);
+            $paragraph="";
+            $wrapped_mode = 1;
+            $self->pushline(($level x (length($t)-1))."\n");
+        } elsif ($markdown and
+                 ($line =~ m/^(#{1,6})( +)(.*?)( +\1)?$/)) {
+            my $titlelevel1 = $1;
+            my $titlespaces = $2;
+            my $title = $3;
+            my $titlelevel2 = $4||"";
+            # Found one line title
+            do_paragraph($self,$paragraph,$wrapped_mode);
+            $wrapped_mode = 0;
+            $paragraph="";
+            my $t = $self->translate($title,
+                                     $self->{ref},
+                                     "Title $titlelevel1",
+                                     "wrap" => 0);
+            $self->pushline($titlelevel1.$titlespaces.$t.$titlelevel2."\n");
+            $wrapped_mode = 1;
+        } elsif ($markdown and
+                 ($paragraph eq "") and
+                 ($line =~ /^((\*\s*){3,}|(-\s*){3,}|(_\s*){3,})$/)) {
+            # Horizontal rule
+            $wrapped_mode = 1;
+            $self->pushline($line."\n");
         } elsif ($line =~ /^-- $/) {
             # Break paragraphs on email signature hint
             do_paragraph($self,$paragraph,$wrapped_mode);
@@ -503,6 +605,15 @@ sub parse {
             do_paragraph($self,$paragraph,$wrapped_mode);
             $paragraph="$line\n";
             $wrapped_mode = 1;
+        } elsif ($tabs eq "split" and $line =~ m/\t/ and $paragraph !~ m/\t/s) {
+            $wrapped_mode = 0;
+            do_paragraph($self,$paragraph,$wrapped_mode);
+            $paragraph = "$line\n";
+            $wrapped_mode = 0;
+        } elsif ($tabs eq "split" and $line !~ m/\t/ and $paragraph =~ m/\t/s) {
+            do_paragraph($self,$paragraph,$wrapped_mode);
+            $paragraph = "$line\n";
+            $wrapped_mode = 1;
         } else {
             if ($line =~ /^\s/) {
                 # A line starting by a space indicates a non-wrap
@@ -530,6 +641,8 @@ sub parse {
         # are considered as verbatim paragraphs
         $wrapped_mode = 0 if (   $paragraph =~ m/^(\*|[0-9]+[.)] )/s
                           or $paragraph =~ m/[ \t][ \t][ \t]/s);
+        $wrapped_mode = 0 if (    $tabs eq "verbatim"
+                              and $paragraph =~ m/\t/s);
         if ($markdown) {
             # Some Markdown markup can (or might) not survive wrapping
             $wrapped_mode = 0 if (