summaryrefslogtreecommitdiff
path: root/perl/Locale/Po4a/Text.pm
diff options
context:
space:
mode:
Diffstat (limited to 'perl/Locale/Po4a/Text.pm')
-rw-r--r--perl/Locale/Po4a/Text.pm337
1 files changed, 322 insertions, 15 deletions
diff --git a/perl/Locale/Po4a/Text.pm b/perl/Locale/Po4a/Text.pm
index 32f9dec..d9fcc88 100644
--- a/perl/Locale/Po4a/Text.pm
+++ b/perl/Locale/Po4a/Text.pm
@@ -100,6 +100,18 @@ Handle some special markup in Markdown-formatted texts.
my $markdown = 0;
+=item B<asciidoc>
+
+Handle documents in the asciidoc format.
+
+=cut
+
+my $asciidoc = 0;
+
+=back
+
+=cut
+
sub initialize {
my $self = shift;
my %options = @_;
@@ -121,6 +133,8 @@ sub initialize {
if (defined $options{'markdown'}) {
$markdown=1;
}
+
+ $asciidoc=1 if (defined $options{'asciidoc'});
}
sub parse {
@@ -143,7 +157,6 @@ sub parse {
$expect_header = 1;
}
- # TODO: preserve original line ends throughout the code instead
chomp($line);
$self->{ref}="$ref";
if ($debianchangelog and
@@ -166,19 +179,294 @@ sub parse {
$line =~ m/^%%?\s*$/) {
# Found end of fortune
do_paragraph($self,$paragraph,$wrapped_mode);
- # FIXME: test if this is still needed when always adding
- # newline in do_paragraph()
$self->pushline("\n") unless ( $wrapped_mode == 0
or $paragraph eq "");
$paragraph="";
$wrapped_mode = 1;
$self->pushline("$line\n");
+ } elsif ( (defined $self->{verbatim})
+ and ($self->{verbatim} == 2)) {
+ # Untranslated blocks
+ $self->pushline($line."\n");
+ if ($asciidoc and
+ ($line =~ m/^(\/{4,}|~{4,})$/)) {
+ undef $self->{verbatim};
+ undef $self->{type};
+ $wrapped_mode = 1;
+ }
} elsif ($line =~ /^\s*$/) {
# Break paragraphs on lines containing only spaces
do_paragraph($self,$paragraph,$wrapped_mode);
$paragraph="";
+ $wrapped_mode = 1 unless defined($self->{verbatim});
+ $self->pushline($line."\n");
+ } elsif ($asciidoc and (not defined($self->{verbatim})) and
+ ($line =~ m/^(\+|--)$/)) {
+ # List Item Continuation or List Block
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph="";
+ $self->pushline($line."\n");
+ } elsif ($asciidoc and (not defined($self->{verbatim})) and
+ ($line =~ m/^(={4,}|-{4,}|~{4,}|\^{4,}|\+{4,})$/) and
+ (defined($paragraph) )and
+ ($paragraph =~ m/^[^\n]*\n$/s) and
+ (length($paragraph) == (length($line)+1))) {
+ # 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($t)-1))."\n");
+ } elsif ($asciidoc and
+ ($line =~ m/^(={1,5})( +)(.*?)( +\1)?$/)) {
+ 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");
+ $wrapped_mode = 1;
+ } elsif ($asciidoc and
+ ($line =~ m/^(\/{4,}|\+{4,}|-{4,}|\.{4,}|\*{4,}|_{4,}|={4,}|~{4,})$/)) {
+ # Found one delimited block
+ my $t = $line;
+ $t =~ s/^(.).*$/$1/;
+ my $type = "delimited block $t";
+ if (defined $self->{verbatim} and ($self->{type} ne $type)) {
+ $paragraph .= "$line\n";
+ } else {
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ if ( (defined $self->{type})
+ and ($self->{type} eq $type)) {
+ undef $self->{type};
+ undef $self->{verbatim};
+ $wrapped_mode = 1;
+ } else {
+ if ($t eq "\/") {
+ # CommentBlock, should not be treated
+ $self->{verbatim} = 2;
+ } elsif ($t eq "+") {
+ # PassthroughBlock
+ $wrapped_mode = 0;
+ $self->{verbatim} = 1;
+ } elsif ($t eq "-") {
+ # ListingBlock
+ $wrapped_mode = 0;
+ $self->{verbatim} = 1;
+ } elsif ($t eq ".") {
+ # LiteralBlock
+ $wrapped_mode = 0;
+ $self->{verbatim} = 1;
+ } elsif ($t eq "*") {
+ # SidebarBlock
+ $wrapped_mode = 1;
+ } elsif ($t eq "_") {
+ # QuoteBlock
+ if ( (defined $self->{type})
+ and ($self->{type} eq "verse")) {
+ $wrapped_mode = 0;
+ $self->{verbatim} = 1;
+ } else {
+ $wrapped_mode = 1;
+ }
+ } elsif ($t eq "=") {
+ # ExampleBlock
+ $wrapped_mode = 1;
+ } elsif ($t eq "~") {
+ # Filter blocks, TBC: not translated
+ $wrapped_mode = 0;
+ $self->{verbatim} = 2;
+ }
+ $self->{type} = $type;
+ }
+ $paragraph="";
+ $self->pushline($line."\n");
+ }
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line =~ m/^\[\[([^\]]*)\]\]$/)) {
+ # Found BlockId
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph="";
+ $wrapped_mode = 1;
+ $self->pushline($line."\n");
+ undef $self->{bullet};
+ undef $self->{indent};
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($paragraph eq "") and
+ ($line =~ m/^((?:NOTE|TIP|IMPORTANT|WARNING|CAUTION):\s+)(.*)$/)) {
+ my $type = $1;
+ my $text = $2;
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph=$text."\n";
+ $wrapped_mode = 1;
+ $self->pushline($type);
+ undef $self->{bullet};
+ undef $self->{indent};
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line =~ m/^\[(NOTE|TIP|IMPORTANT|WARNING|CAUTION|verse|quote)\]$/)) {
+ my $type = $1;
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph="";
$wrapped_mode = 1;
$self->pushline($line."\n");
+ if ($type eq "verse") {
+ $wrapped_mode = 0;
+ }
+ undef $self->{bullet};
+ undef $self->{indent};
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line =~ m/^\[(verse|quote), +(.*)\]$/)) {
+ my $type = $1;
+ my $arg = $2;
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph="";
+ my $t = $self->translate($arg,
+ $self->{ref},
+ "$type",
+ "wrap" => 0);
+ $self->pushline("[$type, $t]\n");
+ $wrapped_mode = 1;
+ if ($type eq "verse") {
+ $wrapped_mode = 0;
+ }
+ $self->{type} = $type;
+ undef $self->{bullet};
+ undef $self->{indent};
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line =~ m/^\[icon="(.*)"\]$/)) {
+ my $arg = $1;
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph="";
+ my $t = $self->translate($arg,
+ $self->{ref},
+ "icon",
+ "wrap" => 0);
+ $self->pushline("[icon=\"$t\"]\n");
+ $wrapped_mode = 1;
+ undef $self->{bullet};
+ undef $self->{indent};
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line =~ m/^\[icons=None, +caption="(.*)"\]$/)) {
+ my $arg = $1;
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph="";
+ my $t = $self->translate($arg,
+ $self->{ref},
+ "caption",
+ "wrap" => 0);
+ $self->pushline("[icons=None, caption=\"$t\"]\n");
+ $wrapped_mode = 1;
+ undef $self->{bullet};
+ undef $self->{indent};
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line =~ m/^(\s*)([*_+`'#[:alnum:]].*)((?:::|;;|\?\?|:-)(?: *\\)?)$/)) {
+ my $indent = $1;
+ my $label = $2;
+ my $labelend = $3;
+ # Found labeled list
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph="";
+ $wrapped_mode = 1;
+ $self->{bullet} = "";
+ $self->{indent} = $indent;
+ my $t = $self->translate($label,
+ $self->{ref},
+ "Labeled list",
+ "wrap" => 0);
+ $self->pushline("$indent$t$labelend\n");
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line =~ m/^(\s*)(\S.*)((?:::|;;)\s+)(.*)$/)) {
+ my $indent = $1;
+ my $label = $2;
+ my $labelend = $3;
+ my $labeltext = $4;
+ # Found Horizontal Labeled Lists
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph=$labeltext."\n";
+ $wrapped_mode = 1;
+ $self->{bullet} = "";
+ $self->{indent} = $indent;
+ my $t = $self->translate($label,
+ $self->{ref},
+ "Labeled list",
+ "wrap" => 0);
+ $self->pushline("$indent$t$labelend");
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line =~ m/^\:(\S.*?)(:\s*)(.*)$/)) {
+ my $attrname = $1;
+ my $attrsep = $2;
+ my $attrvalue = $3;
+ # Found a Attribute entry
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph="";
+ $wrapped_mode = 1;
+ undef $self->{bullet};
+ undef $self->{indent};
+ my $t = $self->translate($attrvalue,
+ $self->{ref},
+ "Attribute :$attrname:",
+ "wrap" => 0);
+ $self->pushline(":$attrname$attrsep$t\n");
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line !~ m/^\.\./) and ($line =~ m/^\.(\S.*)$/)) {
+ my $title = $1;
+ # Found block title
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph="";
+ $wrapped_mode = 1;
+ undef $self->{bullet};
+ undef $self->{indent};
+ my $t = $self->translate($title,
+ $self->{ref},
+ "Block title",
+ "wrap" => 0);
+ $self->pushline(".$t\n");
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line =~ m/^(\s*)((?:[-*o+]|(?:[0-9]+[.\)])|(?:[a-z][.\)])|\([0-9]+\)|\.|\.\.)\s+)(.*)$/)) {
+ my $indent = $1||"";
+ my $bullet = $2;
+ my $text = $3;
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph = $text."\n";
+ $self->{indent} = $indent;
+ $self->{bullet} = $bullet;
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ ($line =~ m/^((?:<?[0-9]+)?> +)(.*)$/)) {
+ my $bullet = $1;
+ my $text = $2;
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph = $text."\n";
+ $self->{indent} = "";
+ $self->{bullet} = $bullet;
+ } elsif ($asciidoc and not defined $self->{verbatim} and
+ (defined $self->{bullet} and $line =~ m/^(\s+)(.*)$/)) {
+ my $indent = $1;
+ my $text = $2;
+ if (not defined $self->{indent}) {
+ $paragraph .= $text."\n";
+ $self->{indent} = $indent;
+ } elsif (length($paragraph) and (length($self->{bullet}) + length($self->{indent}) == length($indent))) {
+ $paragraph .= $text."\n";
+ } else {
+
+ do_paragraph($self,$paragraph,$wrapped_mode);
+ $paragraph = $text."\n";
+ $self->{indent} = $indent;
+ $self->{bullet} = "";
+ }
} elsif ($line =~ /^-- $/) {
# Break paragraphs on email signature hint
do_paragraph($self,$paragraph,$wrapped_mode);
@@ -226,6 +514,9 @@ sub parse {
or $line =~ /"""$/)) { # """ textblock inside macro begin
# Markdown markup needing separation _after_ this line
$end_of_paragraph = 1;
+ } else {
+ undef $self->{bullet};
+ undef $self->{indent};
}
if ($fortunes) {
$line =~ s/%%(.*)$//;
@@ -265,9 +556,19 @@ sub parse {
sub do_paragraph {
my ($self, $paragraph, $wrap) = (shift, shift, shift);
+ my $type = shift || $self->{type} || "Plain text";
return if ($paragraph eq "");
- if ($bullets) {
+# DEBUG
+# my $b;
+# if (defined $self->{bullet}) {
+# $b = $self->{bullet};
+# } else {
+# $b = "UNDEF";
+# }
+# $type .= " verbatim: '".($self->{verbatim}||"NONE")."' bullet: '$b' indent: '".($self->{indent}||"NONE")."' type: '".($self->{type}||"NONE")."'";
+
+ if ($bullets and not $wrap and not defined $self->{verbatim}) {
# Detect bullets
# | * blah blah
# |<spaces> blah
@@ -312,18 +613,24 @@ TEST_BULLET:
}
}
}
- # TODO: detect indented paragraphs
- my $transfinal = $self->translate($paragraph,
- $self->{ref},
- "Plain text",
- "wrap" => $wrap);
-
- # TODO: preserve original line ends throughout the code instead
- chomp $transfinal;
- $transfinal .= "\n";
-
- $self->pushline( $transfinal );
+ my $end = "";
+ if ($wrap) {
+ $paragraph =~ s/^(.*?)(\n*)$/$1/s;
+ $end = $2 || "";
+ }
+ my $t = $self->translate($paragraph,
+ $self->{ref},
+ $type,
+ "wrap" => $wrap);
+ if (defined $self->{bullet}) {
+ my $bullet = $self->{bullet};
+ my $indent1 = $self->{indent};
+ my $indent2 = $indent1.(' ' x length($bullet));
+ $t =~ s/^/$indent1$bullet/s;
+ $t =~ s/\n(.)/\n$indent2$1/sg;
+ }
+ $self->pushline( $t.$end );
}
1;
generated by cgit v1.2.3 (git 2.46.0) at 2024-11-14 17:26:51 +0000