diff options
| author | Tatsuya Kinoshita <tats@debian.org> | 2013-08-01 22:01:06 +0000 |
|---|---|---|
| committer | Tatsuya Kinoshita <tats@debian.org> | 2013-08-01 22:01:06 +0000 |
| commit | a7fedda6fca5b1bbceec5f69abe2a05432151055 (patch) | |
| tree | d553c98c8f2d1d4d84f589ab017ec1a85428de4c /scripts/w3mman/w3mman2html.cgi.in | |
| parent | Merge branch 'bug/shy' (diff) | |
| parent | Correct underline processing and more UTF-8 support for w3mman2html.cgi (diff) | |
| download | w3m-a7fedda6fca5b1bbceec5f69abe2a05432151055.tar.gz w3m-a7fedda6fca5b1bbceec5f69abe2a05432151055.zip | |
Merge branch 'bug/w3mman2html-utf8'
Diffstat (limited to '')
| -rw-r--r-- | scripts/w3mman/w3mman2html.cgi.in | 8 |
1 files changed, 5 insertions, 3 deletions
diff --git a/scripts/w3mman/w3mman2html.cgi.in b/scripts/w3mman/w3mman2html.cgi.in index f430307..2e3576c 100644 --- a/scripts/w3mman/w3mman2html.cgi.in +++ b/scripts/w3mman/w3mman2html.cgi.in @@ -126,12 +126,14 @@ while(<F>) { s/\&/\&/g; s/\</\</g; s/\>/\>/g; + # non ASCII UTF-8 codepoint + my $utf8="[\300-\337][\200-\277]|[\340-\357][\200-\277]{2}|[\360-\367][\200-\277]{3}|[\370-\373][\200-\277]{4}|[\374\375][\200-\277]{5}"; - s@([\200-\377].)(\010{1,2}\1)+@<b>$1</b>@g; + s@($utf8)(\010\1)+@<b>$1</b>@g; s@(\&\w+;|.)(\010\1)+@<b>$1</b>@g; - s@__\010{1,2}((\<b\>)?[\200-\377].(\</b\>)?)@<u>$1</u>@g; + s@_\010((\<b\>)?($utf8)(\</b\>)?)@<u>$1</u>@g; s@_\010((\<b\>)?(\&\w+\;|.)(\</b\>)?)@<u>$1</u>@g; - s@((\<b\>)?[\200-\377].(\</b\>)?)\010{1,2}__@<u>$1</u>@g; + s@((\<b\>)?($utf8)(\</b\>)?)\010_@<u>$1</u>@g; s@((\<b\>)?(\&\w+\;|.)(\</b\>)?)\010_@<u>$1</u>@g; s@.\010(.)@$1@g; |