Bug#1020215: w3m interprets when not in but in

[Tatsuya Kinoshita]

On 2023-01-06 at 19:31 +0100, Robert Alm Nilsson wrote:
> After using my version of w3m every day since I posted this, the only
> problem I have noticed on a few pages is that it doesn't handle titles
> directly under html tags, so this doesn't work:
>
> Hello

Ah, it should be interpreted as in .

I've installed a patch that only picks a first title as a workaround.

  - https://github.com/tats/w3m/commit/b9c022a8101e6325468bf310d6038df11e5fac67

Thanks for your report.

--
Tatsuya Kinoshita


pgpvdGS3k2laz.pgp
Description: PGP signature

Bug#1020215: w3m interprets when not in but in

[Robert Alm Nilsson]

After using my version of w3m every day since I posted this, the only
problem I have noticed on a few pages is that it doesn't handle titles
directly under html tags, so this doesn't work:

Hello

I don't know if we want to care about that case and I don't think it's
valid HTML but I wanted to let you know.

Bug#1020215: w3m interprets when not in but in

[Tatsuya Kinoshita]

Control: tags -1 + patch fixed-upstream pending

On 2023-01-06 at 15:26 +0100, Rene Kita wrote:
> Reviewed and tested by me.
> Tats, could you please include it in your planned snapshot tarball?

Merged, thanks.

--
Tatsuya Kinoshita


pgpAWqia7oxxQ.pgp
Description: PGP signature

Bug#1020215: w3m interprets when not in but in

[Robert Alm Nilsson]

Package: w3m
Version: 0.5.3+git20220429-1+b1

When opening a page in w3m that contains an  element that contains
a  tag, w3m will use that svg title as the page title.

An example of a page to demonstrate this problem is github.com.  The
real title of github.com is "GitHub: Where the world builds software ·
GitHub" but w3m displays the title "Go" (or the name of any other
language) because the front page contains svg images for different
programming languages with title tags.

Here is a patch I made.  I don't know if this is the most optimal way to
solve this but it seems to work in practice and I will use it locally
until there in an upstream fix.

commit f41db326e73fde685c1d0b79e46beec56336995e
Author: Robert Alm Nilsson 
Date:   Sun Sep 18 09:51:29 2022 +0200

Only read title when in head

Before this change, it was possible that w3m would interpret a title tag
under e.g. an svg tag as the page title.

diff --git a/file.c b/file.c
index 9704cea..8e8b280 100644
--- a/file.c
+++ b/file.c
@@ -4816,19 +4816,23 @@ HTMLtagproc1(struct parsed_tag *tag, struct 
html_feed_environ *h_env)
/* obuf->flag |= RB_IGNORE_P; */
return 1;
 case HTML_TITLE:
-   close_anchor(h_env, obuf);
-   process_title(tag);
-   obuf->flag |= RB_TITLE;
-   obuf->end_tag = HTML_N_TITLE;
+   if (obuf->flag & RB_HEAD) {
+   close_anchor(h_env, obuf);
+   process_title(tag);
+   obuf->flag |= RB_TITLE;
+   obuf->end_tag = HTML_N_TITLE;
+   }
return 1;
 case HTML_N_TITLE:
-   if (!(obuf->flag & RB_TITLE))
-   return 1;
-   obuf->flag &= ~RB_TITLE;
-   obuf->end_tag = 0;
-   tmp = process_n_title(tag);
-   if (tmp)
-   HTMLlineproc1(tmp->ptr, h_env);
+   if (obuf->flag | RB_HEAD) {
+   if (!(obuf->flag & RB_TITLE))
+   return 1;
+   obuf->flag &= ~RB_TITLE;
+   obuf->end_tag = 0;
+   tmp = process_n_title(tag);
+   if (tmp)
+   HTMLlineproc1(tmp->ptr, h_env);
+   }
return 1;
 case HTML_TITLE_ALT:
if (parsedtag_get_value(tag, ATTR_TITLE, ))
@@ -5523,9 +5527,13 @@ HTMLtagproc1(struct parsed_tag *tag, struct 
html_feed_environ *h_env)
}
}
 case HTML_N_HEAD:
+   obuf->flag &= ~RB_HEAD;
if (obuf->flag & RB_TITLE)
HTMLlineproc1("", h_env);
+   return 1;
 case HTML_HEAD:
+   obuf->flag |= RB_HEAD;
+   return 1;
 case HTML_N_BODY:
return 1;
 default:
diff --git a/fm.h b/fm.h
index 25857f8..9e12b42 100644
--- a/fm.h
+++ b/fm.h
@@ -675,6 +675,7 @@ struct readbuffer {
 #define RB_DEL 0x10
 #define RB_S   0x20
 #define RB_HTML5   0x40
+#define RB_HEAD0x80
 
 #define RB_GET_ALIGN(obuf) ((obuf)->flag_ALIGN)
 #define RB_SET_ALIGN(obuf,align) do{(obuf)->flag &= ~RB_ALIGN; (obuf)->flag |= 
(align); }while(0)