Bug 73980 – webkit.py gdb pretty printer should consider 8-bit characters of StringImpl.

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Hayato Ito]

A webkit.py gdb pretty printer always uses StringImpl.m_data16 as data buffer. This is wrong assumption and pretty printer for WTFString seems broken.
We should take care of 8bit data buffers, m_data8, as well as m_data16.

[Hayato Ito]

Created attachment 118166 [details]
pretty printers for 8-bit StringImpl and LChar*

[Tony Chang]

Comment on attachment 118166 [details]
pretty printers for 8-bit StringImpl and LChar*

Can you rebase your change? https://bugs.webkit.org/show_bug.cgi?id=73878 added similar support, but it could be extended further to decode the utf8 and handle LChar*.

[Hayato Ito]

Ops. I didn't notice there was a similar unlanded patch. Let me rebase/merge two patches and upload the result.

(In reply to comment #2)
> (From update of attachment 118166 [details])
> Can you rebase your change? https://bugs.webkit.org/show_bug.cgi?id=73878 added similar support, but it could be extended further to decode the utf8 and handle LChar*.

[Hayato Ito]

Created attachment 118301 [details]
rebased

[Hayato Ito]

I rebased it.

I don't think that we can decode m_data8 using 'utf-16' as in https://bugs.webkit.org/show_bug.cgi?id=73878. So in this patch, I return m_data8 without any decoding. I think this is a reasonable assumption.

[Tony Chang]

Comment on attachment 118301 [details]
rebased

View in context: https://bugs.webkit.org/attachment.cgi?id=118301&action=review

> Tools/gdb/webkit.py:60
> +def ustring_to_unicode_string(ptr, length=None):
>      """Convert a pointer to UTF-16 data into a Python Unicode string.

We should try to be consistent with our use of unicode strings or utf8 strings.  Using utf8 might be easier, but I'll defer to your judgement.

> Tools/gdb/webkit.py:71
>      string = struct.pack('H' * length, *char_vals).decode('utf-16', 'replace')
> +    return string + unicode(error_message)

You could use utf8 for both of these.

> Tools/gdb/webkit.py:133
> +            return lstring_to_string(self.val['m_data8'], self.get_length())
> +        return ustring_to_unicode_string(self.val['m_data16'], self.get_length()).encode('utf-8')

And could drop the extra encode('utf-8') here.

[Hayato Ito]

Created attachment 118310 [details]
make it utf-8 string

[Hayato Ito]

Thank you for the review.

I updated the patch so that it uses Python's (non-unicode) string everywhere.

(In reply to comment #6)
> (From update of attachment 118301 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=118301&action=review
> 
> > Tools/gdb/webkit.py:60
> > +def ustring_to_unicode_string(ptr, length=None):
> >      """Convert a pointer to UTF-16 data into a Python Unicode string.
> 
> We should try to be consistent with our use of unicode strings or utf8 strings.  Using utf8 might be easier, but I'll defer to your judgement.
> 
> > Tools/gdb/webkit.py:71
> >      string = struct.pack('H' * length, *char_vals).decode('utf-16', 'replace')
> > +    return string + unicode(error_message)
> 
> You could use utf8 for both of these.
> 
> > Tools/gdb/webkit.py:133
> > +            return lstring_to_string(self.val['m_data8'], self.get_length())
> > +        return ustring_to_unicode_string(self.val['m_data16'], self.get_length()).encode('utf-8')
> 
> And could drop the extra encode('utf-8') here.

[Tony Chang]

Comment on attachment 118310 [details]
make it utf-8 string

Thanks!

[WebKit Review Bot]

Comment on attachment 118310 [details]
make it utf-8 string

Clearing flags on attachment: 118310

Committed r102420: <http://trac.webkit.org/changeset/102420>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.