It's another mess --- I'm not very familiar with Python's string methods
--- but the code below seems to work. I've included a few comments.
def SPHINX_set_introspect_html(self, html, completing=False):
if completing:
self.__introspect_html = html
else:
# I think the following check should be in all *_set_introspect_html(),
# because set_introspect_html('') is called in worksheet.py, apparently
# with every cell evaluation.
if html == '':
self.__introspect_html = html
return
html = html.replace('\\\\', '\\')
from sage.misc.misc import SAGE_ROOT, tmp_dir
# [...snip...]
# I pause here because Sphinx "includes" load.js in the output as many
# times as set_introspect_html() has been called, which I daresay is
# a riddle. Even if we discard the header, this could become inefficient.
os.environ['SAGE_DOC_JSMATH'] = "True"
# [...snip...]
# It seems that Sphinx doesn't always use blockquotes, so I've
# added a few checks. I'm sure this could be much simpler.
html = new_html
i = html.find('<div class="body">')
html = html[i:]
i = html.rfind('</blockquote>')
if i == -1:
i = html.rfind('<div class="sphinxsidebar">')
if i == -1:
pass # worst case
else:
html = html[:i]
i = html.rfind('</div>')
html = html[:i]
i = html.rfind('</div>')
html = html[:i]
i = html.rfind('</div>')
html = html[:i]
else:
html = html[:i+13]
html = html.replace('<pre>', '<pre class="literal-block">')
html = '<div class="docstring">' + html + '</div>'
self.__introspect_html = html
With a dedicated "doc" directory in DOT_SAGE and some help from Python's
hashlib module, we might be able to keep a persistent (yet disposable)
cache of documentation, along the lines of ccache. It may be cheaper
for set_introspect_html() to apply a hash function to the incoming data
than to run Sphinx. (Alternatively, we could pass a time stamp as
another argument.) We could use the hash hex values themselves, plus
some human-readable identifiers, as file names. But I'm just speculating.
Sincerely,
Pat LeSmithe
John H Palmieri wrote:
> On Mar 31, 11:10 pm, Pat LeSmithe <qed...@gmail.com> wrote:
>> After some manual inspection of Sphinx's cmdline.py, I think I've made
>> some [crude] progress (screenshot attached). In cell.py, I replaced
>> Cell.set_introspect_html() with
>
> This is cool. I've modified it a bit: I've deleted some of the html
> and I've added some of the style sheet stuff to server/notebook/
> css.py. I'll try to add a patch to the trac ticket later today.
>
>> This has many problems:
>>
>> * It creates a new temporary directory for every non-completing
>> introspection.
>
> Right, and I think this makes it a bit slower than the ideal. I don't
> know if it's too slow to be usable.
>
>> * It needs CSS class-fication, e.g., with class="introspection", to
>> avoid spontaneously restyling other page elements.
>
> I think I've fixed this.
>
>> * It doesn't touch completions. Perhaps this is OK.
>
> Well, typing
>
> sage.algebras.steenrod_algebra
>
> doesn't even produce a docstring with my version of your patch.
>
>> * ?? output has extra triple quotes.
>
> Hmm. I haven't looked into this.
>
>> * It still doesn't get the math right, though I think it does link to
>> some jsMath code.
>
> I've fixed this.
>
>> * Search doesn't work.
>> * Links to the source and index are broken. I tried, unsuccessfully, to
>> move the sidebar to the right (seehttp://sphinx.pocoo.org/theming.html).
>> * No cross-links. Maybe doctreedir and freshenv are relevant here, too.
>
> I've deleted all of the extra stuff: just show the docstring.
>
> Here's a link to a new picture: (how do I add links to posts?)
>
> <http://sage.math.washington.edu/home/palmieri/misc/sphinx.png>
>
> John
>
> >
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to
sage-devel-unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---
