From kkinder at rackspace.com  Tue Oct 15 13:09:03 2002
From: kkinder at rackspace.com (Ken Kinder)
Date: Tue Oct 15 13:09:03 2002
Subject: [Moin-devel] Formatter Troubles
Message-ID: <1034712664.16262.10.camel@ken>

I've had, and continue to have, a great deal of difficulty writing my
own formatter. I have a mostly working DocBook/XML output formatter, and
here's how it works:

  -- As particular methods are called on the formatter for the starting
and ending of elements, I keep track of context.

  -- Each of these methods returns a blank string.

  -- The document is build through a minidom object that I have to build
dynamicly by painstakingly construct by keeping track of the exact
context of where I am in the document.

  -- I modified wiki.py to call formatter.rawtext() when it has raw
unformatted text instead of using the formatter. That's the only way I
can actually capture what's in between the tags.

  -- Nothing is outputted until document end.

So, the problem is, I have text like

foo *bar*

What I get is:

<strong></strong>foo bar

Through debug variables, I've absolutely confirmed that strong is called
to open and close before "bar" is sent to rawtext.

Why might this be? Are there any other pitfalls you can think of that
are going to continue to bite me as a struggle mapping the way
formatters work to the way building an XML file via DOM work?

Thanks in advance.
-Ken

-- 
Ken Kinder
Requirements Manager; Core Team -- Rackspace Managed Hosting(tm)
210-892-4025 x1095




From kkinder at rackspace.com  Tue Oct 15 13:49:02 2002
From: kkinder at rackspace.com (Ken Kinder)
Date: Tue Oct 15 13:49:02 2002
Subject: [Moin-devel] Current XML Formatter
Message-ID: <1034715048.16262.16.camel@ken>

By the way, the current MoinMoin formatter does not produce valid XML.
That's why you should use DOM to build XML.

This example Wiki page:

  = Section 1 =
  == Subsection 1.1 ==
  Test Paragraph
  = Section 2 =
  Section 2 first '''paragraph'''

  Section 2 second paragraph

Results in:

  <?xml version="1.0" encoding="ISO-8859-1"?>
  <sect1><title>"FooBar"</title>
    <p>
      <anchor name="head-8238deed4280e27b481b25362606dde509438ffb"/>
        <s2 title="Section 1">
          <anchor name="head-17dd117be035422d0402f29c4b0b48e059ae01bf"/>
          <s3 title="Subsection 1.1">
            Test Paragraph 
            <anchor
name="head-464cab2347069f72abf5c4cb7d7f47bf4359fe75"/>
          </s3>
        </s2>
      <s2 title="Section 2">
        Section 2 first <strong>paragraph</strong> 
        </p>  <--- ***** Wowowwow, you have to close s2 first!!
    <p>Section 2 second paragraph </p></sect2></sect1>

This, again, is why you should use DOM -- that you why KNOW the XML is
valid. Using DOM is what I'm trying to do for my docbook formatter.

-- 
Ken Kinder
Requirements Manager; Core Team -- Rackspace Managed Hosting(tm)
210-892-4025 x1095




From lele at seldati.it  Tue Oct 15 14:48:02 2002
From: lele at seldati.it (Lele Gaifax)
Date: Tue Oct 15 14:48:02 2002
Subject: [Moin-devel] Formatter Troubles
In-Reply-To: <1034712664.16262.10.camel@ken>
References: <1034712664.16262.10.camel@ken>
Message-ID: <15788.36130.502570.120345@paci.nautilus>

>>>>> On 15 Oct 2002 15:11:03 -0500, Ken Kinder <kkinder at rackspace.com> said:

    KK> So, the problem is, I have text like

    KK> foo *bar*

    KK> What I get is:

    KK> <strong></strong>foo bar

    KK> Through debug variables, I've absolutely confirmed that strong
    KK> is called to open and close before "bar" is sent to rawtext.

    KK> Why might this be? Are there any other pitfalls you can think
    KK> of that are going to continue to bite me as a struggle mapping
    KK> the way formatters work to the way building an XML file via
    KK> DOM work?

This is the same thing that happens, in some circumstances, under
PyApache. I guess it has something to do with the way moin emits its
text, some time using sys.stdout.write, some time using print
statements: some parts of moin replace sys.stdout to get the output of
"subparts" (macros for example).

I've spent some time to track the problem, with out luck.

bye, lele.
-- 
nickname: Lele Gaifax	| Quando vivro' di quello che ho pensato ieri
real: Emanuele Gaifas	| comincero' ad aver paura di chi mi copia.
email: lele at seldati.it	|		-- Fortunato Depero, 1929.




From tbird20d at yahoo.com  Tue Oct 15 16:43:01 2002
From: tbird20d at yahoo.com (Tim Bird)
Date: Tue Oct 15 16:43:01 2002
Subject: [Moin-devel] Formatter Troubles
In-Reply-To: <15788.36130.502570.120345@paci.nautilus>
Message-ID: <20021015234211.80895.qmail@web20810.mail.yahoo.com>

--- Lele Gaifax <lele at seldati.it> wrote:
> >>>>> On 15 Oct 2002 15:11:03 -0500, Ken Kinder
> <kkinder at rackspace.com> said:
> 
>     KK> So, the problem is, I have text like
> 
>     KK> foo *bar*
> 
>     KK> What I get is:
> 
>     KK> <strong></strong>foo bar
> 
>     KK> Through debug variables, I've absolutely confirmed that
> strong
>     KK> is called to open and close before "bar" is sent to
> rawtext.
> 
>     KK> Why might this be? Are there any other pitfalls you can
> think
>     KK> of that are going to continue to bite me as a struggle
> mapping
>     KK> the way formatters work to the way building an XML file via
>     KK> DOM work?
> 
> This is the same thing that happens, in some circumstances, under
> PyApache. I guess it has something to do with the way moin emits
> its
> text, some time using sys.stdout.write, some time using print
> statements: some parts of moin replace sys.stdout to get the output
> of
> "subparts" (macros for example).
> 
> I've spent some time to track the problem, with out luck.

I'm just guessing, but maybe you need to flush whatever you're
writing to, right after your write:

sys.stdout.flush()

With all the streams being replaced and substituted, I wouldn't
be at all surprised to see stuff left in output buffers and not
making to the network socket in program write order.

Tim Bird

__________________________________________________
Do you Yahoo!?
Faith Hill - Exclusive Performances, Videos & More
http://faith.yahoo.com



From kkinder at rackspace.com  Wed Oct 16 07:14:05 2002
From: kkinder at rackspace.com (Ken Kinder)
Date: Wed Oct 16 07:14:05 2002
Subject: [Moin-devel] Formatter Troubles
In-Reply-To: <20021015234211.80895.qmail@web20810.mail.yahoo.com>
References: <20021015234211.80895.qmail@web20810.mail.yahoo.com>
Message-ID: <1034777796.18832.4.camel@ken>

On Tue, 2002-10-15 at 18:42, Tim Bird wrote:
>> This is the same thing that happens, in some circumstances, under
>> PyApache. I guess it has something to do with the way moin emits
>> its text, some time using sys.stdout.write, some time using print
>> statements: some parts of moin replace sys.stdout to get the output
>> of "subparts" (macros for example).
>> 
>> I've spent some time to track the problem, with out luck.
> 
> I'm just guessing, but maybe you need to flush whatever you're
> writing to, right after your write:
> 
> sys.stdout.flush()
> 
> With all the streams being replaced and substituted, I wouldn't
> be at all surprised to see stuff left in output buffers and not
> making to the network socket in program write order.
> 

No, that really isn't it. I've confirmed that if I have:

   This part is normal. '''This part is bold.'''

The method that invokes the strong highlighting *really*is* called
before the method that prints the text:

  -- formatter.strong is called to open
  -- formatter.strong is called to close
  -- formatter.rawtext (one I added) is called with "This part is
normal. This part is bold."

You see, I don't actually print the text as I go. In order to make sure
I produce proper XML, nothing is actually printed until
formatter.endDocument.

It seems that upon further investigation of the source code, the parser
is calling the formatter on strong tags and then *inserting* whatever it
returns into the text, before calling formatter.rawtext -- although this
answers my question, it doesn't really solve my problem.

I'm asking for advice on what the best way to build a document using
DOM. This process doesn't seem to fit into the formatter method because
of the problem I describe above.

-- 
Ken Kinder
Requirements Manager; Core Team -- Rackspace Managed Hosting(tm)
210-892-4025 x1095