Digital Media Publisher (DMP) Search: "blocked" in content in <h2>s in TOC view

[naglists]

Hiya,

Authoring using Editor 6.0 m090. Building DMP with 6.0 m010 or m020 (I
can't quite tell).

I've got a Styler stylesheet making HTML Help (CHM) files that are then
decompiled and rebuilt into DMP packages. <chapter> and <h1> are chunking.
<h2> and <h3> are not. The search results I am not seeing are when
displaying the Table of Contents hit results (not the defaul, Relevance
results).

If I have the following (pseudo-) structure (omitting some normal stuff for
brevity):

<chapter><title>
..content
<h1><title>
..content
<subhead><title>...
...content

then a search in DMP will find any text in the <h1> or any of its children
or descendants. However, if I have this:

<chapter><title>
..content
<h1><title>
..content
<h2><title>
..content
<subhead><title>...
...content

then a search in DMP will not find anything below the <h2>.

WHEN displaying the Table of Contents (not the default Relevance) results.
The content is all accessible when navigating through the TOC of the DMP
library/book.

What's up with that? Has anyone seen this behavior before? Is there some
setting I need to flip to "enable" DMP to see "through" an element to its
children/descendants?


--
Paul Nagai

[ClayHelberg]

Hi Paul--

I guess I'd want to see more detail about the HTML that your process is generating. If you are generating output for HTML Help and then repurposing it for DMP, it may be the case that your topics are off just a little bit from what DMP expects. (HTML Help is notorious for having some quirky markup requirements.)

Can you publish directly to DMP input image instead of HTML Help (or maybe in addition to HTML Help)? You might have better luck that way.

--Clay

[naglists]

Hey Clay,

Thanks for that idea. All my DMP stuff I built a loooong time ago before PE
had gathered up the capability. So it looks like Publish>DMP creates the
HTML files and a TOC file (topics.ditamap) for input into the DMP build
process and that you still run a "make" using those files. I got this to
build and the content to be included in the DMC package. However, if I
specify tocType="DMP" the content is omitted from the DMC package, so I do
not specify a tocType. Is there a value I should be using?

Also, when I point to the topics.ditamap (or a ditamap.ditamap I created
following the instructions in Help Center), I get all of the content,
including the css folder, the topics.ditamap and the ditamap.ditamap, if
present, dropped into the DMC's TOC ... flat ... with no hierarchy.

Can you share a content_config.xml and a ditamap.ditamap (and/or a
topics.ditamap) from a working DMP image prepared by Publish DMP that
consolidates the content into a "book"?

[ClayHelberg]

Hi Paul--

There are two ways you can do it in Arbortext 6.x. You can produce a "DMP Input Image", which generates the topic files, CSS, etc. but leaves the rest of the project process up to you (including building the actual DMC or WAR package). You can also produce a DMP Image directly, which should do everything for you and give you an installable DMC package. You may have better results if you do the full one-step publish rather than publishing an input image and then postprocessing into a DMC package.

If you have a full dmpmap, including project metadata at the top, you usually don't need a content_config.xml file as well. The easiest way to see a working example is to load one of the standard sample documents (New->Docbook->Sample) and publish as DMP image.

--Clay

[naglists]

Thanks for the additional tip, Clay. I've had to veer off for now.
Discovered I can't fully try this second tip until I upgrade my dev
environment. We currently do not have DMP installed on our PE servers ...
So I can build for input into DMP since that's all just XSL, but I can't
build and image or a multi-document image until I install DMP. That may
need to be the route I take, though ... I haven't figured out how to
include a DMP Input Image in an older content_config.xml type build
environment.

Anyhow ... Vacation is calling, so I'll have to come back to it in the next
gap.

[naglists]

Two updates:

1) With respect to "tuning" our current process of using PE and a Styler
stylesheet to produce HTML Help which we then decompile in order to
recompile using DMP: I have discovered that if I add HTML chunking to h2
and h3, then my search terms will be found and displayed with the results
are sorted by Table of Contents. Fascinating! However, when we do this, we
get the predictable but undesirable result that the right-hand content pane
now only shows tiny bits of content (h2 or h3, and if h2, then only the
content that is introductory / sibling to its descendant h3s). With
chunking only down to the h1, the entire h1 displays and the left-hand
navigation pane/TOC moves the reader about within that h1. I opened a
ticket with PTC hoping to get some tips on how to modify this behavior.

2) With respect to building the DMP input image with PE bypassing HTML Help
and the Microsoft decompiler: So I have upgraded one of my environments and
I now have the PE/DMP capability. I have successfully created a DMP input
image and built that input image into a DMC application, however ... I
can't quite figure out how to get a "book". Rather, I get the equivalent of
the resulting ditamap IN the left-hand TOC. Not a book title that acts like
a folder containing all the chapters of the book. Do I need to create a
ditamap to wrap the ditamap? Or is there some special way to indicate in
the content_config file that is now calling the ditamap that it should show
the whole thing as a unit, not each chapter as a unit. Does that make
sense? That said, the DMP build does not recognize a value ("DITAMAP") for
the attribute @tocType on DocumentSet in content_config, so some of that
may still be a version issue ... that is, my DMP predates full PE/DMP
integration. . It still builds without the attribute but ...

[naglists]

Another update: It turns out you can "see" the exclusion of content below
the lowest HTML chunked level from the search index in the CHM. I am now
thinking the problem we are facing (at least so long as we continue to use
the XML => CHM => HTML => DMP path) is either a limitation of Microsoft
HTML Help or a feature/deficiency of our Styler stylesheet ... whether my
coding or a limitation in its ability is not clear. Hoping to hear from
support on that question soon.

I should probably build a demo in axdocbook (and support may ask me to) to
see what's going on there.


On Fri, Jun 27, 2014 at 4:26 PM, Paul Nagai <->
wrote:

> Two updates:
>
> 1) With respect to "tuning" our current process of using PE and a Styler
> stylesheet to produce HTML Help which we then decompile in order to
> recompile using DMP: I have discovered that if I add HTML chunking to h2
> and h3, then my search terms will be found and displayed with the results
> are sorted by Table of Contents. Fascinating! However, when we do this, we
> get the predictable but undesirable result that the right-hand content pane
> now only shows tiny bits of content (h2 or h3, and if h2, then only the
> content that is introductory / sibling to its descendant h3s). With
> chunking only down to the h1, the entire h1 displays and the left-hand
> navigation pane/TOC moves the reader about within that h1. I opened a
> ticket with PTC hoping to get some tips on how to modify this behavior.
>
> 2) With respect to building the DMP input image with PE bypassing HTML
> Help and the Microsoft decompiler: So I have upgraded one of my
> environments and I now have the PE/DMP capability. I have successfully
> created a DMP input image and built that input image into a DMC
> application, however ... I can't quite figure out how to get a "book".
> Rather, I get the equivalent of the resulting ditamap IN the left-hand TOC.
> Not a book title that acts like a folder containing all the chapters of the
> book. Do I need to create a ditamap to wrap the ditamap? Or is there some
> special way to indicate in the content_config file that is now calling the
> ditamap that it should show the whole thing as a unit, not each chapter as
> a unit. Does that make sense? That said, the DMP build does not recognize a
> value ("DITAMAP") for the attribute @tocType on DocumentSet in
> content_config, so some of that may still be a version issue ... that is,
> my DMP predates full PE/DMP integration. . It still builds without the
> attribute but ...
>
>
>
> On Wed, Apr 9, 2014 at 6:21 PM, Clay Helberg <chelberg@terraxml.com>
> wrote:
>
>> Hi Paul--
>>
>>
>>
>> I guess I'd want to see more detail about the HTML that your process is
>> generating. If you are generating output for HTML Help and then repurposing
>> it for DMP, it may be the case that your topics are off just a little bit
>> from what DMP expects. (HTML Help is notorious for having some quirky
>> markup requirements.)
>>
>>
>>
>> Can you publish directly to DMP input image instead of HTML Help (or
>> maybe in addition to HTML Help)? You might have better luck that way.
>>
>>
>>
>> --Clay
>>
>>
>>
>> *From:* Paul Nagai [
>> in content in <h2>s in TOC view
>>
>>
>>
>>
>>
>> Hiya,
>>
>> Authoring using Editor 6.0 m090. Building DMP with 6.0 m010 or m020 (I
>> can't quite tell).
>>
>>
>>
>> I've got a Styler stylesheet making HTML Help (CHM) files that are then
>> decompiled and rebuilt into DMP packages. <chapter> and <h1> are chun...

[ClayHelberg]

Hi Paul-

Building something from the OOTB sample docs is always a good strategy when trying to determine whether an issue is specific to your custom doctype/code/content or not.

--Clay