ACTION-263: Summarize the options and the recommendations related to HTML parsing workflow in the CMS, see discussion

Hi all, as follow up to Lyon discussion here
http://www.w3.org/2012/11/01-mlw-lt-irc#T09-05-34

http://www.w3.org/2012/11/01-mlw-lt-irc#T10-09-46 marks the near ideal
solution that we arrived at during the coffee break. It is also presented
as such in the atached blurb along with other valid options.


I have summarized the options. This may develop into a best practice
document during 2013.

The attachment text copy pasted DOWN below:

Best regards
dF

Dr. David Filip
=======================
LRC | CNGL | LT-Web | CSIS
University of Limerick, Ireland
telephone: +353-6120-2781
*cellphone: +353-86-0222-158*
facsimile: +353-6120-2734
mailto: david.filip@ul.ie

 ITS categories transfer in CMS <-> LSP scenario



[This blurb is intended as a germ of a best practice document that could be
produced by the WG at a later stage]
Definitions [these will need to be merged with defs from the requirements
grandfathered document] CAT

Computer Aided Translation. Tooling making use of Translation Memory.
LSP

stands for a Language Service Provider, a company to which Localization
Services are being outsourced by corporations and SMBs. In some cases LSPs
can be internal corporate departments, such as Oracle WPTG (World Wide
Product Translation Group).
Localization Buyer

Corporations, small or medium businesses that need to make their content or
products multilingual and work in other than their home markets.
Status Quo:

Localization Buyers store HTML fragments in CDATA sections of XML
documents. This practice is common but far from being commendable, as the
CDATA sections are out of scope of any rules set in the carrier XML
document.

LSPs are used to coping with this bad practice and they normally have
cascaded parsing mechanisms that can handle the CDATA sections in an
intelligible way. However the CDATA can be just anything and so the
welformedness issues are dumped onto the LSP and LSPs stab on parsing the
CDATA as HTML or any other syntax is just a stab into the dark and
backfires every now and then.
Best Practice

The options in case you want to transfer useful metadata onto your
localization service provider are the following.

1)      Send valid ITS 2.0 decorated HTML 5 with an external XML rules file.

a.        This is a valid and conformant way, and all localizable content
is in scope of the externally provided rules. Thre is however the risk of
separation of the rules file that would make the its- prefixed mark up
within HTML useless in most cases.

2)      Use  XLIFF with ITS 2.0 mapping [XLIFF 1.2 is available, XLIFF 2.0
mapping to be finalized within 2013]

a.       This is a clean and conformant solution, but may no tbe feasible
if you do not have localizable content extraction know how or if your
target CAT tool does not support XLIFF

3)      Use XHTML 5 serialization that allows use of XML based ITS scoping
mechanisms in the same file as the content payload.
Technical caveats

TBD

[See current discussion on use of Tidy in PHP etc.]