[Z3-zemantic] Re: big zemantic storage / some changes

[Tres Seaver]

David Pratt wrote:

> Found this googling:
> 
> zope2/Products/ZemanticIndex/README.txt and other sources for some zope2
> testing
> 
> Where is this source being maintained. Can it be checked out?

The code there is a rough sketch I put together on the plane back from
the Paris sprint.  It doesn't actually integrate Zemantic / rdflib yet.
 The Subversion repository for the checkins is in a not-yet-public area,
used at the sprint.

$ svn co <URL>/zope2/Products/ZemanticIndex
A  ZemanticIndex/tests
A  ZemanticIndex/tests/test_cataloguing.py
A  ZemanticIndex/tests/__init__.py
A  ZemanticIndex/__init__.py
A  ZemanticIndex/interfaces.py
A  ZemanticIndex/README.txt
A  ZemanticIndex/index.py
Checked out revision 471.

$ cat ZemanticIndex/README.txt
ZemanticIndex Product README

  Overview

    This product allows a Zemantic triple store to index content within
    a Zope2 catalog.

  Indexing Content

    The index adapts the object being indexed (the "subject", in RDF
    terms) to IZemanticInfo, which returns a sequence of two-tuple
    statement fragments about the subject.  Thses statements are in the
    form, (<predicate>, <object>).  The index then constructs
    three-tuples from these values, in the form (<subject>, <predicate>,
    <object>), using the catalog's UID as the first first term

  Searching Content

    The index expects its 'query' object to be adaptable to
    IZemanticInfo.  It uses the two-tuples supplied by that adapter to
    synthesize a query for matching documents, duing an "and" search by
    default.  The 'operator' passed to the index can be either 'and'
    (perform an intersection), or an 'or' (perform a union).

    As a convenience for HTML-generated queries, the indes will also
    accept its 'query' term as a sequence of (<predicate>, <object>)
    tuples.

I'm attaching the current sources in case anybody would like to move the
project forward (I'm not going to be able to work on it for several
weeks, at least).  Note again that the implementation in 'index.py' is a
false-front, intended only to help flesh out the tests (so that they
would exercise "real" behavior when it eventually switches over to rdflib).


Tres.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ZemanticIndex-20050507.tar.gz
Type: application/gzip
Size: 3357 bytes
Desc: not available
Url : http://codespeak.net/pipermail/z3-zemantic/attachments/20050507/6fc7cdc9/ZemanticIndex-20050507.tar.bin