[Sqlalchemy-users] Pickling (and restoring) objects

[Jonathan Hayward http://JonathansCorner.com]

I am trying to figure out how to use SQLalchemy for a composite class that
includes lists in its fields (most importantly of strings). I suspect two or
three obvious ways of doing it:

1: Do what is usually the wrong way in Python, making a relatively low level
   specification of "This object field is stored in this database." There is
   one table for the "document" class, and a separate table for the "histogram"
   class which is a field in the "document" class.

2: Find out how to solve it the right way, and use a mapper (I'm not clear on
   how) so that the "document" table seamlessly handles not only the strings,
   but i.e. the histogram (the histogram being a Python class).

(3: Some kind of hybrid?)

What I have now is an object that I'd like to be in the database but not in
memory unless it is being used, which contains (besides Strings and Floats) a
histogram, which can be my choice of a wrapper for a hash mapping words to the
positive number of occurrences, or in finalized form a tuple of pairs of (word,
frequency).

Does the mapper ignore fields that aren't specified as part of the schema? If
so, can I make a function to enter an object into memory, which enters all the
String/Float fields automatically, and then does extra work and translates the
word histogram into its own table?

What I have now are:

        document_table = sqlalchemy.Table("document", \
          sqlalchemy.Column("file_contents", sqlalchemy.String), \
          sqlalchemy.Column("filename", sqlalchemy.String), \
          sqlalchemy.Column("html_name", sqlalchemy.String, \
            primary_key = True), \
          sqlalchemy.Column("last_modified", sqlalchemy.Float(20)), \
          sqlalchemy.Column("last_viewed", sqlalchemy.Float(20)), \
          sqlalchemy.Column("permanent_ID", sqlalchemy.String), \
          sqlalchemy.Column("relative_filename", sqlalchemy.String), \
          sqlalchemy.Column("root", sqlalchemy.String), \
          sqlalchemy.Column("section", sqlalchemy.String), \
          sqlalchemy.Column("text", sqlalchemy.String), \
          sqlalchemy.Column("title", sqlalchemy.String), \
          sqlalchemy.Column("tokens", sqlalchemy.String), \
          sqlalchemy.Column("version", sqlalchemy.Integer), \
          )
        document_table_mapper = mapper(document, document_table)
        histogram_table = sqlalchemy.Table("histogram", \
          sqlalchemy.Column("html_name", sqlalchemy.String, \
            foreignkey("document.html_name")), \
          sqlalchemy.Column("word", sqlalchemy.String), \
          sqlalchemy.Column("count", sqlalchemy.Integer),
          )
        histogram_table_mapper = mapper(histogram, histogram_table)

How do I say "Make a document_table row for this document and for each word in
the histogram make a histogram_table row"?

--

Without loading all the results in memory, how do I get a cursorlike object for (if this SQL is wrong, PLEASE correct me) "SELECT document_table.* from document_table, histogram_table WHERE document_table.html_name = histogram_table.html_name AND histogram_table.word = 'xyz' SORT BY histogram.last_viewed DESC;" that will preferably load objects and not just rows?

--
++ Jonathan Hayward, [EMAIL PROTECTED]
** To see an award-winning website with stories, essays, artwork,
** games, and a four-dimensional maze, why not visit my home page?
** All of this is waiting for you at http://JonathansCorner.com

** If you'd like a Google Mail ( gmail.com) account, please tell me!