--- UPDATED to be more clear ---
I have quite the task ahead of me, and I am hoping Alembic and SQLAlchemy can do what I need.
I want to develop a desktop application that helps migrate data from one SQLite database to a completely different model and back again. So migrations right? That’s all well and good, but I need to ensure if I haven’t modeled a specific column/table, that it’s ported properly to a table that will be read later to reconstruct the database.
Example:
DB 1:
Table names
ID
First Name
Last Name
Table address
Street 1
Street 2
City
State
DB2:
Table givenName
ID
name
Table Surname
ID
Name
Say with alembic I have mapped the following:
DB1 names.firstname => DB2 givenName.name
DB1 names.lastname => DB2 surname.name
But say I want to use a migration to port DB1 to DB2, store the unknown data somewhere, and then reconstruct it properly when I go from DB2 -> DB1.
So how I’d envision this is a joiner table of sorts.
DB2
Table Joiner
table_name
column_name
data
The thing is I want this all to be completely dynamic so no piece of information is ever lost.
Now let's add an extra complexity, and I want to construct a generator of sorts so I can simply pass down new XML/JSON declarations. This would define the mappings, and any calls to translators already based in the code (date conversions, etc).
For an example of two database formats that need to be migrated from one to the other see https://cloudup.com/cYzP2lCQjbo
My question is if this is even possible or conceivable with SQLAlchemy and Alembic. How feasible is this? Any thoughts?
