export

I've been musing about the possibility of building a module for mass importing contents from existing sites, as an aid for migration to Drupal. I came across a potential model, an import utility for the CMS SAPID, http://sapid.sourceforge.net/en/doc/import/, code at http://prdownloads.sourceforge.net/sapid/migrator.tar.gz?download. The approach combines web crawling with regular expression-based data extraction. Users input the wrapping code patterns to search for, e.g., main content might be wrapped in and or maybe a div with a given class name.

[Some questions/suggestions that are kinda late in the game, but coming from research I've been doing today....]

To what extent is this API intended to facilitate data exchanges between Drupal and other CMSs?

Have you seen the Portable Site Information PSI specification, http://psilib.sourceforge.net/? It sets out a generalized XML interchange format for CMS data (and provides some tools written in Python). Would it be useful to consider this as a basis for the Import/Export XML format?

A potentially useful reference is the Exorcist softare produced as an offshoot of the Midgard CMS, http://www.midgard-project.org/documentation/exorcist/. It provides a general toolset for data exchanges between CMSs (in Java).

Posted on behalf of dman@drupal.org:

From http://basement.greenash.net.au/soc2006/ImportExportApiModule?show_comments=1#comment_19:

Where is good for discussions? A Drupal.org thread, the issues register, or somewhere here? I'd rather keep it open than emails.

Granted you may not be ready for a code review yet :) I had a quick look and was surprised you are attacking the database directly. Wouldn't it be better to let Drupal continue to handle the serialization, and get (eg) user_load() or node_load() to instantiate the php object, and base your mappings to that structure? By digging directly at the database (clever though your structure mapping system is) you are bypassing a lot of Drupal goodness, especially once you come to node_load().