Data extract requests can be one of four types:

• Spatial extract based on a geographic area
• Bulk extract
• Differential bulk extract
• A request to populate the cache

A spatial extract is initiated from a PC running a custom SwERVE client application and will cause an extract based on a Smallworld GIS 'spatial scanner' to be performed. The client software consists of an application which allows the user to select the data types to be extracted from the Smallworld GIS and the geographic extent of the data to be extracted.

The SwERVE Extract ACP, in this case, will extract the data directly from the GIS (Figure 1), or will, if there is cached data available, take the majority of the data from the cache but will also 'look' in the GIS to check for changes which have been made to that data since the cache was built. Any changed data (added, deleted or modified) will be merged with the cache data before being made available to the client (Figure 2). The extract ACP will extract data based on 1-Km blocks.

Figure 1 Spatial Extract From GIS

Figure 2 Spatial Extract From Cache and GIS

The SwERVE Extract ACP will process a spatial extract request in a manner dependent on the particular mix of data types being requested. There are three types of information which may be extracted, these are:

• Geographic data types
• Non-geographic data types
• Join information (relationships) between data types

The three types are illustrated in Figure 3

Figure 3 Example Geographic, Non-Geographic and Join Data

Of these, the extract of geographic data types is the most straight forward with these being the root for any extract process. The extract of join information will occur where an object from one table is found to join to an object from another table in the currently selected collection. The extract of non-geographic information is handled in one of two ways. The first is where the selected non-geographic data type is extracted based on a join or relationship from one or more of the selected tables. The second is where it is found that no objects from a non-geographic table have been accessed via the join from a geographic table. In this case, all the information found in the GIS for the selected non-geographic table will be extracted. In the example shown in Figure 3, if the collection includes the non-geographic Occupier Details table and the geographic Street Map table, the specific non-geographic Occupier Details data will be extracted via the join from the geographic Street Map data. If, however, the collection only includes the Occupier Details table then all of the data from the Occupier Details table will be extracted.