3.2. Configure a Schema

3.2.1. Source Schemas Column Maps

Generally, colmap editing has these steps:

  • Ingest, to setup the source tables
  • bambry colmap -c, to create the table colmaps
  • edit the table colmaps
  • bambry colmap -b, to create the combined column map
  • Review the combined column map
  • bambry colmap -l, to load the combined colman into the source schema.
  • bambry sync -r, to sync from the source schema records back out to files.

Start by importing the headerstypes bundle from ambry/test/bundle_tests/ingest.example.com/headerstypes. This bundle has one file, header4, that has an extra header line, so the column names are a bit off.

$ ambry import -d ambry/test/bundle_tests/ingest.example.com/headerstypes
$ bambrycd d00headers003
$ mv schema.csv schema-orig.csv
$ mv source-schema.csv source-schema-orig.csv Creating the Column Map

Ingest the header files. Then create a column map file and open it in a spreadsheet program.

$ bambry ingest -t headers
$ bambry colmap -c
$ open colmap_header.csv.

[ TO make these descriptions easier, ‘column’ referrs to a colum in the spreadsheet, while a column in a source or destination table will be called a ‘field’ ]

In the colmap file, the first column is the names of the fields of the destination table, and the rest of the columns are for the field names in an ingested source. In the colmap_header.csv file, you’ll notice that most of the files have the same field names in the same rows, but headers4 is off in its own.

The file can be a bit hard to understand, so think of it like this: Find a column, column 2 or larger, where the first line is the name of a source. Then, go down that column to find a particular field name in the source. Then, look at the name in the first column to find the field name that the source field will be mapped to in the destination.

When the column map is first created, the soruce and destination field names are all the same. In this case, The renter_cost_gt_30 field in the headers1 source table maps to the renter_cost_gt_30 field in the destination table. However, in the headers4 source table, the gvid field is gvid_a, and the title_right_in_the_middle_cost_gt_30_cv_b is too long.

To fix these, move the gvid_a field name up to the row where gvid is in the first column, under headers. This
will cause the gvid_a source field to be mapped to the gvid output column.

For this example, just move all of the headers4 fields up, then delete the corresponding field names in the first column. Refer to source-schema-orig.csv to see the final version.

Only edit field names in the first column. Outside of the first column, only move field names. Never edit the first row. Creating the Column Map

After editing one or more table column maps, you can combine them with

$ bambry colmap -b

This command will create the colmap.csv file, which will hold the source and destination field names for only the fields that changed. Load the Column Map

$ bambry colmap -b

Load the column map into records in the bundle, altering the existing source table records with the changed from the colmap.csv file. Export the final source schema.

$ bambry sync -o

Sync out to create the new source_schema.csv file, which has the changes you made to the headers4 table.

3.2.2. Destination Schemas