The current requirements to import new catalogs come with large technical debt (docker, python...). Compared this with Mix'n'match import where it is enough to provide:
a set of metadata fields about your catalog
a list of records from you catalog, each record consisting of 3-5 fields
The complex process of importing a new catalog (I counted four Python files that need to be edited!) should be replaced or by a method that requires no programming skills. Better define two import formats (based on TSV, NDJSON, YAML..) for metadata about a catalog (1) and for individual catalog records (2). Conversion from the original catalog data to soweego import format can be implemented by any kind of script (Python, Node, Perl, whatever...) instead of forcing data contributors to become programming contributors.
The current requirements to import new catalogs come with large technical debt (docker, python...). Compared this with Mix'n'match import where it is enough to provide:
The complex process of importing a new catalog (I counted four Python files that need to be edited!) should be replaced or by a method that requires no programming skills. Better define two import formats (based on TSV, NDJSON, YAML..) for metadata about a catalog (1) and for individual catalog records (2). Conversion from the original catalog data to soweego import format can be implemented by any kind of script (Python, Node, Perl, whatever...) instead of forcing data contributors to become programming contributors.