From binary to specific inputstream

[bjdmeest]

An open issue to get some discussion going:

The current spec specifies rml:compression X as a way to specify

from this binary file, use decompression algorithm X, and return the first file you find as an inputstream

example:

<#JSON> a rml:LogicalSource;
     rml:source [ a rml:Source, rml:FilePath;
       rml:path "/path/to/data.json.gz";
       rml:compression rml:gzip; # GZip compression
     ];

I have the feeling this can be generalized to specify

from this binary file, use 'input stream retrieval algorithm X', use 'path selector Y', to return an inputstream

Concrete use case is mapping, e.g., XMP metadata from image files

<#XML> a rml:LogicalSource;
     rml:source [ a rml:Source, rml:FilePath;
       rml:path "/path/to/picture.heif";
       rml:inputStream [
           rml:binaryInterpreter rml:heif_xmp
     ];

But this could also be used for the first example:

<#JSON> a rml:LogicalSource;
     rml:source [ a rml:Source, rml:FilePath;
       rml:path "/path/to/data.json.gz";
       rml:inputStream [
           rml:binaryInterpreter rml:gunzip
           rml:streamReference "relative/path/to/specific/file/in/gz/file.json"
     ];

We extend from 'compressed' files to 'all types of binary files', WDYT?