Expected Behavior

Given I have a Shapefile that is encoded via ISO-8859-1 and has a .cst file that denotes this (but no .cpg file)
When I upload this Shapefile using Geonode with a PostgreSQL/PostGIS backend set to UTF-8 encoding
Then the Shapefile is successfully converted from ISO-8859-1 to UTF-8 during the upload process and is successfully created as a new layer in Geonode.

Actual Behavior

The ogr2ogr command throws the following warning and error:

Warning 6: dataset XXX does not support layer creation option ENCODING
ERROR 1: Non UTF-8 content found when writing feature -1 of layer XXX

I can see that there is function which attempts to get the encoding from the .cst if there isn't a .cpg provided:

The problem appears to be that when it gets the encoding from the .cst file it is adding it to the ogr2ogr command via the ENCODING parameter:

This doesn't work because the target in this instance is the PostgreSQL database where the encoding is not managed at the layer (table) level, but rather at the database level. I think instead it should be setting a --config option, and as this is the handler for Shapefiles, it probably makes sense to use the SHAPE_ENCODING parameter, so something like:

if encoding:
  additional_options.append(f"--config SHAPE_ENCODING {encoding}")

Alternatively, the _get_encoding function could be updated so if there isn't a .cpg file then it creates one from the contents of the .cst file (as providing a .cpg file with the ISO-8859-1 encoding is handled correctly by ogr2ogr). This is probably more work though.

The following ZIP file can be used to replicate this issue. I've done some testing using the --config SHAPE_ENCODING method mentioned above and it resolves the issue in this instance.
hospitals_8859.zip

Specifications

• GeoNode version: 4.3.1
• Installation type (vanilla, geonode-project):
• Installation method (manual, docker):
• Platform:
• Additional details: