Function icu_data::ucm::request_mapping_file

pub fn request_mapping_file(mapping: &str) -> Result<String, IcuDataError>

Given the name of an encoding known to ICU, return its raw UCM data as a String.

You should not request encoding filenames (with .ucm), but it’ll still be understood.

Internally, this function considers 105 byte arrays stored as static strings. It takes the name of the mapping and looks it up in CHARSET_LOOKUP, which tells it what byte array contains the mapping. For example, CHARSET_LOOKUP["ibm-737_P100-1997.ucm"] is 58. So, we know that BYTES_58 contains the file. BYTES_58 is defined as:

include_bytes!("../../resources/brotli/ibm-737_P100-1997ibm-775_P100-1996ibm-803_P100-1999ibm-806_P100-1998ibm-808_P100-1999ibm-813_P100-1995ibm-819_P100-1999ibm-833_P100-1995ibm-834_P100-1995ibm-834_X100-1995.ucm.tar.b");

So, we un-Brotli compress the data in BYTES_58 and then send it through a .tar file parser. We iterate through the metadata entries until we find one equal to ibm-737_P100-1997.ucm. We then clone that data to a String type because only the compressed versions are owned in memory and return.

This function returns a Result<_, _> type because users may provide unknown mappings. The only error you should ever receive is IcuDataError::UnknownMappingRequested, all the valid mappings have been tested and decompress on my machine.