Some series files from GEO FTP server have invalid UTF-8 characters

[arteymix]

I've added a workaround in c3ea1be to skip retrieving samples for these cases, but it would be best to attempt other encodings. The XML file clearly indicate UTF-8. This also appears to be an issue with older documents as I've been testing the binary search for pre-2006 records.

Example stacktrace:

2024-12-09 16:25:02,527 ERROR 3707466 [main] u.g.c.l.e.g.s.GeoBrowserImpl.fillDetails(601) | Error while processing MINiML for GSE1889, sample details will not be obtained 
java.io.IOException: Document at https://ftp.ncbi.nlm.nih.gov/geo/series/GSE1nnn/GSE1889/miniml/GSE1889_family.xml.tgz appears to have invalid UTF-8 characters.                                                           at ubic.gemma.core.loader.expression.geo.service.GeoBrowserImpl.lambda$fetchDetailedGeoSeriesDocument$7(GeoBrowserImpl.java:791) ~[gemma-core-1.31.13-SNAPSHOT.jar:?]                                      
        at ubic.gemma.core.util.SimpleRetry.execute(SimpleRetry.java:48) ~[gemma-core-1.31.13-SNAPSHOT.jar:?]                                                                                                      
        at ubic.gemma.core.loader.expression.geo.service.GeoBrowserImpl.execute(GeoBrowserImpl.java:882) ~[gemma-core-1.31.13-SNAPSHOT.jar:?]                                                                      
        at ubic.gemma.core.loader.expression.geo.service.GeoBrowserImpl.fetchDetailedGeoSeriesDocument(GeoBrowserImpl.java:776) ~[gemma-core-1.31.13-SNAPSHOT.jar:?]                                               
        at ubic.gemma.core.loader.expression.geo.service.GeoBrowserImpl.fillDetails(GeoBrowserImpl.java:599) ~[gemma-core-1.31.13-SNAPSHOT.jar:?]                                                                  
        at ubic.gemma.core.loader.expression.geo.service.GeoBrowserImpl.retrieveGeoRecords(GeoBrowserImpl.java:471) ~[gemma-core-1.31.13-SNAPSHOT.jar:?]                                                           
        at ubic.gemma.core.apps.GeoGrabberCli.lambda$getGeoRecords$1(GeoGrabberCli.java:516) ~[gemma-cli-1.31.13-SNAPSHOT.jar:?]                                                                                   
        at ubic.gemma.core.util.SimpleRetry.execute(SimpleRetry.java:48) [gemma-core-1.31.13-SNAPSHOT.jar:?]                                                                                                       
        at ubic.gemma.core.apps.GeoGrabberCli.getGeoRecords(GeoGrabberCli.java:516) [gemma-cli-1.31.13-SNAPSHOT.jar:?]                                                                                 
        at ubic.gemma.core.apps.GeoGrabberCli.browseDatasets(GeoGrabberCli.java:357) [gemma-cli-1.31.13-SNAPSHOT.jar:?]
        at ubic.gemma.core.apps.GeoGrabberCli.doWork(GeoGrabberCli.java:288) [gemma-cli-1.31.13-SNAPSHOT.jar:?]
        at ubic.gemma.core.util.AbstractCLI.executeCommand(AbstractCLI.java:209) [gemma-cli-1.31.13-SNAPSHOT.jar:?]
        at ubic.gemma.core.apps.GemmaCLI.main(GemmaCLI.java:288) [gemma-cli-1.31.13-SNAPSHOT.jar:?]

https://ftp.ncbi.nlm.nih.gov/geo/series/GSE1nnn/GSE1889/miniml/GSE1889_family.xml.tgz

An alternative would be to report these to GEO so they can get their files fixed for good.

[arteymix]

It looks like we have a workaround in EutilsFetch that consists of stripping the <?xml header, might reuse that.