participants.tsv information is silently not embedded?

[yarikoptic]

What is the problem?

on ds000201 which is a bit "damaged" by the crawling (some tarballs content seems should have gone under derivatives - yet to fix), but with a very rich participants.tsv (although without Age but with AgeGroup) etc. Thanks to #2149 quickly found out that no gender is reported for a sample subject although it is specified:

$> datalad -f json_pp plugin extract_metadata type=bids file=sub-9001/ses-1/anat/sub-9001_ses-1_T1w.nii.gz | grep -i Gender

$> grep 9001 participants.tsv                                                                                           
9001	Male	Young	19,78997095	19,78997095	Studerar för närvarande på universitet/högskola	0	1	12	4,75	4,666666667	6	5,6	10:30:00	11:30:00	06:10:00	08:00:00	5 Nästan aldrig	00:00:00	2 Stämmer inte så bra	1 Stämmer inte alls	1 Stämmer inte alls	2 Stämmer inte så bra	3 Stämmer ganska bra	4 Ja, i stort sett tillräckligt	5 Mycket litet problem	08:00:00	3 Varken bra eller dåligt	4	3,571428571	4,142857143	1,285714286	2,428571429	55	14	14	27	31	26	39	42	19	46	41	17	16	6	5	5	28	NA	1,533333333	1,625	1,272727273	1,727272727	Male	Male	SurelyNot	Surely	No	NA	No		Early	Early			Interesting	Good	NA	NA	NA	NA	No	No, but lost focus		5	4	32	11	18	6	5	30	11	16	38	29	21	30	51	43	40	30	37	10	11	29	16	6	2	2	6	0	2	B

note: this was ran with #2151 merged in for reading unicode participants.tsv... didn't check where the problem lies yet or it is the one to blame -- too much of DEBUG output, while no warning if aggregate-metadata ran without any increased level of logging. just reporting for now

[mih]

Does it produce any metadata from participants.tsv? It would only be read when generating dataset-level metadata. But you are pointing it to a single file, without giving it a --dataset.

[mih]

Please see #2326 for a related RF.