fuzzy search on metatdata upload

[adamrp]

It would be awesome if, when metadata files are uploaded, each column is fuzzy-searched against all existing fields, and the system makes suggestions. E.g., "You entered ph. We already have a column called pH. Use pH instead?" and then the user can select "yes" or "no" (default to..... yes? thoughts?)

[antgonza]

+1

[squirrelo]

http://www.postgresql.org/docs/9.1/static/fuzzystrmatch.html

leaving this here

[squirrelo]

More to leave here: https://pypi.python.org/pypi/fuzzywuzzy/0.6.0

from fuzzywuzzy import process

>>> x = [r[0] for r in conn.execute_fetchall('select column_name from qiita.study_sample_columns')]
>>> process.extractBests('text', x, score_cutoff=70, limit=2)
[('texture', 90)]

>>> process.extractBests('altitiue', x, score_cutoff=70, limit=2)
[('altitude', 88), ('latitude', 75)]

>>> process.extractBests('altitiue', x, score_cutoff=80, limit=2)
[('altitude', 88)]

With python-Levenshtein installed it's really, really fast too. Would be two more dependencies, however, but they are pip installable.

[antgonza]

Metadata search has been completely removed from Qiita as this falls into redbiom ... closing.