20130807-delete-patinfo

<< | Index | >>

Summary

• Swissmedicinfo importer does not handle correctly the case when patinfo has to be deleted

Commits

• https://github.com/ngiger/oddb.org/commit/16205f80b1ac4b8bad9e439f3f9c79d5a5217855 Import swissmedicinfo. Delete unused patinfo
• https://github.com/ngiger/sbsm/commit/e351c4a268eba321e08474856553485127d7869a Omitting some tests which probably never ran

<..>

• https://github.com/ngiger/sbsm/commit/487276366a41db773d8c8d5a276b8f0990901bfa Some changes to support travis-CI

Index

• Swissmedicinfo importer does not handle correctly the case when patinfo has to be deleted
• Added travis-CI for sbsm
• Updated travis-CI for spreadsheet

Added travis-CI for sbsm

Added (in small chunks) some commits to add tracis-CI and fixing some problems. Ommitted a bunch of test, where I could not find any commit where they passed.

Open problems:

• Shouldn't we ommit the failing test/test_trans_handler.rb:41 which would require us to have rockit 0.3.8 installed?
• Would it be okay to adapt the expectations to the actual results in various failing tests of test/test_trans_handler.rb?

Updated travis-CI for spreadsheet

• Found out that we alredy had a .travis.yml. Added myself to the notifications
• Used minitest/reportes (Must probably revert this change to test, whether this is the cause for the failure under ruby 1.8.7 and 2.0.0

Swissmedicinfo importer does not handle correctly the case when patinfo has to be deleted

• Yasuhiro made some changes in April to handle the case, that update has to handle the case, that on FI or PI are present.

Restored database with dump from oddb.org. Looked at sequences (59160) should be empty after import.

ch.oddb> registration('59160').sequences.size
-> 2
ch.oddb> registration('60115').sequences.size
-> 2

Reloaded database using /usr/local/bin/pg_create_db.sh (stopped all oddb services before).

Changed manually on line 2686540/2804585 of (xmllinted data/xml/AipsDownload_latest.xml)PI fromm 60115 -> 59160. (vim 2686539:2686550s/60115/59160/g) Rerun the import (added a few puts for better tracing). Now at least a search for "Levofloxacin-Mepha" return PI and FI for each of the 10 results.

The following manual seemed to work registration('59160').each_sequence{ |seq| seq.patinfo = nil; seq.odba_isolated_store }. After adding it to the plugin/text_info.rb, also the patinfo in french ware removed. Reloading database from backup.

Question: Is it correct to only seq.patinfo = nil or must I delete the seq.patinfo-object? No we must call delete on it or it would leave in the database a

breadrumbs of Nasivin and Cardiospermum Salbe Cosmochema are wrongly linked.

Does reimporting both resolve the problem? Yes. This is okay now (online and on my private VM).

Trying to do the delete via bin/cmd says how to delete FI

Redoing an import which should remove patinfo. Seems to be okay. Pushed the commit

Dokumented deleting the Patinfo under http://dev.ywesee.com/Main/CHoddbdeletePI