User:Ianmcorvidae/Geordi Tips and Tricks
Though geordi has the potential to be quite useful, it's not always clear what's possible. This page will (hopefully) provide a general overview of what's available and how to use the tool.
Basic searches search all fields within a document. This isn't always what you want, but there's some ways to trim it down:
To search a specific field, you can prefix whatever you're interested in with the field name. Most notably, perhaps, this allows you to search based on properties of the matching, based on linked subitems (as the "search by linked subitem" search does, but you could search for more than one), or based on mapped fields. For learning the format of this section, see User:Ianmcorvidae/Ingestion. Some examples:
- "_geordi.mapping.release.label.name:Naxos" to search for items that look like releases put out by the label Naxos.
- "_geordi.matchings.current_matching.user:yourusername" to search for things you've matched (technically, those items where your matching is the most recent/current one -- you could use 'matchings' instead of 'current_matching' to search all matchings).
- '_geordi.matchings.current_matching.mbid:"some-mbid"' to search for things matched to the entity "some-mbid" -- but be warned that this only finds things that use that MBID exactly, which might not work correctly when entities have been merged. Work is planned to make this work as you'd expect, however. You can also use the 'type' field in current_matching to limit it to specific entity types, not that MBIDs should be reused anywhere.
If you want to get really fancy, you can also use elasticsearch's query DSL directly by using the "JSON Search" option. Read the documentation there on how to make this happen, and be warned that if your query is malformed you're likely to get a 500 error. But, it can also let you do cool things not yet implemented generally, like changing the number of search results displayed or sorting them by some field.
On item pages, the main interaction is using all those icons that you can see next to various bits of the interface. Specifically: matched items will show an icon corresponding to the type of MusicBrainz entity they're linked to, and the match icon will have a yellow or green background. A yellow background indicates a match performed by some automatic process, while a green background indicates a human-entered match. Some places will have a downward-pointing arrow; this jumps to the corresponding place in the "Link to MusicBrainz" section, which may have more information. Finally, there are two varieties of search icons; the plain black-and-white icon will search within geordi, for other items that also use the same subitem (artist, label, recording, etc.) as the icon corresponds to; the purple-and-yellow search icon will search MusicBrainz, though the parameters it chooses are far from well-tuned at this time -- choose wisely!
The "JSON Source" section provides a view of all available data. Though some of this data is extracted and formatted in the top section, some pieces may not be; if you're looking for extra information when importing, for example, perusing through this section might be useful. If you find anything that might be extracted into the top section, make sure to alert User:Ianmcorvidae of where it is, or use the link to the source code at the top of the page and submit a pull request.
The "Link to MusicBrainz" section lists all subitems that can be matched to MusicBrainz -- though there isn't much added functionality here, note that some subitems may appear only in this section, and not above.
NOTE: Import is far from complete, so I'll fill out this section after it takes something closer to its final form.