diff options
author | Bryan Newbold <bnewbold@archive.org> | 2020-04-10 12:58:31 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2020-04-10 12:58:31 -0700 |
commit | 44f12b817be9d4ccd4b2c0c6ea07ea401ae86ffa (patch) | |
tree | 96198ef9af427718730c9ecc56eb731da54ba1cc /notes | |
parent | ee53062606c6d194c52013063c4ff414fc762a00 (diff) | |
download | fatcat-covid19-44f12b817be9d4ccd4b2c0c6ea07ea401ae86ffa.tar.gz fatcat-covid19-44f12b817be9d4ccd4b2c0c6ea07ea401ae86ffa.zip |
add jupyter notebook and client notes
Diffstat (limited to 'notes')
-rw-r--r-- | notes/clients.md | 29 |
1 files changed, 29 insertions, 0 deletions
diff --git a/notes/clients.md b/notes/clients.md new file mode 100644 index 0000000..807bd7f --- /dev/null +++ b/notes/clients.md @@ -0,0 +1,29 @@ + +## Python Client + +Install (using a virtualenv, conda, whatever): + + pip install jupyter matplotlib elasticsearch-dsl + + + +## Browser Javascript + +In short, *direct, rich, browser search queries will not work* because browser +HTTP GET requests do not allow bodies, and this is the mechanism Elasticsearch +uses to send complex queries. The search.fatcat.wiki endpoint currently blocks +all POST requests for security reasons. + +This includes things like the elasticsearch-js/elasticsearch-browser client, +and use from "notebooks" like observablehq.com, jsfiddle, etc. + +A subset of queries may work using only GET with URL parameters, for example +X-Pack SQL queries. + +The resolution for this would be to implement proper read-only filtering of ES +requests, either with middleware or upgrading ES and configuring access +policies properly. + +Note: would import with something like: + + elasticsearch = require('https://bundle.run/elasticsearch-browser@16.7.1'); |