Currently we just get the hash from browser URL and try to parse node id as the last element:

https://gitlab.iscpif.fr/gargantext/purescript-gargantext/blob/dev/src/Gargantext/Components/Forest/Breadcrumb.purs#L73

But we could have done this:

{ route } <- Store.use
route' <- T.useLive T.unequal route

And now the route' variable is of type Gargantext.Routes -> AppRoute and contains all info about all the ids and sessions involved in that route, already parsed. This is more reliable than parsing the last element of the path - for things like annuaire details etc the last element is from the contexts table, not nodes.

The get breadcrumb branch GraphQL call could use the route' info to give more precise breadcrumb path. Sometimes (e.g. for document it just returns 'Home' because the last element is in contexts). See also #645