I would like to find documents on a path in a tree structure in MongoDB using $graphLookup.
For example, let's say I have this small structure:
0 - root
└── 01 "home"
├── 011 "john"
│ ├── 0111 "documents"
│ └── 0112 "movies"
│ ├── 01121 "Holiday inn"
│ └── 01122 "Harry Potter"
└── 012 "maria"
representing the data:
[
{"_id":"0","name":"root","parent":""},
{"_id":"01","name":"home","parent":"0"},
{"_id":"011","name":"john","parent":"01"},
{"_id":"0111","name":"documents","parent":"011"},
{"_id":"0112","name":"movies","parent":"011"},
{"_id":"01121","name":"Holiday Inn","parent":"0112"},
{"_id":"01122","name":"Harry Potter","parent":"0112"},
{"_id":"012","name":"maria","parent":"01"}
]
Is it possible to use $graphLookup (potentially with other steps in the aggregation) to
- check that a given path (e.g. "root/home/maria") exists?
- to retrieve all documents on a target path (e.g. "root/home/john/movies/Harry Potter")?
In principle it should be possible to do it. But I'm not sure if $graphLookup is able to do it efficiently. That is, by checking that on each step in the recursion we're looking for a document with a specific name. The inefficient alternative would be to generate all paths of length N (starting with "root" folder) and find the matching one. Another inefficient options would be to store the path in the documents themselves, but that would incur a cost of many updates to child documents in case of renames. These are solutions I would very much like to avoid, for obvious reasons. But I'm unable to figure out whether an efficient query can be written for MongoDB.
Complications:
- a folder may have potentially many children, so it may be easier to model the structure using parent references,
- users may rename things, so it wouldn't be efficient to store the path in the documents themselves.
