Question

I’m trying to use rmongodb to fetch information from a MongoDB database for further processing in R. However, I have some difficulties to really get started. This one works:

cursor <- mongo.find(mongo, "people", query=list(last.name="Smith", first.name="John"),
                 fields=list(address=1L, age=1L))
while (mongo.cursor.next(cursor)){
  print(mongo.cursor.value(cursor))}

Now, what if I want to find people whose first name is either “John” or “Bob” or “Catherine”? I tried query=list(last.name="Smith", first.name=c(John, Bob, Catherine)) but this didn’t work out. Replacing = with % didn’t work either.

Another issue is that the database content is nested, which means I have subtrees, subsubtrees etc. For example, for the entry first.name="John", last.name="Smith" I might have subentries like address, age, occupation, and for occupation again I might have categories as subtrees (e.g. years from 2005 to 2012 and for each year I would have an an entry like “unemployed”, “clerk”, “entrepreneur”). So what if I want to find all people with first name “John” who are 40 years old and were unemployed in 2010? What would the query look like?

EDIT as a reply to Stennie: Here’s an example of the structure of my database and the query I’m trying to do. Imagine that alumnis of a university have been subdivided into groups (e.g. “very good students”, “good students” and so on). Each group then contains a list of people that have been assigned to this group along with their details.

(0){..}
   _id  : (Object ID) class id
   groupname: (string) unique name for this group (e.g. "beststudents")
   members[11]
       (0){..}
           persid : (integer) 1
           firstname: (string)
           surname: (string)
           age: (integer)
           occupation: (string)
       (1){..}
           persid : (integer) 2
           firstname: (string)
           surname: (string)
           age: (integer)
           occupation: (string)
#      and so on until (10){..}
(1){..}
   _id  : (Object ID) class id
   groupname: (string) unique name for this group
   members[3]
       (0){..}
           persid : (integer) 1
           firstname: (string)
           surname: (string)
           age: (integer)
           occupation: (string)
#      and so on until (2){..}
# and many more

Now let’s assume that I am interested in the groups with the names “best students” and “good students”, and would like to get “surname” and “occupation” for each member of each of these groups as an R object in order to do some plots, stats or whatever. And maybe I’d also want to refine this request to only get those members that are younger than 40 years old. Now after having read Stennie’s reply, I tried it this way:

cursor <- mongo.find(mongo, "test.people",
          list(groupname=list('$in'=c("beststudents", "goodstudents")),
               members.age=list('$lt'=40) # I haven't tried this with my DB, so I hope this line is right
               ),
          fields=list(members.surname=1L, members.occupation=1L)
        )
count <- mongo.count(mongo, "test.people",
          list(groupname=list('$in'=c("beststudents", "goodstudents")),
               members.age=list('$lt'=40)
               )
        )
surnames <- vector("character", count)
occupations <- vector("character", count)
i <- 1
while (mongo.cursor.next(cursor)) {
  b <- mongo.cursor.value(cursor)
  surnames[i] <- mongo.bson.value(b, "members.surname")
  occupations[i] <- mongo.bson.value(b, "members.occupation")
  i <- i + 1
}
df <- as.data.frame(list(surnames=surnames, occupations=occupations))

There’s no error message after running this, but I get an empty data frame. What’s wrong with this code?

Answer 1

Now, what if I want to find people whose first name is either “John”
or “Bob” or “Catherine”?

You can use the $in operator for this:

cursor <- mongo.find(mongo, "test.people",
   list(last.name="Smith", 
        first.name=list('$in'=c('John','Bob','Catherine'))
   )
)

It would be worth having a read of the MongoDB Advanced Queries page as well as Dot Notation (Reaching Into Objects).

Another issue is that the database content is nested, which means
I have subtrees, subsubtrees etc.

The data structure sounds potentially challenging to manipulate; would need a practical example of a document to try to illustrate the query.

So what if I want to find all people with first name “John” who
are 40 years old and were unemployed in 2010? What would the query look like?

Making some assumptions on the data structure, here is an example of a simple “and” query:

cursor <- mongo.find(mongo, "test.people",
    list(
        first.name='John',
        fy2012.job='unemployed',
        age = 40
    )
)