#
# Copyright 2009-2013 10gen, Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
# PODNAME: MongoDB::Examples
# ABSTRACT: Some examples of MongoDB syntax
__END__
=pod
=encoding UTF-8
=head1 NAME
MongoDB::Examples - Some examples of MongoDB syntax
=head1 VERSION
version v0.704.2.0
=head1 MAPPING SQL TO MONGODB
For developers familiar with SQL, the following chart should help you see how
many common SQL queries could be expressed in MongoDB.
These are Perl-specific examples of translating SQL queries to MongoDB's query
language. To see the JavaScript (or other languages') mappings, see
L<http://dochub.mongodb.org/core/sqlToMongo>.
In the following examples, C<$db> is a L<MongoDB::Database> object which was
retrieved by using C<get_database>. See L<MongoDB::MongoClient> for more.
=over
=item C<CREATE TABLE USERS (a Number, b Number)>
Implicit, can be done explicitly.
=item C<INSERT INTO USERS VALUES(1,1)>
$db->get_collection( 'users' )->insert( { a => 1, b => 1 } );
=item C<SELECT a,b FROM users>
$db->get_collection( 'users')->find( { } )->fields( { a => 1, b => 1 });
=item C<SELECT * FROM users>
$db->get_collection( 'users' )->find;
=item C<SELECT * FROM users WHERE age=33>
$db->get_collection( 'users' )->find( { age => 33 } )
=item C<SELECT a,b FROM users WHERE age=33>
$db->get_collection( 'users' )->find( { age => 33 } )->fields( { a => 1, b => 1 });
=item C<SELECT * FROM users WHERE age=33 ORDER BY name>
$db->get_collection( 'users' )->find( { age => 33 } )->sort( { name => 1 } );
=item C<<SELECT * FROM users WHERE age>33>>
$db->get_collection( 'users' )->find( { age => { '$gt' => 33 } } );
=item C<<SELECT * FROM users WHERE age<33>>
$db->get_collection( 'users' )->find( { age => { '$lt' => 33 } } );
=item C<SELECT * FROM users WHERE name LIKE "%Joe%">
$db->get_collection( 'users' )->find( { name => qr/Joe/ } );
=item C<SELECT * FROM users WHERE name LIKE "Joe%">
$db->get_collection( 'users' )->find( {name => qr/^Joe/ } );
=item C<<SELECT * FROM users WHERE age>33 AND age<=40>>
$db->get_collection( 'users' )->find( { age => { '$gt' => 33, '$lte' => 40 } } );
=item C<SELECT * FROM users ORDER BY name DESC>
$db->get_collection( 'users' )->find->sort( { name => -1 } );
=item C<CREATE INDEX myindexname ON users(name)>
$db->get_collection( 'users' )->ensure_index( { name => 1 } );
=item C<CREATE INDEX myindexname ON users(name,ts DESC)>
$db->get_collection( 'users' )->ensure_index( Tie::IxHash->new( name => 1, ts => -1 ) );
In this example, we must use L<Tie::IxHash> to preserve the ordering of the arguments to
C<ensureIndex>.
=item C<SELECT * FROM users WHERE a=1 and b='q'>
$db->get_collection( 'users' )->find( {a => 1, b => "q" } );
=item C<SELECT * FROM users LIMIT 10 SKIP 20>
$db->get_collection( 'users' )->find->limit(10)->skip(20);
=item C<SELECT * FROM users WHERE a=1 or b=2>
$db->get_collection( 'users' )->find( { '$or' => [ {a => 1 }, { b => 2 } ] } );
=item C<SELECT * FROM users LIMIT 1>
$db->get_collection( 'users' )->find->limit(1);
=item C<EXPLAIN SELECT * FROM users WHERE z=3>
$db->get_collection( 'users' )->find( { z => 3 } )->explain;
=item C<SELECT DISTINCT last_name FROM users>
$db->run_command( { distinct => "users", key => "last_name" } );
=item C<SELECT COUNT(*y) FROM users>
$db->get_collection( 'users' )->count;
=item C<<SELECT COUNT(*y) FROM users where age > 30>>
$db->get_collection( 'users' )->find( { "age" => { '$gt' => 30 } } )->count;
=item C<SELECT COUNT(age) from users>
$db->get_collection( 'users' )->find( { age => { '$exists' => 1 } } )->count;
=item C<UPDATE users SET a=1 WHERE b='q'>
$db->get_collection( 'users' )->update( { b => "q" }, { '$set' => { a => 1 } } );
=item C<UPDATE users SET a=a+2 WHERE b='q'>
$db->get_collection( 'users' )->update( { b => "q" }, { '$inc' => { a => 2 } } );
=item C<DELETE FROM users WHERE z="abc">
$db->get_database( 'users' )->remove( { z => "abc" } );
=back
=head1 DATABASE COMMANDS
If you do something in the MongoDB shell and you would like to translate it to
Perl, the best way is to run the function in the shell without parentheses, which
will print the source. You can then generally translate the source into Perl
fairly easily.
For example, suppose we want to use C<db.foo.validate> in Perl. We could
run:
> db.foo.validate
function (full) {
var cmd = {validate:this.getName()};
if (typeof full == "object") {
Object.extend(cmd, full);
} else {
cmd.full = full;
}
var res = this._db.runCommand(cmd);
if (typeof res.valid == "undefined") {
res.valid = false;
var raw = res.result || res.raw;
if (raw) {
var str = "-" + tojson(raw);
res.valid = !(str.match(/exception/) || str.match(/corrupt/));
var p = /lastExtentSize:(\d+)/;
var r = p.exec(str);
if (r) {
res.lastExtentSize = Number(r[1]);
}
}
}
return res;
}
Thus, we can translate the important parts into Perl:
$db->run_command( { validate => "foo" } );
=head2 Find-and-modify
The find-and-modify command is similar to update (or remove), but it will return
the modified document. It can be useful for implementing queues or locks.
For example, suppose we had a list of things to do, and we wanted to remove the
highest-priority item for processing. We could do a L<MongoDB::Collection/find>
and then a L<MongoDB::Collection/remove>, but that wouldn't be atomic (a write
could occur between the query and the remove). Instead, we can use find and
modify.
my $next_task = $db->run_command({
findAndModify => "todo",
sort => {priority => -1},
remove => 1
});
This will atomically find and pop the next-highest-priority task.
See L<http://www.mongodb.org/display/DOCS/findAndModify+Command> for more
details on find-and-modify.
=head1 AGGREGATION
The aggregation framework is MongoDB's analogy for SQL GROUP BY queries,
but more generic and more powerful. An invocation of the aggregation framework
specifies a series of stages in a pipeline to be executed in order by
the server. Each stage of the pipeline is
drawn from one of the following so-called "pipeline operators":
C<$project>, C<$match>, C<$limit>, C<$skip>, C<$unwind>, C<$group>,
C<$sort>, and C<$geoNear>.
The aggregation framework is the preferred way of performing
most aggregation tasks. New in version 2.2, it has largely
obviated mapReduce (L<http://docs.mongodb.org/manual/reference/command/mapReduce/#dbcmd.mapReduce>),
and group (L<http://docs.mongodb.org/manual/reference/command/group/#dbcmd.group>).
See the MongoDB aggregation framework documentation for more
information (L<http://docs.mongodb.org/manual/aggregation/>).
=head2 $match and $group
The C<$group> pipeline operator is used like GROUP BY in SQL. For example,
suppose we have a number of local businesses stored in a "business" collection.
If we wanted to find the number of coffeeshops in each neighborhood, we
could do:
my $out = $db->get_collection('business')->aggregate(
[
{'$match' => {'type' => 'coffeeshop'}},
{'$group' => {'_id' => '$neighborhood', 'num_coffeshops' => {'$sum' => 1}}}
]
);
The SQL equivalent is C<SELECT neighborhood, COUNT(*) FROM business GROUP BY neighborhood WHERE type = 'coffeeshop'>.
After executing the above aggregation query, C<$out> will contain an array of
result documents such as the following:
[
{
'_id' => 'Soho',
'num_coffeshops' => 23
},
{
'_id' => 'Chinatown',
'num_coffeshops' => 14
},
{
'_id' => 'Upper East Side',
'num_coffeshops' => 10
},
{
'_id' => 'East Village',
'num_coffeshops' => 87
}
]
Note that L<MongoDB::Collection/aggregate> takes an array reference as
an argument. Each element of the array is document which specifies a stage
in the aggregation pipeline. Here our aggregation query consists of a
C<$match> phase followed by a C<$group> phase. Use C<$match> to filter the
documents in the collection prior to aggregation. The C<_id> field in the
C<$group> stage specifies the key to group by; the C<$> in C<'$neighborhood'>
indicates that we are referencing the name of a key. Finally, we use the
C<$sum> operator to add one for every document in a particular neighborhood.
There are other operators, such as C<$avg>, C<$max>, C<$min>, C<$push>, and
C<$addToSet>, which can be used in the C<$group> phase and work much like
C<$sum>.
=head2 $project and $unwind
Now let's look at a more complex example of the aggregation framework that
makes use of the C<$project> and C<$unwind> pipeline operators. Suppose
we have a collection called 'courses' which contains information on college
courses. An example document in the collection looks like this:
{
'_id' => 'CSCI0170',
'name' => 'Computer Science 17',
'description' => 'An Integrated Introduction to Computer Science',
'instructor_id' => 29823498,
'instructor_name' => 'A. Greenwald',
'students' => [
{ 'student_id' => 91736114, 'student_name' => 'D. Storch' },
{ 'student_id' => 89100891, 'student_name' => 'J. Rassi' }
]
}
We wish to generate a report containing one document per student that indicates
the courses in which each student is enrolled. The following call to
L<MongoDB::Collection/aggregate> will do the trick:
my $out = $db->get_collection('courses')->aggregate([
{'$unwind' => '$students'},
{'$project' => {
'_id' => 0,
'course' => '$_id',
'student_id' => '$students.student_id',
}
},
{'$group' => {
'_id' => '$student_id',
'courses' => {'$addToSet' => '$course'}
}
}
]);
The output documents will each have a student ID number and an array of the
courses in which that student is enrolled:
[
{
'_id' => 91736114,
'courses' => ['CSCI0170', 'CSCI0220', 'APMA1650', 'HIST1230']
}
{
'_id' => 89100891,
'courses' => ['CSCI0170', 'CSCI1670', 'CSCI1690']
}
]
The C<$unwind> stage of the aggregation query "peels off" elements of the courses
array one-by-one and places them in their own documents. After this phase completes,
there is a separate document for each (course, student) pair. The C<$project> stage
then throws out unecessary fields and keeps the ones we are interested in. It also
pulls the student ID field out of its subdocument and creates a top-level field
with the key C<student_id>. Last, we group by student ID, using C<$addToSet> in
order to add the unique courses for each student to the C<courses> array.
=head2 $sort, $skip, and $limit
The C<$sort>, C<$skip>, and C<$limit> pipeline operators work much like their
companion methods in L<MongoDB::Cursor>. Returning to the previous students and
courses example, suppose that we were particularly interested in the student with
the ID that is numerically third-to-highest. We could retrieve the course list for that
student by adding C<$sort>, C<$skip>, and C<$limit> phases to the pipeline:
my $out = $db->get_collection('courses')->aggregate([
{'$unwind' => '$students'},
{'$project' => {
'_id' => 0,
'course' => '$_id',
'student_id' => '$students.student_id',
}
},
{'$group' => {
'_id' => '$student_id',
'courses' => {'$addToSet' => '$course'}
}
},
{'$sort' => {'_id' => -1}},
{'$skip' => 2},
{'$limit' => 1}
]);
=head2 Group
In addition to the aggregation framework, MongoDB offers a few special
commands for common aggregation tasks: C<group>, C<distinct>, and C<count>.
Returning to the coffeeshop example, the same result could be obtained using
C<group> with the following code:
my $reduce = <<REDUCE;
function(doc, prev) {
if (doc.type == "coffeeshop") {
prev["num coffeeshops"]++;
}
}
REDUCE
my $result = $db->run_command({group => {
'ns' => "business",
'key' => {"neighborhood" => 1},
'initial' => {"num coffeeshops" => 0},
'$reduce' => MongoDB::Code->new(code => $reduce)
Modern code should generally prefer the C<$group> aggregation pipeline
operator to the C<group> database command.
=head2 Distinct
The distinct command returns all values for a given key in a collection. For
example, suppose we had a collection with the following documents (C<_id> value
ignored):
{ 'name' => 'a', code => 1 }
{ 'name' => 'b', code => 1 }
{ 'name' => 'c', code => 2 }
{ 'name' => 'd', code => "3" }
If we wanted to see all of values in the "code" field, we could run:
my $result = $db->run_command([
"distinct" => "collection_name",
"key" => "code",
"query" => { }
]);
Notice that the arguments are in an array, to ensure that their order is
preserved. You could also use a L<Tie::IxHash>.
C<query> is an optional argument, which can be used to only run C<distinct> on
specific documents. It takes a hash (or L<Tie::IxHash> or array) in the same
form as L<MongoDB::Collection/"find($query)">.
Running C<distinct> on the above collection would give you:
{
'ok' => '1',
'values' => [
1,
2,
"3"
]
};
=head2 MapReduce
For some special purpose aggregation tasks, the aggregation framework may not
be sufficient. In this case, the database server can execute special MapReduce
jobs written in JavaScript. Be warned: MapReduce is generally slower than the
aggregation framework and should be avoided unless your application requires
the flexibility that it provides.
This example counts the number of occurrences of each tag in a collection. Each
document contains a "tags" array that contains zero or more strings.
my $map = <<MAP;
function() {
this.tags.forEach(function(tag) {
emit(tag, {count : 1});
});
}
MAP
my $reduce = <<REDUCE;
function(prev, current) {
result = {count : 0};
current.forEach(function(item) {
result.count += item.count;
});
return result;
}
REDUCE
my $cmd = Tie::IxHash->new("mapreduce" => "foo",
"map" => $map,
"reduce" => $reduce);
my $result = $db->run_command($cmd);
See the MongoDB documentation on MapReduce for more information
(L<http://docs.mongodb.org/manual/core/map-reduce>).
=head1 QUERYING
=head2 Nested Fields
MongoDB allows you to store deeply nested structures and then query for fields
within them using I<dot-notation>. For example, suppose we have a users
collection with documents that look like:
{
"userId" => 12345,
"address" => {
"street" => "123 Main St",
"city" => "Springfield",
"state" => "MN",
"zip" => "43213"
}
}
If we want to query for all users from Springfield, we can do:
my $cursor = $users->find({"address.city" => "Springfield"});
This will search documents for an "address" field that is a subdocument and a
"city" field within the subdocument.
=head1 UPDATING
=head2 Positional Operator
In MongoDB 1.3.4 and later, you can use positional operator, C<$>, to update
elements of an array. For instance, suppose you have an array of user
information and you want to update a user's name.
A sample document in JavaScript:
{
"users" : [
{
"name" : "bill",
"age" : 60
},
{
"name" : "fred",
"age" : 29
},
]
}
The update:
$coll->update({"users.name" => "fred"}, {'users.$.name' => "george"});
This will update the array so that the element containing C<"name" =E<gt> "fred">
now has C<"name" =E<gt> "george">.
=head1 AUTHORS
=over 4
=item *
David Golden <david.golden@mongodb.org>
=item *
Mike Friedman <friedo@mongodb.com>
=item *
Kristina Chodorow <kristina@mongodb.org>
=item *
Florian Ragwitz <rafl@debian.org>
=back
=head1 COPYRIGHT AND LICENSE
This software is Copyright (c) 2014 by MongoDB, Inc..
This is free software, licensed under:
The Apache License, Version 2.0, January 2004
=cut