Count implementation

CHANGELOG.md

@@ -21,6 +21,8 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
   key specified with DDL schema or in `_ddl_sharding_key` space.
   NOTE: CRUD methods delete(), get() and update() requires that sharding key
   must be a part of primary key.
+* `crud.count()` function to calculate the number of tuples
+  in the space according to conditions.
 
 ### Fixed
 

README.md

@@ -87,6 +87,7 @@ Table below describe what operations supports custom sharding key:
 | `replace()` / `replace_object()` | Yes                        |
 | `upsert()` / `upsert_object()`   | Yes                        |
 | `select()` / `pairs()`           | Yes                        |
+| `count()`                        | Yes                        |
 | `update()`                       | Yes                        |
 | `upsert()` / `upsert_object()`   | Yes                        |
 | `replace() / replace_object()`   | Yes                        |
@@ -586,6 +587,48 @@ crud.len('customers')
 ...
 ```
 
+### Count
+
+`CRUD` supports multi-conditional count, treating a cluster as a single space.
+The same as with `select()` the conditions may include field names or numbers,
+as well as index names. The recommended first condition is a TREE index; this
+helps to reduce the number of tuples to scan. Otherwise a full scan is performed.
+If compared with `len()`, `count()` method scans the entire space to count the
+tuples according user conditions. This method does yield that's why result may
+be approximate. Number of tuples before next `yield()` is under control with
+option `yield_every`.
+
+```lua
+local result, err = crud.count(space_name, conditions, opts)
+```
+
+where:
+
+* `space_name` (`string`) - name of the space
+* `conditions` (`?table`) - array of [conditions](#select-conditions)
+* `opts`:
+  * `yield_every` (`?number`) - number of tuples processed to yield after,
+    `yield_every` should be > 0, default value is 100
+  * `timeout` (`?number`) - `vshard.call` timeout (in seconds), default value is 2
+  * `bucket_id` (`?number|cdata`) - bucket ID
+  * `force_map_call` (`?boolean`) - if `true`
+    then the map call is performed without any optimizations even,
+    default value is `false`
+  * `mode` (`?string`, `read` or `write`) - if `write` is specified then `count` is
+    performed on master, default value is `read`
+  * `prefer_replica` (`?boolean`) - if `true` then the preferred target is one of
+    the replicas, default value is `false`
+  * `balance` (`?boolean`) - use replica according to
+    [vshard load balancing policy](https://www.tarantool.io/en/doc/latest/reference/reference_rock/vshard/vshard_api/#router-api-call),
+    default value is `false`
+
+```lua
+crud.count('customers', {{'<=', 'age', 35}})
+---
+- 5
+...
+```
+
 ### Call options for crud methods
 
 Combinations of `mode`, `prefer_replica` and `balance` options lead to:

crud.lua

@@ -11,6 +11,7 @@ local delete = require('crud.delete')
 local select = require('crud.select')
 local truncate = require('crud.truncate')
 local len = require('crud.len')
+local count = require('crud.count')
 local borders = require('crud.borders')
 local sharding_key = require('crud.common.sharding_key')
 local utils = require('crud.common.utils')
@@ -76,6 +77,10 @@ crud.truncate = truncate.call
 -- @function len
 crud.len = len.call
 
+-- @refer count.call
+-- @function count
+crud.count = count.call
+
 -- @refer borders.min
 -- @function min
 crud.min = borders.min
@@ -113,6 +118,7 @@ function crud.init_storage()
     select.init()
     truncate.init()
     len.init()
+    count.init()
     borders.init()
     sharding_key.init()
 end

crud/common/sharding.lua

@@ -2,12 +2,24 @@ local vshard = require('vshard')
 local errors = require('errors')
 
 local BucketIDError = errors.new_class("BucketIDError", {capture_stack = false})
+local GetReplicasetsError = errors.new_class('GetReplicasetsError', {capture_stack = false})
 
 local utils = require('crud.common.utils')
 local sharding_key_module = require('crud.common.sharding_key')
 
 local sharding = {}
 
+function sharding.get_replicasets_by_bucket_id(bucket_id)
+    local replicaset, err = vshard.router.route(bucket_id)
+    if replicaset == nil then
+        return nil, GetReplicasetsError:new("Failed to get replicaset for bucket_id %s: %s", bucket_id, err.err)
+    end
+
+    return {
+        [replicaset.uuid] = replicaset,
+    }
+end
+
 function sharding.key_get_bucket_id(key, specified_bucket_id)
     if specified_bucket_id ~= nil then
         return specified_bucket_id

crud/select/plan.lua → crud/compare/plan.lua

@@ -12,12 +12,15 @@ if has_keydef then
     keydef_lib = compat.require('tuple.keydef', 'key_def')
 end
 
-local select_plan = {}
+local plan = {}
 
 local IndexTypeError = errors.new_class('IndexTypeError', {capture_stack = false})
 local FilterFieldsError = errors.new_class('FilterFieldsError', {capture_stack = false})
 local NoIndexesError = errors.new_class('NoIndexesError', {capture_stack = false})
 
+-- TREE index helps reducing the number of tuples to scan.
+-- Otherwise a full scan is performed. So at least one of
+-- condition indexes or primary index should be of type TREE
 local function index_is_allowed(index)
     return index.type == 'TREE'
 end
@@ -158,7 +161,7 @@ local function construct_after_tuple_by_fields(fieldno_map, field_names, tuple)
     return transformed_tuple
 end
 
-function select_plan.new(space, conditions, opts)
+function plan.new(space, conditions, opts)
     dev_checks('table', '?table', {
         first = '?number',
         after_tuple = '?table|cdata',
@@ -297,9 +300,9 @@ function select_plan.new(space, conditions, opts)
     return plan
 end
 
-select_plan.internal = {
+plan.internal = {
     get_sharding_key_from_scan_value = get_sharding_key_from_scan_value,
     extract_sharding_key_from_conditions = extract_sharding_key_from_conditions
 }
 
-return select_plan
+return plan