-
Notifications
You must be signed in to change notification settings - Fork 12
/
aggregate.go
233 lines (206 loc) · 4.57 KB
/
aggregate.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
package datatable
import (
"bytes"
"encoding/gob"
"fmt"
"github.com/cespare/xxhash"
"github.com/datasweet/datatable/serie"
"github.com/pkg/errors"
)
// GroupBy defines the group by configuration
// Name is the name of the output column
// Type is the type of the output column
// Keyer is our main function to aggregate
type GroupBy struct {
Name string
Type ColumnType
Keyer func(row Row) (interface{}, bool)
}
// AggregationType defines the avalaible aggregation
type AggregationType uint8
const (
Avg AggregationType = iota
Count
CountDistinct
Cusum
Max
Min
Median
Stddev
Sum
Variance
)
func (a AggregationType) String() string {
switch a {
case Avg:
return "avg"
case Count:
return "count"
case CountDistinct:
return "count_distinct"
case Cusum:
return "cusum"
case Max:
return "max"
case Min:
return "min"
case Median:
return "median"
case Stddev:
return "stddev"
case Sum:
return "sum"
case Variance:
return "variance"
default:
panic("unkwown aggregation type")
}
}
// AggregateBy defines the aggregation
type AggregateBy struct {
Type AggregationType
Field string
As string
}
// GroupBy splits our datatable by group
func (dt *DataTable) GroupBy(by ...GroupBy) (*Groups, error) {
if len(by) == 0 {
return nil, ErrNoGroupBy
}
var groups []*group
gindex := make(map[uint64]int)
for pos := 0; pos < dt.nrows; pos++ {
row := dt.Row(pos)
buf := bytes.NewBuffer(nil)
enc := gob.NewEncoder(buf)
buckets := make([]interface{}, len(by))
for i, k := range by {
k := &k
if v, ok := k.Keyer(row); ok {
buckets[i] = v
enc.Encode(v)
}
}
hash := xxhash.Sum64(buf.Bytes())
if at, ok := gindex[hash]; ok {
groups[at].Rows = append(groups[at].Rows, pos)
} else {
gindex[hash] = len(groups)
groups = append(groups, &group{
Key: hash,
Buckets: buckets,
Rows: []int{pos},
})
}
}
return &Groups{dt: dt, groups: groups, by: by}, nil
}
// Aggregate aggregates some field
func (dt *DataTable) Aggregate(by ...AggregateBy) (*DataTable, error) {
g := &Groups{
dt: dt,
groups: []*group{
&group{TakeAll: true},
},
}
return g.Aggregate(by...)
}
// Groups
type Groups struct {
dt *DataTable
by []GroupBy
groups []*group
}
type group struct {
Key uint64
Buckets []interface{}
Rows []int
TakeAll bool
}
// Aggregate our groups
func (g *Groups) Aggregate(aggs ...AggregateBy) (*DataTable, error) {
if g == nil {
return nil, ErrNoGroups
}
if g.dt == nil {
return nil, ErrNilDatatable
}
// check cols
series := make(map[string]serie.Serie)
for _, agg := range aggs {
col := g.dt.Column(agg.Field)
if col == nil {
err := errors.Errorf("column '%s' not found", agg.Field)
return nil, errors.Wrap(err, ErrColumnNotFound.Error())
}
switch agg.Type {
case Avg, Count, CountDistinct, Cusum, Max, Min, Median, Stddev, Sum, Variance:
series[agg.Field] = col.(*column).serie
default:
return nil, ErrUnknownAgg
}
}
out := New(g.dt.name)
// create columns
for _, by := range g.by {
typ := by.Type
if len(typ) == 0 {
typ = Raw
}
if err := out.AddColumn(by.Name, typ); err != nil {
err = errors.Wrapf(err, "can't add column '%s'", by.Name)
return nil, errors.Wrap(err, ErrCantAddColumn.Error())
}
}
for _, agg := range aggs {
name := agg.As
if len(name) == 0 {
name = fmt.Sprintf("%s %s", agg.Type, agg.Field)
}
typ := Float64
switch agg.Type {
case Count, CountDistinct:
typ = Int64
default:
}
if err := out.AddColumn(name, typ); err != nil {
err = errors.Wrapf(err, "can't add column '%s'", name)
return nil, errors.Wrap(err, ErrCantAddColumn.Error())
}
}
// aggregate the series
for _, group := range g.groups {
values := make([]interface{}, 0, len(group.Buckets)+len(aggs))
values = append(values, group.Buckets...)
for _, agg := range aggs {
serie := series[agg.Field]
if !group.TakeAll {
serie = serie.Pick(group.Rows...)
}
switch agg.Type {
case Avg:
values = append(values, serie.Avg())
case Count:
values = append(values, serie.Count())
case CountDistinct:
values = append(values, serie.CountDistinct())
case Cusum:
values = append(values, serie.Cusum())
case Max:
values = append(values, serie.Max())
case Min:
values = append(values, serie.Min())
case Median:
values = append(values, serie.Median())
case Stddev:
values = append(values, serie.Stddev())
case Sum:
values = append(values, serie.Sum())
case Variance:
values = append(values, serie.Variance())
}
}
out.AppendRow(values...)
}
return out, nil
}