-
Notifications
You must be signed in to change notification settings - Fork 153
/
result.go
338 lines (297 loc) · 9.51 KB
/
result.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
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
package flux
import (
"io"
"github.com/influxdata/flux/array"
"github.com/influxdata/flux/iocounter"
"github.com/influxdata/flux/semantic"
"github.com/influxdata/flux/values"
)
type Result interface {
Name() string
// Tables returns a TableIterator for iterating through results
Tables() TableIterator
}
type TableIterator interface {
Do(f func(Table) error) error
}
// Table represents a set of streamed data with a common schema.
// The contents of the table can be read exactly once.
//
// This data structure is not thread-safe.
type Table interface {
// Key returns the set of data that is common among all rows
// in the table.
Key() GroupKey
// Cols contains metadata about the column schema.
Cols() []ColMeta
// Do calls f to process the data contained within the table.
// This must only be called once and implementations should return
// an error if this is called multiple times.
Do(f func(ColReader) error) error
// Done indicates that this table is no longer needed and that the
// underlying processor that produces the table may discard any
// buffers that need to be processed. If the table has already been
// read with Do, this happens automatically.
// This is also not required if the table is empty.
// It should be safe to always call this function and call it multiple
// times.
Done()
// Empty returns whether the table contains no records.
// This should not return true when the table is empty because of an error.
Empty() bool
}
// BufferedTable is an implementation of Table that has all of its
// data buffered.
type BufferedTable interface {
Table
// Buffer returns the i'th buffer in the buffered table.
// This allows accessing the buffered table contents without
// using the Table.
Buffer(i int) ColReader
// BufferN returns the number of buffers in this table.
BufferN() int
// Copy will return a copy of the BufferedTable without
// consuming the Table itself. If this Table has already
// been consumed by the Do method, then this will panic.
Copy() BufferedTable
}
// ColMeta contains the information about the column metadata.
type ColMeta struct {
// Label is the name of the column. The label is unique per table.
Label string
// Type is the type of the column. Only basic types are allowed.
Type ColType
}
// ColType is the type for a column. This covers only basic data types.
type ColType int
const (
TInvalid ColType = iota
TBool
TInt
TUInt
TFloat
TString
TTime
)
// ColumnType returns the column type when given a semantic.Type.
// It returns flux.TInvalid if the Type is not a valid column type.
func ColumnType(typ semantic.MonoType) ColType {
switch typ.Nature() {
case semantic.Bool:
return TBool
case semantic.Int:
return TInt
case semantic.UInt:
return TUInt
case semantic.Float:
return TFloat
case semantic.String:
return TString
case semantic.Time:
return TTime
default:
return TInvalid
}
}
func SemanticType(typ ColType) semantic.MonoType {
switch typ {
case TBool:
return semantic.BasicBool
case TInt:
return semantic.BasicInt
case TUInt:
return semantic.BasicUint
case TFloat:
return semantic.BasicFloat
case TString:
return semantic.BasicString
case TTime:
return semantic.BasicTime
default:
return semantic.MonoType{}
}
}
// String returns a string representation of the column type.
func (t ColType) String() string {
switch t {
case TInvalid:
return "invalid"
case TBool:
return "bool"
case TInt:
return "int"
case TUInt:
return "uint"
case TFloat:
return "float"
case TString:
return "string"
case TTime:
return "time"
default:
return "unknown"
}
}
// ColReader allows access to reading arrow buffers of column data.
// All data the ColReader exposes is guaranteed to be in memory.
// A ColReader that is produced when processing a Table will be
// released once it goes out of scope. Retain can be used to keep
// a reference to the buffered memory.
type ColReader interface {
Key() GroupKey
// Cols returns a list of column metadata.
Cols() []ColMeta
// Len returns the length of the slices.
// All slices will have the same length.
Len() int
Bools(j int) *array.Boolean
Ints(j int) *array.Int
UInts(j int) *array.Uint
Floats(j int) *array.Float
Strings(j int) *array.String
Times(j int) *array.Int
// Retain will retain this buffer to avoid having the
// memory consumed by it freed.
Retain()
// Release will release a reference to this buffer.
Release()
}
type GroupKey interface {
Cols() []ColMeta
Values() []values.Value
HasCol(label string) bool
LabelValue(label string) values.Value
IsNull(j int) bool
ValueBool(j int) bool
ValueUInt(j int) uint64
ValueInt(j int) int64
ValueFloat(j int) float64
ValueString(j int) string
ValueDuration(j int) values.Duration
ValueTime(j int) values.Time
Value(j int) values.Value
Equal(o GroupKey) bool
Less(o GroupKey) bool
String() string
// EqualTrueNulls should be functionally identical to Equal,
// with the exception that two nulls should never be treated
// as equal - the idea being that, by strict technical definition,
// null should not equal null. Therefore, this implementation is
// closer to the `true` null behavior.
//
// This is useful for transformations like join()
// that repurpose group keys for something other than grouping.
EqualTrueNulls(o GroupKey) bool
}
// GroupKeys provides a sortable collection of group keys.
type GroupKeys []GroupKey
func (a GroupKeys) Len() int { return len(a) }
func (a GroupKeys) Less(i, j int) bool { return a[i].Less(a[j]) }
func (a GroupKeys) Swap(i, j int) { a[i], a[j] = a[j], a[i] }
// String returns a string representation of the keys
func (a GroupKeys) String() string {
var result string
for _, key := range a {
result += key.String() + "\n"
}
return result
}
// ResultDecoder can decode a result from a reader.
type ResultDecoder interface {
// Decode decodes data from r into a result.
Decode(r io.Reader) (Result, error)
}
// ResultEncoder can encode a result into a writer.
type ResultEncoder interface {
// Encode encodes data from the result into w.
// Returns the number of bytes written to w and any error.
Encode(w io.Writer, result Result) (int64, error)
}
// MultiResultDecoder can decode multiple results from a reader.
type MultiResultDecoder interface {
// Decode decodes multiple results from r.
Decode(r io.ReadCloser) (ResultIterator, error)
}
// MultiResultEncoder can encode multiple results into a writer.
type MultiResultEncoder interface {
// Encode writes multiple results from r into w.
// Returns the number of bytes written to w and any error resulting from the encoding process.
// It is up to the specific implementation for whether it will encode any errors that occur
// from the ResultIterator.
Encode(w io.Writer, results ResultIterator) (int64, error)
}
// EncoderError is an interface that any error produced from
// a ResultEncoder implementation should conform to.
// It allows for differentiation
// between errors that occur in results, and errors that occur while encoding results.
type EncoderError interface {
IsEncoderError() bool
}
// isEncoderError reports whether or not the underlying cause of
// an error is a valid EncoderError.
func isEncoderError(err error) bool {
encErr, ok := err.(EncoderError)
return ok && encErr.IsEncoderError()
}
// DelimitedMultiResultEncoder encodes multiple results using a trailing delimiter.
// The delimiter is written after every result.
//
// If an error is encountered when iterating and the error is an encoder error,
// the error will be returned. Otherwise, the error is assumed to
// have arisen from query execution, and said error will be encoded with the
// EncodeError method of the Encoder field.
//
// If the io.Writer implements flusher, it will be flushed after each delimiter.
type DelimitedMultiResultEncoder struct {
Delimiter []byte
Encoder interface {
ResultEncoder
// EncodeError encodes an error on the writer.
EncodeError(w io.Writer, err error) error
}
}
type flusher interface {
Flush()
}
// Encode will encode the results into the writer using the Encoder and separating each entry
// by the Delimiter. If an error occurs while processing the ResultIterator or is returned from
// the underlying Encoder, Encode will return the error if nothing has yet been written to the
// Writer. If something has been written to the Writer, then an error will only be returned
// when the error is an EncoderError.
func (e *DelimitedMultiResultEncoder) Encode(w io.Writer, results ResultIterator) (int64, error) {
wc := &iocounter.Writer{Writer: w}
for results.More() {
result := results.Next()
if _, err := e.Encoder.Encode(wc, result); err != nil {
// If we have an error that's from encoding or if we have not
// yet written any data to the writer, return the error.
if isEncoderError(err) || wc.Count() == 0 {
return wc.Count(), err
}
// Otherwise, the error happened during query execution and we
// are stuck encoding it.
err := e.Encoder.EncodeError(wc, err)
return wc.Count(), err
}
if _, err := wc.Write(e.Delimiter); err != nil {
return wc.Count(), err
}
// Flush the writer after each result.
if f, ok := w.(flusher); ok {
f.Flush()
}
}
results.Release()
// If we have any outlying errors in results, encode them
// If we have an error in the result and we have not written
// to the writer, then return the error as-is. Otherwise, encode
// it the same way we do above.
if err := results.Err(); err != nil {
if wc.Count() == 0 {
return 0, err
}
err := e.Encoder.EncodeError(wc, err)
return wc.Count(), err
}
return wc.Count(), nil
}