forked from dataux/dataux
/
esresults.go
375 lines (338 loc) · 10.8 KB
/
esresults.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
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
package elasticsearch
import (
"database/sql/driver"
"encoding/json"
"fmt"
"strings"
u "github.com/araddon/gou"
"github.com/araddon/qlbridge/datasource"
"github.com/araddon/qlbridge/exec"
"github.com/araddon/qlbridge/rel"
"github.com/araddon/qlbridge/value"
)
var (
// Ensure we implement Scanner
_ exec.Task = (*ResultReader)(nil)
//_ datasource.Scanner = (*ResultReader)(nil)
)
// Elasticsearch ResultProvider, adapts the elasticsearch http json
// to dataux/driver values
//
type ResultReader struct {
*exec.TaskBase
finalized bool
hasprojection bool
cursor int
proj *rel.Projection
Docs []u.JsonHelper
Vals [][]driver.Value
cols []string
Total int
Aggs u.JsonHelper
ScrollId string
Req *SqlToEs
}
// A wrapper, allowing us to implement sql/driver Next() interface
// which is different than qlbridge/datasource Next()
type ResultReaderNext struct {
*ResultReader
}
func NewResultReader(req *SqlToEs) *ResultReader {
m := &ResultReader{}
m.TaskBase = exec.NewTaskBase(req.ctx)
m.Req = req
return m
}
func (m *ResultReader) Close() error { return nil }
func (m *ResultReader) buildProjection() {
if m.hasprojection {
return
}
m.hasprojection = true
m.proj = rel.NewProjection()
cols := m.proj.Columns
sql := m.Req.sel
if sql.Star {
// Select Each field, grab fields from Table Schema
for _, fld := range m.Req.tbl.Fields {
//u.Infof("found %#v", fld)
cols = append(cols, rel.NewResultColumn(fld.Name, len(cols), nil, fld.Type))
}
} else if sql.CountStar() {
// Count *
cols = append(cols, rel.NewResultColumn("count", len(cols), nil, value.IntType))
} else if len(m.Aggs) > 0 {
if m.Req.hasSingleValue {
for _, col := range sql.Columns {
if col.CountStar() {
cols = append(cols, rel.NewResultColumn(col.Key(), len(cols), col, value.IntType))
} else {
u.Debugf("why Aggs? %#v", col)
cols = append(cols, rel.NewResultColumn(col.Key(), len(cols), col, value.IntType))
}
}
} else if m.Req.hasMultiValue {
// MultiValue returns are resultsets that have multiple rows for a single expression, ie top 10 terms for this field, etc
// if len(sql.GroupBy) > 0 {
// We store the Field Name Here
//u.Debugf("why MultiValue Aggs? %#v", m.Req)
cols = append(cols, rel.NewResultColumn("field_name", len(cols), nil, value.StringType))
cols = append(cols, rel.NewResultColumn("key", len(cols), nil, value.StringType)) // the value of the field
cols = append(cols, rel.NewResultColumn("count", len(cols), nil, value.IntType))
}
} else {
for _, col := range m.Req.sel.Columns {
if fld, ok := m.Req.tbl.FieldMap[col.SourceField]; ok {
//u.Debugf("column: %#v", col)
cols = append(cols, rel.NewResultColumn(col.SourceField, len(cols), col, fld.Type))
} else {
if fld, ok := m.Req.tbl.FieldMap[col.As]; ok {
//u.Debugf("column: %#v", col)
cols = append(cols, rel.NewResultColumn(col.SourceField, len(cols), col, fld.Type))
} else {
//u.Debugf("col %#v", col)
//u.Debugf("Could not find: %v sourcefield?%v fields:%v", col.As, col.SourceField, m.Req.tbl.Columns())
cols = append(cols, rel.NewResultColumn(col.SourceField, len(cols), col, value.StringType))
}
}
}
}
m.proj.Columns = cols
colNames := make([]string, len(cols))
for i, col := range cols {
colNames[i] = col.Name
}
m.cols = colNames
u.Debugf("leaving Columns: %v", len(m.proj.Columns))
}
func (m *ResultReader) Columns() []string {
m.buildProjection()
return m.cols
}
// Run()
//
// Normally, finalize is responsible for ensuring schema, setu
// but in the case of elasticsearch, since it is a non-streaming
// response, we build out values in advance
func (m *ResultReader) Run() error {
sigChan := m.SigChan()
outCh := m.MessageOut()
m.finalized = true
m.buildProjection()
//defer context.Recover()
defer func() {
close(outCh) // closing output channels is the signal to stop
//m.TaskBase.Close()
u.Debugf("nice, finalize ResultReader out: %p row ct %v", outCh, len(m.Vals))
}()
sql := m.Req.sel
m.Vals = make([][]driver.Value, 0)
colNames := make(map[string]int, len(m.proj.Columns))
for i, col := range m.proj.Columns {
colNames[col.As] = i
}
if sql.Star {
// ??
} else if sql.CountStar() {
// Count *
vals := make([]driver.Value, 1)
vals[0] = m.Total
m.Vals = append(m.Vals, vals)
} else if len(m.Aggs) > 0 {
//u.Debugf("aggs: multi?%v single?%v", m.Req.hasMultiValue, m.Req.hasSingleValue)
if m.Req.hasMultiValue && m.Req.hasSingleValue {
return fmt.Errorf("Must not mix single value and multi-value aggs")
}
if m.Req.hasSingleValue {
vals := make([]driver.Value, len(sql.Columns))
for i, col := range sql.Columns {
fldName := col.Key()
if col.Expr != nil {
u.Debugf("col: %v", col.Expr.String())
}
if col.CountStar() {
u.Debugf("found count star")
vals[i] = m.Total
} else {
u.Debugf("looking for col: %v=%v", fldName, m.Aggs.Get(fldName+"/value"))
vals[i] = m.Aggs.Get(fldName + "/value")
}
}
u.Debugf("write result: %v", vals)
m.Vals = append(m.Vals, vals)
} else if m.Req.hasMultiValue {
// MultiValue returns are resultsets that have multiple rows for a single expression, ie top 10 terms for this field, etc
if len(sql.GroupBy) > 0 {
//for i, col := range sql.Columns {
for i, _ := range sql.GroupBy {
fldName := fmt.Sprintf("group_by_%d", i)
u.Debugf("looking for col: %v %v", fldName, m.Aggs.Get(fldName+"/results"))
results := m.Aggs.Helpers(fldName + "/buckets")
for _, result := range results {
vals := make([]driver.Value, 3)
vals[0] = fldName
vals[1] = result.String("key")
vals[2] = result.Int("doc_count")
m.Vals = append(m.Vals, vals)
}
u.Debugf("missing value? %v", m.Aggs.Get(fldName))
// by, _ := json.MarshalIndent(m.Aggs.Get(fldName), " ", " ")
// vals[1] = by
}
} else {
// MultiValue are generally aggregates
for _, col := range sql.Columns {
fldName := col.As
u.Debugf("looking for col: %v %v", fldName, m.Aggs.Get(fldName+"/results"))
results := m.Aggs.Helpers(fldName + "/buckets")
for _, result := range results {
vals := make([]driver.Value, 3)
vals[0] = fldName
vals[1] = result.String("key")
vals[2] = result.Int("doc_count")
m.Vals = append(m.Vals, vals)
}
}
}
}
} else {
if err := m.pageDocs(); err != nil {
u.Errorf("error: %v", err)
}
}
for i, vals := range m.Vals {
//u.Debugf("new row ct: %v cols:%v vals:%v", len(m.Vals), colNames, vals)
//msg := &datasource.SqlDriverMessage{vals, len(m.Vals)}
msg := datasource.NewSqlDriverMessageMap(uint64(i), vals, colNames)
//u.Infof("In source Scanner iter %#v", msg)
select {
case <-sigChan:
return nil
case outCh <- msg:
// continue
}
}
return nil
}
func (m *ResultReader) pageDocs() error {
metaFields := map[string]byte{"_id": 1, "_type": 1, "_score": 1}
// If we have projected fields using "fields=field1,field2" in Elasticsearch
// then it will change the response format to
//
// gou.JsonHelper{"_index":"github_push", "_type":"event", "_id":"b307c13d856d95b4990f89b5df2fd667", "_score":1,
// "fields":map[string]interface {}{"repository.name":[]interface {}{"fluentd-ui"}, "actor":[]interface {}{"uu59"}}}
//
// INSTEAD of
//
// gou.JsonHelper{"_index":"github_push", "_type":"event", "_id":"b307c13d856d95b4990f89b5df2fd667", "_score":1,
// "_source":map[string]interface {}{"repository.name":[]interface {}{"fluentd-ui"}, "actor":[]interface {}{"uu59"}}}
keyPath := "_source."
useFields := false
cols := m.proj.Columns
if len(cols) == 0 {
u.Errorf("wat? no cols? %v", cols)
}
for _, doc := range m.Docs {
if len(doc) > 0 {
// by, _ := json.MarshalIndent(doc, " ", " ")
// u.Debugf("doc: %v", string(by))
if useFields {
doc = doc.Helper("fields")
if len(doc) < 1 {
u.Warnf("could not find fields? %#v", doc)
continue
}
}
vals := make([]driver.Value, len(m.proj.Columns))
fldI := 0
for _, col := range cols {
key := keyPath + col.Name
keyCandiates := []string{key}
if _, ok := metaFields[col.Name]; ok {
key = col.Name
keyCandiates[0] = key
} else if strings.Contains(col.Col.SourceOriginal, ".") {
keyCandiates = append(keyCandiates, keyPath+col.Col.SourceOriginal)
}
//u.Debugf("%#v", col.Col)
for _, key := range keyCandiates {
if docVal := doc.Get(key); docVal == nil {
//u.Debugf("skipping %q in %v", key, doc.Keys())
continue
}
//u.Debugf("looking for? %v %#v", key, col.Col)
//u.Debugf("looking for %q", key)
if useFields {
//u.Debugf("use fields: '%s' type=%v Strings()='%v' doc=%#v", col.Name, col.Type.String(), doc.Strings(key), doc)
switch col.Type {
case value.StringType:
if docVals := doc.Strings(key); len(docVals) > 0 {
vals[fldI] = docVals[0]
} else {
u.Warnf("no vals for %v? %#v", col.Name, docVals)
}
case value.TimeType:
if docVals := doc.Strings(key); len(docVals) > 0 {
vals[fldI] = docVals[0]
} else {
u.Warnf("no vals? %#v", docVals)
}
case value.IntType, value.NumberType:
if docVals := doc.List(key); len(docVals) > 0 {
vals[fldI] = docVals[0]
} else {
u.Warnf("no vals? %#v", docVals)
}
case value.ByteSliceType:
u.Debugf("blob? %v", key)
if docVal := doc.Get(key); docVal != nil {
by, _ := json.Marshal(docVal)
vals[fldI] = string(by)
}
default:
u.Warnf("unrecognized type: %v %T", key, col.Type)
}
} else {
//u.Debugf("col.type %v type %v", key, col.Type.String())
switch col.Type {
case value.StringType:
vals[fldI] = ""
strVal := doc.String(key)
//u.Debugf("strval: %s? or %s? =%q", key, col.As, strVal)
if strVal != "" {
vals[fldI] = strVal
} else {
jhVal := doc.Helper(key)
if len(jhVal) > 0 {
//u.Debugf("looking for? key:%v type:%s val:%s", key, col.Type.String(), jhVal)
jsonBytes, err := json.Marshal(jhVal)
if err == nil {
vals[fldI] = string(jsonBytes)
}
}
}
case value.TimeType:
vals[fldI] = doc.String(key)
case value.IntType:
vals[fldI] = doc.Int64(key)
case value.NumberType:
vals[fldI] = doc.Float64(key)
case value.ByteSliceType:
//u.Debugf("blob? %v", key)
if docVal := doc.Get(key); docVal != nil {
by, _ := json.Marshal(docVal)
vals[fldI] = string(by)
}
default:
u.Warnf("unrecognized type: %v %T", col.Name, col.Type)
}
}
}
fldI++
}
//u.Debugf("vals: %#v", vals)
m.Vals = append(m.Vals, vals)
}
}
return nil
}