blob: 7657f841d632b96dd4e206e7487235188d94160b [file] [log] [blame]
khenaidooab1f7bd2019-11-14 14:00:27 -05001// Copyright 2015 The Prometheus Authors
2// Licensed under the Apache License, Version 2.0 (the "License");
3// you may not use this file except in compliance with the License.
4// You may obtain a copy of the License at
5//
6// http://www.apache.org/licenses/LICENSE-2.0
7//
8// Unless required by applicable law or agreed to in writing, software
9// distributed under the License is distributed on an "AS IS" BASIS,
10// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
11// See the License for the specific language governing permissions and
12// limitations under the License.
13
14package expfmt
15
16import (
17 "fmt"
18 "io"
19 "math"
20 "mime"
21 "net/http"
22
23 dto "github.com/prometheus/client_model/go"
24
25 "github.com/matttproud/golang_protobuf_extensions/pbutil"
26 "github.com/prometheus/common/model"
27)
28
29// Decoder types decode an input stream into metric families.
30type Decoder interface {
31 Decode(*dto.MetricFamily) error
32}
33
34// DecodeOptions contains options used by the Decoder and in sample extraction.
35type DecodeOptions struct {
36 // Timestamp is added to each value from the stream that has no explicit timestamp set.
37 Timestamp model.Time
38}
39
40// ResponseFormat extracts the correct format from a HTTP response header.
41// If no matching format can be found FormatUnknown is returned.
42func ResponseFormat(h http.Header) Format {
43 ct := h.Get(hdrContentType)
44
45 mediatype, params, err := mime.ParseMediaType(ct)
46 if err != nil {
47 return FmtUnknown
48 }
49
50 const textType = "text/plain"
51
52 switch mediatype {
53 case ProtoType:
54 if p, ok := params["proto"]; ok && p != ProtoProtocol {
55 return FmtUnknown
56 }
57 if e, ok := params["encoding"]; ok && e != "delimited" {
58 return FmtUnknown
59 }
60 return FmtProtoDelim
61
62 case textType:
63 if v, ok := params["version"]; ok && v != TextVersion {
64 return FmtUnknown
65 }
66 return FmtText
67 }
68
69 return FmtUnknown
70}
71
72// NewDecoder returns a new decoder based on the given input format.
73// If the input format does not imply otherwise, a text format decoder is returned.
74func NewDecoder(r io.Reader, format Format) Decoder {
75 switch format {
76 case FmtProtoDelim:
77 return &protoDecoder{r: r}
78 }
79 return &textDecoder{r: r}
80}
81
82// protoDecoder implements the Decoder interface for protocol buffers.
83type protoDecoder struct {
84 r io.Reader
85}
86
87// Decode implements the Decoder interface.
88func (d *protoDecoder) Decode(v *dto.MetricFamily) error {
89 _, err := pbutil.ReadDelimited(d.r, v)
90 if err != nil {
91 return err
92 }
93 if !model.IsValidMetricName(model.LabelValue(v.GetName())) {
94 return fmt.Errorf("invalid metric name %q", v.GetName())
95 }
96 for _, m := range v.GetMetric() {
97 if m == nil {
98 continue
99 }
100 for _, l := range m.GetLabel() {
101 if l == nil {
102 continue
103 }
104 if !model.LabelValue(l.GetValue()).IsValid() {
105 return fmt.Errorf("invalid label value %q", l.GetValue())
106 }
107 if !model.LabelName(l.GetName()).IsValid() {
108 return fmt.Errorf("invalid label name %q", l.GetName())
109 }
110 }
111 }
112 return nil
113}
114
115// textDecoder implements the Decoder interface for the text protocol.
116type textDecoder struct {
117 r io.Reader
118 p TextParser
119 fams []*dto.MetricFamily
120}
121
122// Decode implements the Decoder interface.
123func (d *textDecoder) Decode(v *dto.MetricFamily) error {
124 // TODO(fabxc): Wrap this as a line reader to make streaming safer.
125 if len(d.fams) == 0 {
126 // No cached metric families, read everything and parse metrics.
127 fams, err := d.p.TextToMetricFamilies(d.r)
128 if err != nil {
129 return err
130 }
131 if len(fams) == 0 {
132 return io.EOF
133 }
134 d.fams = make([]*dto.MetricFamily, 0, len(fams))
135 for _, f := range fams {
136 d.fams = append(d.fams, f)
137 }
138 }
139
140 *v = *d.fams[0]
141 d.fams = d.fams[1:]
142
143 return nil
144}
145
146// SampleDecoder wraps a Decoder to extract samples from the metric families
147// decoded by the wrapped Decoder.
148type SampleDecoder struct {
149 Dec Decoder
150 Opts *DecodeOptions
151
152 f dto.MetricFamily
153}
154
155// Decode calls the Decode method of the wrapped Decoder and then extracts the
156// samples from the decoded MetricFamily into the provided model.Vector.
157func (sd *SampleDecoder) Decode(s *model.Vector) error {
158 err := sd.Dec.Decode(&sd.f)
159 if err != nil {
160 return err
161 }
162 *s, err = extractSamples(&sd.f, sd.Opts)
163 return err
164}
165
166// ExtractSamples builds a slice of samples from the provided metric
khenaidood948f772021-08-11 17:49:24 -0400167// families. If an error occurs during sample extraction, it continues to
khenaidooab1f7bd2019-11-14 14:00:27 -0500168// extract from the remaining metric families. The returned error is the last
169// error that has occurred.
170func ExtractSamples(o *DecodeOptions, fams ...*dto.MetricFamily) (model.Vector, error) {
171 var (
172 all model.Vector
173 lastErr error
174 )
175 for _, f := range fams {
176 some, err := extractSamples(f, o)
177 if err != nil {
178 lastErr = err
179 continue
180 }
181 all = append(all, some...)
182 }
183 return all, lastErr
184}
185
186func extractSamples(f *dto.MetricFamily, o *DecodeOptions) (model.Vector, error) {
187 switch f.GetType() {
188 case dto.MetricType_COUNTER:
189 return extractCounter(o, f), nil
190 case dto.MetricType_GAUGE:
191 return extractGauge(o, f), nil
192 case dto.MetricType_SUMMARY:
193 return extractSummary(o, f), nil
194 case dto.MetricType_UNTYPED:
195 return extractUntyped(o, f), nil
196 case dto.MetricType_HISTOGRAM:
197 return extractHistogram(o, f), nil
198 }
199 return nil, fmt.Errorf("expfmt.extractSamples: unknown metric family type %v", f.GetType())
200}
201
202func extractCounter(o *DecodeOptions, f *dto.MetricFamily) model.Vector {
203 samples := make(model.Vector, 0, len(f.Metric))
204
205 for _, m := range f.Metric {
206 if m.Counter == nil {
207 continue
208 }
209
210 lset := make(model.LabelSet, len(m.Label)+1)
211 for _, p := range m.Label {
212 lset[model.LabelName(p.GetName())] = model.LabelValue(p.GetValue())
213 }
214 lset[model.MetricNameLabel] = model.LabelValue(f.GetName())
215
216 smpl := &model.Sample{
217 Metric: model.Metric(lset),
218 Value: model.SampleValue(m.Counter.GetValue()),
219 }
220
221 if m.TimestampMs != nil {
222 smpl.Timestamp = model.TimeFromUnixNano(*m.TimestampMs * 1000000)
223 } else {
224 smpl.Timestamp = o.Timestamp
225 }
226
227 samples = append(samples, smpl)
228 }
229
230 return samples
231}
232
233func extractGauge(o *DecodeOptions, f *dto.MetricFamily) model.Vector {
234 samples := make(model.Vector, 0, len(f.Metric))
235
236 for _, m := range f.Metric {
237 if m.Gauge == nil {
238 continue
239 }
240
241 lset := make(model.LabelSet, len(m.Label)+1)
242 for _, p := range m.Label {
243 lset[model.LabelName(p.GetName())] = model.LabelValue(p.GetValue())
244 }
245 lset[model.MetricNameLabel] = model.LabelValue(f.GetName())
246
247 smpl := &model.Sample{
248 Metric: model.Metric(lset),
249 Value: model.SampleValue(m.Gauge.GetValue()),
250 }
251
252 if m.TimestampMs != nil {
253 smpl.Timestamp = model.TimeFromUnixNano(*m.TimestampMs * 1000000)
254 } else {
255 smpl.Timestamp = o.Timestamp
256 }
257
258 samples = append(samples, smpl)
259 }
260
261 return samples
262}
263
264func extractUntyped(o *DecodeOptions, f *dto.MetricFamily) model.Vector {
265 samples := make(model.Vector, 0, len(f.Metric))
266
267 for _, m := range f.Metric {
268 if m.Untyped == nil {
269 continue
270 }
271
272 lset := make(model.LabelSet, len(m.Label)+1)
273 for _, p := range m.Label {
274 lset[model.LabelName(p.GetName())] = model.LabelValue(p.GetValue())
275 }
276 lset[model.MetricNameLabel] = model.LabelValue(f.GetName())
277
278 smpl := &model.Sample{
279 Metric: model.Metric(lset),
280 Value: model.SampleValue(m.Untyped.GetValue()),
281 }
282
283 if m.TimestampMs != nil {
284 smpl.Timestamp = model.TimeFromUnixNano(*m.TimestampMs * 1000000)
285 } else {
286 smpl.Timestamp = o.Timestamp
287 }
288
289 samples = append(samples, smpl)
290 }
291
292 return samples
293}
294
295func extractSummary(o *DecodeOptions, f *dto.MetricFamily) model.Vector {
296 samples := make(model.Vector, 0, len(f.Metric))
297
298 for _, m := range f.Metric {
299 if m.Summary == nil {
300 continue
301 }
302
303 timestamp := o.Timestamp
304 if m.TimestampMs != nil {
305 timestamp = model.TimeFromUnixNano(*m.TimestampMs * 1000000)
306 }
307
308 for _, q := range m.Summary.Quantile {
309 lset := make(model.LabelSet, len(m.Label)+2)
310 for _, p := range m.Label {
311 lset[model.LabelName(p.GetName())] = model.LabelValue(p.GetValue())
312 }
313 // BUG(matt): Update other names to "quantile".
314 lset[model.LabelName(model.QuantileLabel)] = model.LabelValue(fmt.Sprint(q.GetQuantile()))
315 lset[model.MetricNameLabel] = model.LabelValue(f.GetName())
316
317 samples = append(samples, &model.Sample{
318 Metric: model.Metric(lset),
319 Value: model.SampleValue(q.GetValue()),
320 Timestamp: timestamp,
321 })
322 }
323
324 lset := make(model.LabelSet, len(m.Label)+1)
325 for _, p := range m.Label {
326 lset[model.LabelName(p.GetName())] = model.LabelValue(p.GetValue())
327 }
328 lset[model.MetricNameLabel] = model.LabelValue(f.GetName() + "_sum")
329
330 samples = append(samples, &model.Sample{
331 Metric: model.Metric(lset),
332 Value: model.SampleValue(m.Summary.GetSampleSum()),
333 Timestamp: timestamp,
334 })
335
336 lset = make(model.LabelSet, len(m.Label)+1)
337 for _, p := range m.Label {
338 lset[model.LabelName(p.GetName())] = model.LabelValue(p.GetValue())
339 }
340 lset[model.MetricNameLabel] = model.LabelValue(f.GetName() + "_count")
341
342 samples = append(samples, &model.Sample{
343 Metric: model.Metric(lset),
344 Value: model.SampleValue(m.Summary.GetSampleCount()),
345 Timestamp: timestamp,
346 })
347 }
348
349 return samples
350}
351
352func extractHistogram(o *DecodeOptions, f *dto.MetricFamily) model.Vector {
353 samples := make(model.Vector, 0, len(f.Metric))
354
355 for _, m := range f.Metric {
356 if m.Histogram == nil {
357 continue
358 }
359
360 timestamp := o.Timestamp
361 if m.TimestampMs != nil {
362 timestamp = model.TimeFromUnixNano(*m.TimestampMs * 1000000)
363 }
364
365 infSeen := false
366
367 for _, q := range m.Histogram.Bucket {
368 lset := make(model.LabelSet, len(m.Label)+2)
369 for _, p := range m.Label {
370 lset[model.LabelName(p.GetName())] = model.LabelValue(p.GetValue())
371 }
372 lset[model.LabelName(model.BucketLabel)] = model.LabelValue(fmt.Sprint(q.GetUpperBound()))
373 lset[model.MetricNameLabel] = model.LabelValue(f.GetName() + "_bucket")
374
375 if math.IsInf(q.GetUpperBound(), +1) {
376 infSeen = true
377 }
378
379 samples = append(samples, &model.Sample{
380 Metric: model.Metric(lset),
381 Value: model.SampleValue(q.GetCumulativeCount()),
382 Timestamp: timestamp,
383 })
384 }
385
386 lset := make(model.LabelSet, len(m.Label)+1)
387 for _, p := range m.Label {
388 lset[model.LabelName(p.GetName())] = model.LabelValue(p.GetValue())
389 }
390 lset[model.MetricNameLabel] = model.LabelValue(f.GetName() + "_sum")
391
392 samples = append(samples, &model.Sample{
393 Metric: model.Metric(lset),
394 Value: model.SampleValue(m.Histogram.GetSampleSum()),
395 Timestamp: timestamp,
396 })
397
398 lset = make(model.LabelSet, len(m.Label)+1)
399 for _, p := range m.Label {
400 lset[model.LabelName(p.GetName())] = model.LabelValue(p.GetValue())
401 }
402 lset[model.MetricNameLabel] = model.LabelValue(f.GetName() + "_count")
403
404 count := &model.Sample{
405 Metric: model.Metric(lset),
406 Value: model.SampleValue(m.Histogram.GetSampleCount()),
407 Timestamp: timestamp,
408 }
409 samples = append(samples, count)
410
411 if !infSeen {
412 // Append an infinity bucket sample.
413 lset := make(model.LabelSet, len(m.Label)+2)
414 for _, p := range m.Label {
415 lset[model.LabelName(p.GetName())] = model.LabelValue(p.GetValue())
416 }
417 lset[model.LabelName(model.BucketLabel)] = model.LabelValue("+Inf")
418 lset[model.MetricNameLabel] = model.LabelValue(f.GetName() + "_bucket")
419
420 samples = append(samples, &model.Sample{
421 Metric: model.Metric(lset),
422 Value: count.Value,
423 Timestamp: timestamp,
424 })
425 }
426 }
427
428 return samples
429}