-
Notifications
You must be signed in to change notification settings - Fork 1.2k
/
normalize.go
356 lines (325 loc) · 9.4 KB
/
normalize.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
// Unless explicitly stated otherwise all files in this repository are licensed
// under the Apache License Version 2.0.
// This product includes software developed at Datadog (https://www.datadoghq.com/).
// Copyright 2016-present Datadog, Inc.
package traceutil
import (
"errors"
"strings"
"sync"
"unicode"
"unicode/utf8"
)
const (
// DefaultSpanName is the default name we assign a span if it's missing and we have no reasonable fallback
DefaultSpanName = "unnamed_operation"
// DefaultServiceName is the default name we assign a service if it's missing and we have no reasonable fallback
DefaultServiceName = "unnamed-service"
)
const (
// MaxNameLen the maximum length a name can have
MaxNameLen = 100
// MaxServiceLen the maximum length a service can have
MaxServiceLen = 100
// MaxResourceLen the maximum length a resource can have
MaxResourceLen = 5000
)
var (
// ErrEmpty specifies that the passed input was empty.
ErrEmpty = errors.New("empty")
// ErrTooLong signifies that the input was too long.
ErrTooLong = errors.New("too long")
// ErrInvalid signifies that the input was invalid.
ErrInvalid = errors.New("invalid")
)
// NormalizeName normalizes a span name and returns an error describing the reason
// (if any) why the name was modified.
func NormalizeName(name string) (string, error) {
if name == "" {
return DefaultSpanName, ErrEmpty
}
var err error
if len(name) > MaxNameLen {
name = TruncateUTF8(name, MaxNameLen)
err = ErrTooLong
}
name, ok := normMetricNameParse(name)
if !ok {
return DefaultSpanName, ErrInvalid
}
return name, err
}
// NormalizeService normalizes a span service and returns an error describing the reason
// (if any) why the name was modified.
func NormalizeService(svc string, lang string) (string, error) {
if svc == "" {
return fallbackService(lang), ErrEmpty
}
var err error
if len(svc) > MaxServiceLen {
svc = TruncateUTF8(svc, MaxServiceLen)
err = ErrTooLong
}
// We are normalizing just the tag value.
s := NormalizeTagValue(svc)
if s == "" {
return fallbackService(lang), ErrInvalid
}
return s, err
}
// NormalizePeerService normalizes a span's peer.service and returns an error describing the reason
// (if any) why the name was modified.
func NormalizePeerService(svc string) (string, error) {
if svc == "" {
return "", nil
}
var err error
if len(svc) > MaxServiceLen {
svc = TruncateUTF8(svc, MaxServiceLen)
err = ErrTooLong
}
// We are normalizing just the tag value.
s := NormalizeTagValue(svc)
if s == "" {
return "", ErrInvalid
}
return s, err
}
// fallbackServiceNames is a cache of default service names to use
// when the span's service is unset or invalid.
var fallbackServiceNames sync.Map
// fallbackService returns the fallback service name for a service
// belonging to language lang.
func fallbackService(lang string) string {
if lang == "" {
return DefaultServiceName
}
if v, ok := fallbackServiceNames.Load(lang); ok {
return v.(string)
}
var str strings.Builder
str.WriteString("unnamed-")
str.WriteString(lang)
str.WriteString("-service")
fallbackServiceNames.Store(lang, str.String())
return str.String()
}
const maxTagLength = 200
// NormalizeTag applies some normalization to ensure the full tag_key:tag_value string matches the backend requirements.
func NormalizeTag(v string) string {
return normalize(v, true)
}
// NormalizeTagValue applies some normalization to ensure the tag value matches the backend requirements.
// It should be used for cases where we have just the tag_value as the input (instead of tag_key:tag_value).
func NormalizeTagValue(v string) string {
return normalize(v, false)
}
func normalize(v string, allowDigitStartChar bool) string {
// Fast path: Check if the tag is valid and only contains ASCII characters,
// if yes return it as-is right away. For most use-cases this reduces CPU usage.
if isNormalizedASCIITag(v, allowDigitStartChar) {
return v
}
// the algorithm works by creating a set of cuts marking start and end offsets in v
// that have to be replaced with underscore (_)
if len(v) == 0 {
return ""
}
var (
trim int // start character (if trimming)
cuts [][2]int // sections to discard: (start, end) pairs
chars int // number of characters processed
)
var (
i int // current byte
r rune // current rune
jump int // tracks how many bytes the for range advances on its next iteration
)
tag := []byte(v)
for i, r = range v {
jump = utf8.RuneLen(r) // next i will be i+jump
if r == utf8.RuneError {
// On invalid UTF-8, the for range advances only 1 byte (see: https://golang.org/ref/spec#For_range (point 2)).
// However, utf8.RuneError is equivalent to unicode.ReplacementChar so we should rely on utf8.DecodeRune to tell
// us whether this is an actual error or just a unicode.ReplacementChar that was present in the string.
_, width := utf8.DecodeRune(tag[i:])
jump = width
}
// fast path; all letters (and colons) are ok
switch {
case r >= 'a' && r <= 'z' || r == ':':
chars++
goto end
case r >= 'A' && r <= 'Z':
// lower-case
tag[i] += 'a' - 'A'
chars++
goto end
}
if unicode.IsUpper(r) {
// lowercase this character
if low := unicode.ToLower(r); utf8.RuneLen(r) == utf8.RuneLen(low) {
// but only if the width of the lowercased character is the same;
// there are some rare edge-cases where this is not the case, such
// as \u017F (ſ)
utf8.EncodeRune(tag[i:], low)
r = low
}
}
switch {
case unicode.IsLetter(r):
chars++
// If it's not a unicode letter, and it's the first char, and digits are allowed for the start char,
// we should goto end because the remaining cases are not valid for a start char.
case allowDigitStartChar && chars == 0:
trim = i + jump
goto end
case unicode.IsDigit(r) || r == '.' || r == '/' || r == '-':
chars++
default:
// illegal character
chars++
if n := len(cuts); n > 0 && cuts[n-1][1] >= i {
// merge intersecting cuts
cuts[n-1][1] += jump
} else {
// start a new cut
cuts = append(cuts, [2]int{i, i + jump})
}
}
end:
if i+jump >= 2*maxTagLength {
// bail early if the tag contains a lot of non-letter/digit characters.
// If a tag is test🍣🍣[...]🍣, then it's unlikely to be a properly formatted tag
break
}
if chars >= maxTagLength {
// we've reached the maximum
break
}
}
tag = tag[trim : i+jump] // trim start and end
if len(cuts) == 0 {
// tag was ok, return it as it is
return string(tag)
}
delta := trim // cut offsets delta
for _, cut := range cuts {
// start and end of cut, including delta from previous cuts:
start, end := cut[0]-delta, cut[1]-delta
if end >= len(tag) {
// this cut includes the end of the string; discard it
// completely and finish the loop.
tag = tag[:start]
break
}
// replace the beginning of the cut with '_'
tag[start] = '_'
if end-start == 1 {
// nothing to discard
continue
}
// discard remaining characters in the cut
copy(tag[start+1:], tag[end:])
// shorten the slice
tag = tag[:len(tag)-(end-start)+1]
// count the new delta for future cuts
delta += cut[1] - cut[0] - 1
}
return string(tag)
}
// This code is borrowed from dd-go metric normalization
// fast isAlpha for ascii
func isAlpha(b byte) bool {
return (b >= 'a' && b <= 'z') || (b >= 'A' && b <= 'Z')
}
// fast isAlphaNumeric for ascii
func isAlphaNum(b byte) bool {
return isAlpha(b) || (b >= '0' && b <= '9')
}
// normMetricNameParse normalizes metric names with a parser instead of using
// garbage-creating string replacement routines.
func normMetricNameParse(name string) (string, bool) {
if name == "" || len(name) > MaxNameLen {
return name, false
}
var i, ptr int
var resa [MaxNameLen]byte
res := resa[:0]
// skip non-alphabetic characters
for ; i < len(name) && !isAlpha(name[i]); i++ {
continue
}
// if there were no alphabetic characters it wasn't valid
if i == len(name) {
return "", false
}
for ; i < len(name); i++ {
switch {
case isAlphaNum(name[i]):
res = append(res, name[i])
ptr++
case name[i] == '.':
// we skipped all non-alpha chars up front so we have seen at least one
switch res[ptr-1] {
// overwrite underscores that happen before periods
case '_':
res[ptr-1] = '.'
default:
res = append(res, '.')
ptr++
}
default:
// we skipped all non-alpha chars up front so we have seen at least one
switch res[ptr-1] {
// no double underscores, no underscores after periods
case '.', '_':
default:
res = append(res, '_')
ptr++
}
}
}
if res[ptr-1] == '_' {
res = res[:ptr-1]
}
return string(res), true
}
func isNormalizedASCIITag(tag string, checkValidStartChar bool) bool {
if len(tag) == 0 {
return true
}
if len(tag) > maxTagLength {
return false
}
i := 0
if checkValidStartChar {
if !isValidASCIIStartChar(tag[0]) {
return false
}
i++
}
for ; i < len(tag); i++ {
b := tag[i]
// TODO: Attempt to optimize this check using SIMD/vectorization.
if isValidASCIITagChar(b) {
continue
}
if b == '_' {
// an underscore is only okay if followed by a valid non-underscore character
i++
if i == len(tag) || !isValidASCIITagChar(tag[i]) {
return false
}
} else {
return false
}
}
return true
}
func isValidASCIIStartChar(c byte) bool {
return ('a' <= c && c <= 'z') || c == ':'
}
func isValidASCIITagChar(c byte) bool {
return isValidASCIIStartChar(c) || ('0' <= c && c <= '9') || c == '.' || c == '/' || c == '-'
}