forked from russellhaering/goxmldsig
-
Notifications
You must be signed in to change notification settings - Fork 0
/
canonicalize.go
254 lines (209 loc) · 6.85 KB
/
canonicalize.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
package dsig
import (
"sort"
"strings"
"github.com/beevik/etree"
)
// Canonicalizer is an implementation of a canonicalization algorithm.
type Canonicalizer interface {
Canonicalize(el *etree.Element) ([]byte, error)
Algorithm() AlgorithmID
}
type c14N10ExclusiveCanonicalizer struct {
InclusiveNamespaces map[string]struct{}
}
// MakeC14N10ExclusiveCanonicalizerWithPrefixList constructs an exclusive Canonicalizer
// from a PrefixList in NMTOKENS format (a white space separated list).
func MakeC14N10ExclusiveCanonicalizerWithPrefixList(prefixList string) Canonicalizer {
prefixes := strings.Fields(prefixList)
prefixSet := make(map[string]struct{}, len(prefixes))
for _, prefix := range prefixes {
prefixSet[prefix] = struct{}{}
}
return &c14N10ExclusiveCanonicalizer{
InclusiveNamespaces: prefixSet,
}
}
// Canonicalize transforms the input Element into a serialized XML document in canonical form.
func (c *c14N10ExclusiveCanonicalizer) Canonicalize(el *etree.Element) ([]byte, error) {
scope := make(map[string]c14nSpace)
return canonicalSerialize(excCanonicalPrep(el, scope, c.InclusiveNamespaces))
}
func (c *c14N10ExclusiveCanonicalizer) Algorithm() AlgorithmID {
return CanonicalXML10ExclusiveAlgorithmId
}
type c14N11Canonicalizer struct{}
// MakeC14N11Canonicalizer constructs an inclusive canonicalizer.
func MakeC14N11Canonicalizer() Canonicalizer {
return &c14N11Canonicalizer{}
}
// Canonicalize transforms the input Element into a serialized XML document in canonical form.
func (c *c14N11Canonicalizer) Canonicalize(el *etree.Element) ([]byte, error) {
scope := make(map[string]struct{})
return canonicalSerialize(canonicalPrep(el, scope))
}
func (c *c14N11Canonicalizer) Algorithm() AlgorithmID {
return CanonicalXML11AlgorithmId
}
func composeAttr(space, key string) string {
if space != "" {
return space + ":" + key
}
return key
}
type attrsByKey []etree.Attr
func (a attrsByKey) Len() int {
return len(a)
}
func (a attrsByKey) Swap(i, j int) {
a[i], a[j] = a[j], a[i]
}
func (a attrsByKey) Less(i, j int) bool {
// As I understand it: any "xmlns" attribute should come first, followed by any
// any "xmlns:prefix" attributes, presumably ordered by prefix. Lastly any other
// attributes in lexicographical order.
if a[i].Space == "" && a[i].Key == "xmlns" {
return true
}
if a[j].Space == "" && a[j].Key == "xmlns" {
return false
}
if a[i].Space == "xmlns" {
if a[j].Space == "xmlns" {
return a[i].Key < a[j].Key
}
return true
}
if a[j].Space == "xmlns" {
return false
}
return composeAttr(a[i].Space, a[i].Key) < composeAttr(a[j].Space, a[j].Key)
}
type c14nSpace struct {
a etree.Attr
used bool
}
const nsSpace = "xmlns"
// excCanonicalPrep accepts an *etree.Element and recursively transforms it into one
// which is ready for serialization to exclusive canonical form. Specifically this
// entails:
//
// 1. Stripping re-declarations of namespaces
// 2. Stripping unused namespaces
// 3. Sorting attributes into canonical order.
//
// NOTE(russell_h): Currently this function modifies the passed element.
func excCanonicalPrep(el *etree.Element, _nsAlreadyDeclared map[string]c14nSpace, inclusiveNamespaces map[string]struct{}) *etree.Element {
//Copy alreadyDeclared map (only contains namespaces)
nsAlreadyDeclared := make(map[string]c14nSpace, len(_nsAlreadyDeclared))
for k := range _nsAlreadyDeclared {
nsAlreadyDeclared[k] = _nsAlreadyDeclared[k]
}
//Track the namespaces used on the current element
nsUsedHere := make(map[string]struct{})
//Make sure to track the element namespace for the case:
//<foo:bar xmlns:foo="..."/>
if el.Space != "" {
nsUsedHere[el.Space] = struct{}{}
}
toRemove := make([]string, 0, 0)
for _, a := range el.Attr {
switch a.Space {
case nsSpace:
//For simplicity, remove all xmlns attribues; to be added in one pass
//later. Otherwise, we need another map/set to track xmlns attributes
//that we left alone.
toRemove = append(toRemove, a.Space+":"+a.Key)
if _, ok := nsAlreadyDeclared[a.Key]; !ok {
//If we're not tracking ancestor state already for this namespace, add
//it to the map
nsAlreadyDeclared[a.Key] = c14nSpace{a: a, used: false}
}
// This algorithm accepts a set of namespaces which should be treated
// in an inclusive fashion. Specifically that means we should keep the
// declaration of that namespace closest to the root of the tree. We can
// accomplish that be pretending it was used by this element.
_, inclusive := inclusiveNamespaces[a.Key]
if inclusive {
nsUsedHere[a.Key] = struct{}{}
}
default:
//We only track namespaces, so ignore attributes without one.
if a.Space != "" {
nsUsedHere[a.Space] = struct{}{}
}
}
}
//Remove all attributes so that we can add them with much-simpler logic
for _, attrK := range toRemove {
el.RemoveAttr(attrK)
}
//For all namespaces used on the current element, declare them if they were
//not declared (and used) in an ancestor.
for k := range nsUsedHere {
spc := nsAlreadyDeclared[k]
//If previously unused, mark as used
if !spc.used {
el.Attr = append(el.Attr, spc.a)
spc.used = true
//Assignment here is only to update the pre-existing `used` tracking value
nsAlreadyDeclared[k] = spc
}
}
//Canonicalize all children, passing down the ancestor tracking map
for _, child := range el.ChildElements() {
excCanonicalPrep(child, nsAlreadyDeclared, inclusiveNamespaces)
}
//Sort attributes lexicographically
sort.Sort(attrsByKey(el.Attr))
return el.Copy()
}
// canonicalPrep accepts an *etree.Element and transforms it into one which is ready
// for serialization into inclusive canonical form. Specifically this
// entails:
//
// 1. Stripping re-declarations of namespaces
// 2. Sorting attributes into canonical order
//
// Inclusive canonicalization does not strip unused namespaces.
//
// TODO(russell_h): This is very similar to excCanonicalPrep - perhaps they should
// be unified into one parameterized function?
func canonicalPrep(el *etree.Element, seenSoFar map[string]struct{}) *etree.Element {
_seenSoFar := make(map[string]struct{})
for k, v := range seenSoFar {
_seenSoFar[k] = v
}
ne := el.Copy()
sort.Sort(attrsByKey(ne.Attr))
if len(ne.Attr) != 0 {
for _, attr := range ne.Attr {
if attr.Space != nsSpace {
continue
}
key := attr.Space + ":" + attr.Key
if _, seen := _seenSoFar[key]; seen {
ne.RemoveAttr(attr.Space + ":" + attr.Key)
} else {
_seenSoFar[key] = struct{}{}
}
}
}
for i, token := range ne.Child {
childElement, ok := token.(*etree.Element)
if ok {
ne.Child[i] = canonicalPrep(childElement, _seenSoFar)
}
}
return ne
}
func canonicalSerialize(el *etree.Element) ([]byte, error) {
doc := etree.NewDocument()
doc.SetRoot(el)
doc.WriteSettings = etree.WriteSettings{
CanonicalAttrVal: true,
CanonicalEndTags: true,
CanonicalText: true,
}
return doc.WriteToBytes()
}