|
|
1
|
+// Licensed under the Apache License, Version 2.0,
|
|
|
2
|
+// <http://apache.org/licenses/LICENSE-2.0> or the MIT license
|
|
|
3
|
+// <http://opensource.org/licenses/MIT>, at your option. This file may not be
|
|
|
4
|
+// copied, modified, or distributed except according to those terms.
|
|
|
5
|
+
|
|
|
6
|
+use std::borrow::Cow;
|
|
|
7
|
+use thiserror::Error;
|
|
|
8
|
+
|
|
|
9
|
+#[derive(Error, Debug, PartialEq, Eq)]
|
|
|
10
|
+pub enum UnescapeError {
|
|
|
11
|
+ #[error("space in an unquoted string")]
|
|
|
12
|
+ SpaceInUnquotedString,
|
|
|
13
|
+ #[error("unterminated escape sequence or string")]
|
|
|
14
|
+ Unterminated,
|
|
|
15
|
+ #[error("found an unescaped quote")]
|
|
|
16
|
+ UnescapedQuote,
|
|
|
17
|
+ #[error("found an invalid escape sequence")]
|
|
|
18
|
+ InvalidEscape,
|
|
|
19
|
+ #[error("an octal-escaped value exceeds 255")]
|
|
|
20
|
+ OctalOverflow,
|
|
|
21
|
+}
|
|
|
22
|
+
|
|
|
23
|
+/// Adaptation of tor's unescape_string (src/lib/encoding/cstring.c).
|
|
|
24
|
+/// The main difference is that we allow unquoted strings.
|
|
|
25
|
+pub fn tor_unescape<'a>(buf: &'a [u8]) -> Result<Cow<'a, [u8]>, UnescapeError> {
|
|
|
26
|
+ match buf.first() {
|
|
|
27
|
+ Some(b'"') => {
|
|
|
28
|
+ // Quoted string, consume the item and continue with the function.
|
|
|
29
|
+ }
|
|
|
30
|
+ _ => {
|
|
|
31
|
+ if buf.contains(&b' ') {
|
|
|
32
|
+ return Err(UnescapeError::SpaceInUnquotedString);
|
|
|
33
|
+ }
|
|
|
34
|
+ return Ok(Cow::Borrowed(buf));
|
|
|
35
|
+ }
|
|
|
36
|
+ }
|
|
|
37
|
+
|
|
|
38
|
+ if buf.len() == 1 || buf.last() != Some(&b'\"') {
|
|
|
39
|
+ return Err(UnescapeError::Unterminated);
|
|
|
40
|
+ }
|
|
|
41
|
+
|
|
|
42
|
+ // This is used only if we did an actual change.
|
|
|
43
|
+ let mut out = None;
|
|
|
44
|
+
|
|
|
45
|
+ // We already consumed the initial quote.
|
|
|
46
|
+ let mut i = 1;
|
|
|
47
|
+ while i < buf.len() {
|
|
|
48
|
+ let c = buf[i];
|
|
|
49
|
+ i += 1;
|
|
|
50
|
+ match c {
|
|
|
51
|
+ b'\"' => {
|
|
|
52
|
+ if i != buf.len() {
|
|
|
53
|
+ return Err(UnescapeError::UnescapedQuote);
|
|
|
54
|
+ }
|
|
|
55
|
+ return Ok(out
|
|
|
56
|
+ .map(|b| Cow::Owned(b))
|
|
|
57
|
+ .unwrap_or_else(|| Cow::Borrowed(&buf[1..buf.len() - 1])));
|
|
|
58
|
+ }
|
|
|
59
|
+ 0 | b'\n' => return Err(UnescapeError::Unterminated),
|
|
|
60
|
+ b'\\' => {
|
|
|
61
|
+ let esc = buf.get(i).ok_or(UnescapeError::Unterminated)?;
|
|
|
62
|
+ i += 1;
|
|
|
63
|
+ let out = out.get_or_insert_with(|| {
|
|
|
64
|
+ let mut b = Vec::with_capacity(buf.len() - 2);
|
|
|
65
|
+ b.extend(&buf[1..i - 2]);
|
|
|
66
|
+ b
|
|
|
67
|
+ });
|
|
|
68
|
+ out.push(match esc {
|
|
|
69
|
+ b'n' => Ok(b'\n'),
|
|
|
70
|
+ b'r' => Ok(b'\r'),
|
|
|
71
|
+ b't' => Ok(b'\t'),
|
|
|
72
|
+ b'x' | b'X' => unescape_hex(buf, &mut i),
|
|
|
73
|
+ b'0'..=b'7' => unescape_octal(buf, &mut i),
|
|
|
74
|
+ b'\'' | b'"' | b'\\' => Ok(*esc),
|
|
|
75
|
+ _ => Err(UnescapeError::InvalidEscape),
|
|
|
76
|
+ }?);
|
|
|
77
|
+ }
|
|
|
78
|
+ _ => {
|
|
|
79
|
+ if let Some(b) = out.as_mut() {
|
|
|
80
|
+ b.push(c);
|
|
|
81
|
+ }
|
|
|
82
|
+ }
|
|
|
83
|
+ }
|
|
|
84
|
+ }
|
|
|
85
|
+
|
|
|
86
|
+ // If we fall out of the loop we never saw a closing quote.
|
|
|
87
|
+ Err(UnescapeError::Unterminated)
|
|
|
88
|
+}
|
|
|
89
|
+
|
|
|
90
|
+fn unescape_hex(buf: &[u8], i: &mut usize) -> Result<u8, UnescapeError> {
|
|
|
91
|
+ // The C code expects exactly two hex digits.
|
|
|
92
|
+ *i += 2;
|
|
|
93
|
+ str::from_utf8(&buf[*i - 2..*i])
|
|
|
94
|
+ .ok()
|
|
|
95
|
+ .and_then(|s| u8::from_str_radix(s, 16).ok())
|
|
|
96
|
+ .ok_or(UnescapeError::InvalidEscape)
|
|
|
97
|
+}
|
|
|
98
|
+
|
|
|
99
|
+fn unescape_octal(buf: &[u8], i: &mut usize) -> Result<u8, UnescapeError> {
|
|
|
100
|
+ *i -= 1;
|
|
|
101
|
+ let mut len = 1;
|
|
|
102
|
+ while (*i + len) < buf.len() && len < 3 {
|
|
|
103
|
+ let b = buf[*i + len];
|
|
|
104
|
+ if b >= b'0' && b <= b'7' {
|
|
|
105
|
+ len += 1;
|
|
|
106
|
+ } else {
|
|
|
107
|
+ break;
|
|
|
108
|
+ }
|
|
|
109
|
+ }
|
|
|
110
|
+
|
|
|
111
|
+ *i += len;
|
|
|
112
|
+ // This should never fail with InvalidEscape, as we have just validated the
|
|
|
113
|
+ // octal values.
|
|
|
114
|
+ str::from_utf8(&buf[*i - len..*i])
|
|
|
115
|
+ .ok()
|
|
|
116
|
+ .and_then(|s| u16::from_str_radix(s, 8).ok())
|
|
|
117
|
+ .ok_or(UnescapeError::InvalidEscape)
|
|
|
118
|
+ .and_then(|v| v.try_into().map_err(|_| UnescapeError::OctalOverflow))
|
|
|
119
|
+}
|
|
|
120
|
+
|
|
|
121
|
+#[cfg(test)]
|
|
|
122
|
+mod tests {
|
|
|
123
|
+ use super::*;
|
|
|
124
|
+
|
|
|
125
|
+ #[test]
|
|
|
126
|
+ fn simple() {
|
|
|
127
|
+ assert_eq!(&*tor_unescape(b"test").unwrap(), b"test");
|
|
|
128
|
+ assert_eq!(&*tor_unescape(b"\"test\"").unwrap(), b"test");
|
|
|
129
|
+ }
|
|
|
130
|
+
|
|
|
131
|
+ #[test]
|
|
|
132
|
+ fn quote_in_unquoted() {
|
|
|
133
|
+ assert_eq!(&*tor_unescape(b"te\"st").unwrap(), b"te\"st");
|
|
|
134
|
+ assert_eq!(&*tor_unescape(b"test\"").unwrap(), b"test\"");
|
|
|
135
|
+ }
|
|
|
136
|
+
|
|
|
137
|
+ #[test]
|
|
|
138
|
+ fn empty() {
|
|
|
139
|
+ assert_eq!(&*tor_unescape(b"").unwrap(), &[]);
|
|
|
140
|
+ assert_eq!(&*tor_unescape(b"\"\"").unwrap(), &[]);
|
|
|
141
|
+ }
|
|
|
142
|
+
|
|
|
143
|
+ #[test]
|
|
|
144
|
+ fn unescape_simple() {
|
|
|
145
|
+ assert_eq!(&*tor_unescape(b"\"\\n\"").unwrap(), b"\n");
|
|
|
146
|
+ assert_eq!(&*tor_unescape(b"\"\\r\"").unwrap(), b"\r");
|
|
|
147
|
+ assert_eq!(&*tor_unescape(b"\"\\t\"").unwrap(), b"\t");
|
|
|
148
|
+ assert_eq!(&*tor_unescape(b"\"\\r\\n\"").unwrap(), b"\r\n");
|
|
|
149
|
+ assert_eq!(&*tor_unescape(b"\"'\"").unwrap(), b"'");
|
|
|
150
|
+ assert_eq!(&*tor_unescape(b"\"\\'\"").unwrap(), b"'");
|
|
|
151
|
+ assert_eq!(&*tor_unescape(b"\"\\\"\"").unwrap(), b"\"");
|
|
|
152
|
+ assert_eq!(&*tor_unescape(b"\"\\\\\"").unwrap(), b"\\");
|
|
|
153
|
+ }
|
|
|
154
|
+
|
|
|
155
|
+ #[test]
|
|
|
156
|
+ fn unescape_hex() {
|
|
|
157
|
+ assert_eq!(&*tor_unescape(b"\"\\x20\"").unwrap(), b" ");
|
|
|
158
|
+ assert_eq!(&*tor_unescape(b"\"\\x20test\"").unwrap(), b" test");
|
|
|
159
|
+ assert_eq!(&*tor_unescape(b"\"test\\x20\"").unwrap(), b"test ");
|
|
|
160
|
+ assert_eq!(&*tor_unescape(b"\"te\\x20st\"").unwrap(), b"te st");
|
|
|
161
|
+
|
|
|
162
|
+ assert_eq!(&*tor_unescape(b"\"\\X20\"").unwrap(), b" ");
|
|
|
163
|
+ assert_eq!(&*tor_unescape(b"\"\\X20test\"").unwrap(), b" test");
|
|
|
164
|
+ assert_eq!(&*tor_unescape(b"\"test\\X20\"").unwrap(), b"test ");
|
|
|
165
|
+ assert_eq!(&*tor_unescape(b"\"te\\X20st\"").unwrap(), b"te st");
|
|
|
166
|
+
|
|
|
167
|
+ assert_eq!(&*tor_unescape(b"\"\\x00\"").unwrap(), b"\0");
|
|
|
168
|
+ assert_eq!(&*tor_unescape(b"\"test\\x00\"").unwrap(), b"test\0");
|
|
|
169
|
+ assert_eq!(&*tor_unescape(b"\"\\x00test\"").unwrap(), b"\0test");
|
|
|
170
|
+ assert_eq!(&*tor_unescape(b"\"\\X00\"").unwrap(), b"\0");
|
|
|
171
|
+ assert_eq!(&*tor_unescape(b"\"TEST\\X00\"").unwrap(), b"TEST\0");
|
|
|
172
|
+ assert_eq!(&*tor_unescape(b"\"\\X00TEST\"").unwrap(), b"\0TEST");
|
|
|
173
|
+ }
|
|
|
174
|
+
|
|
|
175
|
+ #[test]
|
|
|
176
|
+ fn unescape_octal() {
|
|
|
177
|
+ assert_eq!(&*tor_unescape(b"\"\\0\"").unwrap(), b"\0");
|
|
|
178
|
+ assert_eq!(&*tor_unescape(b"\"\\00\"").unwrap(), b"\0");
|
|
|
179
|
+ assert_eq!(&*tor_unescape(b"\"\\000\"").unwrap(), b"\0");
|
|
|
180
|
+
|
|
|
181
|
+ assert_eq!(&*tor_unescape(b"\"\\2\"").unwrap(), b"\x02");
|
|
|
182
|
+ assert_eq!(&*tor_unescape(b"\"\\02\"").unwrap(), b"\x02");
|
|
|
183
|
+ assert_eq!(&*tor_unescape(b"\"\\002\"").unwrap(), b"\x02");
|
|
|
184
|
+
|
|
|
185
|
+ assert_eq!(&*tor_unescape(b"\"\\40\"").unwrap(), b" ");
|
|
|
186
|
+ assert_eq!(&*tor_unescape(b"\"\\040\"").unwrap(), b" ");
|
|
|
187
|
+
|
|
|
188
|
+ assert_eq!(&*tor_unescape(b"\"\\40test\"").unwrap(), b" test");
|
|
|
189
|
+ assert_eq!(&*tor_unescape(b"\"\\040test\"").unwrap(), b" test");
|
|
|
190
|
+ assert_eq!(&*tor_unescape(b"\"\\40test\\0\"").unwrap(), b" test\0");
|
|
|
191
|
+ }
|
|
|
192
|
+
|
|
|
193
|
+ #[test]
|
|
|
194
|
+ fn invalid_unicode() {
|
|
|
195
|
+ // Raw invalid sequence, without quotes
|
|
|
196
|
+ assert_eq!(*tor_unescape(b"\xF5").unwrap(), [0xF5u8]);
|
|
|
197
|
+ // Raw invalid sequence, with quotes
|
|
|
198
|
+ assert_eq!(*tor_unescape(b"\"\xF5\"").unwrap(), [0xF5u8]);
|
|
|
199
|
+ // Escaped invalid values, we will unescape but keep them as they are.
|
|
|
200
|
+ assert_eq!(*tor_unescape(b"\"\\xF5\"").unwrap(), [0xF5u8]);
|
|
|
201
|
+ assert_eq!(*tor_unescape(b"\"\\365\"").unwrap(), [0xF5u8]);
|
|
|
202
|
+ }
|
|
|
203
|
+
|
|
|
204
|
+ #[test]
|
|
|
205
|
+ fn unquoted_space() {
|
|
|
206
|
+ assert_eq!(
|
|
|
207
|
+ tor_unescape(b"test test").unwrap_err(),
|
|
|
208
|
+ UnescapeError::SpaceInUnquotedString
|
|
|
209
|
+ );
|
|
|
210
|
+ }
|
|
|
211
|
+
|
|
|
212
|
+ #[test]
|
|
|
213
|
+ fn unterminated() {
|
|
|
214
|
+ assert_eq!(
|
|
|
215
|
+ tor_unescape(b"\"test").unwrap_err(),
|
|
|
216
|
+ UnescapeError::Unterminated
|
|
|
217
|
+ );
|
|
|
218
|
+ assert_eq!(
|
|
|
219
|
+ tor_unescape(b"\"test\\\"").unwrap_err(),
|
|
|
220
|
+ UnescapeError::Unterminated
|
|
|
221
|
+ );
|
|
|
222
|
+ assert_eq!(
|
|
|
223
|
+ tor_unescape(b"\"test\n\"").unwrap_err(),
|
|
|
224
|
+ UnescapeError::Unterminated
|
|
|
225
|
+ );
|
|
|
226
|
+ assert_eq!(
|
|
|
227
|
+ tor_unescape(b"\"test\n").unwrap_err(),
|
|
|
228
|
+ UnescapeError::Unterminated
|
|
|
229
|
+ );
|
|
|
230
|
+ assert_eq!(
|
|
|
231
|
+ tor_unescape(b"\"test\0\"").unwrap_err(),
|
|
|
232
|
+ UnescapeError::Unterminated
|
|
|
233
|
+ );
|
|
|
234
|
+ assert_eq!(
|
|
|
235
|
+ tor_unescape(b"\"test\0").unwrap_err(),
|
|
|
236
|
+ UnescapeError::Unterminated
|
|
|
237
|
+ );
|
|
|
238
|
+
|
|
|
239
|
+ // Not having a final quote shortcircuits other errors.
|
|
|
240
|
+ assert_eq!(
|
|
|
241
|
+ tor_unescape(b"\"test \" test").unwrap_err(),
|
|
|
242
|
+ UnescapeError::Unterminated
|
|
|
243
|
+ );
|
|
|
244
|
+ }
|
|
|
245
|
+
|
|
|
246
|
+ #[test]
|
|
|
247
|
+ fn unescaped_quote() {
|
|
|
248
|
+ assert_eq!(
|
|
|
249
|
+ tor_unescape(b"\"test \" test\"").unwrap_err(),
|
|
|
250
|
+ UnescapeError::UnescapedQuote
|
|
|
251
|
+ );
|
|
|
252
|
+ }
|
|
|
253
|
+
|
|
|
254
|
+ #[test]
|
|
|
255
|
+ fn invalid_escape() {
|
|
|
256
|
+ assert_eq!(
|
|
|
257
|
+ tor_unescape(b"\"\\z\"").unwrap_err(),
|
|
|
258
|
+ UnescapeError::InvalidEscape,
|
|
|
259
|
+ );
|
|
|
260
|
+
|
|
|
261
|
+ assert_eq!(
|
|
|
262
|
+ tor_unescape(b"\"\\xy\"").unwrap_err(),
|
|
|
263
|
+ UnescapeError::InvalidEscape,
|
|
|
264
|
+ );
|
|
|
265
|
+ assert_eq!(
|
|
|
266
|
+ tor_unescape(b"\"\\XY\"").unwrap_err(),
|
|
|
267
|
+ UnescapeError::InvalidEscape,
|
|
|
268
|
+ );
|
|
|
269
|
+ assert_eq!(
|
|
|
270
|
+ tor_unescape(b"\"\\x1Y\"").unwrap_err(),
|
|
|
271
|
+ UnescapeError::InvalidEscape,
|
|
|
272
|
+ );
|
|
|
273
|
+ assert_eq!(
|
|
|
274
|
+ tor_unescape(b"\"\\x1\"").unwrap_err(),
|
|
|
275
|
+ UnescapeError::InvalidEscape,
|
|
|
276
|
+ );
|
|
|
277
|
+ assert_eq!(
|
|
|
278
|
+ tor_unescape(b"\"\\x\\xF5z\"").unwrap_err(),
|
|
|
279
|
+ UnescapeError::InvalidEscape,
|
|
|
280
|
+ );
|
|
|
281
|
+ }
|
|
|
282
|
+
|
|
|
283
|
+ #[test]
|
|
|
284
|
+ fn octal_overflow() {
|
|
|
285
|
+ assert_eq!(
|
|
|
286
|
+ tor_unescape(b"\"\\777\"").unwrap_err(),
|
|
|
287
|
+ UnescapeError::OctalOverflow,
|
|
|
288
|
+ );
|
|
|
289
|
+ }
|
|
|
290
|
+} |