PR c++/83490
[official-gcc.git] / libiberty / rust-demangle.c
blob3d7d0905b140860501a31e52d10cfb92bbfc75a2
1 /* Demangler for the Rust programming language
2 Copyright (C) 2016-2017 Free Software Foundation, Inc.
3 Written by David Tolnay (dtolnay@gmail.com).
5 This file is part of the libiberty library.
6 Libiberty is free software; you can redistribute it and/or
7 modify it under the terms of the GNU Library General Public
8 License as published by the Free Software Foundation; either
9 version 2 of the License, or (at your option) any later version.
11 In addition to the permissions in the GNU Library General Public
12 License, the Free Software Foundation gives you unlimited permission
13 to link the compiled version of this file into combinations with other
14 programs, and to distribute those combinations without any restriction
15 coming from the use of this file. (The Library Public License
16 restrictions do apply in other respects; for example, they cover
17 modification of the file, and distribution when not linked into a
18 combined executable.)
20 Libiberty is distributed in the hope that it will be useful,
21 but WITHOUT ANY WARRANTY; without even the implied warranty of
22 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
23 Library General Public License for more details.
25 You should have received a copy of the GNU Library General Public
26 License along with libiberty; see the file COPYING.LIB.
27 If not, see <http://www.gnu.org/licenses/>. */
30 #ifdef HAVE_CONFIG_H
31 #include "config.h"
32 #endif
34 #include "safe-ctype.h"
36 #include <sys/types.h>
37 #include <string.h>
38 #include <stdio.h>
40 #ifdef HAVE_STRING_H
41 #include <string.h>
42 #else
43 extern size_t strlen(const char *s);
44 extern int strncmp(const char *s1, const char *s2, size_t n);
45 extern void *memset(void *s, int c, size_t n);
46 #endif
48 #include <demangle.h>
49 #include "libiberty.h"
52 /* Mangled Rust symbols look like this:
53 _$LT$std..sys..fd..FileDesc$u20$as$u20$core..ops..Drop$GT$::drop::hc68340e1baa4987a
55 The original symbol is:
56 <std::sys::fd::FileDesc as core::ops::Drop>::drop
58 The last component of the path is a 64-bit hash in lowercase hex,
59 prefixed with "h". Rust does not have a global namespace between
60 crates, an illusion which Rust maintains by using the hash to
61 distinguish things that would otherwise have the same symbol.
63 Any path component not starting with a XID_Start character is
64 prefixed with "_".
66 The following escape sequences are used:
68 "," => $C$
69 "@" => $SP$
70 "*" => $BP$
71 "&" => $RF$
72 "<" => $LT$
73 ">" => $GT$
74 "(" => $LP$
75 ")" => $RP$
76 " " => $u20$
77 "\"" => $u22$
78 "'" => $u27$
79 "+" => $u2b$
80 ";" => $u3b$
81 "[" => $u5b$
82 "]" => $u5d$
83 "{" => $u7b$
84 "}" => $u7d$
85 "~" => $u7e$
87 A double ".." means "::" and a single "." means "-".
89 The only characters allowed in the mangled symbol are a-zA-Z0-9 and _.:$ */
91 static const char *hash_prefix = "::h";
92 static const size_t hash_prefix_len = 3;
93 static const size_t hash_len = 16;
95 static int is_prefixed_hash (const char *start);
96 static int looks_like_rust (const char *sym, size_t len);
97 static int unescape (const char **in, char **out, const char *seq, char value);
99 /* INPUT: sym: symbol that has been through C++ (gnu v3) demangling
101 This function looks for the following indicators:
103 1. The hash must consist of "h" followed by 16 lowercase hex digits.
105 2. As a sanity check, the hash must use between 5 and 15 of the 16
106 possible hex digits. This is true of 99.9998% of hashes so once
107 in your life you may see a false negative. The point is to
108 notice path components that could be Rust hashes but are
109 probably not, like "haaaaaaaaaaaaaaaa". In this case a false
110 positive (non-Rust symbol has an important path component
111 removed because it looks like a Rust hash) is worse than a false
112 negative (the rare Rust symbol is not demangled) so this sets
113 the balance in favor of false negatives.
115 3. There must be no characters other than a-zA-Z0-9 and _.:$
117 4. There must be no unrecognized $-sign sequences.
119 5. There must be no sequence of three or more dots in a row ("..."). */
122 rust_is_mangled (const char *sym)
124 size_t len, len_without_hash;
126 if (!sym)
127 return 0;
129 len = strlen (sym);
130 if (len <= hash_prefix_len + hash_len)
131 /* Not long enough to contain "::h" + hash + something else */
132 return 0;
134 len_without_hash = len - (hash_prefix_len + hash_len);
135 if (!is_prefixed_hash (sym + len_without_hash))
136 return 0;
138 return looks_like_rust (sym, len_without_hash);
141 /* A hash is the prefix "::h" followed by 16 lowercase hex digits. The
142 hex digits must comprise between 5 and 15 (inclusive) distinct
143 digits. */
145 static int
146 is_prefixed_hash (const char *str)
148 const char *end;
149 char seen[16];
150 size_t i;
151 int count;
153 if (strncmp (str, hash_prefix, hash_prefix_len))
154 return 0;
155 str += hash_prefix_len;
157 memset (seen, 0, sizeof(seen));
158 for (end = str + hash_len; str < end; str++)
159 if (*str >= '0' && *str <= '9')
160 seen[*str - '0'] = 1;
161 else if (*str >= 'a' && *str <= 'f')
162 seen[*str - 'a' + 10] = 1;
163 else
164 return 0;
166 /* Count how many distinct digits seen */
167 count = 0;
168 for (i = 0; i < 16; i++)
169 if (seen[i])
170 count++;
172 return count >= 5 && count <= 15;
175 static int
176 looks_like_rust (const char *str, size_t len)
178 const char *end = str + len;
180 while (str < end)
181 switch (*str)
183 case '$':
184 if (!strncmp (str, "$C$", 3))
185 str += 3;
186 else if (!strncmp (str, "$SP$", 4)
187 || !strncmp (str, "$BP$", 4)
188 || !strncmp (str, "$RF$", 4)
189 || !strncmp (str, "$LT$", 4)
190 || !strncmp (str, "$GT$", 4)
191 || !strncmp (str, "$LP$", 4)
192 || !strncmp (str, "$RP$", 4))
193 str += 4;
194 else if (!strncmp (str, "$u20$", 5)
195 || !strncmp (str, "$u22$", 5)
196 || !strncmp (str, "$u27$", 5)
197 || !strncmp (str, "$u2b$", 5)
198 || !strncmp (str, "$u3b$", 5)
199 || !strncmp (str, "$u5b$", 5)
200 || !strncmp (str, "$u5d$", 5)
201 || !strncmp (str, "$u7b$", 5)
202 || !strncmp (str, "$u7d$", 5)
203 || !strncmp (str, "$u7e$", 5))
204 str += 5;
205 else
206 return 0;
207 break;
208 case '.':
209 /* Do not allow three or more consecutive dots */
210 if (!strncmp (str, "...", 3))
211 return 0;
212 /* Fall through */
213 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
214 case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
215 case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
216 case 's': case 't': case 'u': case 'v': case 'w': case 'x':
217 case 'y': case 'z':
218 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
219 case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
220 case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
221 case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
222 case 'Y': case 'Z':
223 case '0': case '1': case '2': case '3': case '4': case '5':
224 case '6': case '7': case '8': case '9':
225 case '_':
226 case ':':
227 str++;
228 break;
229 default:
230 return 0;
233 return 1;
237 INPUT: sym: symbol for which rust_is_mangled(sym) returned 1.
239 The input is demangled in-place because the mangled name is always
240 longer than the demangled one. */
242 void
243 rust_demangle_sym (char *sym)
245 const char *in;
246 char *out;
247 const char *end;
249 if (!sym)
250 return;
252 in = sym;
253 out = sym;
254 end = sym + strlen (sym) - (hash_prefix_len + hash_len);
256 while (in < end)
257 switch (*in)
259 case '$':
260 if (!(unescape (&in, &out, "$C$", ',')
261 || unescape (&in, &out, "$SP$", '@')
262 || unescape (&in, &out, "$BP$", '*')
263 || unescape (&in, &out, "$RF$", '&')
264 || unescape (&in, &out, "$LT$", '<')
265 || unescape (&in, &out, "$GT$", '>')
266 || unescape (&in, &out, "$LP$", '(')
267 || unescape (&in, &out, "$RP$", ')')
268 || unescape (&in, &out, "$u20$", ' ')
269 || unescape (&in, &out, "$u22$", '\"')
270 || unescape (&in, &out, "$u27$", '\'')
271 || unescape (&in, &out, "$u2b$", '+')
272 || unescape (&in, &out, "$u3b$", ';')
273 || unescape (&in, &out, "$u5b$", '[')
274 || unescape (&in, &out, "$u5d$", ']')
275 || unescape (&in, &out, "$u7b$", '{')
276 || unescape (&in, &out, "$u7d$", '}')
277 || unescape (&in, &out, "$u7e$", '~'))) {
278 /* unexpected escape sequence, not looks_like_rust. */
279 goto fail;
281 break;
282 case '_':
283 /* If this is the start of a path component and the next
284 character is an escape sequence, ignore the underscore. The
285 mangler inserts an underscore to make sure the path
286 component begins with a XID_Start character. */
287 if ((in == sym || in[-1] == ':') && in[1] == '$')
288 in++;
289 else
290 *out++ = *in++;
291 break;
292 case '.':
293 if (in[1] == '.')
295 /* ".." becomes "::" */
296 *out++ = ':';
297 *out++ = ':';
298 in += 2;
300 else
302 /* "." becomes "-" */
303 *out++ = '-';
304 in++;
306 break;
307 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
308 case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
309 case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
310 case 's': case 't': case 'u': case 'v': case 'w': case 'x':
311 case 'y': case 'z':
312 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
313 case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
314 case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
315 case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
316 case 'Y': case 'Z':
317 case '0': case '1': case '2': case '3': case '4': case '5':
318 case '6': case '7': case '8': case '9':
319 case ':':
320 *out++ = *in++;
321 break;
322 default:
323 /* unexpected character in symbol, not looks_like_rust. */
324 goto fail;
326 goto done;
328 fail:
329 *out++ = '?'; /* This is pretty lame, but it's hard to do better. */
330 done:
331 *out = '\0';
334 static int
335 unescape (const char **in, char **out, const char *seq, char value)
337 size_t len = strlen (seq);
339 if (strncmp (*in, seq, len))
340 return 0;
342 **out = value;
344 *in += len;
345 *out += 1;
347 return 1;