|
| 1 | +//===--- OxCamlDemangle.cpp -------------------------------------*- C++ -*-===// |
| 2 | +// |
| 3 | +// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | +// See https://llvm.org/LICENSE.txt for license information. |
| 5 | +// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| 6 | +// |
| 7 | +//===----------------------------------------------------------------------===// |
| 8 | +// |
| 9 | +// This file defines a demangler for the new mangling scheme devised for OxCaml |
| 10 | +// |
| 11 | +//===----------------------------------------------------------------------===// |
| 12 | + |
| 13 | +#include <cassert> |
| 14 | + |
| 15 | +#include "llvm/Demangle/Demangle.h" |
| 16 | +#include "llvm/Demangle/StringView.h" |
| 17 | +#include "llvm/Demangle/Utility.h" |
| 18 | + |
| 19 | +using llvm::itanium_demangle::OutputBuffer; |
| 20 | +using llvm::itanium_demangle::StringView; |
| 21 | + |
| 22 | +#define ERROR (~((unsigned)0)) |
| 23 | + |
| 24 | +static unsigned ConsumeUnsignedDecimal(StringView& sv) { |
| 25 | + unsigned res = 0, i = 0; |
| 26 | + while(sv[i] >= '0' && sv[i] <= '9') { |
| 27 | + res = res * 10 + (sv[i] - '0'); |
| 28 | + i++; |
| 29 | + } |
| 30 | + sv = sv.dropFront(i); |
| 31 | + if(i == 0) |
| 32 | + return ERROR; |
| 33 | + return res; |
| 34 | +} |
| 35 | + |
| 36 | +static unsigned ConsumeUnsigned26(StringView& sv) { |
| 37 | + unsigned res = 0, i = 0; |
| 38 | + while(sv[i] >= 'A' && sv[i] <= 'Z') { |
| 39 | + res = res * 26 + (sv[i] - 'A'); |
| 40 | + i++; |
| 41 | + } |
| 42 | + sv = sv.dropFront(i); |
| 43 | + if(i == 0) |
| 44 | + return ERROR; |
| 45 | + return res; |
| 46 | +} |
| 47 | + |
| 48 | +static bool islowerhex(char c) { |
| 49 | + return (c >= '0' && c <= '9') || (c >= 'a' && c <= 'f'); |
| 50 | +} |
| 51 | + |
| 52 | +static unsigned lowerhex(char c) { |
| 53 | + if(c >= '0' && c <= '9') |
| 54 | + return c - '0'; |
| 55 | + else { |
| 56 | + assert(c >= 'a' && c <= 'f'); |
| 57 | + return c - 'a' + 10; |
| 58 | + } |
| 59 | +} |
| 60 | + |
| 61 | +// Decode unicode-escaped identifier (format: u<len><coded>_<raw>) |
| 62 | +// Returns true on success, false on error |
| 63 | +static bool DecodeUnicodeEscaped(StringView& Mangled, OutputBuffer& Demangled) { |
| 64 | + unsigned len = ConsumeUnsignedDecimal(Mangled); |
| 65 | + if(len == ERROR || len <= 0 || len > Mangled.size()) |
| 66 | + return false; |
| 67 | + |
| 68 | + size_t split = Mangled.find('_'); |
| 69 | + if(split >= len) |
| 70 | + return false; |
| 71 | + |
| 72 | + StringView coded = Mangled.substr(0, split); |
| 73 | + StringView raw = Mangled.substr(split+1, len-split-1); |
| 74 | + |
| 75 | + while(!coded.empty()) { |
| 76 | + unsigned chunklen = ConsumeUnsigned26(coded); |
| 77 | + if(chunklen == ERROR || chunklen > raw.size()) |
| 78 | + return false; |
| 79 | + Demangled << raw.substr(0, chunklen); |
| 80 | + raw = raw.dropFront(chunklen); |
| 81 | + |
| 82 | + unsigned i; |
| 83 | + for(i = 0; i+1 < coded.size() && islowerhex(coded[i]); i+=2) { |
| 84 | + if(!islowerhex(coded[i+1])) |
| 85 | + return false; |
| 86 | + char c = (char)(lowerhex(coded[i]) << 4 | lowerhex(coded[i+1])); |
| 87 | + Demangled << c; |
| 88 | + } |
| 89 | + coded = coded.dropFront(i); |
| 90 | + } |
| 91 | + |
| 92 | + if(!raw.empty()) |
| 93 | + Demangled << raw; |
| 94 | + |
| 95 | + Mangled = Mangled.dropFront(len); |
| 96 | + return true; |
| 97 | +} |
| 98 | + |
| 99 | +// Decode identifier (either plain or unicode-escaped) |
| 100 | +// Handles: <len><text> or u<len><coded>_<raw> |
| 101 | +// Returns true on success, false on error |
| 102 | +static bool DecodeIdentifier(StringView& Mangled, OutputBuffer& Demangled) { |
| 103 | + if(Mangled.consumeFront('u')) { |
| 104 | + // Unicode-escaped identifier |
| 105 | + return DecodeUnicodeEscaped(Mangled, Demangled); |
| 106 | + } else { |
| 107 | + // Plain identifier with length prefix |
| 108 | + unsigned len = ConsumeUnsignedDecimal(Mangled); |
| 109 | + if(len == ERROR || len <= 0 || len > Mangled.size()) |
| 110 | + return false; |
| 111 | + Demangled << Mangled.substr(0, len); |
| 112 | + Mangled = Mangled.dropFront(len); |
| 113 | + return true; |
| 114 | + } |
| 115 | +} |
| 116 | + |
| 117 | +// Decode anonymous location (format: filename_line_col) |
| 118 | +// Anonymous functions/modules are encoded as: fn(filename:line:col) |
| 119 | +// Returns true on success, false on error |
| 120 | +static bool DecodeAnonymousLocation(StringView& Mangled, OutputBuffer& Demangled) { |
| 121 | + // Allocate temporary buffer based on remaining mangled string size |
| 122 | + // The decoded identifier will be at most the size of the remaining mangled string |
| 123 | + size_t buffer_size = Mangled.size(); |
| 124 | + if(buffer_size == 0) |
| 125 | + return false; |
| 126 | + |
| 127 | + char *temp_buf = static_cast<char *>(std::malloc(buffer_size)); |
| 128 | + if(temp_buf == nullptr) |
| 129 | + std::terminate(); |
| 130 | + |
| 131 | + OutputBuffer TempDemangled(temp_buf, buffer_size); |
| 132 | + |
| 133 | + if(!DecodeIdentifier(Mangled, TempDemangled)) { |
| 134 | + std::free(temp_buf); |
| 135 | + return false; |
| 136 | + } |
| 137 | + |
| 138 | + size_t temp_len = TempDemangled.getCurrentPosition(); |
| 139 | + |
| 140 | + // Parse filename_line_col format by finding the last two underscores |
| 141 | + size_t first_underscore = 0, second_underscore = 0; |
| 142 | + int underscore_count = 0; |
| 143 | + |
| 144 | + for(size_t j = temp_len; j > 0; j--) { |
| 145 | + if(temp_buf[j-1] == '_') { |
| 146 | + underscore_count++; |
| 147 | + if(underscore_count == 1) |
| 148 | + second_underscore = j - 1; |
| 149 | + else if(underscore_count == 2) { |
| 150 | + first_underscore = j - 1; |
| 151 | + break; |
| 152 | + } |
| 153 | + } |
| 154 | + } |
| 155 | + |
| 156 | + // Output in format fn(filename:line:col) |
| 157 | + if(underscore_count >= 2) { |
| 158 | + Demangled << "fn("; |
| 159 | + for(size_t j = 0; j < first_underscore; j++) |
| 160 | + Demangled << temp_buf[j]; |
| 161 | + Demangled << ':'; |
| 162 | + for(size_t j = first_underscore + 1; j < second_underscore; j++) |
| 163 | + Demangled << temp_buf[j]; |
| 164 | + Demangled << ':'; |
| 165 | + for(size_t j = second_underscore + 1; j < temp_len; j++) |
| 166 | + Demangled << temp_buf[j]; |
| 167 | + Demangled << ')'; |
| 168 | + } else { |
| 169 | + // Fallback: just output the identifier as-is |
| 170 | + for(size_t j = 0; j < temp_len; j++) |
| 171 | + Demangled << temp_buf[j]; |
| 172 | + } |
| 173 | + |
| 174 | + std::free(temp_buf); |
| 175 | + return true; |
| 176 | +} |
| 177 | + |
| 178 | +char *llvm::oxcamlDemangle(const char *MangledName) { |
| 179 | + StringView Mangled(MangledName); |
| 180 | + if(!Mangled.consumeFront("_O")) |
| 181 | + return nullptr; |
| 182 | + |
| 183 | + // Allocate the buffer at a reasonable size, as OutputBuffer allocates 992 |
| 184 | + // bytes when starting from an empty buffer |
| 185 | + char *DemangledBuffer; |
| 186 | + DemangledBuffer = static_cast<char *>(std::malloc(Mangled.size())); |
| 187 | + if (DemangledBuffer == nullptr) |
| 188 | + std::terminate(); |
| 189 | + OutputBuffer Demangled(DemangledBuffer, Mangled.size()); |
| 190 | + |
| 191 | +#define ENDONERROR() do { \ |
| 192 | + std::free(Demangled.getBuffer()); \ |
| 193 | + return nullptr; \ |
| 194 | +} while(0) |
| 195 | + |
| 196 | + // Parse path items |
| 197 | + while(!Mangled.empty()) { |
| 198 | + // Check for terminating underscore |
| 199 | + if(Mangled[0] == '_') { |
| 200 | + // End of symbol path, rest is unique id |
| 201 | + break; |
| 202 | + } |
| 203 | + |
| 204 | + // Handle each path_item type |
| 205 | + switch(Mangled[0]) { |
| 206 | + case 'M': // Module |
| 207 | + if(!Demangled.empty()) |
| 208 | + Demangled << '.'; |
| 209 | + Mangled = Mangled.dropFront(1); |
| 210 | + if(!DecodeIdentifier(Mangled, Demangled)) |
| 211 | + ENDONERROR(); |
| 212 | + break; |
| 213 | + |
| 214 | + case 'F': // NamedFunction |
| 215 | + if(!Demangled.empty()) |
| 216 | + Demangled << '.'; |
| 217 | + Mangled = Mangled.dropFront(1); |
| 218 | + if(!DecodeIdentifier(Mangled, Demangled)) |
| 219 | + ENDONERROR(); |
| 220 | + break; |
| 221 | + |
| 222 | + case 'L': // AnonymousFunction |
| 223 | + if(!Demangled.empty()) |
| 224 | + Demangled << '.'; |
| 225 | + Mangled = Mangled.dropFront(1); |
| 226 | + if(!DecodeAnonymousLocation(Mangled, Demangled)) |
| 227 | + ENDONERROR(); |
| 228 | + break; |
| 229 | + |
| 230 | + case 'S': // AnonymousModule |
| 231 | + if(!Demangled.empty()) |
| 232 | + Demangled << '.'; |
| 233 | + Mangled = Mangled.dropFront(1); |
| 234 | + if(!DecodeAnonymousLocation(Mangled, Demangled)) |
| 235 | + ENDONERROR(); |
| 236 | + break; |
| 237 | + |
| 238 | + case 'P': // PartialFunction (no dot separator) |
| 239 | + Mangled = Mangled.dropFront(1); |
| 240 | + Demangled << "(partially_applied)"; |
| 241 | + break; |
| 242 | + |
| 243 | + default: |
| 244 | + // No prefix means Module (legacy compatibility) |
| 245 | + if(!Demangled.empty()) |
| 246 | + Demangled << '.'; |
| 247 | + if(!DecodeIdentifier(Mangled, Demangled)) |
| 248 | + ENDONERROR(); |
| 249 | + break; |
| 250 | + } |
| 251 | + } |
| 252 | + |
| 253 | + Demangled << '\0'; |
| 254 | + |
| 255 | + return Demangled.getBuffer(); |
| 256 | +} |
0 commit comments