1218885Sdim//===-- Path.cpp - Implement OS Path Concept --------------------*- C++ -*-===//
2218885Sdim//
3218885Sdim//                     The LLVM Compiler Infrastructure
4218885Sdim//
5218885Sdim// This file is distributed under the University of Illinois Open Source
6218885Sdim// License. See LICENSE.TXT for details.
7218885Sdim//
8218885Sdim//===----------------------------------------------------------------------===//
9218885Sdim//
10218885Sdim//  This header file implements the operating system Path concept.
11218885Sdim//
12218885Sdim//===----------------------------------------------------------------------===//
13218885Sdim
14218885Sdim#include "llvm/Support/Path.h"
15218885Sdim#include "llvm/Config/config.h"
16249423Sdim#include "llvm/Support/Endian.h"
17218885Sdim#include "llvm/Support/FileSystem.h"
18218885Sdim#include <cassert>
19218885Sdim#include <cstring>
20218885Sdim#include <ostream>
21218885Sdimusing namespace llvm;
22218885Sdimusing namespace sys;
23221345Sdimnamespace {
24221345Sdimusing support::ulittle32_t;
25221345Sdim}
26218885Sdim
27218885Sdim//===----------------------------------------------------------------------===//
28218885Sdim//=== WARNING: Implementation here must contain only TRULY operating system
29218885Sdim//===          independent code.
30218885Sdim//===----------------------------------------------------------------------===//
31218885Sdim
32218885Sdimbool Path::operator==(const Path &that) const {
33218885Sdim  return path == that.path;
34218885Sdim}
35218885Sdim
36218885Sdimbool Path::operator<(const Path& that) const {
37218885Sdim  return path < that.path;
38218885Sdim}
39218885Sdim
40218885SdimLLVMFileType
41218885Sdimsys::IdentifyFileType(const char *magic, unsigned length) {
42218885Sdim  assert(magic && "Invalid magic number string");
43218885Sdim  assert(length >=4 && "Invalid magic number length");
44218885Sdim  switch ((unsigned char)magic[0]) {
45218885Sdim    case 0xDE:  // 0x0B17C0DE = BC wraper
46218885Sdim      if (magic[1] == (char)0xC0 && magic[2] == (char)0x17 &&
47218885Sdim          magic[3] == (char)0x0B)
48218885Sdim        return Bitcode_FileType;
49218885Sdim      break;
50218885Sdim    case 'B':
51218885Sdim      if (magic[1] == 'C' && magic[2] == (char)0xC0 && magic[3] == (char)0xDE)
52218885Sdim        return Bitcode_FileType;
53218885Sdim      break;
54218885Sdim    case '!':
55218885Sdim      if (length >= 8)
56218885Sdim        if (memcmp(magic,"!<arch>\n",8) == 0)
57218885Sdim          return Archive_FileType;
58218885Sdim      break;
59218885Sdim
60218885Sdim    case '\177':
61218885Sdim      if (magic[1] == 'E' && magic[2] == 'L' && magic[3] == 'F') {
62239462Sdim        bool Data2MSB = magic[5] == 2;
63239462Sdim        unsigned high = Data2MSB ? 16 : 17;
64239462Sdim        unsigned low  = Data2MSB ? 17 : 16;
65239462Sdim        if (length >= 18 && magic[high] == 0)
66239462Sdim          switch (magic[low]) {
67218885Sdim            default: break;
68218885Sdim            case 1: return ELF_Relocatable_FileType;
69218885Sdim            case 2: return ELF_Executable_FileType;
70218885Sdim            case 3: return ELF_SharedObject_FileType;
71218885Sdim            case 4: return ELF_Core_FileType;
72218885Sdim          }
73218885Sdim      }
74218885Sdim      break;
75218885Sdim
76218885Sdim    case 0xCA:
77218885Sdim      if (magic[1] == char(0xFE) && magic[2] == char(0xBA) &&
78218885Sdim          magic[3] == char(0xBE)) {
79218885Sdim        // This is complicated by an overlap with Java class files.
80218885Sdim        // See the Mach-O section in /usr/share/file/magic for details.
81218885Sdim        if (length >= 8 && magic[7] < 43)
82218885Sdim          // FIXME: Universal Binary of any type.
83218885Sdim          return Mach_O_DynamicallyLinkedSharedLib_FileType;
84218885Sdim      }
85218885Sdim      break;
86218885Sdim
87221345Sdim      // The two magic numbers for mach-o are:
88221345Sdim      // 0xfeedface - 32-bit mach-o
89221345Sdim      // 0xfeedfacf - 64-bit mach-o
90218885Sdim    case 0xFE:
91221345Sdim    case 0xCE:
92221345Sdim    case 0xCF: {
93218885Sdim      uint16_t type = 0;
94218885Sdim      if (magic[0] == char(0xFE) && magic[1] == char(0xED) &&
95234353Sdim          magic[2] == char(0xFA) &&
96221345Sdim          (magic[3] == char(0xCE) || magic[3] == char(0xCF))) {
97218885Sdim        /* Native endian */
98218885Sdim        if (length >= 16) type = magic[14] << 8 | magic[15];
99221345Sdim      } else if ((magic[0] == char(0xCE) || magic[0] == char(0xCF)) &&
100221345Sdim                 magic[1] == char(0xFA) && magic[2] == char(0xED) &&
101221345Sdim                 magic[3] == char(0xFE)) {
102218885Sdim        /* Reverse endian */
103218885Sdim        if (length >= 14) type = magic[13] << 8 | magic[12];
104218885Sdim      }
105218885Sdim      switch (type) {
106218885Sdim        default: break;
107218885Sdim        case 1: return Mach_O_Object_FileType;
108218885Sdim        case 2: return Mach_O_Executable_FileType;
109218885Sdim        case 3: return Mach_O_FixedVirtualMemorySharedLib_FileType;
110218885Sdim        case 4: return Mach_O_Core_FileType;
111218885Sdim        case 5: return Mach_O_PreloadExecutable_FileType;
112218885Sdim        case 6: return Mach_O_DynamicallyLinkedSharedLib_FileType;
113218885Sdim        case 7: return Mach_O_DynamicLinker_FileType;
114218885Sdim        case 8: return Mach_O_Bundle_FileType;
115218885Sdim        case 9: return Mach_O_DynamicallyLinkedSharedLibStub_FileType;
116226633Sdim        case 10: return Mach_O_DSYMCompanion_FileType;
117218885Sdim      }
118218885Sdim      break;
119218885Sdim    }
120218885Sdim    case 0xF0: // PowerPC Windows
121218885Sdim    case 0x83: // Alpha 32-bit
122218885Sdim    case 0x84: // Alpha 64-bit
123218885Sdim    case 0x66: // MPS R4000 Windows
124218885Sdim    case 0x50: // mc68K
125218885Sdim    case 0x4c: // 80386 Windows
126218885Sdim      if (magic[1] == 0x01)
127218885Sdim        return COFF_FileType;
128218885Sdim
129218885Sdim    case 0x90: // PA-RISC Windows
130218885Sdim    case 0x68: // mc68K Windows
131218885Sdim      if (magic[1] == 0x02)
132218885Sdim        return COFF_FileType;
133218885Sdim      break;
134221345Sdim
135221345Sdim    case 0x4d: // Possible MS-DOS stub on Windows PE file
136221345Sdim      if (magic[1] == 0x5a) {
137221345Sdim        uint32_t off = *reinterpret_cast<const ulittle32_t *>(magic + 0x3c);
138221345Sdim        // PE/COFF file, either EXE or DLL.
139221345Sdim        if (off < length && memcmp(magic + off, "PE\0\0",4) == 0)
140221345Sdim          return COFF_FileType;
141221345Sdim      }
142221345Sdim      break;
143221345Sdim
144218885Sdim    case 0x64: // x86-64 Windows.
145218885Sdim      if (magic[1] == char(0x86))
146218885Sdim        return COFF_FileType;
147218885Sdim      break;
148218885Sdim
149218885Sdim    default:
150218885Sdim      break;
151218885Sdim  }
152218885Sdim  return Unknown_FileType;
153218885Sdim}
154218885Sdim
155218885Sdimbool
156218885SdimPath::isArchive() const {
157234353Sdim  fs::file_magic type;
158218885Sdim  if (fs::identify_magic(str(), type))
159218885Sdim    return false;
160234353Sdim  return type == fs::file_magic::archive;
161218885Sdim}
162218885Sdim
163218885Sdimbool
164218885SdimPath::isDynamicLibrary() const {
165234353Sdim  fs::file_magic type;
166218885Sdim  if (fs::identify_magic(str(), type))
167218885Sdim    return false;
168218885Sdim  switch (type) {
169218885Sdim    default: return false;
170234353Sdim    case fs::file_magic::macho_fixed_virtual_memory_shared_lib:
171234353Sdim    case fs::file_magic::macho_dynamically_linked_shared_lib:
172234353Sdim    case fs::file_magic::macho_dynamically_linked_shared_lib_stub:
173234353Sdim    case fs::file_magic::elf_shared_object:
174234353Sdim    case fs::file_magic::pecoff_executable:  return true;
175218885Sdim  }
176218885Sdim}
177218885Sdim
178218885Sdimbool
179218885SdimPath::isObjectFile() const {
180234353Sdim  fs::file_magic type;
181234353Sdim  if (fs::identify_magic(str(), type) || type == fs::file_magic::unknown)
182218885Sdim    return false;
183218885Sdim  return true;
184218885Sdim}
185218885Sdim
186218885SdimPath
187218885SdimPath::FindLibrary(std::string& name) {
188218885Sdim  std::vector<sys::Path> LibPaths;
189218885Sdim  GetSystemLibraryPaths(LibPaths);
190218885Sdim  for (unsigned i = 0; i < LibPaths.size(); ++i) {
191218885Sdim    sys::Path FullPath(LibPaths[i]);
192218885Sdim    FullPath.appendComponent("lib" + name + LTDL_SHLIB_EXT);
193218885Sdim    if (FullPath.isDynamicLibrary())
194218885Sdim      return FullPath;
195218885Sdim    FullPath.eraseSuffix();
196218885Sdim    FullPath.appendSuffix("a");
197218885Sdim    if (FullPath.isArchive())
198218885Sdim      return FullPath;
199218885Sdim  }
200218885Sdim  return sys::Path();
201218885Sdim}
202218885Sdim
203218885SdimStringRef Path::GetDLLSuffix() {
204218885Sdim  return &(LTDL_SHLIB_EXT[1]);
205218885Sdim}
206218885Sdim
207218885Sdimvoid
208218885SdimPath::appendSuffix(StringRef suffix) {
209218885Sdim  if (!suffix.empty()) {
210218885Sdim    path.append(".");
211218885Sdim    path.append(suffix);
212218885Sdim  }
213218885Sdim}
214218885Sdim
215218885Sdimbool
216218885SdimPath::isBitcodeFile() const {
217234353Sdim  fs::file_magic type;
218218885Sdim  if (fs::identify_magic(str(), type))
219218885Sdim    return false;
220234353Sdim  return type == fs::file_magic::bitcode;
221218885Sdim}
222218885Sdim
223218885Sdimbool Path::hasMagicNumber(StringRef Magic) const {
224218885Sdim  std::string actualMagic;
225218885Sdim  if (getMagicNumber(actualMagic, static_cast<unsigned>(Magic.size())))
226218885Sdim    return Magic == actualMagic;
227218885Sdim  return false;
228218885Sdim}
229218885Sdim
230218885Sdimstatic void getPathList(const char*path, std::vector<Path>& Paths) {
231218885Sdim  const char* at = path;
232218885Sdim  const char* delim = strchr(at, PathSeparator);
233218885Sdim  Path tmpPath;
234218885Sdim  while (delim != 0) {
235218885Sdim    std::string tmp(at, size_t(delim-at));
236218885Sdim    if (tmpPath.set(tmp))
237218885Sdim      if (tmpPath.canRead())
238218885Sdim        Paths.push_back(tmpPath);
239218885Sdim    at = delim + 1;
240218885Sdim    delim = strchr(at, PathSeparator);
241218885Sdim  }
242218885Sdim
243218885Sdim  if (*at != 0)
244218885Sdim    if (tmpPath.set(std::string(at)))
245218885Sdim      if (tmpPath.canRead())
246218885Sdim        Paths.push_back(tmpPath);
247218885Sdim}
248218885Sdim
249218885Sdimstatic StringRef getDirnameCharSep(StringRef path, const char *Sep) {
250218885Sdim  assert(Sep[0] != '\0' && Sep[1] == '\0' &&
251218885Sdim         "Sep must be a 1-character string literal.");
252218885Sdim  if (path.empty())
253218885Sdim    return ".";
254218885Sdim
255218885Sdim  // If the path is all slashes, return a single slash.
256218885Sdim  // Otherwise, remove all trailing slashes.
257218885Sdim
258218885Sdim  signed pos = static_cast<signed>(path.size()) - 1;
259218885Sdim
260218885Sdim  while (pos >= 0 && path[pos] == Sep[0])
261218885Sdim    --pos;
262218885Sdim
263218885Sdim  if (pos < 0)
264218885Sdim    return path[0] == Sep[0] ? Sep : ".";
265218885Sdim
266218885Sdim  // Any slashes left?
267218885Sdim  signed i = 0;
268218885Sdim
269218885Sdim  while (i < pos && path[i] != Sep[0])
270218885Sdim    ++i;
271218885Sdim
272218885Sdim  if (i == pos) // No slashes?  Return "."
273218885Sdim    return ".";
274218885Sdim
275218885Sdim  // There is at least one slash left.  Remove all trailing non-slashes.
276218885Sdim  while (pos >= 0 && path[pos] != Sep[0])
277218885Sdim    --pos;
278218885Sdim
279218885Sdim  // Remove any trailing slashes.
280218885Sdim  while (pos >= 0 && path[pos] == Sep[0])
281218885Sdim    --pos;
282218885Sdim
283218885Sdim  if (pos < 0)
284218885Sdim    return path[0] == Sep[0] ? Sep : ".";
285218885Sdim
286218885Sdim  return path.substr(0, pos+1);
287218885Sdim}
288218885Sdim
289218885Sdim// Include the truly platform-specific parts of this class.
290218885Sdim#if defined(LLVM_ON_UNIX)
291218885Sdim#include "Unix/Path.inc"
292218885Sdim#endif
293218885Sdim#if defined(LLVM_ON_WIN32)
294218885Sdim#include "Windows/Path.inc"
295218885Sdim#endif
296