path: root/lib/Archive/ArchiveReader.cpp

//===-- ArchiveReader.cpp - Read LLVM archive files -------------*- C++ -*-===//
// 
//                     The LLVM Compiler Infrastructure
//
// This file was developed by Reid Spencer and is distributed under the 
// University of Illinois Open Source License. See LICENSE.TXT for details.
// 
//===----------------------------------------------------------------------===//
//
// Builds up standard unix archive files (.a) containing LLVM bytecode.
//
//===----------------------------------------------------------------------===//

#include "ArchiveInternals.h"
#include "llvm/Bytecode/Reader.h"

using namespace llvm;

namespace {

/// Read a variable-bit-rate encoded unsigned integer
inline unsigned readInteger(const char*&At, const char*End) {
  unsigned Shift = 0;
  unsigned Result = 0;
  
  do {
    if (At == End) 
      throw std::string("Ran out of data reading vbr_uint!");
    Result |= (unsigned)((*At++) & 0x7F) << Shift;
    Shift += 7;
  } while (At[-1] & 0x80);
  return Result;
}

}

// Completely parse the Archive's symbol table and populate symTab member var.
void
Archive::parseSymbolTable(const void* data, unsigned size) {
  const char* At = (const char*) data;
  const char* End = At + size;
  while (At < End) {
    unsigned offset = readInteger(At, End);
    unsigned length = readInteger(At, End);
    if (At + length > End)
      throw std::string("malformed symbol table");
    // we don't care if it can't be inserted (duplicate entry)
    symTab.insert(std::make_pair(std::string(At,length),offset));
    At += length;
  }
  symTabSize = size;
}

// This member parses an ArchiveMemberHeader that is presumed to be pointed to
// by At. The At pointer is updated to the byte just after the header, which
// can be variable in size. 
ArchiveMember*
Archive::parseMemberHeader(const char*& At, const char* End) {
  assert(At + sizeof(ArchiveMemberHeader) < End && "Not enough data");

  // Cast archive member header
  ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At;
  At += sizeof(ArchiveMemberHeader);

  // Instantiate the ArchiveMember to be filled
  ArchiveMember* member = new ArchiveMember(this);

  // Extract the size and determine if the file is 
  // compressed or not (negative length).
  int flags = 0;
  int MemberSize = atoi(Hdr->size);
  if (MemberSize < 0) {
    flags |= ArchiveMember::CompressedFlag;
    MemberSize = -MemberSize;
  }

  // Check the size of the member for sanity
  if (At + MemberSize > End)
    throw std::string("invalid member length in archive file");

  // Check the member signature
  if (!Hdr->checkSignature())
    throw std::string("invalid file member signature");

  // Convert and check the member name
  // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol 
  // table. The special name "//" and 14 blanks is for a string table, used 
  // for long file names. This library doesn't generate either of those but
  // it will accept them. If the name starts with #1/ and the remainder is 
  // digits, then those digits specify the length of the name that is 
  // stored immediately following the header. The special name 
  // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bytecode. 
  // Anything else is a regular, short filename that is terminated with 
  // a '/' and blanks.

  std::string pathname;
  unsigned index;
  switch (Hdr->name[0]) {
    case '#':
      if (Hdr->name[1] == '1' && Hdr->name[2] == '/') {
        if (isdigit(Hdr->name[3])) {
          unsigned len = atoi(&Hdr->name[3]);
          pathname.assign(At,len);
          At += len + 1; // terminated by \n
          flags |= ArchiveMember::HasLongFilenameFlag;
        } else
          throw std::string("invalid long filename");
      } else if (Hdr->name[1] == '_' && 
                 (0==memcmp(Hdr->name,ARFILE_LLVM_SYMTAB_NAME,16))) {
        // The member is using a long file name (>15 chars) format.
        // This format is standard for 4.4BSD and Mac OSX operating
        // systems. LLVM uses it similarly. In this format, the
        // remainder of the name field (after #1/) specifies the
        // length of the file name which occupy the first bytes of
        // the member's data. The pathname already has the #1/ stripped.
        pathname.assign(ARFILE_LLVM_SYMTAB_NAME);
        flags |= ArchiveMember::LLVMSymbolTableFlag;
      }
      break;
    case '/':
      if (Hdr->name[1]== '/') {
        if (0==memcmp(Hdr->name,ARFILE_STRTAB_NAME,16)) {
          pathname.assign(ARFILE_STRTAB_NAME);
          flags |= ArchiveMember::StringTableFlag;
        } else {
          throw std::string("invalid string table name");
        }
      } else if (Hdr->name[1] == ' ') {
        if (0==memcmp(Hdr->name,ARFILE_SYMTAB_NAME,16)) {
          pathname.assign(ARFILE_SYMTAB_NAME);
          flags |= ArchiveMember::ForeignSymbolTableFlag;
        } else {
          throw std::string("invalid foreign symbol table name");
        }
      } else if (isdigit(Hdr->name[1])) {
        unsigned index = atoi(&Hdr->name[1]);
        if (index < strtab.length()) {
          const char* namep = strtab.c_str() + index;
          const char* endp = strtab.c_str() + strtab.length();
          const char* p = namep;
          const char* last_p = p;
          while (p < endp) {
            if (*p == '\n' && *last_p == '/') {
              pathname.assign(namep,last_p-namep);
              flags |= ArchiveMember::HasLongFilenameFlag;
              break;
            }
            last_p = p;
            p++;
          }
          if (p >= endp)
            throw std::string("missing name termiantor in string table");
        } else {
          throw std::string("name index beyond string table");
        }
      }
      break;

    default:
      char* slash = (char*) memchr(Hdr->name,'/',16);
      if (slash == 0)
        throw std::string("missing name terminator");
      pathname.assign(Hdr->name,slash-Hdr->name);
      break;
  }

  // Determine if this is a bytecode file
  switch (sys::IdentifyFileType(At,4)) {
    case sys::BytecodeFileType:
      flags |= ArchiveMember::BytecodeFlag;
      break;
    case sys::CompressedBytecodeFileType:
      flags |= ArchiveMember::CompressedBytecodeFlag;
      flags &= ~ArchiveMember::CompressedFlag;
      break;
    default:
      flags &= ~(ArchiveMember::BytecodeFlag|
                 ArchiveMember::CompressedBytecodeFlag);
      break;
  }

  // Fill in fields of the ArchiveMember
  member->next = 0;
  member->prev = 0;
  member->parent = this;
  member->path.setFile(pathname);
  member->info.fileSize = MemberSize;
  member->info.modTime.fromEpochTime(atoi(Hdr->date));
  sscanf(Hdr->mode,"%o",&(member->info.mode));
  member->info.user = atoi(Hdr->uid);
  member->info.group = atoi(Hdr->gid);
  member->flags = flags;
  member->data = At;

  return member;
}

void
Archive::checkSignature() {
  // Check the magic string at file's header
  if (mapfile->size() < 8 || memcmp(base, ARFILE_MAGIC,8))
    throw std::string("invalid signature for an archive file");
}

// This function loads the entire