Function cmListFileLexer_ReadBOM

Source/LexerParser/cmListFileLexer.c:2740–2801

--------------------------------------------------------------------------*/

/*--------------------------------------------------------------------------*/
static cmListFileLexer_BOM cmListFileLexer_ReadBOM(FILE* f,
                                                   unsigned char readBuffer[4],
                                                   size_t* readSize)
{
  /* Read the up to four bytes that might correspond to a BOM. In case these
     bytes turn out not to represent a BOM, save them for later consumption in
     order to avoid seeking the file (which might not be seekable, e.g., if
     it's a pipe). */
  unsigned char* b = readBuffer;

  size_t n = fread(b, 1, 2, f);
  *readSize = n; /* Initialize first and then accumulate */

  if (n == 2) {
    if (b[0] == 0xEF && b[1] == 0xBB) {
      n = fread(b + 2, 1, 1, f);
      *readSize += n;

      if (n == 1) {
        if (b[2] == 0xBF) {
          *readSize = 0; /* We consumed the BOM: discard it */
          return cmListFileLexer_BOM_UTF8;
        }
      }
    } else if (b[0] == 0xFE && b[1] == 0xFF) {
      *readSize = 0; /* We consumed the BOM: discard it */
      /* UTF-16 BE */
      return cmListFileLexer_BOM_UTF16BE;
    } else if (b[0] == 0 && b[1] == 0) {
      n = fread(b + 2, 1, 2, f);
      *readSize += n;

      if (n == 2) {
        if (b[2] == 0xFE && b[3] == 0xFF) {
          *readSize = 0; /* We consumed the BOM: discard it */
          return cmListFileLexer_BOM_UTF32BE;
        }
      }
    } else if (b[0] == 0xFF && b[1] == 0xFE) {
      n = fread(b + 2, 1, 2, f);
      *readSize += n;

      if (n == 2 && b[2] == 0 && b[3] == 0) {
        *readSize = 0; /* We consumed the BOM: discard it */
        return cmListFileLexer_BOM_UTF32LE;
      }

      /* In case we were able to subsequently read only a single byte out of two
         (i.e., three in total), the file must be corrupt and the BOM cannot
         represent a UTF-16-LE BOM since each code unit must consist of two
         bytes. This avoids incorrectly detecting an incomplete UTF-32-LE BOM as
         UTF-16-LE input. */
      if (n % 2 == 0) {
        *readSize = n; /* We consumed the read bytes as BOM only partially */
        memmove(b, b + 2, n);
        return cmListFileLexer_BOM_UTF16LE;
      }
    }