Fossil

  const char *zFilename /* The full name of the file being committed. */
){
  int bReverse;           /* UTF-16 byte order is reversed? */
  int fUnicode;           /* return value of starts_with_utf16_bom() */
  int lookFlags;          /* output flags from looks_like_utf8/utf16() */
  int fHasNul;            /* the blob contains one or more NUL chars */
  int fHasCrLf;           /* the blob contains one or more CR/LF pairs */
  int fHasLong;           /* the blob contains an overly long line */
  char *zMsg;             /* Warning message */
  Blob fname;             /* Relative pathname of the file */
  static int allOk = 0;   /* Set to true to disable this routine */

  if( allOk ) return 0;
  fUnicode = could_be_utf16(p, &bReverse);
  if( fUnicode ){
    lookFlags = looks_like_utf16(p, bReverse);
  }else{
    lookFlags = looks_like_utf8(p);
  }
  fHasNul = (lookFlags & LOOK_NUL);
  fHasCrLf = (lookFlags & LOOK_CRLF);
  fHasLong = (lookFlags & LOOK_LONG);
  if( fHasNul || fHasLong || fHasCrLf || fUnicode ){
    const char *zWarning;
    const char *zDisable;
    const char *zConvert = "c=convert/";
    Blob ans;
    char cReply;

    if( fHasNul || fHasLong ){
      if( binOk ){
        return 0; /* We don't want binary warnings for this file. */
      }
      if( !fHasNul && fHasLong ){
        zWarning = "long lines";
      }else{
        zWarning = "binary data";
      }
      zDisable = "\"binary-glob\" setting";
      zConvert = ""; /* We cannot convert binary files. */
    }else if( fHasCrLf && fUnicode ){

#define LOOK_NONE    ((int)0x00000000) /* Nothing special was found. */
#define LOOK_NUL     ((int)0x00000001) /* One or more NUL chars were found. */
#define LOOK_CR      ((int)0x00000002) /* One or more CR chars were found. */
#define LOOK_LONE_CR ((int)0x00000004) /* An unpaired CR char was found. */
#define LOOK_LF      ((int)0x00000008) /* One or more LF chars were found. */
#define LOOK_LONE_LF ((int)0x00000010) /* An unpaired CR char was found. */
#define LOOK_CRLF    ((int)0x00000020) /* One or more CR/LF pairs were found. */
#define LOOK_LONG    ((int)0x00000040) /* An over length line was found. */
#define LOOK_ODD     ((int)0x00000080) /* An odd number of bytes was found. */
#define LOOK_SHORT   ((int)0x00000100) /* Unable to perform full check. */
#define LOOK_BINARY  (LOOK_NUL | LOOK_LONG | LOOK_SHORT) /* May be binary. */
#endif /* INTERFACE */

/*
** Maximum length of a line in a text file, in bytes.  (2**13 = 8192 bytes)
*/
#define LENGTH_MASK_SZ  13
#define LENGTH_MASK     ((1<<LENGTH_MASK_SZ)-1)

    int c2 = c;
    c = *++z; ++j;
    if( c==0 ){
      flags |= LOOK_NUL;  /* NUL character in a file -> binary */
    }else if( c=='\n' ){
      flags |= LOOK_LF;
      if( c2=='\r' ){
        flags |= (LOOK_CR | LOOK_CRLF);  /* Found LF preceded by CR */
      }else{
        flags |= LOOK_LONE_LF;
      }
      if( j>LENGTH_MASK ){
        flags |= LOOK_LONG;  /* Very long line -> binary */
      }
      j = 0;
    }else if( c=='\r' ){
      flags |= LOOK_CR;
      if( n<=1 || z[1]!='\n' ){
        flags |= LOOK_LONE_CR;  /* More chars, next char is not LF */
      }
    }
  }
  if( j>LENGTH_MASK ){
    flags |= LOOK_LONG;  /* Very long line -> binary */
  }
  return flags;
}

/*
** Define the type needed to represent a Unicode (UTF-16) character.
*/
#ifndef WCHAR_T
#  ifdef _WIN32
#    define WCHAR_T wchar_t
#  else
#    define WCHAR_T unsigned short
#  endif
#endif

/*
** Maximum length of a line in a text file, in UTF-16 characters.  (4096)
** The number of bytes represented by this value cannot exceed LENGTH_MASK
** bytes, because that is the line buffer size used by the diff engine.
*/
#define UTF16_LENGTH_MASK_SZ   (LENGTH_MASK_SZ-(sizeof(WCHAR_T)-sizeof(char)))
#define UTF16_LENGTH_MASK      ((1<<UTF16_LENGTH_MASK_SZ)-1)

/*
** This macro is used to swap the byte order of a UTF-16 character in the
** looks_like_utf16() function.
*/
#define UTF16_SWAP(ch)         ((((ch) << 8) & 0xFF00) | (((ch) >> 8) & 0xFF))
#define UTF16_SWAP_IF(expr,ch) ((expr) ? UTF16_SWAP((ch)) : (ch))

/*
** This function attempts to scan each logical line within the blob to
** determine the type of content it appears to contain.  The return value
** is a combination of one or more of the LOOK_XXX flags (see above):
**
** !LOOK_BINARY -- The content appears to consist entirely of text; however,

  if( n==0 ) return flags;  /* Empty file -> text */
  if( n%sizeof(WCHAR_T) ){
    flags |= LOOK_ODD;  /* Odd number of bytes -> binary (UTF-8?) */
    if( n<sizeof(WCHAR_T) ) return flags;  /* One byte -> binary (UTF-8?) */
  }
  c = *z;
  if( bReverse ){
    c = UTF16_SWAP(c);
  }
  if( c==0 ){
    flags |= LOOK_NUL;  /* NUL character in a file -> binary */
  }else if( c=='\r' ){
    flags |= LOOK_CR;
    if( n<=sizeof(WCHAR_T) || UTF16_SWAP_IF(bReverse, z[1])!='\n' ){
      flags |= LOOK_LONE_CR;  /* More chars, next char is not LF */
    }
  }
  j = (c!='\n');
  if( !j ) flags |= (LOOK_LF | LOOK_LONE_LF);  /* Found LF as first char */
  while( 1 ){
    int c2 = c;
    n -= sizeof(WCHAR_T);
    if( n<sizeof(WCHAR_T) ) break;
    c = *++z;
    if( bReverse ){
      c = UTF16_SWAP(c);
    }
    ++j;
    if( c==0 ){
      flags |= LOOK_NUL;  /* NUL character in a file -> binary */



    }else if( c=='\n' ){
      flags |= LOOK_LF;
      if( c2=='\r' ){
        flags |= (LOOK_CR | LOOK_CRLF);  /* Found LF preceded by CR */
      }else{
        flags |= LOOK_LONE_LF;
      }
      if( j>UTF16_LENGTH_MASK ){
        flags |= LOOK_LONG;  /* Very long line -> binary */
      }
      j = 0;
    }else if( c=='\r' ){
      flags |= LOOK_CR;
      if( n<=sizeof(WCHAR_T) || UTF16_SWAP_IF(bReverse, z[1])!='\n' ){
        flags |= LOOK_LONE_CR;  /* More chars, next char is not LF */
      }
    }


  }
  if( j>UTF16_LENGTH_MASK ){
    flags |= LOOK_LONG;  /* Very long line -> binary */
  }
  return flags;
}

/*
** This function returns an array of bytes representing the byte-order-mark
** for UTF-8.

*/
void looks_like_utf_test_cmd(void){
  Blob blob;     /* the contents of the specified file */
  int fUtf8;     /* return value of starts_with_utf8_bom() */
  int fUtf16;    /* return value of starts_with_utf16_bom() */
  int fUnicode;  /* return value of could_be_utf16() */
  int lookFlags; /* output flags from looks_like_utf8/utf16() */
  int bRevUtf16 = 0; /* non-zero -> UTF-16 byte order reversed */
  int bRevUnicode = 0; /* non-zero -> UTF-16 byte order reversed */
  if( g.argc!=3 ) usage("FILENAME");
  blob_read_from_file(&blob, g.argv[2]);
  fUtf8 = starts_with_utf8_bom(&blob, 0);
  fUtf16 = starts_with_utf16_bom(&blob, 0, &bRevUtf16);
  fUnicode = could_be_utf16(&blob, &bRevUnicode);
  lookFlags = fUnicode ? looks_like_utf16(&blob, bRevUnicode) :
                         looks_like_utf8(&blob);
  fossil_print("File \"%s\" has %d bytes.\n",g.argv[2],blob_size(&blob));
  fossil_print("Starts with UTF-8 BOM: %s\n",fUtf8?"yes":"no");
  fossil_print("Starts with UTF-16 BOM: %s\n",
               fUtf16?(bRevUtf16?"reversed":"yes"):"no");
  fossil_print("Looks like UTF-%s: %s\n",fUnicode?"16":"8",
               (lookFlags&LOOK_BINARY)?"no":"yes");
  fossil_print("Has flag LOOK_NUL: %s\n",(lookFlags&LOOK_NUL)?"yes":"no");
  fossil_print("Has flag LOOK_CR: %s\n",(lookFlags&LOOK_CR)?"yes":"no");
  fossil_print("Has flag LOOK_LONE_CR: %s\n",
               (lookFlags&LOOK_LONE_CR)?"yes":"no");
  fossil_print("Has flag LOOK_LF: %s\n",(lookFlags&LOOK_LF)?"yes":"no");
  fossil_print("Has flag LOOK_LONE_LF: %s\n",
               (lookFlags&LOOK_LONE_LF)?"yes":"no");
  fossil_print("Has flag LOOK_CRLF: %s\n",(lookFlags&LOOK_CRLF)?"yes":"no");
  fossil_print("Has flag LOOK_LONG: %s\n",(lookFlags&LOOK_LONG)?"yes":"no");
  fossil_print("Has flag LOOK_ODD: %s\n",(lookFlags&LOOK_ODD)?"yes":"no");
  fossil_print("Has flag LOOK_SHORT: %s\n",(lookFlags&LOOK_SHORT)?"yes":"no");
  blob_reset(&blob);
}