/*
** Issue a warning and give the user an opportunity to abandon out
** if a Unicode (UTF-16) byte-order-mark (BOM) or a \r\n line ending
** is seen in a text file.
*/
static void commit_warning(const Blob *p, int crnlOk, const char *zFilename){
  int eType;              /* return value of looks_like_text() */
  int fUnicode;           /* return value of starts_with_utf16_bom() */
  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;
  eType = looks_like_text(p);
  fUnicode = starts_with_utf16_bom(p);
  if( eType==-1 || fUnicode ){
    const char *zWarning;
    Blob ans;
    char cReply;

    if( eType==-1 && fUnicode ){
      zWarning = "Unicode and CR/NL line endings";
    }else if( eType==-1 ){
      if( crnlOk ){
        return; /* We don't want CR/NL warnings for this file. */
      }
      zWarning = "CR/NL line endings";
    }else{
      zWarning = "Unicode";
    }
    file_relative_name(zFilename, &fname, 0);
    blob_zero(&ans);
    zMsg = mprintf(
         "%s contains %s; commit anyhow (a=all/y/N)?",
         blob_str(&fname), zWarning);
    prompt_user(zMsg, &ans);
    fossil_free(zMsg);
    cReply = blob_str(&ans)[0];
    if( cReply=='a' || cReply=='A' ){
      allOk = 1;
    }else if( cReply!='y' && cReply!='Y' ){
      fossil_fatal("Abandoning commit due to %s in %s",
                   zWarning, blob_str(&fname));
    }
    blob_reset(&ans);
    blob_reset(&fname);
  }
}

/*

*/
#define DIFF_CANNOT_COMPUTE_BINARY \
    "cannot compute difference between binary files\n"

#define DIFF_CANNOT_COMPUTE_SYMLINK \
    "cannot compute difference between symlink and regular file\n"

#define looks_like_binary(blob) (looks_like_text((blob)) == 0)
#endif /* INTERFACE */

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

** values are:
**
**  (1) -- The content appears to consist entirely of text, with lines
**         delimited by line-feed characters; however, the encoding may
**         not be UTF-8.
**
**  (0) -- The content appears to be binary because it contains embedded
**         NUL (\000) characters or an extremely long line.  Since this
**         function does not understand UTF-16, it may falsely consider
**         UTF-16 text to be binary.
**
** (-1) -- The content appears to consist entirely of text, with lines
**         delimited by carriage-return, line-feed pairs; however, the
**         encoding may not be UTF-8.
**


*/
int looks_like_text(const Blob *pContent){
  const char *z = blob_buffer(pContent);
  unsigned int n = blob_size(pContent);
  int j, c;
  int result = 1;  /* Assume text with no CR/NL */

  /* Check individual lines.
  */
  if( n==0 ) return result;  /* Empty file -> text */
  c = *z;
  if( c==0 ) return 0;  /* \000 byte in a file -> binary */







  j = (c!='\n');
  while( --n>0 ){
    c = *++z; ++j;
    if( c==0 ) return 0;  /* \000 byte in a file -> binary */
    if( c=='\n' ){
      if( z[-1]=='\r' ){
        result = -1;  /* Contains CR/NL, continue */
      }
      if( j>LENGTH_MASK ){
        return 0;  /* Very long line -> binary */
      }
      j = 0;
    }
  }
  if( j>LENGTH_MASK ){
    return 0;  /* Very long line -> binary */
  }
  return result;  /* No problems seen -> not binary */
}

/*
** This function returns non-zero if the blob starts with a UTF-16le or
** UTF-16be byte-order-mark (BOM).
*/
int starts_with_utf16_bom(const Blob *pContent){
  const char *z = blob_buffer(pContent);
  int c1, c2;

  if( blob_size(pContent)<2 ) return 0;
  c1 = z[0]; c2 = z[1];
  if( (c1==(char)0xff) && (c2==(char)0xfe) ){
    return 1;
  }else if( (c1==(char)0xff) && (c2==(char)0xfe) ){
    return 1;
  }
  return 0;
}

/*
** Return true if two DLine elements are identical.
*/
static int same_dline(DLine *pA, DLine *pB){
  return pA->h==pB->h && memcmp(pA->z,pB->z,pA->h & LENGTH_MASK)==0;
}