comparison src/coding.c @ 17835:f36ffb6f1208

Name change through the code: coding-category-internal => coding-category-emacs-mule, XXX_coding_internal => XXX_coding_emacs_mule, coding_type_internal => coding_type_emacs_mule, coding_type_automatic => coding_type_undecided, CODING_CATEGORY_MASK_INTERNAL => CODING_CATEGORY_MASK_EMACS_MULE, CODING_CATEGORY_IDX_INTERNAL => CODING_CATEGORY_IDX_EMACS_MULE, CODING_EOL_AUTOMATIC => CODING_EOL_UNDECIDED.

17834:e154b943bbba

Boston, MA 02111-1307, USA.  */

/*** TABLE OF CONTENTS ***

  1. Preamble
  2. Emacs' internal format handlers
  3. ISO2022 handlers
  4. Shift-JIS and BIG5 handlers
  5. End-of-line handlers
  6. C library functions
  7. Emacs Lisp library functions

/*** GENERAL NOTE on CODING SYSTEM ***

  Coding system is an encoding mechanism of one or more character
  sets.  Here's a list of coding systems which Emacs can handle.  When
  we say "decode", it means converting some other coding system to
  Emacs' internal format, and when we say "encode", it means
  converting Emacs' internal format to some other coding system.

  0. Emacs' internal format

  Emacs itself holds a multi-lingual character in a buffer and a string
  in a special format.  Details are described in the section 2.

  1. ISO2022

  which appropriate flag bits for the category XXX is set.  The flag
  bits are defined in macros CODING_CATEGORY_MASK_XXX.  Below is the
  template of these functions.  */
#if 0
int
detect_coding_internal (src, src_end)
     unsigned char *src, *src_end;
{
  ...
}
#endif

/*** GENERAL NOTES on `decode_coding_XXX ()' functions ***

  These functions decode SRC_BYTES length text at SOURCE encoded in
  CODING to Emacs' internal format.  The resulting text goes to a
  place pointed by DESTINATION, the length of which should not exceed
  DST_BYTES.  The bytes actually processed is returned as *CONSUMED.
  The return value is the length of the decoded text.  Below is a
  template of these functions.  */
#if 0
decode_coding_XXX (coding, source, destination, src_bytes, dst_bytes, consumed)
     struct coding_system *coding;
     unsigned char *source, *destination;
     int src_bytes, dst_bytes;

}
#endif

/*** GENERAL NOTES on `encode_coding_XXX ()' functions ***

  These functions encode SRC_BYTES length text at SOURCE of Emacs
  internal format to CODING.  The resulting text goes to a place
  pointed by DESTINATION, the length of which should not exceed
  DST_BYTES.  The bytes actually processed is returned as *CONSUMED.
  The return value is the length of the encoded text.  Below is a
  template of these functions.  */
#if 0
encode_coding_XXX (coding, source, destination, src_bytes, dst_bytes, consumed)
     struct coding_system *coding;
     unsigned char *source, *destination;
     int src_bytes, dst_bytes;

/* Table of coding-systems currently assigned to each coding-category.  */
Lisp_Object coding_category_table[CODING_CATEGORY_IDX_MAX];

/* Table of names of symbol for each coding-category.  */
char *coding_category_name[CODING_CATEGORY_IDX_MAX] = {
  "coding-category-internal",
  "coding-category-sjis",
  "coding-category-iso-7",
  "coding-category-iso-8-1",
  "coding-category-iso-8-2",
  "coding-category-iso-else",

/* Alist of charsets vs revision number.  */
Lisp_Object Vcharset_revision_alist;


/*** 2. Emacs internal format handlers ***/

/* Emacs' internal format for encoding multiple character sets is a
   kind of multi-byte encoding, i.e. encoding a character by a sequence
   of one-byte codes of variable length.  ASCII characters and control
   characters (e.g. `tab', `newline') are represented by one-byte as

      return 0;		       	\
  } while (0)

/* See the above "GENERAL NOTES on `detect_coding_XXX ()' functions".
   Check if a text is encoded in Emacs' internal format.  If it is,
   return CODING_CATEGORY_MASK_INTERNAL, else return 0.  */

int
detect_coding_internal (src, src_end)
     unsigned char *src, *src_end;
{
  unsigned char c;
  int composing = 0;

	default:
	label_end_of_switch:
	  break;
	}
    }
  return CODING_CATEGORY_MASK_INTERNAL;
}


/*** 3. ISO2022 handlers ***/

	  if (coding->flags & CODING_FLAG_ISO_INIT_AT_BOL)
	    bcopy (coding->spec.iso2022.initial_designation,
		   coding->spec.iso2022.current_designation,
		   sizeof coding->spec.iso2022.initial_designation);
	  if (coding->eol_type == CODING_EOL_LF
	      || coding->eol_type == CODING_EOL_AUTOMATIC)
	    *dst++ = ISO_CODE_LF;
	  else if (coding->eol_type == CODING_EOL_CRLF)
	    *dst++ = ISO_CODE_CR, *dst++ = ISO_CODE_LF;
	  else
	    *dst++ = ISO_CODE_CR;

	    }
	  /* fall down to treat '\r' as '\n' ...  */

	case EMACS_linefeed_code:
	  if (coding->eol_type == CODING_EOL_LF
	      || coding->eol_type == CODING_EOL_AUTOMATIC)
	    *dst++ = '\n';
	  else if (coding->eol_type == CODING_EOL_CRLF)
	    *dst++ = '\r', *dst++ = '\n';
	  else
	    *dst++ = '\r';

    return 0;

  switch (coding->eol_type)
    {
    case CODING_EOL_LF:
    case CODING_EOL_AUTOMATIC:
      produced = (src_bytes > dst_bytes) ? dst_bytes : src_bytes;
      bcopy (source, destination, produced);
      if (coding->selective)
	{
	  int i = produced;

   from that symbol.

   `element[0]' contains information to be set in `coding->type'.  The
   value and its meaning is as follows:

   0 -- coding_system_internal
   1 -- coding_system_sjis
   2 -- coding_system_iso2022
   3 -- coding_system_big5
   4 -- coding_system_ccl
   nil -- coding_system_no_conversion
   t -- coding_system_automatic

   `element[4]' contains information to be set in `coding->flags' and
   `coding->spec'.  The meaning varies by `coding->type'.

   If `coding->type' is `coding_type_iso2022', element[4] is a vector

  if (!VECTORP (coding_system)
      || XVECTOR (coding_system)->size != 5)
    goto label_invalid_coding_system;

  if (VECTORP (eol_type))
    coding->eol_type = CODING_EOL_AUTOMATIC;
  else if (XFASTINT (eol_type) == 1)
    coding->eol_type = CODING_EOL_CRLF;
  else if (XFASTINT (eol_type) == 2)
    coding->eol_type = CODING_EOL_CR;
  else

  type = XVECTOR (coding_system)->contents[0];
  switch (XFASTINT (type))
    {
    case 0:
      coding->type = coding_type_internal;
      break;

    case 1:
      coding->type = coding_type_sjis;
      break;

      coding->require_flushing = 1;
      break;

    default:
      if (EQ (type, Qt))
	coding->type = coding_type_automatic;
      else
	coding->type = coding_type_no_conversion;
      break;
    }
  return 0;

   is one of Emacs' internal format, ISO2022, SJIS, and BIG5.  But,
   it's impossible to distinguish some coding systems accurately
   because they use the same range of codes.  So, at first, coding
   systems are categorized into 7, those are:

   o coding-category-internal

   	The category for a coding system which has the same code range
	as Emacs' internal format.  Assigned the coding-system (Lisp
	symbol) `internal' by default.

   o coding-category-sjis

	The category for a coding system which has the same code range
	as SJIS.  Assigned the coding-system (Lisp

    /* C is an ISO2022 specific control code of C1,
       or the first byte of SJIS's 2-byte character code,
       or a leading code of Emacs.  */
    mask = (detect_coding_iso2022 (src, src_end)
	    | detect_coding_sjis (src, src_end)
	    | detect_coding_internal (src, src_end));

  else if (c < 0xA0)
    /* C is the first byte of SJIS character code,
       or a leading-code of Emacs.  */
    mask = (detect_coding_sjis (src, src_end)
	    | detect_coding_internal (src, src_end));

  else
    /* C is a character of ISO2022 in graphic plane right,
       or a SJIS's 1-byte character code (i.e. JISX0201),
       or the first byte of BIG5's 2-byte code.  */

  setup_coding_system (XSYMBOL (coding_category_table[idx])->value, coding);
}

/* Detect how end-of-line of a text of length SRC_BYTES pointed by SRC
   is encoded.  Return one of CODING_EOL_LF, CODING_EOL_CRLF,
   CODING_EOL_CR, and CODING_EOL_AUTOMATIC.  */

int
detect_eol_type (src, src_bytes)
     unsigned char *src;
     int src_bytes;

	    return CODING_EOL_CRLF;
	  else
	    return CODING_EOL_CR;
	}
    }
  return CODING_EOL_AUTOMATIC;
}

/* Detect how end-of-line of a text of length SRC_BYTES pointed by SRC
   is encoded.  If it detects an appropriate format of end-of-line, it
   sets the information in *CODING.  */

     int src_bytes;
{
  Lisp_Object val;
  int eol_type = detect_eol_type (src, src_bytes);

  if (eol_type == CODING_EOL_AUTOMATIC)
    /*  We found no end-of-line in the source text.  */
    return;

  val = Fget (coding->symbol, Qeol_type);
  if (VECTORP (val) && XVECTOR (val)->size == 3)

    {
      *consumed = 0;
      return 0;
    }

  if (coding->type == coding_type_automatic)
    detect_coding (coding, source, src_bytes);

  if (coding->eol_type == CODING_EOL_AUTOMATIC)
    detect_eol (coding, source, src_bytes);

  coding->carryover_size = 0;
  switch (coding->type)
    {

      produced = (src_bytes > dst_bytes) ? dst_bytes : src_bytes;
      bcopy (source, destination, produced);
      *consumed = produced;
      break;

    case coding_type_internal:
    case coding_type_automatic:
      if (coding->eol_type == CODING_EOL_LF
	  ||  coding->eol_type == CODING_EOL_AUTOMATIC)
	goto label_no_conversion;
      produced = decode_eol (coding, source, destination,
			     src_bytes, dst_bytes, consumed);
      break;

	    }
	}
      *consumed = produced;
      break;

    case coding_type_internal:
    case coding_type_automatic:
      if (coding->eol_type == CODING_EOL_LF
	  ||  coding->eol_type == CODING_EOL_AUTOMATIC)
	goto label_no_conversion;
      produced = encode_eol (coding, source, destination,
			     src_bytes, dst_bytes, consumed);
      break;

DEFUN ("detect-coding-region", Fdetect_coding_region, Sdetect_coding_region,
       2, 2, 0,
  "Detect coding-system of the text in the region between START and END.\n\
Return a list of possible coding-systems ordered by priority.\n\
If only ASCII characters are found, it returns `automatic-conversion'\n\
 or its subsidiary coding-system according to a detected end-of-line format.")
  (b, e)
     Lisp_Object b, e;
{
  int coding_mask, eol_type;

  coding_mask = detect_coding_mask (POS_ADDR (beg), end - beg);
  eol_type  = detect_eol_type (POS_ADDR (beg), end - beg);

  if (coding_mask == CODING_CATEGORY_MASK_ANY)
    {
      val = intern ("automatic-conversion");
      if (eol_type != CODING_EOL_AUTOMATIC)
	{
	  Lisp_Object val2 = Fget (val, Qeol_type);
	  if (VECTORP (val2))
	    val = XVECTOR (val2)->contents[eol_type];
	}

	 coding-systems.  */
      val2 = val;
      val = Qnil;
      for (; !NILP (val2); val2 = XCONS (val2)->cdr)
	{
	  if (eol_type == CODING_EOL_AUTOMATIC)
	    val = Fcons (XCONS (val2)->car, val);
	  else
	    {
	      Lisp_Object val3 = Fget (XCONS (val2)->car, Qeol_type);
	      if (VECTORP (val3))

     int encodep;
{
  register unsigned char *beg_addr = *begp, *end_addr = *endp;

  if (coding->eol_type != CODING_EOL_LF
      && coding->eol_type != CODING_EOL_AUTOMATIC)
    /* Since we anyway have to convert end-of-line format, it is not
       worth skipping at most 100 bytes or so.  */
    return;

  if (encodep)			/* for encoding */
    {
      switch (coding->type)
	{
	case coding_type_no_conversion:
	case coding_type_internal:
	case coding_type_automatic:
	  /* We need no conversion.  */
	  *begp = *endp;
	  return;
	case coding_type_ccl:
	  /* We can't skip any data.  */

	{
	case coding_type_no_conversion:
	  /* We need no conversion.  */
	  *begp = *endp;
	  return;
	case coding_type_internal:
	  if (coding->eol_type == CODING_EOL_LF)
	    {
	      /* We need no conversion.  */
	      *begp = *endp;
	      return;

init_coding_once ()
{
  int i;

  /* Emacs internal format specific initialize routine.  */ 
  for (i = 0; i <= 0x20; i++)
    emacs_code_class[i] = EMACS_control_code;
  emacs_code_class[0x0A] = EMACS_linefeed_code;
  emacs_code_class[0x0D] = EMACS_carriage_return_code;
  for (i = 0x21 ; i < 0x7F; i++)

17835:f36ffb6f1208

Boston, MA 02111-1307, USA.  */

/*** TABLE OF CONTENTS ***

  1. Preamble
  2. Emacs' internal format (emacs-mule) handlers
  3. ISO2022 handlers
  4. Shift-JIS and BIG5 handlers
  5. End-of-line handlers
  6. C library functions
  7. Emacs Lisp library functions

/*** GENERAL NOTE on CODING SYSTEM ***

  Coding system is an encoding mechanism of one or more character
  sets.  Here's a list of coding systems which Emacs can handle.  When
  we say "decode", it means converting some other coding system to
  Emacs' internal format (emacs-internal), and when we say "encode",
  it means converting the coding system emacs-mule to some other
  coding system.

  0. Emacs' internal format (emacs-mule)

  Emacs itself holds a multi-lingual character in a buffer and a string
  in a special format.  Details are described in the section 2.

  1. ISO2022

  which appropriate flag bits for the category XXX is set.  The flag
  bits are defined in macros CODING_CATEGORY_MASK_XXX.  Below is the
  template of these functions.  */
#if 0
int
detect_coding_emacs_mule (src, src_end)
     unsigned char *src, *src_end;
{
  ...
}
#endif

/*** GENERAL NOTES on `decode_coding_XXX ()' functions ***

  These functions decode SRC_BYTES length text at SOURCE encoded in
  CODING to Emacs' internal format (emacs-mule).  The resulting text
  goes to a place pointed by DESTINATION, the length of which should
  not exceed DST_BYTES.  The bytes actually processed is returned as
  *CONSUMED.  The return value is the length of the decoded text.
  Below is a template of these functions.  */
#if 0
decode_coding_XXX (coding, source, destination, src_bytes, dst_bytes, consumed)
     struct coding_system *coding;
     unsigned char *source, *destination;
     int src_bytes, dst_bytes;

}
#endif

/*** GENERAL NOTES on `encode_coding_XXX ()' functions ***

  These functions encode SRC_BYTES length text at SOURCE of Emacs'
  internal format (emacs-mule) to CODING.  The resulting text goes to
  a place pointed by DESTINATION, the length of which should not
  exceed DST_BYTES.  The bytes actually processed is returned as
  *CONSUMED.  The return value is the length of the encoded text.
  Below is a template of these functions.  */
#if 0
encode_coding_XXX (coding, source, destination, src_bytes, dst_bytes, consumed)
     struct coding_system *coding;
     unsigned char *source, *destination;
     int src_bytes, dst_bytes;

/* Table of coding-systems currently assigned to each coding-category.  */
Lisp_Object coding_category_table[CODING_CATEGORY_IDX_MAX];

/* Table of names of symbol for each coding-category.  */
char *coding_category_name[CODING_CATEGORY_IDX_MAX] = {
  "coding-category-emacs-mule",
  "coding-category-sjis",
  "coding-category-iso-7",
  "coding-category-iso-8-1",
  "coding-category-iso-8-2",
  "coding-category-iso-else",

/* Alist of charsets vs revision number.  */
Lisp_Object Vcharset_revision_alist;


/*** 2. Emacs internal format (emacs-mule) handlers ***/

/* Emacs' internal format for encoding multiple character sets is a
   kind of multi-byte encoding, i.e. encoding a character by a sequence
   of one-byte codes of variable length.  ASCII characters and control
   characters (e.g. `tab', `newline') are represented by one-byte as

      return 0;		       	\
  } while (0)

/* See the above "GENERAL NOTES on `detect_coding_XXX ()' functions".
   Check if a text is encoded in Emacs' internal format.  If it is,
   return CODING_CATEGORY_MASK_EMASC_MULE, else return 0.  */

int
detect_coding_emacs_mule (src, src_end)
     unsigned char *src, *src_end;
{
  unsigned char c;
  int composing = 0;

	default:
	label_end_of_switch:
	  break;
	}
    }
  return CODING_CATEGORY_MASK_EMACS_MULE;
}


/*** 3. ISO2022 handlers ***/

	  if (coding->flags & CODING_FLAG_ISO_INIT_AT_BOL)
	    bcopy (coding->spec.iso2022.initial_designation,
		   coding->spec.iso2022.current_designation,
		   sizeof coding->spec.iso2022.initial_designation);
	  if (coding->eol_type == CODING_EOL_LF
	      || coding->eol_type == CODING_EOL_UNDECIDED)
	    *dst++ = ISO_CODE_LF;
	  else if (coding->eol_type == CODING_EOL_CRLF)
	    *dst++ = ISO_CODE_CR, *dst++ = ISO_CODE_LF;
	  else
	    *dst++ = ISO_CODE_CR;

	    }
	  /* fall down to treat '\r' as '\n' ...  */

	case EMACS_linefeed_code:
	  if (coding->eol_type == CODING_EOL_LF
	      || coding->eol_type == CODING_EOL_UNDECIDED)
	    *dst++ = '\n';
	  else if (coding->eol_type == CODING_EOL_CRLF)
	    *dst++ = '\r', *dst++ = '\n';
	  else
	    *dst++ = '\r';

    return 0;

  switch (coding->eol_type)
    {
    case CODING_EOL_LF:
    case CODING_EOL_UNDECIDED:
      produced = (src_bytes > dst_bytes) ? dst_bytes : src_bytes;
      bcopy (source, destination, produced);
      if (coding->selective)
	{
	  int i = produced;

   from that symbol.

   `element[0]' contains information to be set in `coding->type'.  The
   value and its meaning is as follows:

   0 -- coding_type_emacs_mule
   1 -- coding_type_sjis
   2 -- coding_type_iso2022
   3 -- coding_type_big5
   4 -- coding_type_ccl encoder/decoder written in CCL
   nil -- coding_type_no_conversion
   t -- coding_type_undecided (automatic conversion on decoding,
   			       no-conversion on encoding)

   `element[4]' contains information to be set in `coding->flags' and
   `coding->spec'.  The meaning varies by `coding->type'.

   If `coding->type' is `coding_type_iso2022', element[4] is a vector

  if (!VECTORP (coding_system)
      || XVECTOR (coding_system)->size != 5)
    goto label_invalid_coding_system;

  if (VECTORP (eol_type))
    coding->eol_type = CODING_EOL_UNDECIDED;
  else if (XFASTINT (eol_type) == 1)
    coding->eol_type = CODING_EOL_CRLF;
  else if (XFASTINT (eol_type) == 2)
    coding->eol_type = CODING_EOL_CR;
  else

  type = XVECTOR (coding_system)->contents[0];
  switch (XFASTINT (type))
    {
    case 0:
      coding->type = coding_type_emacs_mule;
      break;

    case 1:
      coding->type = coding_type_sjis;
      break;

      coding->require_flushing = 1;
      break;

    default:
      if (EQ (type, Qt))
	coding->type = coding_type_undecided;
      else
	coding->type = coding_type_no_conversion;
      break;
    }
  return 0;

   is one of Emacs' internal format, ISO2022, SJIS, and BIG5.  But,
   it's impossible to distinguish some coding systems accurately
   because they use the same range of codes.  So, at first, coding
   systems are categorized into 7, those are:

   o coding-category-emacs-mule

   	The category for a coding system which has the same code range
	as Emacs' internal format.  Assigned the coding-system (Lisp
	symbol) `emacs-mule' by default.

   o coding-category-sjis

	The category for a coding system which has the same code range
	as SJIS.  Assigned the coding-system (Lisp

    /* C is an ISO2022 specific control code of C1,
       or the first byte of SJIS's 2-byte character code,
       or a leading code of Emacs.  */
    mask = (detect_coding_iso2022 (src, src_end)
	    | detect_coding_sjis (src, src_end)
	    | detect_coding_emacs_mule (src, src_end));

  else if (c < 0xA0)
    /* C is the first byte of SJIS character code,
       or a leading-code of Emacs.  */
    mask = (detect_coding_sjis (src, src_end)
	    | detect_coding_emacs_mule (src, src_end));

  else
    /* C is a character of ISO2022 in graphic plane right,
       or a SJIS's 1-byte character code (i.e. JISX0201),
       or the first byte of BIG5's 2-byte code.  */

  setup_coding_system (XSYMBOL (coding_category_table[idx])->value, coding);
}

/* Detect how end-of-line of a text of length SRC_BYTES pointed by SRC
   is encoded.  Return one of CODING_EOL_LF, CODING_EOL_CRLF,
   CODING_EOL_CR, and CODING_EOL_UNDECIDED.  */

int
detect_eol_type (src, src_bytes)
     unsigned char *src;
     int src_bytes;

	    return CODING_EOL_CRLF;
	  else
	    return CODING_EOL_CR;
	}
    }
  return CODING_EOL_UNDECIDED;
}

/* Detect how end-of-line of a text of length SRC_BYTES pointed by SRC
   is encoded.  If it detects an appropriate format of end-of-line, it
   sets the information in *CODING.  */

     int src_bytes;
{
  Lisp_Object val;
  int eol_type = detect_eol_type (src, src_bytes);

  if (eol_type == CODING_EOL_UNDECIDED)
    /*  We found no end-of-line in the source text.  */
    return;

  val = Fget (coding->symbol, Qeol_type);
  if (VECTORP (val) && XVECTOR (val)->size == 3)

    {
      *consumed = 0;
      return 0;
    }

  if (coding->type == coding_type_undecided)
    detect_coding (coding, source, src_bytes);

  if (coding->eol_type == CODING_EOL_UNDECIDED)
    detect_eol (coding, source, src_bytes);

  coding->carryover_size = 0;
  switch (coding->type)
    {

      produced = (src_bytes > dst_bytes) ? dst_bytes : src_bytes;
      bcopy (source, destination, produced);
      *consumed = produced;
      break;

    case coding_type_emacs_mule:
    case coding_type_undecided:
      if (coding->eol_type == CODING_EOL_LF
	  ||  coding->eol_type == CODING_EOL_UNDECIDED)
	goto label_no_conversion;
      produced = decode_eol (coding, source, destination,
			     src_bytes, dst_bytes, consumed);
      break;

	    }
	}
      *consumed = produced;
      break;

    case coding_type_emacs_mule:
    case coding_type_undecided:
      if (coding->eol_type == CODING_EOL_LF
	  ||  coding->eol_type == CODING_EOL_UNDECIDED)
	goto label_no_conversion;
      produced = encode_eol (coding, source, destination,
			     src_bytes, dst_bytes, consumed);
      break;

DEFUN ("detect-coding-region", Fdetect_coding_region, Sdetect_coding_region,
       2, 2, 0,
  "Detect coding-system of the text in the region between START and END.\n\
Return a list of possible coding-systems ordered by priority.\n\
If only ASCII characters are found, it returns `undecided'\n\
 or its subsidiary coding-system according to a detected end-of-line format.")
  (b, e)
     Lisp_Object b, e;
{
  int coding_mask, eol_type;

  coding_mask = detect_coding_mask (POS_ADDR (beg), end - beg);
  eol_type  = detect_eol_type (POS_ADDR (beg), end - beg);

  if (coding_mask == CODING_CATEGORY_MASK_ANY)
    {
      val = intern ("undecided");
      if (eol_type != CODING_EOL_UNDECIDED)
	{
	  Lisp_Object val2 = Fget (val, Qeol_type);
	  if (VECTORP (val2))
	    val = XVECTOR (val2)->contents[eol_type];
	}

	 coding-systems.  */
      val2 = val;
      val = Qnil;
      for (; !NILP (val2); val2 = XCONS (val2)->cdr)
	{
	  if (eol_type == CODING_EOL_UNDECIDED)
	    val = Fcons (XCONS (val2)->car, val);
	  else
	    {
	      Lisp_Object val3 = Fget (XCONS (val2)->car, Qeol_type);
	      if (VECTORP (val3))

     int encodep;
{
  register unsigned char *beg_addr = *begp, *end_addr = *endp;

  if (coding->eol_type != CODING_EOL_LF
      && coding->eol_type != CODING_EOL_UNDECIDED)
    /* Since we anyway have to convert end-of-line format, it is not
       worth skipping at most 100 bytes or so.  */
    return;

  if (encodep)			/* for encoding */
    {
      switch (coding->type)
	{
	case coding_type_no_conversion:
	case coding_type_emacs_mule:
	case coding_type_undecided:
	  /* We need no conversion.  */
	  *begp = *endp;
	  return;
	case coding_type_ccl:
	  /* We can't skip any data.  */

	{
	case coding_type_no_conversion:
	  /* We need no conversion.  */
	  *begp = *endp;
	  return;
	case coding_type_emacs_mule:
	  if (coding->eol_type == CODING_EOL_LF)
	    {
	      /* We need no conversion.  */
	      *begp = *endp;
	      return;

init_coding_once ()
{
  int i;

  /* Emacs' internal format specific initialize routine.  */ 
  for (i = 0; i <= 0x20; i++)
    emacs_code_class[i] = EMACS_control_code;
  emacs_code_class[0x0A] = EMACS_linefeed_code;
  emacs_code_class[0x0D] = EMACS_carriage_return_code;
  for (i = 0x21 ; i < 0x7F; i++)