xmlrole.c

一个用visual studio C++开发的XML文件解析和生成程序.

/*
Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
See the file copying.txt for copying permission.
*/

#include "xmldef.h"
#include "xmlrole.h"
#include "ascii.h"

/* Doesn't check:

 that ,| are not mixed in a model group
 content of literals

*/

static const char KW_ANY[] = { ASCII_A, ASCII_N, ASCII_Y, '\0' };
static const char KW_ATTLIST[] = { ASCII_A, ASCII_T, ASCII_T, ASCII_L, ASCII_I, ASCII_S, ASCII_T, '\0' };
static const char KW_CDATA[] = { ASCII_C, ASCII_D, ASCII_A, ASCII_T, ASCII_A, '\0' };
static const char KW_DOCTYPE[] = { ASCII_D, ASCII_O, ASCII_C, ASCII_T, ASCII_Y, ASCII_P, ASCII_E, '\0' };
static const char KW_ELEMENT[] = { ASCII_E, ASCII_L, ASCII_E, ASCII_M, ASCII_E, ASCII_N, ASCII_T, '\0' };
static const char KW_EMPTY[] = { ASCII_E, ASCII_M, ASCII_P, ASCII_T, ASCII_Y, '\0' };
static const char KW_ENTITIES[] = { ASCII_E, ASCII_N, ASCII_T, ASCII_I, ASCII_T, ASCII_I, ASCII_E, ASCII_S, '\0' };
static const char KW_ENTITY[] = { ASCII_E, ASCII_N, ASCII_T, ASCII_I, ASCII_T, ASCII_Y, '\0' };
static const char KW_FIXED[] = { ASCII_F, ASCII_I, ASCII_X, ASCII_E, ASCII_D, '\0' };
static const char KW_ID[] = { ASCII_I, ASCII_D, '\0' };
static const char KW_IDREF[] = { ASCII_I, ASCII_D, ASCII_R, ASCII_E, ASCII_F, '\0' };
static const char KW_IDREFS[] = { ASCII_I, ASCII_D, ASCII_R, ASCII_E, ASCII_F, ASCII_S, '\0' };
static const char KW_IGNORE[] = { ASCII_I, ASCII_G, ASCII_N, ASCII_O, ASCII_R, ASCII_E, '\0' };
static const char KW_IMPLIED[] = { ASCII_I, ASCII_M, ASCII_P, ASCII_L, ASCII_I, ASCII_E, ASCII_D, '\0' };
static const char KW_INCLUDE[] = { ASCII_I, ASCII_N, ASCII_C, ASCII_L, ASCII_U, ASCII_D, ASCII_E, '\0' };
static const char KW_NDATA[] = { ASCII_N, ASCII_D, ASCII_A, ASCII_T, ASCII_A, '\0' };
static const char KW_NMTOKEN[] = { ASCII_N, ASCII_M, ASCII_T, ASCII_O, ASCII_K, ASCII_E, ASCII_N, '\0' };
static const char KW_NMTOKENS[] = { ASCII_N, ASCII_M, ASCII_T, ASCII_O, ASCII_K, ASCII_E, ASCII_N, ASCII_S, '\0' };
static const char KW_NOTATION[] = { ASCII_N, ASCII_O, ASCII_T, ASCII_A, ASCII_T, ASCII_I, ASCII_O, ASCII_N, '\0' };
static const char KW_PCDATA[] = { ASCII_P, ASCII_C, ASCII_D, ASCII_A, ASCII_T, ASCII_A, '\0' };
static const char KW_PUBLIC[] = { ASCII_P, ASCII_U, ASCII_B, ASCII_L, ASCII_I, ASCII_C, '\0' };
static const char KW_REQUIRED[] = { ASCII_R, ASCII_E, ASCII_Q, ASCII_U, ASCII_I, ASCII_R, ASCII_E, ASCII_D, '\0' };
static const char KW_SYSTEM[] = { ASCII_S, ASCII_Y, ASCII_S, ASCII_T, ASCII_E, ASCII_M, '\0' };

#ifndef MIN_BYTES_PER_CHAR
#define MIN_BYTES_PER_CHAR(enc) ((enc)->minBytesPerChar)
#endif

#ifdef XML_DTD
#define setTopLevel(state) \
  ((state)->handler = ((state)->documentEntity \
                       ? internalSubset \
                       : externalSubset1))
#else /* not XML_DTD */
#define setTopLevel(state) ((state)->handler = internalSubset)
#endif /* not XML_DTD */

typedef int PROLOG_HANDLER(PROLOG_STATE *state,
			   int tok,
			   const char *ptr,
			   const char *end,
			   const ENCODING *enc);

static PROLOG_HANDLER
  prolog0, prolog1, prolog2,
  doctype0, doctype1, doctype2, doctype3, doctype4, doctype5,
  internalSubset,
  entity0, entity1, entity2, entity3, entity4, entity5, entity6,
  entity7, entity8, entity9,
  notation0, notation1, notation2, notation3, notation4,
  attlist0, attlist1, attlist2, attlist3, attlist4, attlist5, attlist6,
  attlist7, attlist8, attlist9,
  element0, element1, element2, element3, element4, element5, element6,
  element7,
#ifdef XML_DTD
  externalSubset0, externalSubset1,
  condSect0, condSect1, condSect2,
#endif /* XML_DTD */
  declClose,
  error;

static
int common(PROLOG_STATE *state, int tok);

static
int prolog0(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    state->handler = prolog1;
    return XML_ROLE_NONE;
  case XML_TOK_XML_DECL:
    state->handler = prolog1;
    return XML_ROLE_XML_DECL;
  case XML_TOK_PI:
    state->handler = prolog1;
    return XML_ROLE_NONE;
  case XML_TOK_COMMENT:
    state->handler = prolog1;
  case XML_TOK_BOM:
    return XML_ROLE_NONE;
  case XML_TOK_DECL_OPEN:
    if (!XmlNameMatchesAscii(enc,
			     ptr + 2 * MIN_BYTES_PER_CHAR(enc),
			     end,
			     KW_DOCTYPE))
      break;
    state->handler = doctype0;
    return XML_ROLE_NONE;
  case XML_TOK_INSTANCE_START:
    state->handler = error;
    return XML_ROLE_INSTANCE_START;
  }
  return common(state, tok);
}

static
int prolog1(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_PI:
  case XML_TOK_COMMENT:
  case XML_TOK_BOM:
    return XML_ROLE_NONE;
  case XML_TOK_DECL_OPEN:
    if (!XmlNameMatchesAscii(enc,
			     ptr + 2 * MIN_BYTES_PER_CHAR(enc),
			     end,
			     KW_DOCTYPE))
      break;
    state->handler = doctype0;
    return XML_ROLE_NONE;
  case XML_TOK_INSTANCE_START:
    state->handler = error;
    return XML_ROLE_INSTANCE_START;
  }
  return common(state, tok);
}

static
int prolog2(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_PI:
  case XML_TOK_COMMENT:
    return XML_ROLE_NONE;
  case XML_TOK_INSTANCE_START:
    state->handler = error;
    return XML_ROLE_INSTANCE_START;
  }
  return common(state, tok);
}

static
int doctype0(PROLOG_STATE *state,
	     int tok,
	     const char *ptr,
	     const char *end,
	     const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
  case XML_TOK_PREFIXED_NAME:
    state->handler = doctype1;
    return XML_ROLE_DOCTYPE_NAME;
  }
  return common(state, tok);
}

static
int doctype1(PROLOG_STATE *state,
	     int tok,
	     const char *ptr,
	     const char *end,
	     const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_OPEN_BRACKET:
    state->handler = internalSubset;
    return XML_ROLE_NONE;
  case XML_TOK_DECL_CLOSE:
    state->handler = prolog2;
    return XML_ROLE_DOCTYPE_CLOSE;
  case XML_TOK_NAME:
    if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
      state->handler = doctype3;
      return XML_ROLE_NONE;
    }
    if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
      state->handler = doctype2;
      return XML_ROLE_NONE;
    }
    break;
  }
  return common(state, tok);
}

static
int doctype2(PROLOG_STATE *state,
	     int tok,
	     const char *ptr,
	     const char *end,
	     const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_LITERAL:
    state->handler = doctype3;
    return XML_ROLE_DOCTYPE_PUBLIC_ID;
  }
  return common(state, tok);
}

static
int doctype3(PROLOG_STATE *state,
	     int tok,
	     const char *ptr,
	     const char *end,
	     const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_LITERAL:
    state->handler = doctype4;
    return XML_ROLE_DOCTYPE_SYSTEM_ID;
  }
  return common(state, tok);
}

static
int doctype4(PROLOG_STATE *state,
	     int tok,
	     const char *ptr,
	     const char *end,
	     const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_OPEN_BRACKET:
    state->handler = internalSubset;
    return XML_ROLE_NONE;
  case XML_TOK_DECL_CLOSE:
    state->handler = prolog2;
    return XML_ROLE_DOCTYPE_CLOSE;
  }
  return common(state, tok);
}

static
int doctype5(PROLOG_STATE *state,
	     int tok,
	     const char *ptr,
	     const char *end,
	     const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_DECL_CLOSE:
    state->handler = prolog2;
    return XML_ROLE_DOCTYPE_CLOSE;
  }
  return common(state, tok);
}

static
int internalSubset(PROLOG_STATE *state,
		   int tok,
		   const char *ptr,
		   const char *end,
		   const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_DECL_OPEN:
    if (XmlNameMatchesAscii(enc,
			    ptr + 2 * MIN_BYTES_PER_CHAR(enc),
			    end,
			    KW_ENTITY)) {
      state->handler = entity0;
      return XML_ROLE_NONE;
    }
    if (XmlNameMatchesAscii(enc,
			    ptr + 2 * MIN_BYTES_PER_CHAR(enc),
			    end,
			    KW_ATTLIST)) {
      state->handler = attlist0;
      return XML_ROLE_NONE;
    }
    if (XmlNameMatchesAscii(enc,
			    ptr + 2 * MIN_BYTES_PER_CHAR(enc),
			    end,
			    KW_ELEMENT)) {
      state->handler = element0;
      return XML_ROLE_NONE;
    }
    if (XmlNameMatchesAscii(enc,
			    ptr + 2 * MIN_BYTES_PER_CHAR(enc),
			    end,
			    KW_NOTATION)) {
      state->handler = notation0;
      return XML_ROLE_NONE;
    }
    break;
  case XML_TOK_PI:
  case XML_TOK_COMMENT:
    return XML_ROLE_NONE;
  case XML_TOK_PARAM_ENTITY_REF:
    return XML_ROLE_PARAM_ENTITY_REF;
  case XML_TOK_CLOSE_BRACKET:
    state->handler = doctype5;
    return XML_ROLE_NONE;
  }
  return common(state, tok);
}

#ifdef XML_DTD

static
int externalSubset0(PROLOG_STATE *state,
		    int tok,
		    const char *ptr,
		    const char *end,
		    const ENCODING *enc)
{
  state->handler = externalSubset1;
  if (tok == XML_TOK_XML_DECL)
    return XML_ROLE_TEXT_DECL;
  return externalSubset1(state, tok, ptr, end, enc);
}

static
int externalSubset1(PROLOG_STATE *state,
		    int tok,
		    const char *ptr,
		    const char *end,
		    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_COND_SECT_OPEN:
    state->handler = condSect0;
    return XML_ROLE_NONE;
  case XML_TOK_COND_SECT_CLOSE:
    if (state->includeLevel == 0)
      break;
    state->includeLevel -= 1;
    return XML_ROLE_NONE;
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_CLOSE_BRACKET:
    break;
  case XML_TOK_NONE:
    if (state->includeLevel)
      break;
    return XML_ROLE_NONE;
  default:
    return internalSubset(state, tok, ptr, end, enc);
  }
  return common(state, tok);
}

#endif /* XML_DTD */

static
int entity0(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_PERCENT:
    state->handler = entity1;
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    state->handler = entity2;
    return XML_ROLE_GENERAL_ENTITY_NAME;
  }
  return common(state, tok);
}

static
int entity1(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    state->handler = entity7;
    return XML_ROLE_PARAM_ENTITY_NAME;
  }
  return common(state, tok);
}

static
int entity2(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
      state->handler = entity4;
      return XML_ROLE_NONE;
    }
    if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
      state->handler = entity3;
      return XML_ROLE_NONE;
    }
    break;
  case XML_TOK_LITERAL:
    state->handler = declClose;
    return XML_ROLE_ENTITY_VALUE;
  }
  return common(state, tok);
}

static
int entity3(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_LITERAL:
    state->handler = entity4;
    return XML_ROLE_ENTITY_PUBLIC_ID;
  }
  return common(state, tok);
}


static
int entity4(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_LITERAL:
    state->handler = entity5;
    return XML_ROLE_ENTITY_SYSTEM_ID;
  }
  return common(state, tok);
}

static
int entity5(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_DECL_CLOSE:
    setTopLevel(state);
    return XML_ROLE_EXTERNAL_GENERAL_ENTITY_NO_NOTATION;
  case XML_TOK_NAME:
    if (XmlNameMatchesAscii(enc, ptr, end, KW_NDATA)) {
      state->handler = entity6;
      return XML_ROLE_NONE;
    }
    break;
  }
  return common(state, tok);
}

static
int entity6(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    state->handler = declClose;
    return XML_ROLE_ENTITY_NOTATION_NAME;
  }
  return common(state, tok);
}

static
int entity7(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
      state->handler = entity9;
      return XML_ROLE_NONE;
    }
    if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
      state->handler = entity8;
      return XML_ROLE_NONE;
    }
    break;
  case XML_TOK_LITERAL:
    state->handler = declClose;
    return XML_ROLE_ENTITY_VALUE;
  }
  return common(state, tok);
}

static
int entity8(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_LITERAL:
    state->handler = entity9;
    return XML_ROLE_ENTITY_PUBLIC_ID;
  }
  return common(state, tok);
}

static
int entity9(PROLOG_STATE *state,
	    int tok,
	    const char *ptr,
	    const char *end,
	    const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_LITERAL:
    state->handler = declClose;
    return XML_ROLE_ENTITY_SYSTEM_ID;
  }
  return common(state, tok);
}

static
int notation0(PROLOG_STATE *state,
	      int tok,
	      const char *ptr,
	      const char *end,
	      const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    state->handler = notation1;
    return XML_ROLE_NOTATION_NAME;
  }
  return common(state, tok);
}

static
int notation1(PROLOG_STATE *state,
	      int tok,
	      const char *ptr,
	      const char *end,
	      const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S:
    return XML_ROLE_NONE;
  case XML_TOK_NAME:
    if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
      state->handler = notation3;
      return XML_ROLE_NONE;
    }
    if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
      state->handler = notation2;
      return XML_ROLE_NONE;
    }
    break;
  }
  return common(state, tok);
}

static
int notation2(PROLOG_STATE *state,
	      int tok,
	      const char *ptr,
	      const char *end,
	      const ENCODING *enc)
{
  switch (tok) {
  case XML_TOK_PROLOG_S: