pjs/parser/expat/lib/xmlrole.c

1331 строка
32 KiB
C
Исходник Обычный вид История

/* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
See the file COPYING for copying permission.
1999-02-27 04:41:01 +03:00
*/
#include <stddef.h>
#ifdef COMPILED_FROM_DSP
#include "winconfig.h"
#elif defined(MACOS_CLASSIC)
#include "macconfig.h"
#elif defined(__amigaos4__)
#include "amigaconfig.h"
#else
#ifdef HAVE_EXPAT_CONFIG_H
#include <expat_config.h>
#endif
#endif /* ndef COMPILED_FROM_DSP */
#include "expat_external.h"
#include "internal.h"
1999-02-27 04:41:01 +03:00
#include "xmlrole.h"
#include "ascii.h"
1999-02-27 04:41:01 +03:00
/* 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' };
1999-02-27 04:41:01 +03:00
#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 PTRCALL PROLOG_HANDLER(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc);
1999-02-27 04:41:01 +03:00
static PROLOG_HANDLER
prolog0, prolog1, prolog2,
doctype0, doctype1, doctype2, doctype3, doctype4, doctype5,
internalSubset,
entity0, entity1, entity2, entity3, entity4, entity5, entity6,
entity7, entity8, entity9, entity10,
1999-02-27 04:41:01 +03:00
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 */
1999-02-27 04:41:01 +03:00
declClose,
error;
static int FASTCALL common(PROLOG_STATE *state, int tok);
1999-02-27 04:41:01 +03:00
static int PTRCALL
prolog0(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
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_PI;
1999-02-27 04:41:01 +03:00
case XML_TOK_COMMENT:
state->handler = prolog1;
return XML_ROLE_COMMENT;
1999-02-27 04:41:01 +03:00
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))
1999-02-27 04:41:01 +03:00
break;
state->handler = doctype0;
return XML_ROLE_DOCTYPE_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_INSTANCE_START:
state->handler = error;
return XML_ROLE_INSTANCE_START;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
prolog1(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_NONE;
case XML_TOK_PI:
return XML_ROLE_PI;
1999-02-27 04:41:01 +03:00
case XML_TOK_COMMENT:
return XML_ROLE_COMMENT;
1999-02-27 04:41:01 +03:00
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))
1999-02-27 04:41:01 +03:00
break;
state->handler = doctype0;
return XML_ROLE_DOCTYPE_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_INSTANCE_START:
state->handler = error;
return XML_ROLE_INSTANCE_START;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
prolog2(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_NONE;
case XML_TOK_PI:
return XML_ROLE_PI;
1999-02-27 04:41:01 +03:00
case XML_TOK_COMMENT:
return XML_ROLE_COMMENT;
1999-02-27 04:41:01 +03:00
case XML_TOK_INSTANCE_START:
state->handler = error;
return XML_ROLE_INSTANCE_START;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
doctype0(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_DOCTYPE_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
1999-02-27 04:41:01 +03:00
state->handler = doctype1;
return XML_ROLE_DOCTYPE_NAME;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
doctype1(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_DOCTYPE_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_OPEN_BRACKET:
state->handler = internalSubset;
return XML_ROLE_DOCTYPE_INTERNAL_SUBSET;
1999-02-27 04:41:01 +03:00
case XML_TOK_DECL_CLOSE:
state->handler = prolog2;
return XML_ROLE_DOCTYPE_CLOSE;
case XML_TOK_NAME:
if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
1999-02-27 04:41:01 +03:00
state->handler = doctype3;
return XML_ROLE_DOCTYPE_NONE;
1999-02-27 04:41:01 +03:00
}
if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
1999-02-27 04:41:01 +03:00
state->handler = doctype2;
return XML_ROLE_DOCTYPE_NONE;
1999-02-27 04:41:01 +03:00
}
break;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
doctype2(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_DOCTYPE_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_LITERAL:
state->handler = doctype3;
return XML_ROLE_DOCTYPE_PUBLIC_ID;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
doctype3(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_DOCTYPE_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_LITERAL:
state->handler = doctype4;
return XML_ROLE_DOCTYPE_SYSTEM_ID;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
doctype4(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_DOCTYPE_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_OPEN_BRACKET:
state->handler = internalSubset;
return XML_ROLE_DOCTYPE_INTERNAL_SUBSET;
1999-02-27 04:41:01 +03:00
case XML_TOK_DECL_CLOSE:
state->handler = prolog2;
return XML_ROLE_DOCTYPE_CLOSE;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
doctype5(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_DOCTYPE_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_DECL_CLOSE:
state->handler = prolog2;
return XML_ROLE_DOCTYPE_CLOSE;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
internalSubset(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
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)) {
1999-02-27 04:41:01 +03:00
state->handler = entity0;
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
}
if (XmlNameMatchesAscii(enc,
ptr + 2 * MIN_BYTES_PER_CHAR(enc),
end,
KW_ATTLIST)) {
1999-02-27 04:41:01 +03:00
state->handler = attlist0;
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
}
if (XmlNameMatchesAscii(enc,
ptr + 2 * MIN_BYTES_PER_CHAR(enc),
end,
KW_ELEMENT)) {
1999-02-27 04:41:01 +03:00
state->handler = element0;
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
}
if (XmlNameMatchesAscii(enc,
ptr + 2 * MIN_BYTES_PER_CHAR(enc),
end,
KW_NOTATION)) {
1999-02-27 04:41:01 +03:00
state->handler = notation0;
return XML_ROLE_NOTATION_NONE;
1999-02-27 04:41:01 +03:00
}
break;
case XML_TOK_PI:
return XML_ROLE_PI;
1999-02-27 04:41:01 +03:00
case XML_TOK_COMMENT:
return XML_ROLE_COMMENT;
1999-02-27 04:41:01 +03:00
case XML_TOK_PARAM_ENTITY_REF:
return XML_ROLE_PARAM_ENTITY_REF;
case XML_TOK_CLOSE_BRACKET:
state->handler = doctype5;
return XML_ROLE_DOCTYPE_NONE;
case XML_TOK_NONE:
return XML_ROLE_NONE;
1999-02-27 04:41:01 +03:00
}
return common(state, tok);
}
#ifdef XML_DTD
static int PTRCALL
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 PTRCALL
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);
1999-02-27 04:41:01 +03:00
}
#endif /* XML_DTD */
static int PTRCALL
entity0(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_PERCENT:
state->handler = entity1;
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
state->handler = entity2;
return XML_ROLE_GENERAL_ENTITY_NAME;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
entity1(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
state->handler = entity7;
return XML_ROLE_PARAM_ENTITY_NAME;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
entity2(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
1999-02-27 04:41:01 +03:00
state->handler = entity4;
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
}
if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
1999-02-27 04:41:01 +03:00
state->handler = entity3;
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
}
break;
case XML_TOK_LITERAL:
state->handler = declClose;
state->role_none = XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
return XML_ROLE_ENTITY_VALUE;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
entity3(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_LITERAL:
state->handler = entity4;
return XML_ROLE_ENTITY_PUBLIC_ID;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
entity4(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_LITERAL:
state->handler = entity5;
return XML_ROLE_ENTITY_SYSTEM_ID;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
entity5(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_DECL_CLOSE:
setTopLevel(state);
return XML_ROLE_ENTITY_COMPLETE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
if (XmlNameMatchesAscii(enc, ptr, end, KW_NDATA)) {
1999-02-27 04:41:01 +03:00
state->handler = entity6;
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
}
break;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
entity6(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
state->handler = declClose;
state->role_none = XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
return XML_ROLE_ENTITY_NOTATION_NAME;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
entity7(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
1999-02-27 04:41:01 +03:00
state->handler = entity9;
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
}
if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
1999-02-27 04:41:01 +03:00
state->handler = entity8;
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
}
break;
case XML_TOK_LITERAL:
state->handler = declClose;
state->role_none = XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
return XML_ROLE_ENTITY_VALUE;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
entity8(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_LITERAL:
state->handler = entity9;
return XML_ROLE_ENTITY_PUBLIC_ID;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
entity9(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_LITERAL:
state->handler = entity10;
1999-02-27 04:41:01 +03:00
return XML_ROLE_ENTITY_SYSTEM_ID;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
entity10(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ENTITY_NONE;
case XML_TOK_DECL_CLOSE:
setTopLevel(state);
return XML_ROLE_ENTITY_COMPLETE;
}
return common(state, tok);
}
static int PTRCALL
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_NOTATION_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
state->handler = notation1;
return XML_ROLE_NOTATION_NAME;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
notation1(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_NOTATION_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
if (XmlNameMatchesAscii(enc, ptr, end, KW_SYSTEM)) {
1999-02-27 04:41:01 +03:00
state->handler = notation3;
return XML_ROLE_NOTATION_NONE;
1999-02-27 04:41:01 +03:00
}
if (XmlNameMatchesAscii(enc, ptr, end, KW_PUBLIC)) {
1999-02-27 04:41:01 +03:00
state->handler = notation2;
return XML_ROLE_NOTATION_NONE;
1999-02-27 04:41:01 +03:00
}
break;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
notation2(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_NOTATION_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_LITERAL:
state->handler = notation4;
return XML_ROLE_NOTATION_PUBLIC_ID;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
notation3(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_NOTATION_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_LITERAL:
state->handler = declClose;
state->role_none = XML_ROLE_NOTATION_NONE;
1999-02-27 04:41:01 +03:00
return XML_ROLE_NOTATION_SYSTEM_ID;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
notation4(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_NOTATION_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_LITERAL:
state->handler = declClose;
state->role_none = XML_ROLE_NOTATION_NONE;
1999-02-27 04:41:01 +03:00
return XML_ROLE_NOTATION_SYSTEM_ID;
case XML_TOK_DECL_CLOSE:
setTopLevel(state);
1999-02-27 04:41:01 +03:00
return XML_ROLE_NOTATION_NO_SYSTEM_ID;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
attlist0(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
1999-02-27 04:41:01 +03:00
state->handler = attlist1;
return XML_ROLE_ATTLIST_ELEMENT_NAME;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
attlist1(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_DECL_CLOSE:
setTopLevel(state);
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
1999-02-27 04:41:01 +03:00
state->handler = attlist2;
return XML_ROLE_ATTRIBUTE_NAME;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
attlist2(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
{
static const char * const types[] = {
KW_CDATA,
KW_ID,
KW_IDREF,
KW_IDREFS,
KW_ENTITY,
KW_ENTITIES,
KW_NMTOKEN,
KW_NMTOKENS,
1999-02-27 04:41:01 +03:00
};
int i;
for (i = 0; i < (int)(sizeof(types)/sizeof(types[0])); i++)
if (XmlNameMatchesAscii(enc, ptr, end, types[i])) {
state->handler = attlist8;
return XML_ROLE_ATTRIBUTE_TYPE_CDATA + i;
}
1999-02-27 04:41:01 +03:00
}
if (XmlNameMatchesAscii(enc, ptr, end, KW_NOTATION)) {
1999-02-27 04:41:01 +03:00
state->handler = attlist5;
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
}
break;
case XML_TOK_OPEN_PAREN:
state->handler = attlist3;
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
attlist3(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NMTOKEN:
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
1999-02-27 04:41:01 +03:00
state->handler = attlist4;
return XML_ROLE_ATTRIBUTE_ENUM_VALUE;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
attlist4(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_CLOSE_PAREN:
state->handler = attlist8;
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_OR:
state->handler = attlist3;
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
attlist5(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_OPEN_PAREN:
state->handler = attlist6;
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
attlist6(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
state->handler = attlist7;
return XML_ROLE_ATTRIBUTE_NOTATION_VALUE;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
attlist7(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_CLOSE_PAREN:
state->handler = attlist8;
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_OR:
state->handler = attlist6;
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
/* default value */
static int PTRCALL
attlist8(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_POUND_NAME:
if (XmlNameMatchesAscii(enc,
ptr + MIN_BYTES_PER_CHAR(enc),
end,
KW_IMPLIED)) {
1999-02-27 04:41:01 +03:00
state->handler = attlist1;
return XML_ROLE_IMPLIED_ATTRIBUTE_VALUE;
}
if (XmlNameMatchesAscii(enc,
ptr + MIN_BYTES_PER_CHAR(enc),
end,
KW_REQUIRED)) {
1999-02-27 04:41:01 +03:00
state->handler = attlist1;
return XML_ROLE_REQUIRED_ATTRIBUTE_VALUE;
}
if (XmlNameMatchesAscii(enc,
ptr + MIN_BYTES_PER_CHAR(enc),
end,
KW_FIXED)) {
1999-02-27 04:41:01 +03:00
state->handler = attlist9;
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
}
break;
case XML_TOK_LITERAL:
state->handler = attlist1;
return XML_ROLE_DEFAULT_ATTRIBUTE_VALUE;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
attlist9(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ATTLIST_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_LITERAL:
state->handler = attlist1;
return XML_ROLE_FIXED_ATTRIBUTE_VALUE;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
element0(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
1999-02-27 04:41:01 +03:00
state->handler = element1;
return XML_ROLE_ELEMENT_NAME;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
element1(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
if (XmlNameMatchesAscii(enc, ptr, end, KW_EMPTY)) {
1999-02-27 04:41:01 +03:00
state->handler = declClose;
state->role_none = XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
return XML_ROLE_CONTENT_EMPTY;
}
if (XmlNameMatchesAscii(enc, ptr, end, KW_ANY)) {
1999-02-27 04:41:01 +03:00
state->handler = declClose;
state->role_none = XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
return XML_ROLE_CONTENT_ANY;
}
break;
case XML_TOK_OPEN_PAREN:
state->handler = element2;
state->level = 1;
return XML_ROLE_GROUP_OPEN;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
element2(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_POUND_NAME:
if (XmlNameMatchesAscii(enc,
ptr + MIN_BYTES_PER_CHAR(enc),
end,
KW_PCDATA)) {
1999-02-27 04:41:01 +03:00
state->handler = element3;
return XML_ROLE_CONTENT_PCDATA;
}
break;
case XML_TOK_OPEN_PAREN:
state->level = 2;
state->handler = element6;
return XML_ROLE_GROUP_OPEN;
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
1999-02-27 04:41:01 +03:00
state->handler = element7;
return XML_ROLE_CONTENT_ELEMENT;
case XML_TOK_NAME_QUESTION:
state->handler = element7;
return XML_ROLE_CONTENT_ELEMENT_OPT;
case XML_TOK_NAME_ASTERISK:
state->handler = element7;
return XML_ROLE_CONTENT_ELEMENT_REP;
case XML_TOK_NAME_PLUS:
state->handler = element7;
return XML_ROLE_CONTENT_ELEMENT_PLUS;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
element3(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_CLOSE_PAREN:
state->handler = declClose;
state->role_none = XML_ROLE_ELEMENT_NONE;
return XML_ROLE_GROUP_CLOSE;
1999-02-27 04:41:01 +03:00
case XML_TOK_CLOSE_PAREN_ASTERISK:
state->handler = declClose;
state->role_none = XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
return XML_ROLE_GROUP_CLOSE_REP;
case XML_TOK_OR:
state->handler = element4;
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
element4(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
1999-02-27 04:41:01 +03:00
state->handler = element5;
return XML_ROLE_CONTENT_ELEMENT;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
element5(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_CLOSE_PAREN_ASTERISK:
state->handler = declClose;
state->role_none = XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
return XML_ROLE_GROUP_CLOSE_REP;
case XML_TOK_OR:
state->handler = element4;
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
element6(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_OPEN_PAREN:
state->level += 1;
return XML_ROLE_GROUP_OPEN;
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
1999-02-27 04:41:01 +03:00
state->handler = element7;
return XML_ROLE_CONTENT_ELEMENT;
case XML_TOK_NAME_QUESTION:
state->handler = element7;
return XML_ROLE_CONTENT_ELEMENT_OPT;
case XML_TOK_NAME_ASTERISK:
state->handler = element7;
return XML_ROLE_CONTENT_ELEMENT_REP;
case XML_TOK_NAME_PLUS:
state->handler = element7;
return XML_ROLE_CONTENT_ELEMENT_PLUS;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
element7(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return XML_ROLE_ELEMENT_NONE;
1999-02-27 04:41:01 +03:00
case XML_TOK_CLOSE_PAREN:
state->level -= 1;
if (state->level == 0) {
1999-02-27 04:41:01 +03:00
state->handler = declClose;
state->role_none = XML_ROLE_ELEMENT_NONE;
}
1999-02-27 04:41:01 +03:00
return XML_ROLE_GROUP_CLOSE;
case XML_TOK_CLOSE_PAREN_ASTERISK:
state->level -= 1;
if (state->level == 0) {
1999-02-27 04:41:01 +03:00
state->handler = declClose;
state->role_none = XML_ROLE_ELEMENT_NONE;
}
1999-02-27 04:41:01 +03:00
return XML_ROLE_GROUP_CLOSE_REP;
case XML_TOK_CLOSE_PAREN_QUESTION:
state->level -= 1;
if (state->level == 0) {
1999-02-27 04:41:01 +03:00
state->handler = declClose;
state->role_none = XML_ROLE_ELEMENT_NONE;
}
1999-02-27 04:41:01 +03:00
return XML_ROLE_GROUP_CLOSE_OPT;
case XML_TOK_CLOSE_PAREN_PLUS:
state->level -= 1;
if (state->level == 0) {
1999-02-27 04:41:01 +03:00
state->handler = declClose;
state->role_none = XML_ROLE_ELEMENT_NONE;
}
1999-02-27 04:41:01 +03:00
return XML_ROLE_GROUP_CLOSE_PLUS;
case XML_TOK_COMMA:
state->handler = element6;
return XML_ROLE_GROUP_SEQUENCE;
case XML_TOK_OR:
state->handler = element6;
return XML_ROLE_GROUP_CHOICE;
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
#ifdef XML_DTD
static int PTRCALL
condSect0(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_INCLUDE)) {
state->handler = condSect1;
return XML_ROLE_NONE;
}
if (XmlNameMatchesAscii(enc, ptr, end, KW_IGNORE)) {
state->handler = condSect2;
return XML_ROLE_NONE;
}
break;
}
return common(state, tok);
}
static int PTRCALL
condSect1(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 = externalSubset1;
state->includeLevel += 1;
return XML_ROLE_NONE;
}
return common(state, tok);
}
static int PTRCALL
condSect2(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 = externalSubset1;
return XML_ROLE_IGNORE_SECT;
}
return common(state, tok);
}
#endif /* XML_DTD */
static int PTRCALL
declClose(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
switch (tok) {
case XML_TOK_PROLOG_S:
return state->role_none;
1999-02-27 04:41:01 +03:00
case XML_TOK_DECL_CLOSE:
setTopLevel(state);
return state->role_none;
1999-02-27 04:41:01 +03:00
}
return common(state, tok);
1999-02-27 04:41:01 +03:00
}
static int PTRCALL
error(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
1999-02-27 04:41:01 +03:00
{
return XML_ROLE_NONE;
}
static int FASTCALL
common(PROLOG_STATE *state, int tok)
1999-02-27 04:41:01 +03:00
{
#ifdef XML_DTD
if (!state->documentEntity && tok == XML_TOK_PARAM_ENTITY_REF)
return XML_ROLE_INNER_PARAM_ENTITY_REF;
#endif
1999-02-27 04:41:01 +03:00
state->handler = error;
return XML_ROLE_ERROR;
}
void
XmlPrologStateInit(PROLOG_STATE *state)
1999-02-27 04:41:01 +03:00
{
state->handler = prolog0;
#ifdef XML_DTD
state->documentEntity = 1;
state->includeLevel = 0;
state->inEntityValue = 0;
#endif /* XML_DTD */
1999-02-27 04:41:01 +03:00
}
#ifdef XML_DTD
void
XmlPrologStateInitExternalEntity(PROLOG_STATE *state)
{
state->handler = externalSubset0;
state->documentEntity = 0;
state->includeLevel = 0;
}
#endif /* XML_DTD */