xmlrole.c revision 7f235c0d101d71224c8e498ac8f7fde1f98596c7
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen/*
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenThe contents of this file are subject to the Mozilla Public License
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenVersion 1.1 (the "License"); you may not use this file except in
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowencompliance with the License. You may obtain a copy of the License at
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenhttp://www.mozilla.org/MPL/
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenSoftware distributed under the License is distributed on an "AS IS"
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenbasis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See the
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenLicense for the specific language governing rights and limitations
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenunder the License.
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenThe Original Code is expat.
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenThe Initial Developer of the Original Code is James Clark.
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenPortions created by James Clark are Copyright (C) 1998, 1999
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenJames Clark. All Rights Reserved.
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenContributor(s):
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenAlternatively, the contents of this file may be used under the terms
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenof the GNU General Public License (the "GPL"), in which case the
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenprovisions of the GPL are applicable instead of those above. If you
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenwish to allow use of your version of this file only under the terms of
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenthe GPL and not to allow others to use your version of this file under
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenthe MPL, indicate your decision by deleting the provisions above and
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenreplace them with the notice and other provisions required by the
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie BowenGPL. If you do not delete the provisions above, a recipient may use
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenyour version of this file under either the MPL or the GPL.
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen*/
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen#include "xmldef.h"
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen#include "xmlrole.h"
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen/* Doesn't check:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen that ,| are not mixed in a model group
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen content of literals
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen*/
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen#ifndef MIN_BYTES_PER_CHAR
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen#define MIN_BYTES_PER_CHAR(enc) ((enc)->minBytesPerChar)
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen#endif
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowentypedef int PROLOG_HANDLER(struct prolog_state *state,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen int tok,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *ptr,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *end,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const ENCODING *enc);
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic PROLOG_HANDLER
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen prolog0, prolog1, prolog2,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen doctype0, doctype1, doctype2, doctype3, doctype4, doctype5,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen internalSubset,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen entity0, entity1, entity2, entity3, entity4, entity5, entity6,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen entity7, entity8, entity9,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen notation0, notation1, notation2, notation3, notation4,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen attlist0, attlist1, attlist2, attlist3, attlist4, attlist5, attlist6,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen attlist7, attlist8, attlist9,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen element0, element1, element2, element3, element4, element5, element6,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen element7,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen declClose,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen error;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenint syntaxError(PROLOG_STATE *);
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenint prolog0(PROLOG_STATE *state,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen int tok,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *ptr,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *end,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const ENCODING *enc)
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen{
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen switch (tok) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PROLOG_S:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = prolog1;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_XML_DECL:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = prolog1;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_XML_DECL;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PI:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = prolog1;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_COMMENT:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = prolog1;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_BOM:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_DECL_OPEN:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen if (!XmlNameMatchesAscii(enc,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen ptr + 2 * MIN_BYTES_PER_CHAR(enc),
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen "DOCTYPE"))
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen break;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = doctype0;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_INSTANCE_START:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = error;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_INSTANCE_START;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen }
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return syntaxError(state);
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen}
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenint prolog1(PROLOG_STATE *state,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen int tok,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *ptr,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *end,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const ENCODING *enc)
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen{
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen switch (tok) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PROLOG_S:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PI:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_COMMENT:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_BOM:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_DECL_OPEN:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen if (!XmlNameMatchesAscii(enc,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen ptr + 2 * MIN_BYTES_PER_CHAR(enc),
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen "DOCTYPE"))
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen break;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = doctype0;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_INSTANCE_START:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = error;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_INSTANCE_START;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen }
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return syntaxError(state);
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen}
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenint prolog2(PROLOG_STATE *state,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen int tok,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *ptr,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *end,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const ENCODING *enc)
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen{
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen switch (tok) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PROLOG_S:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PI:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_COMMENT:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_INSTANCE_START:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = error;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_INSTANCE_START;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen }
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return syntaxError(state);
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen}
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenint doctype0(PROLOG_STATE *state,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen int tok,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *ptr,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *end,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const ENCODING *enc)
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen{
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen switch (tok) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PROLOG_S:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_NAME:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PREFIXED_NAME:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = doctype1;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_DOCTYPE_NAME;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen }
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return syntaxError(state);
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen}
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenint doctype1(PROLOG_STATE *state,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen int tok,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *ptr,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *end,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const ENCODING *enc)
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen{
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen switch (tok) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PROLOG_S:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_OPEN_BRACKET:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = internalSubset;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_DECL_CLOSE:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = prolog2;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_DOCTYPE_CLOSE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_NAME:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen if (XmlNameMatchesAscii(enc, ptr, "SYSTEM")) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = doctype3;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen }
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = doctype2;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen }
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen break;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen }
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return syntaxError(state);
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen}
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenint doctype2(PROLOG_STATE *state,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen int tok,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *ptr,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *end,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const ENCODING *enc)
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen{
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen switch (tok) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PROLOG_S:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_LITERAL:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = doctype3;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_DOCTYPE_PUBLIC_ID;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen }
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return syntaxError(state);
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen}
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenint doctype3(PROLOG_STATE *state,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen int tok,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *ptr,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *end,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const ENCODING *enc)
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen{
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen switch (tok) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PROLOG_S:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_LITERAL:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = doctype4;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_DOCTYPE_SYSTEM_ID;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen }
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return syntaxError(state);
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen}
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenint doctype4(PROLOG_STATE *state,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen int tok,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *ptr,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *end,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const ENCODING *enc)
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen{
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen switch (tok) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PROLOG_S:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_OPEN_BRACKET:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = internalSubset;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_NONE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_DECL_CLOSE:
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen state->handler = prolog2;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return XML_ROLE_DOCTYPE_CLOSE;
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen }
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen return syntaxError(state);
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen}
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenstatic
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowenint doctype5(PROLOG_STATE *state,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen int tok,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *ptr,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const char *end,
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen const ENCODING *enc)
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen{
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen switch (tok) {
5c124de5c36bfc236d55578429df5f048f0d0a07Jamie Bowen case XML_TOK_PROLOG_S:
return XML_ROLE_NONE;
case XML_TOK_DECL_CLOSE:
state->handler = prolog2;
return XML_ROLE_DOCTYPE_CLOSE;
}
return syntaxError(state);
}
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),
"ENTITY")) {
state->handler = entity0;
return XML_ROLE_NONE;
}
if (XmlNameMatchesAscii(enc,
ptr + 2 * MIN_BYTES_PER_CHAR(enc),
"ATTLIST")) {
state->handler = attlist0;
return XML_ROLE_NONE;
}
if (XmlNameMatchesAscii(enc,
ptr + 2 * MIN_BYTES_PER_CHAR(enc),
"ELEMENT")) {
state->handler = element0;
return XML_ROLE_NONE;
}
if (XmlNameMatchesAscii(enc,
ptr + 2 * MIN_BYTES_PER_CHAR(enc),
"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 syntaxError(state);
}
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 syntaxError(state);
}
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 syntaxError(state);
}
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, "SYSTEM")) {
state->handler = entity4;
return XML_ROLE_NONE;
}
if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) {
state->handler = entity3;
return XML_ROLE_NONE;
}
break;
case XML_TOK_LITERAL:
state->handler = declClose;
return XML_ROLE_ENTITY_VALUE;
}
return syntaxError(state);
}
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 syntaxError(state);
}
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 syntaxError(state);
}
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:
state->handler = internalSubset;
return XML_ROLE_NONE;
case XML_TOK_NAME:
if (XmlNameMatchesAscii(enc, ptr, "NDATA")) {
state->handler = entity6;
return XML_ROLE_NONE;
}
break;
}
return syntaxError(state);
}
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 syntaxError(state);
}
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, "SYSTEM")) {
state->handler = entity9;
return XML_ROLE_NONE;
}
if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) {
state->handler = entity8;
return XML_ROLE_NONE;
}
break;
case XML_TOK_LITERAL:
state->handler = declClose;
return XML_ROLE_ENTITY_VALUE;
}
return syntaxError(state);
}
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 syntaxError(state);
}
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 syntaxError(state);
}
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 syntaxError(state);
}
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, "SYSTEM")) {
state->handler = notation3;
return XML_ROLE_NONE;
}
if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) {
state->handler = notation2;
return XML_ROLE_NONE;
}
break;
}
return syntaxError(state);
}
static
int notation2(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 = notation4;
return XML_ROLE_NOTATION_PUBLIC_ID;
}
return syntaxError(state);
}
static
int notation3(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_NOTATION_SYSTEM_ID;
}
return syntaxError(state);
}
static
int notation4(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_NOTATION_SYSTEM_ID;
case XML_TOK_DECL_CLOSE:
state->handler = internalSubset;
return XML_ROLE_NOTATION_NO_SYSTEM_ID;
}
return syntaxError(state);
}
static
int attlist0(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 = attlist1;
return XML_ROLE_ATTLIST_ELEMENT_NAME;
}
return syntaxError(state);
}
static
int attlist1(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 = internalSubset;
return XML_ROLE_NONE;
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
state->handler = attlist2;
return XML_ROLE_ATTRIBUTE_NAME;
}
return syntaxError(state);
}
static
int attlist2(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:
{
static const char *types[] = {
"CDATA",
"ID",
"IDREF",
"IDREFS",
"ENTITY",
"ENTITIES",
"NMTOKEN",
"NMTOKENS",
};
int i;
for (i = 0; i < (int)(sizeof(types)/sizeof(types[0])); i++)
if (XmlNameMatchesAscii(enc, ptr, types[i])) {
state->handler = attlist8;
return XML_ROLE_ATTRIBUTE_TYPE_CDATA + i;
}
}
if (XmlNameMatchesAscii(enc, ptr, "NOTATION")) {
state->handler = attlist5;
return XML_ROLE_NONE;
}
break;
case XML_TOK_OPEN_PAREN:
state->handler = attlist3;
return XML_ROLE_NONE;
}
return syntaxError(state);
}
static
int attlist3(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_NMTOKEN:
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
state->handler = attlist4;
return XML_ROLE_ATTRIBUTE_ENUM_VALUE;
}
return syntaxError(state);
}
static
int attlist4(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_CLOSE_PAREN:
state->handler = attlist8;
return XML_ROLE_NONE;
case XML_TOK_OR:
state->handler = attlist3;
return XML_ROLE_NONE;
}
return syntaxError(state);
}
static
int attlist5(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_PAREN:
state->handler = attlist6;
return XML_ROLE_NONE;
}
return syntaxError(state);
}
static
int attlist6(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 = attlist7;
return XML_ROLE_ATTRIBUTE_NOTATION_VALUE;
}
return syntaxError(state);
}
static
int attlist7(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_CLOSE_PAREN:
state->handler = attlist8;
return XML_ROLE_NONE;
case XML_TOK_OR:
state->handler = attlist6;
return XML_ROLE_NONE;
}
return syntaxError(state);
}
/* default value */
static
int attlist8(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_POUND_NAME:
if (XmlNameMatchesAscii(enc,
ptr + MIN_BYTES_PER_CHAR(enc),
"IMPLIED")) {
state->handler = attlist1;
return XML_ROLE_IMPLIED_ATTRIBUTE_VALUE;
}
if (XmlNameMatchesAscii(enc,
ptr + MIN_BYTES_PER_CHAR(enc),
"REQUIRED")) {
state->handler = attlist1;
return XML_ROLE_REQUIRED_ATTRIBUTE_VALUE;
}
if (XmlNameMatchesAscii(enc,
ptr + MIN_BYTES_PER_CHAR(enc),
"FIXED")) {
state->handler = attlist9;
return XML_ROLE_NONE;
}
break;
case XML_TOK_LITERAL:
state->handler = attlist1;
return XML_ROLE_DEFAULT_ATTRIBUTE_VALUE;
}
return syntaxError(state);
}
static
int attlist9(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 = attlist1;
return XML_ROLE_FIXED_ATTRIBUTE_VALUE;
}
return syntaxError(state);
}
static
int element0(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 = element1;
return XML_ROLE_ELEMENT_NAME;
}
return syntaxError(state);
}
static
int element1(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, "EMPTY")) {
state->handler = declClose;
return XML_ROLE_CONTENT_EMPTY;
}
if (XmlNameMatchesAscii(enc, ptr, "ANY")) {
state->handler = declClose;
return XML_ROLE_CONTENT_ANY;
}
break;
case XML_TOK_OPEN_PAREN:
state->handler = element2;
state->level = 1;
return XML_ROLE_GROUP_OPEN;
}
return syntaxError(state);
}
static
int element2(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_POUND_NAME:
if (XmlNameMatchesAscii(enc,
ptr + MIN_BYTES_PER_CHAR(enc),
"PCDATA")) {
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:
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 syntaxError(state);
}
static
int element3(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_CLOSE_PAREN:
case XML_TOK_CLOSE_PAREN_ASTERISK:
state->handler = declClose;
return XML_ROLE_GROUP_CLOSE_REP;
case XML_TOK_OR:
state->handler = element4;
return XML_ROLE_NONE;
}
return syntaxError(state);
}
static
int element4(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 = element5;
return XML_ROLE_CONTENT_ELEMENT;
}
return syntaxError(state);
}
static
int element5(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_CLOSE_PAREN_ASTERISK:
state->handler = declClose;
return XML_ROLE_GROUP_CLOSE_REP;
case XML_TOK_OR:
state->handler = element4;
return XML_ROLE_NONE;
}
return syntaxError(state);
}
static
int element6(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_PAREN:
state->level += 1;
return XML_ROLE_GROUP_OPEN;
case XML_TOK_NAME:
case XML_TOK_PREFIXED_NAME:
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 syntaxError(state);
}
static
int element7(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_CLOSE_PAREN:
state->level -= 1;
if (state->level == 0)
state->handler = declClose;
return XML_ROLE_GROUP_CLOSE;
case XML_TOK_CLOSE_PAREN_ASTERISK:
state->level -= 1;
if (state->level == 0)
state->handler = declClose;
return XML_ROLE_GROUP_CLOSE_REP;
case XML_TOK_CLOSE_PAREN_QUESTION:
state->level -= 1;
if (state->level == 0)
state->handler = declClose;
return XML_ROLE_GROUP_CLOSE_OPT;
case XML_TOK_CLOSE_PAREN_PLUS:
state->level -= 1;
if (state->level == 0)
state->handler = declClose;
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 syntaxError(state);
}
static
int declClose(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 = internalSubset;
return XML_ROLE_NONE;
}
return syntaxError(state);
}
#if 0
static
int ignore(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
{
switch (tok) {
case XML_TOK_DECL_CLOSE:
state->handler = internalSubset;
return 0;
default:
return XML_ROLE_NONE;
}
return syntaxError(state);
}
#endif
static
int error(PROLOG_STATE *state,
int tok,
const char *ptr,
const char *end,
const ENCODING *enc)
{
return XML_ROLE_NONE;
}
static
int syntaxError(PROLOG_STATE *state)
{
state->handler = error;
return XML_ROLE_ERROR;
}
void XmlPrologStateInit(PROLOG_STATE *state)
{
state->handler = prolog0;
}