michael@0: /* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
michael@0:    See the file COPYING for copying permission.
michael@0: */
michael@0: 
michael@0: #ifndef XmlTok_INCLUDED
michael@0: #define XmlTok_INCLUDED 1
michael@0: 
michael@0: #ifdef __cplusplus
michael@0: extern "C" {
michael@0: #endif
michael@0: 
michael@0: /* The following token may be returned by XmlContentTok */
michael@0: #define XML_TOK_TRAILING_RSQB -5 /* ] or ]] at the end of the scan; might be
michael@0:                                     start of illegal ]]> sequence */
michael@0: /* The following tokens may be returned by both XmlPrologTok and
michael@0:    XmlContentTok.
michael@0: */
michael@0: #define XML_TOK_NONE -4          /* The string to be scanned is empty */
michael@0: #define XML_TOK_TRAILING_CR -3   /* A CR at the end of the scan;
michael@0:                                     might be part of CRLF sequence */
michael@0: #define XML_TOK_PARTIAL_CHAR -2  /* only part of a multibyte sequence */
michael@0: #define XML_TOK_PARTIAL -1       /* only part of a token */
michael@0: #define XML_TOK_INVALID 0
michael@0: 
michael@0: /* The following tokens are returned by XmlContentTok; some are also
michael@0:    returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok.
michael@0: */
michael@0: #define XML_TOK_START_TAG_WITH_ATTS 1
michael@0: #define XML_TOK_START_TAG_NO_ATTS 2
michael@0: #define XML_TOK_EMPTY_ELEMENT_WITH_ATTS 3 /* empty element tag <e/> */
michael@0: #define XML_TOK_EMPTY_ELEMENT_NO_ATTS 4
michael@0: #define XML_TOK_END_TAG 5
michael@0: #define XML_TOK_DATA_CHARS 6
michael@0: #define XML_TOK_DATA_NEWLINE 7
michael@0: #define XML_TOK_CDATA_SECT_OPEN 8
michael@0: #define XML_TOK_ENTITY_REF 9
michael@0: #define XML_TOK_CHAR_REF 10               /* numeric character reference */
michael@0: 
michael@0: /* The following tokens may be returned by both XmlPrologTok and
michael@0:    XmlContentTok.
michael@0: */
michael@0: #define XML_TOK_PI 11                     /* processing instruction */
michael@0: #define XML_TOK_XML_DECL 12               /* XML decl or text decl */
michael@0: #define XML_TOK_COMMENT 13
michael@0: #define XML_TOK_BOM 14                    /* Byte order mark */
michael@0: 
michael@0: /* The following tokens are returned only by XmlPrologTok */
michael@0: #define XML_TOK_PROLOG_S 15
michael@0: #define XML_TOK_DECL_OPEN 16              /* <!foo */
michael@0: #define XML_TOK_DECL_CLOSE 17             /* > */
michael@0: #define XML_TOK_NAME 18
michael@0: #define XML_TOK_NMTOKEN 19
michael@0: #define XML_TOK_POUND_NAME 20             /* #name */
michael@0: #define XML_TOK_OR 21                     /* | */
michael@0: #define XML_TOK_PERCENT 22
michael@0: #define XML_TOK_OPEN_PAREN 23
michael@0: #define XML_TOK_CLOSE_PAREN 24
michael@0: #define XML_TOK_OPEN_BRACKET 25
michael@0: #define XML_TOK_CLOSE_BRACKET 26
michael@0: #define XML_TOK_LITERAL 27
michael@0: #define XML_TOK_PARAM_ENTITY_REF 28
michael@0: #define XML_TOK_INSTANCE_START 29
michael@0: 
michael@0: /* The following occur only in element type declarations */
michael@0: #define XML_TOK_NAME_QUESTION 30          /* name? */
michael@0: #define XML_TOK_NAME_ASTERISK 31          /* name* */
michael@0: #define XML_TOK_NAME_PLUS 32              /* name+ */
michael@0: #define XML_TOK_COND_SECT_OPEN 33         /* <![ */
michael@0: #define XML_TOK_COND_SECT_CLOSE 34        /* ]]> */
michael@0: #define XML_TOK_CLOSE_PAREN_QUESTION 35   /* )? */
michael@0: #define XML_TOK_CLOSE_PAREN_ASTERISK 36   /* )* */
michael@0: #define XML_TOK_CLOSE_PAREN_PLUS 37       /* )+ */
michael@0: #define XML_TOK_COMMA 38
michael@0: 
michael@0: /* The following token is returned only by XmlAttributeValueTok */
michael@0: #define XML_TOK_ATTRIBUTE_VALUE_S 39
michael@0: 
michael@0: /* The following token is returned only by XmlCdataSectionTok */
michael@0: #define XML_TOK_CDATA_SECT_CLOSE 40
michael@0: 
michael@0: /* With namespace processing this is returned by XmlPrologTok for a
michael@0:    name with a colon.
michael@0: */
michael@0: #define XML_TOK_PREFIXED_NAME 41
michael@0: 
michael@0: #ifdef XML_DTD
michael@0: #define XML_TOK_IGNORE_SECT 42
michael@0: #endif /* XML_DTD */
michael@0: 
michael@0: #ifdef XML_DTD
michael@0: #define XML_N_STATES 4
michael@0: #else /* not XML_DTD */
michael@0: #define XML_N_STATES 3
michael@0: #endif /* not XML_DTD */
michael@0: 
michael@0: #define XML_PROLOG_STATE 0
michael@0: #define XML_CONTENT_STATE 1
michael@0: #define XML_CDATA_SECTION_STATE 2
michael@0: #ifdef XML_DTD
michael@0: #define XML_IGNORE_SECTION_STATE 3
michael@0: #endif /* XML_DTD */
michael@0: 
michael@0: #define XML_N_LITERAL_TYPES 2
michael@0: #define XML_ATTRIBUTE_VALUE_LITERAL 0
michael@0: #define XML_ENTITY_VALUE_LITERAL 1
michael@0: 
michael@0: /* The size of the buffer passed to XmlUtf8Encode must be at least this. */
michael@0: #define XML_UTF8_ENCODE_MAX 4
michael@0: /* The size of the buffer passed to XmlUtf16Encode must be at least this. */
michael@0: #define XML_UTF16_ENCODE_MAX 2
michael@0: 
michael@0: typedef struct position {
michael@0:   /* first line and first column are 0 not 1 */
michael@0:   XML_Size lineNumber;
michael@0:   XML_Size columnNumber;
michael@0: } POSITION;
michael@0: 
michael@0: typedef struct {
michael@0:   const char *name;
michael@0:   const char *valuePtr;
michael@0:   const char *valueEnd;
michael@0:   char normalized;
michael@0: } ATTRIBUTE;
michael@0: 
michael@0: struct encoding;
michael@0: typedef struct encoding ENCODING;
michael@0: 
michael@0: typedef int (PTRCALL *SCANNER)(const ENCODING *,
michael@0:                                const char *,
michael@0:                                const char *,
michael@0:                                const char **);
michael@0: 
michael@0: struct encoding {
michael@0:   SCANNER scanners[XML_N_STATES];
michael@0:   SCANNER literalScanners[XML_N_LITERAL_TYPES];
michael@0:   int (PTRCALL *sameName)(const ENCODING *,
michael@0:                           const char *,
michael@0:                           const char *);
michael@0:   int (PTRCALL *nameMatchesAscii)(const ENCODING *,
michael@0:                                   const char *,
michael@0:                                   const char *,
michael@0:                                   const char *);
michael@0:   int (PTRFASTCALL *nameLength)(const ENCODING *, const char *);
michael@0:   const char *(PTRFASTCALL *skipS)(const ENCODING *, const char *);
michael@0:   int (PTRCALL *getAtts)(const ENCODING *enc,
michael@0:                          const char *ptr,
michael@0:                          int attsMax,
michael@0:                          ATTRIBUTE *atts);
michael@0:   int (PTRFASTCALL *charRefNumber)(const ENCODING *enc, const char *ptr);
michael@0:   int (PTRCALL *predefinedEntityName)(const ENCODING *,
michael@0:                                       const char *,
michael@0:                                       const char *);
michael@0:   void (PTRCALL *updatePosition)(const ENCODING *,
michael@0:                                  const char *ptr,
michael@0:                                  const char *end,
michael@0:                                  POSITION *);
michael@0:   int (PTRCALL *isPublicId)(const ENCODING *enc,
michael@0:                             const char *ptr,
michael@0:                             const char *end,
michael@0:                             const char **badPtr);
michael@0:   void (PTRCALL *utf8Convert)(const ENCODING *enc,
michael@0:                               const char **fromP,
michael@0:                               const char *fromLim,
michael@0:                               char **toP,
michael@0:                               const char *toLim);
michael@0:   void (PTRCALL *utf16Convert)(const ENCODING *enc,
michael@0:                                const char **fromP,
michael@0:                                const char *fromLim,
michael@0:                                unsigned short **toP,
michael@0:                                const unsigned short *toLim);
michael@0:   int minBytesPerChar;
michael@0:   char isUtf8;
michael@0:   char isUtf16;
michael@0: };
michael@0: 
michael@0: /* Scan the string starting at ptr until the end of the next complete
michael@0:    token, but do not scan past eptr.  Return an integer giving the
michael@0:    type of token.
michael@0: 
michael@0:    Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
michael@0: 
michael@0:    Return XML_TOK_PARTIAL when the string does not contain a complete
michael@0:    token; nextTokPtr will not be set.
michael@0: 
michael@0:    Return XML_TOK_INVALID when the string does not start a valid
michael@0:    token; nextTokPtr will be set to point to the character which made
michael@0:    the token invalid.
michael@0: 
michael@0:    Otherwise the string starts with a valid token; nextTokPtr will be
michael@0:    set to point to the character following the end of that token.
michael@0: 
michael@0:    Each data character counts as a single token, but adjacent data
michael@0:    characters may be returned together.  Similarly for characters in
michael@0:    the prolog outside literals, comments and processing instructions.
michael@0: */
michael@0: 
michael@0: 
michael@0: #define XmlTok(enc, state, ptr, end, nextTokPtr) \
michael@0:   (((enc)->scanners[state])(enc, ptr, end, nextTokPtr))
michael@0: 
michael@0: #define XmlPrologTok(enc, ptr, end, nextTokPtr) \
michael@0:    XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr)
michael@0: 
michael@0: #define XmlContentTok(enc, ptr, end, nextTokPtr) \
michael@0:    XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr)
michael@0: 
michael@0: #define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \
michael@0:    XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr)
michael@0: 
michael@0: #ifdef XML_DTD
michael@0: 
michael@0: #define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \
michael@0:    XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr)
michael@0: 
michael@0: #endif /* XML_DTD */
michael@0: 
michael@0: /* This is used for performing a 2nd-level tokenization on the content
michael@0:    of a literal that has already been returned by XmlTok.
michael@0: */
michael@0: #define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \
michael@0:   (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr))
michael@0: 
michael@0: #define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \
michael@0:    XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr)
michael@0: 
michael@0: #define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \
michael@0:    XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr)
michael@0: 
michael@0: #define XmlSameName(enc, ptr1, ptr2) (((enc)->sameName)(enc, ptr1, ptr2))
michael@0: 
michael@0: #define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \
michael@0:   (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2))
michael@0: 
michael@0: #define XmlNameLength(enc, ptr) \
michael@0:   (((enc)->nameLength)(enc, ptr))
michael@0: 
michael@0: #define XmlSkipS(enc, ptr) \
michael@0:   (((enc)->skipS)(enc, ptr))
michael@0: 
michael@0: #define XmlGetAttributes(enc, ptr, attsMax, atts) \
michael@0:   (((enc)->getAtts)(enc, ptr, attsMax, atts))
michael@0: 
michael@0: #define XmlCharRefNumber(enc, ptr) \
michael@0:   (((enc)->charRefNumber)(enc, ptr))
michael@0: 
michael@0: #define XmlPredefinedEntityName(enc, ptr, end) \
michael@0:   (((enc)->predefinedEntityName)(enc, ptr, end))
michael@0: 
michael@0: #define XmlUpdatePosition(enc, ptr, end, pos) \
michael@0:   (((enc)->updatePosition)(enc, ptr, end, pos))
michael@0: 
michael@0: #define XmlIsPublicId(enc, ptr, end, badPtr) \
michael@0:   (((enc)->isPublicId)(enc, ptr, end, badPtr))
michael@0: 
michael@0: #define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \
michael@0:   (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim))
michael@0: 
michael@0: #define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \
michael@0:   (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim))
michael@0: 
michael@0: typedef struct {
michael@0:   ENCODING initEnc;
michael@0:   const ENCODING **encPtr;
michael@0: } INIT_ENCODING;
michael@0: 
michael@0: int XmlParseXmlDecl(int isGeneralTextEntity,
michael@0:                     const ENCODING *enc,
michael@0:                     const char *ptr,
michael@0:                     const char *end,
michael@0:                     const char **badPtr,
michael@0:                     const char **versionPtr,
michael@0:                     const char **versionEndPtr,
michael@0:                     const char **encodingNamePtr,
michael@0:                     const ENCODING **namedEncodingPtr,
michael@0:                     int *standalonePtr);
michael@0: 
michael@0: int XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name);
michael@0: const ENCODING *XmlGetUtf8InternalEncoding(void);
michael@0: const ENCODING *XmlGetUtf16InternalEncoding(void);
michael@0: int FASTCALL XmlUtf8Encode(int charNumber, char *buf);
michael@0: int FASTCALL XmlUtf16Encode(int charNumber, unsigned short *buf);
michael@0: int XmlSizeOfUnknownEncoding(void);
michael@0: 
michael@0: 
michael@0: typedef int (XMLCALL *CONVERTER) (void *userData, const char *p);
michael@0: 
michael@0: ENCODING *
michael@0: XmlInitUnknownEncoding(void *mem,
michael@0:                        int *table,
michael@0:                        CONVERTER convert,
michael@0:                        void *userData);
michael@0: 
michael@0: int XmlParseXmlDeclNS(int isGeneralTextEntity,
michael@0:                       const ENCODING *enc,
michael@0:                       const char *ptr,
michael@0:                       const char *end,
michael@0:                       const char **badPtr,
michael@0:                       const char **versionPtr,
michael@0:                       const char **versionEndPtr,
michael@0:                       const char **encodingNamePtr,
michael@0:                       const ENCODING **namedEncodingPtr,
michael@0:                       int *standalonePtr);
michael@0: 
michael@0: int XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name);
michael@0: const ENCODING *XmlGetUtf8InternalEncodingNS(void);
michael@0: const ENCODING *XmlGetUtf16InternalEncodingNS(void);
michael@0: ENCODING *
michael@0: XmlInitUnknownEncodingNS(void *mem,
michael@0:                          int *table,
michael@0:                          CONVERTER convert,
michael@0:                          void *userData);
michael@0: #ifdef __cplusplus
michael@0: }
michael@0: #endif
michael@0: 
michael@0: #endif /* not XmlTok_INCLUDED */