123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327 |
- /*
- Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
- See the file copying.txt for copying permission.
- */
- #ifndef XMLTOK_H_INCLUDED
- #define XMLTOK_H_INCLUDED
- #include "xmlrpc-c/c_util.h"
- #ifdef __cplusplus
- extern "C" {
- #endif
- /* The following token may be returned by XmlContentTok */
- #define XML_TOK_TRAILING_RSQB -5 /* ] or ]] at the end of the scan; might be start of
- illegal ]]> sequence */
- /* The following tokens may be returned by both XmlPrologTok and XmlContentTok */
- #define XML_TOK_NONE -4 /* The string to be scanned is empty */
- #define XML_TOK_TRAILING_CR -3 /* A CR at the end of the scan;
- might be part of CRLF sequence */
- #define XML_TOK_PARTIAL_CHAR -2 /* only part of a multibyte sequence */
- #define XML_TOK_PARTIAL -1 /* only part of a token */
- #define XML_TOK_INVALID 0
- /* The following tokens are returned by XmlContentTok; some are also
- returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok */
- #define XML_TOK_START_TAG_WITH_ATTS 1
- #define XML_TOK_START_TAG_NO_ATTS 2
- #define XML_TOK_EMPTY_ELEMENT_WITH_ATTS 3 /* empty element tag <e/> */
- #define XML_TOK_EMPTY_ELEMENT_NO_ATTS 4
- #define XML_TOK_END_TAG 5
- #define XML_TOK_DATA_CHARS 6
- #define XML_TOK_DATA_NEWLINE 7
- #define XML_TOK_CDATA_SECT_OPEN 8
- #define XML_TOK_ENTITY_REF 9
- #define XML_TOK_CHAR_REF 10 /* numeric character reference */
- /* The following tokens may be returned by both XmlPrologTok and XmlContentTok */
- #define XML_TOK_PI 11 /* processing instruction */
- #define XML_TOK_XML_DECL 12 /* XML decl or text decl */
- #define XML_TOK_COMMENT 13
- #define XML_TOK_BOM 14 /* Byte order mark */
- /* The following tokens are returned only by XmlPrologTok */
- #define XML_TOK_PROLOG_S 15
- #define XML_TOK_DECL_OPEN 16 /* <!foo */
- #define XML_TOK_DECL_CLOSE 17 /* > */
- #define XML_TOK_NAME 18
- #define XML_TOK_NMTOKEN 19
- #define XML_TOK_POUND_NAME 20 /* #name */
- #define XML_TOK_OR 21 /* | */
- #define XML_TOK_PERCENT 22
- #define XML_TOK_OPEN_PAREN 23
- #define XML_TOK_CLOSE_PAREN 24
- #define XML_TOK_OPEN_BRACKET 25
- #define XML_TOK_CLOSE_BRACKET 26
- #define XML_TOK_LITERAL 27
- #define XML_TOK_PARAM_ENTITY_REF 28
- #define XML_TOK_INSTANCE_START 29
- /* The following occur only in element type declarations */
- #define XML_TOK_NAME_QUESTION 30 /* name? */
- #define XML_TOK_NAME_ASTERISK 31 /* name* */
- #define XML_TOK_NAME_PLUS 32 /* name+ */
- #define XML_TOK_COND_SECT_OPEN 33 /* <![ */
- #define XML_TOK_COND_SECT_CLOSE 34 /* ]]> */
- #define XML_TOK_CLOSE_PAREN_QUESTION 35 /* )? */
- #define XML_TOK_CLOSE_PAREN_ASTERISK 36 /* )* */
- #define XML_TOK_CLOSE_PAREN_PLUS 37 /* )+ */
- #define XML_TOK_COMMA 38
- /* The following token is returned only by XmlAttributeValueTok */
- #define XML_TOK_ATTRIBUTE_VALUE_S 39
- /* The following token is returned only by XmlCdataSectionTok */
- #define XML_TOK_CDATA_SECT_CLOSE 40
- /* With namespace processing this is returned by XmlPrologTok
- for a name with a colon. */
- #define XML_TOK_PREFIXED_NAME 41
- #define XML_TOK_IGNORE_SECT 42
- #define XML_N_STATES 4
- #define XML_PROLOG_STATE 0
- #define XML_CONTENT_STATE 1
- #define XML_CDATA_SECTION_STATE 2
- #define XML_IGNORE_SECTION_STATE 3
- #define XML_N_LITERAL_TYPES 2
- #define XML_ATTRIBUTE_VALUE_LITERAL 0
- #define XML_ENTITY_VALUE_LITERAL 1
- /* The size of the buffer passed to XmlUtf8Encode must be at least this. */
- #define XML_UTF8_ENCODE_MAX 4
- /* The size of the buffer passed to XmlUtf16Encode must be at least this. */
- #define XML_UTF16_ENCODE_MAX 2
- typedef struct position {
- /* first line and first column are 0 not 1 */
- unsigned long lineNumber;
- unsigned long columnNumber;
- } POSITION;
- typedef struct {
- const char *name;
- const char *valuePtr;
- const char *valueEnd;
- char normalized;
- } ATTRIBUTE;
- struct encoding;
- typedef struct encoding ENCODING;
- struct encoding {
- int (*scanners[XML_N_STATES])(const ENCODING *,
- const char *,
- const char *,
- const char **);
- int (*literalScanners[XML_N_LITERAL_TYPES])(const ENCODING *,
- const char *,
- const char *,
- const char **);
- int (*sameName)(const ENCODING *,
- const char *, const char *);
- int (*nameMatchesAscii)(const ENCODING *,
- const char *, const char *, const char *);
- size_t (*nameLength)(const ENCODING *, const char *);
- const char *(*skipS)(const ENCODING *, const char *);
- int (*getAtts)(const ENCODING *enc, const char *ptr,
- int attsMax, ATTRIBUTE *atts);
- int (*charRefNumber)(const ENCODING *enc, const char *ptr);
- int (*predefinedEntityName)(const ENCODING *, const char *, const char *);
- void (*updatePosition)(const ENCODING *,
- const char *ptr,
- const char *end,
- POSITION *);
- int (*isPublicId)(const ENCODING *enc, const char *ptr, const char *end,
- const char **badPtr);
- void (*utf8Convert)(const ENCODING *enc,
- const char **fromP,
- const char *fromLim,
- char **toP,
- const char *toLim);
- void (*utf16Convert)(const ENCODING *enc,
- const char **fromP,
- const char *fromLim,
- unsigned short **toP,
- const unsigned short *toLim);
- int minBytesPerChar;
- char isUtf8;
- char isUtf16;
- };
- /*
- Scan the string starting at ptr until the end of the next complete token,
- but do not scan past eptr. Return an integer giving the type of token.
- Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
- Return XML_TOK_PARTIAL when the string does not contain a complete token;
- nextTokPtr will not be set.
- Return XML_TOK_INVALID when the string does not start a valid token; nextTokPtr
- will be set to point to the character which made the token invalid.
- Otherwise the string starts with a valid token; nextTokPtr will be set to point
- to the character following the end of that token.
- Each data character counts as a single token, but adjacent data characters
- may be returned together. Similarly for characters in the prolog outside
- literals, comments and processing instructions.
- */
- #define XmlTok(enc, state, ptr, end, nextTokPtr) \
- (((enc)->scanners[state])(enc, ptr, end, nextTokPtr))
- #define XmlPrologTok(enc, ptr, end, nextTokPtr) \
- XmlTok(enc, XML_PROLOG_STATE, ptr, end, nextTokPtr)
- #define XmlContentTok(enc, ptr, end, nextTokPtr) \
- XmlTok(enc, XML_CONTENT_STATE, ptr, end, nextTokPtr)
- #define XmlCdataSectionTok(enc, ptr, end, nextTokPtr) \
- XmlTok(enc, XML_CDATA_SECTION_STATE, ptr, end, nextTokPtr)
- #define XmlIgnoreSectionTok(enc, ptr, end, nextTokPtr) \
- XmlTok(enc, XML_IGNORE_SECTION_STATE, ptr, end, nextTokPtr)
- /* This is used for performing a 2nd-level tokenization on
- the content of a literal that has already been returned by XmlTok. */
- #define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \
- (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr))
- #define XmlAttributeValueTok(enc, ptr, end, nextTokPtr) \
- XmlLiteralTok(enc, XML_ATTRIBUTE_VALUE_LITERAL, ptr, end, nextTokPtr)
- #define XmlEntityValueTok(enc, ptr, end, nextTokPtr) \
- XmlLiteralTok(enc, XML_ENTITY_VALUE_LITERAL, ptr, end, nextTokPtr)
- #define XmlSameName(enc, ptr1, ptr2) (((enc)->sameName)(enc, ptr1, ptr2))
- #define XmlNameMatchesAscii(enc, ptr1, end1, ptr2) \
- (((enc)->nameMatchesAscii)(enc, ptr1, end1, ptr2))
- #define XmlNameLength(enc, ptr) \
- (((enc)->nameLength)(enc, ptr))
- #define XmlSkipS(enc, ptr) \
- (((enc)->skipS)(enc, ptr))
- #define XmlGetAttributes(enc, ptr, attsMax, atts) \
- (((enc)->getAtts)(enc, ptr, attsMax, atts))
- #define XmlCharRefNumber(enc, ptr) \
- (((enc)->charRefNumber)(enc, ptr))
- #define XmlPredefinedEntityName(enc, ptr, end) \
- (((enc)->predefinedEntityName)(enc, ptr, end))
- #define XmlUpdatePosition(enc, ptr, end, pos) \
- (((enc)->updatePosition)(enc, ptr, end, pos))
- #define XmlIsPublicId(enc, ptr, end, badPtr) \
- (((enc)->isPublicId)(enc, ptr, end, badPtr))
- #define XmlUtf8Convert(enc, fromP, fromLim, toP, toLim) \
- (((enc)->utf8Convert)(enc, fromP, fromLim, toP, toLim))
- #define XmlUtf16Convert(enc, fromP, fromLim, toP, toLim) \
- (((enc)->utf16Convert)(enc, fromP, fromLim, toP, toLim))
- typedef struct {
- ENCODING initEnc;
- const ENCODING **encPtr;
- } INIT_ENCODING;
- XMLRPC_DLLEXPORT
- int
- xmlrpc_XmlParseXmlDecl(int const isGeneralTextEntity,
- const ENCODING * const enc,
- const char * const ptr,
- const char * const end,
- const char ** const badPtr,
- const char ** const versionPtr,
- const char ** const encodingNamePtr,
- const ENCODING ** const namedEncodingPtr,
- int * const standalonePtr);
- XMLRPC_DLLEXPORT
- const ENCODING *
- xmlrpc_XmlGetUtf8InternalEncoding(void);
- XMLRPC_DLLEXPORT
- const ENCODING *
- xmlrpc_XmlGetUtf16InternalEncoding(void);
- XMLRPC_DLLEXPORT
- int
- xmlrpc_XmlInitEncoding(INIT_ENCODING * const p,
- const ENCODING ** const encPtr,
- const char * const name);
- XMLRPC_DLLEXPORT
- int
- xmlrpc_XmlUtf8Encode(int const c,
- char * const buf);
- XMLRPC_DLLEXPORT
- int
- xmlrpc_XmlUtf16Encode(int const charNum,
- unsigned short * const buf);
- XMLRPC_DLLEXPORT
- int
- xmlrpc_XmlSizeOfUnknownEncoding(void);
- XMLRPC_DLLEXPORT
- ENCODING *
- xmlrpc_XmlInitUnknownEncoding(void * const mem,
- int * const table,
- int (*convert)(void *userData, const char *p),
- void * const userData);
- XMLRPC_DLLEXPORT
- int
- xmlrpc_XmlParseXmlDeclNS(int const isGeneralTextEntity,
- const ENCODING * const enc,
- const char * const ptr,
- const char * const end,
- const char ** const badPtr,
- const char ** const versionPtr,
- const char ** const encodingNamePtr,
- const ENCODING ** const namedEncodingPtr,
- int * const standalonePtr);
- XMLRPC_DLLEXPORT
- int
- xmlrpc_XmlInitEncodingNS(INIT_ENCODING * const p,
- const ENCODING ** const encPtr,
- const char * const name);
- XMLRPC_DLLEXPORT
- const ENCODING *
- xmlrpc_XmlGetUtf8InternalEncodingNS(void);
- XMLRPC_DLLEXPORT
- const ENCODING *
- xmlrpc_XmlGetUtf16InternalEncodingNS(void);
- XMLRPC_DLLEXPORT
- ENCODING *
- xmlrpc_XmlInitUnknownEncodingNS(void * const mem,
- int * const table,
- int (*convert)(void *userData, const char *p),
- void * const userData);
- #ifdef __cplusplus
- }
- #endif
- #endif /* not XmlTok_INCLUDED */
|