kc3-lang/libxml2/HTMLparser.h

• Show log

  Commit

• Hash : 32bc74ef
  Author :
  Date : 2000-07-14T14:49:25
  

  - doc/encoding.html doc/xml.html: added I18N doc
  - encoding.[ch] HTMLtree.[ch] parser.c HTMLparser.c: I18N encoding
    improvements, both parser and filters, added ASCII & HTML,
    fixed the ISO-Latin-1 one
  - xmllint.c testHTML.c: added/made visible --encode
  - debugXML.c : cleanup
  - most .c files: applied patches due to warning on Windows and
    when using Sun Pro cc compiler
  - xpath.c : cleanup memleaks
  - nanoftp.c : added a TESTING preprocessor flag for standalong
    compile so that people can report bugs more easilly
  - nanohttp.c : ditched socklen_t which was a portability mess
    and replaced it with unsigned int.
  - tree.[ch]: added xmlHasProp()
  - TODO: updated
  - test/ : added more test for entities, NS, encoding, HTML, wap
  - configure.in: preparing for 2.2.0 release
  Daniel
  

Download

• HTMLparser.h

• /*
   * HTMLparser.h : inf=terface for an HTML 4.0 non-verifying parser
   *
   * See Copyright for the status of this software.
   *
   * Daniel.Veillard@w3.org
   */
  
  #ifndef __HTML_PARSER_H__
  #define __HTML_PARSER_H__
  #include <libxml/parser.h>
  
  #ifdef __cplusplus
  extern "C" {
  #endif
  
  /*
   * Most of the back-end structures from XML and HTML are shared
   */
  typedef xmlParserCtxt htmlParserCtxt;
  typedef xmlParserCtxtPtr htmlParserCtxtPtr;
  typedef xmlParserNodeInfo htmlParserNodeInfo;
  typedef xmlSAXHandler htmlSAXHandler;
  typedef xmlSAXHandlerPtr htmlSAXHandlerPtr;
  typedef xmlParserInput htmlParserInput;
  typedef xmlParserInputPtr htmlParserInputPtr;
  typedef xmlDocPtr htmlDocPtr;
  typedef xmlNodePtr htmlNodePtr;
  
  /*
   * Internal description of an HTML element
   */
  typedef struct _htmlElemDesc htmlElemDesc;
  typedef htmlElemDesc *htmlElemDescPtr;
  struct _htmlElemDesc {
      const char *name;	/* The tag name */
      int startTag;       /* Whether the start tag can be implied */
      int endTag;         /* Whether the end tag can be implied */
      int empty;          /* Is this an empty element ? */
      int depr;           /* Is this a deprecated element ? */
      int dtd;            /* 1: only in Loose DTD, 2: only Frameset one */
      const char *desc;   /* the description */
  };
  
  /*
   * Internal description of an HTML entity
   */
  typedef struct _htmlEntityDesc htmlEntityDesc;
  typedef htmlEntityDesc *htmlEntityDescPtr;
  struct _htmlEntityDesc {
      int value;		/* the UNICODE value for the character */
      const char *name;	/* The entity name */
      const char *desc;   /* the description */
  };
  
  /*
   * There is only few public functions.
   */
  htmlElemDescPtr		htmlTagLookup	(const xmlChar *tag);
  htmlEntityDescPtr	htmlEntityLookup(const xmlChar *name);
  
  int			htmlIsAutoClosed(htmlDocPtr doc,
  					 htmlNodePtr elem);
  int			htmlAutoCloseTag(htmlDocPtr doc,
  					 const xmlChar *name,
  					 htmlNodePtr elem);
  htmlEntityDescPtr	htmlParseEntityRef(htmlParserCtxtPtr ctxt,
  					 xmlChar **str);
  int			htmlParseCharRef(htmlParserCtxtPtr ctxt);
  void			htmlParseElement(htmlParserCtxtPtr ctxt);
  
  htmlDocPtr		htmlSAXParseDoc	(xmlChar *cur,
  					 const char *encoding,
  					 htmlSAXHandlerPtr sax,
  					 void *userData);
  htmlDocPtr		htmlParseDoc	(xmlChar *cur,
  					 const char *encoding);
  htmlDocPtr		htmlSAXParseFile(const char *filename,
  					 const char *encoding,
  					 htmlSAXHandlerPtr sax,
  					 void *userData);
  htmlDocPtr		htmlParseFile	(const char *filename,
  					 const char *encoding);
  int			UTF8ToHtml	(unsigned char* out,
  					 int *outlen,
  					 const unsigned char* in,
  					 int *inlen);
  
  /**
   * Interfaces for the Push mode
   */
  void			htmlFreeParserCtxt	(htmlParserCtxtPtr ctxt);
  htmlParserCtxtPtr	htmlCreatePushParserCtxt(htmlSAXHandlerPtr sax,
  						 void *user_data,
  						 const char *chunk,
  						 int size,
  						 const char *filename,
  						 xmlCharEncoding enc);
  int			htmlParseChunk		(htmlParserCtxtPtr ctxt,
  						 const char *chunk,
  						 int size,
  						 int terminate);
  #ifdef __cplusplus
  }
  #endif
  
  #endif /* __HTML_PARSER_H__ */
  

Download