1  /*
  2   * Summary: interface for an HTML 4.0 non-verifying parser
  3   * Description: this module implements an HTML 4.0 non-verifying parser
  4   *              with API compatible with the XML parser ones. It should
  5   *              be able to parse "real world" HTML, even if severely
  6   *              broken from a specification point of view.
  7   *
  8   * Copy: See Copyright for the status of this software.
  9   *
 10   * Author: Daniel Veillard
 11   */
 12  
 13  #ifndef __HTML_PARSER_H__
 14  #define __HTML_PARSER_H__
 15  #include <libxml/xmlversion.h>
 16  #include <libxml/encoding.h>
 17  #include <libxml/parser.h>
 18  #include <libxml/xmlIO.h>
 19  #include <libxml/xmlstring.h>
 20  
 21  #ifdef LIBXML_HTML_ENABLED
 22  
 23  #ifdef __cplusplus
 24  extern "C" {
 25  #endif
 26  
 27  /*
 28   * Most of the back-end structures from XML and HTML are shared.
 29   */
 30  typedef xmlParserCtxt htmlParserCtxt;
 31  typedef xmlParserCtxtPtr htmlParserCtxtPtr;
 32  typedef xmlParserNodeInfo htmlParserNodeInfo;
 33  typedef xmlSAXHandler htmlSAXHandler;
 34  typedef xmlSAXHandlerPtr htmlSAXHandlerPtr;
 35  typedef xmlParserInput htmlParserInput;
 36  typedef xmlParserInputPtr htmlParserInputPtr;
 37  typedef xmlDocPtr htmlDocPtr;
 38  typedef xmlNodePtr htmlNodePtr;
 39  
 40  /*
 41   * Internal description of an HTML element, representing HTML 4.01
 42   * and XHTML 1.0 (which share the same structure).
 43   */
 44  typedef struct _htmlElemDesc htmlElemDesc;
 45  typedef htmlElemDesc *htmlElemDescPtr;
 46  struct _htmlElemDesc {
 47      const char *name;	/* The tag name */
 48      char startTag;      /* Whether the start tag can be implied */
 49      char endTag;        /* Whether the end tag can be implied */
 50      char saveEndTag;    /* Whether the end tag should be saved */
 51      char empty;         /* Is this an empty element ? */
 52      char depr;          /* Is this a deprecated element ? */
 53      char dtd;           /* 1: only in Loose DTD, 2: only Frameset one */
 54      char isinline;      /* is this a block 0 or inline 1 element */
 55      const char *desc;   /* the description */
 56  
 57  /* NRK Jan.2003
 58   * New fields encapsulating HTML structure
 59   *
 60   * Bugs:
 61   *	This is a very limited representation.  It fails to tell us when
 62   *	an element *requires* subelements (we only have whether they're
 63   *	allowed or not), and it doesn't tell us where CDATA and PCDATA
 64   *	are allowed.  Some element relationships are not fully represented:
 65   *	these are flagged with the word MODIFIER
 66   */
 67      const char** subelts;		/* allowed sub-elements of this element */
 68      const char* defaultsubelt;	/* subelement for suggested auto-repair
 69  					   if necessary or NULL */
 70      const char** attrs_opt;		/* Optional Attributes */
 71      const char** attrs_depr;		/* Additional deprecated attributes */
 72      const char** attrs_req;		/* Required attributes */
 73  };
 74  
 75  /*
 76   * Internal description of an HTML entity.
 77   */
 78  typedef struct _htmlEntityDesc htmlEntityDesc;
 79  typedef htmlEntityDesc *htmlEntityDescPtr;
 80  struct _htmlEntityDesc {
 81      unsigned int value;	/* the UNICODE value for the character */
 82      const char *name;	/* The entity name */
 83      const char *desc;   /* the description */
 84  };
 85  
 86  /*
 87   * There is only few public functions.
 88   */
 89  XMLPUBFUN const htmlElemDesc * XMLCALL
 90  			htmlTagLookup	(const xmlChar *tag);
 91  XMLPUBFUN const htmlEntityDesc * XMLCALL
 92  			htmlEntityLookup(const xmlChar *name);
 93  XMLPUBFUN const htmlEntityDesc * XMLCALL
 94  			htmlEntityValueLookup(unsigned int value);
 95  
 96  XMLPUBFUN int XMLCALL
 97  			htmlIsAutoClosed(htmlDocPtr doc,
 98  					 htmlNodePtr elem);
 99  XMLPUBFUN int XMLCALL
100  			htmlAutoCloseTag(htmlDocPtr doc,
101  					 const xmlChar *name,
102  					 htmlNodePtr elem);
103  XMLPUBFUN const htmlEntityDesc * XMLCALL
104  			htmlParseEntityRef(htmlParserCtxtPtr ctxt,
105  					 const xmlChar **str);
106  XMLPUBFUN int XMLCALL
107  			htmlParseCharRef(htmlParserCtxtPtr ctxt);
108  XMLPUBFUN void XMLCALL
109  			htmlParseElement(htmlParserCtxtPtr ctxt);
110  
111  XMLPUBFUN htmlParserCtxtPtr XMLCALL
112  			htmlNewParserCtxt(void);
113  
114  XMLPUBFUN htmlParserCtxtPtr XMLCALL
115  			htmlCreateMemoryParserCtxt(const char *buffer,
116  						   int size);
117  
118  XMLPUBFUN int XMLCALL
119  			htmlParseDocument(htmlParserCtxtPtr ctxt);
120  XMLPUBFUN htmlDocPtr XMLCALL
121  			htmlSAXParseDoc	(const xmlChar *cur,
122  					 const char *encoding,
123  					 htmlSAXHandlerPtr sax,
124  					 void *userData);
125  XMLPUBFUN htmlDocPtr XMLCALL
126  			htmlParseDoc	(const xmlChar *cur,
127  					 const char *encoding);
128  XMLPUBFUN htmlDocPtr XMLCALL
129  			htmlSAXParseFile(const char *filename,
130  					 const char *encoding,
131  					 htmlSAXHandlerPtr sax,
132  					 void *userData);
133  XMLPUBFUN htmlDocPtr XMLCALL
134  			htmlParseFile	(const char *filename,
135  					 const char *encoding);
136  XMLPUBFUN int XMLCALL
137  			UTF8ToHtml	(unsigned char *out,
138  					 int *outlen,
139  					 const unsigned char *in,
140  					 int *inlen);
141  XMLPUBFUN int XMLCALL
142  			htmlEncodeEntities(unsigned char *out,
143  					 int *outlen,
144  					 const unsigned char *in,
145  					 int *inlen, int quoteChar);
146  XMLPUBFUN int XMLCALL
147  			htmlIsScriptAttribute(const xmlChar *name);
148  XMLPUBFUN int XMLCALL
149  			htmlHandleOmittedElem(int val);
150  
151  #ifdef LIBXML_PUSH_ENABLED
152  /**
153   * Interfaces for the Push mode.
154   */
155  XMLPUBFUN htmlParserCtxtPtr XMLCALL
156  			htmlCreatePushParserCtxt(htmlSAXHandlerPtr sax,
157  						 void *user_data,
158  						 const char *chunk,
159  						 int size,
160  						 const char *filename,
161  						 xmlCharEncoding enc);
162  XMLPUBFUN int XMLCALL
163  			htmlParseChunk		(htmlParserCtxtPtr ctxt,
164  						 const char *chunk,
165  						 int size,
166  						 int terminate);
167  #endif /* LIBXML_PUSH_ENABLED */
168  
169  XMLPUBFUN void XMLCALL
170  			htmlFreeParserCtxt	(htmlParserCtxtPtr ctxt);
171  
172  /*
173   * New set of simpler/more flexible APIs
174   */
175  /**
176   * xmlParserOption:
177   *
178   * This is the set of XML parser options that can be passed down
179   * to the xmlReadDoc() and similar calls.
180   */
181  typedef enum {
182      HTML_PARSE_RECOVER  = 1<<0, /* Relaxed parsing */
183      HTML_PARSE_NODEFDTD = 1<<2, /* do not default a doctype if not found */
184      HTML_PARSE_NOERROR	= 1<<5,	/* suppress error reports */
185      HTML_PARSE_NOWARNING= 1<<6,	/* suppress warning reports */
186      HTML_PARSE_PEDANTIC	= 1<<7,	/* pedantic error reporting */
187      HTML_PARSE_NOBLANKS	= 1<<8,	/* remove blank nodes */
188      HTML_PARSE_NONET	= 1<<11,/* Forbid network access */
189      HTML_PARSE_NOIMPLIED= 1<<13,/* Do not add implied html/body... elements */
190      HTML_PARSE_COMPACT  = 1<<16,/* compact small text nodes */
191      HTML_PARSE_IGNORE_ENC=1<<21 /* ignore internal document encoding hint */
192  } htmlParserOption;
193  
194  XMLPUBFUN void XMLCALL
195  		htmlCtxtReset		(htmlParserCtxtPtr ctxt);
196  XMLPUBFUN int XMLCALL
197  		htmlCtxtUseOptions	(htmlParserCtxtPtr ctxt,
198  					 int options);
199  XMLPUBFUN htmlDocPtr XMLCALL
200  		htmlReadDoc		(const xmlChar *cur,
201  					 const char *URL,
202  					 const char *encoding,
203  					 int options);
204  XMLPUBFUN htmlDocPtr XMLCALL
205  		htmlReadFile		(const char *URL,
206  					 const char *encoding,
207  					 int options);
208  XMLPUBFUN htmlDocPtr XMLCALL
209  		htmlReadMemory		(const char *buffer,
210  					 int size,
211  					 const char *URL,
212  					 const char *encoding,
213  					 int options);
214  XMLPUBFUN htmlDocPtr XMLCALL
215  		htmlReadFd		(int fd,
216  					 const char *URL,
217  					 const char *encoding,
218  					 int options);
219  XMLPUBFUN htmlDocPtr XMLCALL
220  		htmlReadIO		(xmlInputReadCallback ioread,
221  					 xmlInputCloseCallback ioclose,
222  					 void *ioctx,
223  					 const char *URL,
224  					 const char *encoding,
225  					 int options);
226  XMLPUBFUN htmlDocPtr XMLCALL
227  		htmlCtxtReadDoc		(xmlParserCtxtPtr ctxt,
228  					 const xmlChar *cur,
229  					 const char *URL,
230  					 const char *encoding,
231  					 int options);
232  XMLPUBFUN htmlDocPtr XMLCALL
233  		htmlCtxtReadFile		(xmlParserCtxtPtr ctxt,
234  					 const char *filename,
235  					 const char *encoding,
236  					 int options);
237  XMLPUBFUN htmlDocPtr XMLCALL
238  		htmlCtxtReadMemory		(xmlParserCtxtPtr ctxt,
239  					 const char *buffer,
240  					 int size,
241  					 const char *URL,
242  					 const char *encoding,
243  					 int options);
244  XMLPUBFUN htmlDocPtr XMLCALL
245  		htmlCtxtReadFd		(xmlParserCtxtPtr ctxt,
246  					 int fd,
247  					 const char *URL,
248  					 const char *encoding,
249  					 int options);
250  XMLPUBFUN htmlDocPtr XMLCALL
251  		htmlCtxtReadIO		(xmlParserCtxtPtr ctxt,
252  					 xmlInputReadCallback ioread,
253  					 xmlInputCloseCallback ioclose,
254  					 void *ioctx,
255  					 const char *URL,
256  					 const char *encoding,
257  					 int options);
258  
259  /* NRK/Jan2003: further knowledge of HTML structure
260   */
261  typedef enum {
262    HTML_NA = 0 ,		/* something we don't check at all */
263    HTML_INVALID = 0x1 ,
264    HTML_DEPRECATED = 0x2 ,
265    HTML_VALID = 0x4 ,
266    HTML_REQUIRED = 0xc /* VALID bit set so ( & HTML_VALID ) is TRUE */
267  } htmlStatus ;
268  
269  /* Using htmlElemDesc rather than name here, to emphasise the fact
270     that otherwise there's a lookup overhead
271  */
272  XMLPUBFUN htmlStatus XMLCALL htmlAttrAllowed(const htmlElemDesc*, const xmlChar*, int) ;
273  XMLPUBFUN int XMLCALL htmlElementAllowedHere(const htmlElemDesc*, const xmlChar*) ;
274  XMLPUBFUN htmlStatus XMLCALL htmlElementStatusHere(const htmlElemDesc*, const htmlElemDesc*) ;
275  XMLPUBFUN htmlStatus XMLCALL htmlNodeStatus(const htmlNodePtr, int) ;
276  /**
277   * htmlDefaultSubelement:
278   * @elt: HTML element
279   *
280   * Returns the default subelement for this element
281   */
282  #define htmlDefaultSubelement(elt) elt->defaultsubelt
283  /**
284   * htmlElementAllowedHereDesc:
285   * @parent: HTML parent element
286   * @elt: HTML element
287   *
288   * Checks whether an HTML element description may be a
289   * direct child of the specified element.
290   *
291   * Returns 1 if allowed; 0 otherwise.
292   */
293  #define htmlElementAllowedHereDesc(parent,elt) \
294  	htmlElementAllowedHere((parent), (elt)->name)
295  /**
296   * htmlRequiredAttrs:
297   * @elt: HTML element
298   *
299   * Returns the attributes required for the specified element.
300   */
301  #define htmlRequiredAttrs(elt) (elt)->attrs_req
302  
303  
304  #ifdef __cplusplus
305  }
306  #endif
307  
308  #endif /* LIBXML_HTML_ENABLED */
309  #endif /* __HTML_PARSER_H__ */