Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 1 | /* |
| 2 | * HTMLparser.h : inf=terface for an HTML 4.0 non-verifying parser |
| 3 | * |
| 4 | * See Copyright for the status of this software. |
| 5 | * |
| 6 | * Daniel.Veillard@w3.org |
| 7 | */ |
| 8 | |
| 9 | #ifndef __HTML_PARSER_H__ |
| 10 | #define __HTML_PARSER_H__ |
| 11 | #include "parser.h" |
| 12 | |
Daniel Veillard | f600e25 | 1999-12-18 15:32:46 +0000 | [diff] [blame] | 13 | #ifdef __cplusplus |
Daniel Veillard | 5cb5ab8 | 1999-12-21 15:35:29 +0000 | [diff] [blame^] | 14 | extern "C" { |
Daniel Veillard | f600e25 | 1999-12-18 15:32:46 +0000 | [diff] [blame] | 15 | #endif |
| 16 | |
Daniel Veillard | 5233ffc | 1999-07-06 22:25:25 +0000 | [diff] [blame] | 17 | /* |
| 18 | * Most of the back-end structures from XML and HTML are shared |
| 19 | */ |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 20 | typedef xmlParserCtxt htmlParserCtxt; |
| 21 | typedef xmlParserCtxtPtr htmlParserCtxtPtr; |
| 22 | typedef xmlParserNodeInfo htmlParserNodeInfo; |
| 23 | typedef xmlSAXHandler htmlSAXHandler; |
| 24 | typedef xmlSAXHandlerPtr htmlSAXHandlerPtr; |
| 25 | typedef xmlParserInput htmlParserInput; |
| 26 | typedef xmlParserInputPtr htmlParserInputPtr; |
| 27 | typedef xmlDocPtr htmlDocPtr; |
| 28 | typedef xmlNodePtr htmlNodePtr; |
| 29 | |
Daniel Veillard | 5233ffc | 1999-07-06 22:25:25 +0000 | [diff] [blame] | 30 | /* |
| 31 | * Internal description of an HTML element |
| 32 | */ |
| 33 | typedef struct htmlElemDesc { |
Daniel Veillard | b96e643 | 1999-08-29 21:02:19 +0000 | [diff] [blame] | 34 | const char *name; /* The tag name */ |
Daniel Veillard | 5233ffc | 1999-07-06 22:25:25 +0000 | [diff] [blame] | 35 | int startTag; /* Whether the start tag can be implied */ |
| 36 | int endTag; /* Whether the end tag can be implied */ |
| 37 | int empty; /* Is this an empty element ? */ |
| 38 | int depr; /* Is this a deprecated element ? */ |
| 39 | int dtd; /* 1: only in Loose DTD, 2: only Frameset one */ |
| 40 | const char *desc; /* the description */ |
| 41 | } htmlElemDesc, *htmlElemDescPtr; |
| 42 | |
| 43 | /* |
| 44 | * Internal description of an HTML entity |
| 45 | */ |
| 46 | typedef struct htmlEntityDesc { |
| 47 | int value; /* the UNICODE value for the character */ |
Daniel Veillard | b96e643 | 1999-08-29 21:02:19 +0000 | [diff] [blame] | 48 | const char *name; /* The entity name */ |
Daniel Veillard | 5233ffc | 1999-07-06 22:25:25 +0000 | [diff] [blame] | 49 | const char *desc; /* the description */ |
| 50 | } htmlEntityDesc, *htmlEntityDescPtr; |
| 51 | |
| 52 | /* |
| 53 | * There is only few public functions. |
| 54 | */ |
Daniel Veillard | 5cb5ab8 | 1999-12-21 15:35:29 +0000 | [diff] [blame^] | 55 | htmlElemDescPtr htmlTagLookup (const xmlChar *tag); |
| 56 | htmlEntityDescPtr htmlEntityLookup(const xmlChar *name); |
Daniel Veillard | 82150d8 | 1999-07-07 07:32:15 +0000 | [diff] [blame] | 57 | |
Daniel Veillard | 5cb5ab8 | 1999-12-21 15:35:29 +0000 | [diff] [blame^] | 58 | int htmlIsAutoClosed(htmlDocPtr doc, |
| 59 | htmlNodePtr elem); |
| 60 | int htmlAutoCloseTag(htmlDocPtr doc, |
| 61 | const xmlChar *name, |
| 62 | htmlNodePtr elem); |
| 63 | htmlEntityDescPtr htmlParseEntityRef(htmlParserCtxtPtr ctxt, |
| 64 | xmlChar **str); |
| 65 | int htmlParseCharRef(htmlParserCtxtPtr ctxt); |
| 66 | void htmlParseElement(htmlParserCtxtPtr ctxt); |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 67 | |
Daniel Veillard | 5cb5ab8 | 1999-12-21 15:35:29 +0000 | [diff] [blame^] | 68 | htmlDocPtr htmlSAXParseDoc (xmlChar *cur, |
| 69 | const char *encoding, |
| 70 | htmlSAXHandlerPtr sax, |
| 71 | void *userData); |
| 72 | htmlDocPtr htmlParseDoc (xmlChar *cur, |
| 73 | const char *encoding); |
| 74 | htmlDocPtr htmlSAXParseFile(const char *filename, |
| 75 | const char *encoding, |
| 76 | htmlSAXHandlerPtr sax, |
| 77 | void *userData); |
| 78 | htmlDocPtr htmlParseFile (const char *filename, |
| 79 | const char *encoding); |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 80 | |
Daniel Veillard | f600e25 | 1999-12-18 15:32:46 +0000 | [diff] [blame] | 81 | #ifdef __cplusplus |
| 82 | } |
| 83 | #endif |
| 84 | |
Daniel Veillard | be70ff7 | 1999-07-05 16:50:46 +0000 | [diff] [blame] | 85 | #endif /* __HTML_PARSER_H__ */ |