htmlparser.c

xml开源解析代码.版本为libxml2-2.6.29,可支持GB3212.网络消息发送XML时很有用.

			xmlParserInputGrow(ctxt->input, INPUT_CHUNK);
		    if (((c & 0xf8) != 0xf0) ||
			((cur[3] & 0xc0) != 0x80))
			goto encoding_error;
		    /* 4-byte code */
		    *len = 4;
		    val = (cur[0] & 0x7) << 18;
		    val |= (cur[1] & 0x3f) << 12;
		    val |= (cur[2] & 0x3f) << 6;
		    val |= cur[3] & 0x3f;
		} else {
		  /* 3-byte code */
		    *len = 3;
		    val = (cur[0] & 0xf) << 12;
		    val |= (cur[1] & 0x3f) << 6;
		    val |= cur[2] & 0x3f;
		}
	    } else {
	      /* 2-byte code */
		*len = 2;
		val = (cur[0] & 0x1f) << 6;
		val |= cur[1] & 0x3f;
	    }
	    if (!IS_CHAR(val)) {
	        htmlParseErrInt(ctxt, XML_ERR_INVALID_CHAR,
				"Char 0x%X out of allowed range\n", val);
	    }    
	    return(val);
	} else {
	    /* 1-byte code */
	    *len = 1;
	    return((int) *ctxt->input->cur);
	}
    }
    /*
     * Assume it's a fixed length encoding (1) with
     * a compatible encoding for the ASCII set, since
     * XML constructs only use < 128 chars
     */
    *len = 1;
    if ((int) *ctxt->input->cur < 0x80)
	return((int) *ctxt->input->cur);

    /*
     * Humm this is bad, do an automatic flow conversion
     */
    xmlSwitchEncoding(ctxt, XML_CHAR_ENCODING_8859_1);
    ctxt->charset = XML_CHAR_ENCODING_UTF8;
    return(xmlCurrentChar(ctxt, len));

encoding_error:
    /*
     * If we detect an UTF8 error that probably mean that the
     * input encoding didn't get properly advertized in the
     * declaration header. Report the error and switch the encoding
     * to ISO-Latin-1 (if you don't like this policy, just declare the
     * encoding !)
     */
    {
        char buffer[150];

	if (ctxt->input->end - ctxt->input->cur >= 4) {
	    snprintf(buffer, 149, "Bytes: 0x%02X 0x%02X 0x%02X 0x%02X\n",
			    ctxt->input->cur[0], ctxt->input->cur[1],
			    ctxt->input->cur[2], ctxt->input->cur[3]);
	} else {
	    snprintf(buffer, 149, "Bytes: 0x%02X\n", ctxt->input->cur[0]);
	}
	htmlParseErr(ctxt, XML_ERR_INVALID_ENCODING,
		     "Input is not proper UTF-8, indicate encoding !\n",
		     BAD_CAST buffer, NULL);
    }

    ctxt->charset = XML_CHAR_ENCODING_8859_1; 
    *len = 1;
    return((int) *ctxt->input->cur);
}

/**
 * htmlSkipBlankChars:
 * @ctxt:  the HTML parser context
 *
 * skip all blanks character found at that point in the input streams.
 *
 * Returns the number of space chars skipped
 */

static int
htmlSkipBlankChars(xmlParserCtxtPtr ctxt) {
    int res = 0;

    while (IS_BLANK_CH(*(ctxt->input->cur))) {
	if ((*ctxt->input->cur == 0) &&
	    (xmlParserInputGrow(ctxt->input, INPUT_CHUNK) <= 0)) {
		xmlPopInput(ctxt);
	} else {
	    if (*(ctxt->input->cur) == '\n') {
		ctxt->input->line++; ctxt->input->col = 1;
	    } else ctxt->input->col++;
	    ctxt->input->cur++;
	    ctxt->nbChars++;
	    if (*ctxt->input->cur == 0)
		xmlParserInputGrow(ctxt->input, INPUT_CHUNK);
	}
	res++;
    }
    return(res);
}



/************************************************************************
 *									*
 * 		The list of HTML elements and their properties		*
 *									*
 ************************************************************************/

/*
 *  Start Tag: 1 means the start tag can be ommited
 *  End Tag:   1 means the end tag can be ommited
 *             2 means it's forbidden (empty elements)
 *             3 means the tag is stylistic and should be closed easily
 *  Depr:      this element is deprecated
 *  DTD:       1 means that this element is valid only in the Loose DTD
 *             2 means that this element is valid only in the Frameset DTD
 *
 * Name,Start Tag,End Tag,Save End,Empty,Deprecated,DTD,inline,Description
	, subElements , impliedsubelt , Attributes, userdata
 */

/* Definitions and a couple of vars for HTML Elements */

#define FONTSTYLE "tt", "i", "b", "u", "s", "strike", "big", "small"
#define NB_FONTSTYLE 8
#define PHRASE "em", "strong", "dfn", "code", "samp", "kbd", "var", "cite", "abbr", "acronym"
#define NB_PHRASE 10
#define SPECIAL "a", "img", "applet", "embed", "object", "font", "basefont", "br", "script", "map", "q", "sub", "sup", "span", "bdo", "iframe"
#define NB_SPECIAL 16
#define INLINE PCDATA FONTSTYLE PHRASE SPECIAL FORMCTRL
#define NB_INLINE NB_PCDATA + NB_FONTSTYLE + NB_PHRASE + NB_SPECIAL + NB_FORMCTRL
#define BLOCK HEADING, LIST "pre", "p", "dl", "div", "center", "noscript", "noframes", "blockquote", "form", "isindex", "hr", "table", "fieldset", "address"
#define NB_BLOCK NB_HEADING + NB_LIST + 14
#define FORMCTRL "input", "select", "textarea", "label", "button"
#define NB_FORMCTRL 5
#define PCDATA
#define NB_PCDATA 0
#define HEADING "h1", "h2", "h3", "h4", "h5", "h6"
#define NB_HEADING 6
#define LIST "ul", "ol", "dir", "menu"
#define NB_LIST 4
#define MODIFIER
#define NB_MODIFIER 0
#define FLOW BLOCK,INLINE
#define NB_FLOW NB_BLOCK + NB_INLINE
#define EMPTY NULL


static const char* const html_flow[] = { FLOW, NULL } ;
static const char* const html_inline[] = { INLINE, NULL } ;

/* placeholders: elts with content but no subelements */
static const char* const html_pcdata[] = { NULL } ;
#define html_cdata html_pcdata


/* ... and for HTML Attributes */

#define COREATTRS "id", "class", "style", "title"
#define NB_COREATTRS 4
#define I18N "lang", "dir"
#define NB_I18N 2
#define EVENTS "onclick", "ondblclick", "onmousedown", "onmouseup", "onmouseover", "onmouseout", "onkeypress", "onkeydown", "onkeyup"
#define NB_EVENTS 9
#define ATTRS COREATTRS,I18N,EVENTS
#define NB_ATTRS NB_NB_COREATTRS + NB_I18N + NB_EVENTS
#define CELLHALIGN "align", "char", "charoff"
#define NB_CELLHALIGN 3
#define CELLVALIGN "valign"
#define NB_CELLVALIGN 1

static const char* const html_attrs[] = { ATTRS, NULL } ;
static const char* const core_i18n_attrs[] = { COREATTRS, I18N, NULL } ;
static const char* const core_attrs[] = { COREATTRS, NULL } ;
static const char* const i18n_attrs[] = { I18N, NULL } ;


/* Other declarations that should go inline ... */
static const char* const a_attrs[] = { ATTRS, "charset", "type", "name",
	"href", "hreflang", "rel", "rev", "accesskey", "shape", "coords",
	"tabindex", "onfocus", "onblur", NULL } ;
static const char* const target_attr[] = { "target", NULL } ;
static const char* const rows_cols_attr[] = { "rows", "cols", NULL } ;
static const char* const alt_attr[] = { "alt", NULL } ;
static const char* const src_alt_attrs[] = { "src", "alt", NULL } ;
static const char* const href_attrs[] = { "href", NULL } ;
static const char* const clear_attrs[] = { "clear", NULL } ;
static const char* const inline_p[] = { INLINE, "p", NULL } ;

static const char* const flow_param[] = { FLOW, "param", NULL } ;
static const char* const applet_attrs[] = { COREATTRS , "codebase",
		"archive", "alt", "name", "height", "width", "align",
		"hspace", "vspace", NULL } ;
static const char* const area_attrs[] = { "shape", "coords", "href", "nohref",
	"tabindex", "accesskey", "onfocus", "onblur", NULL } ;
static const char* const basefont_attrs[] =
	{ "id", "size", "color", "face", NULL } ;
static const char* const quote_attrs[] = { ATTRS, "cite", NULL } ;
static const char* const body_contents[] = { FLOW, "ins", "del", NULL } ;
static const char* const body_attrs[] = { ATTRS, "onload", "onunload", NULL } ;
static const char* const body_depr[] = { "background", "bgcolor", "text",
	"link", "vlink", "alink", NULL } ;
static const char* const button_attrs[] = { ATTRS, "name", "value", "type",
	"disabled", "tabindex", "accesskey", "onfocus", "onblur", NULL } ;


static const char* const col_attrs[] = { ATTRS, "span", "width", CELLHALIGN, CELLVALIGN, NULL } ;
static const char* const col_elt[] = { "col", NULL } ;
static const char* const edit_attrs[] = { ATTRS, "datetime", "cite", NULL } ;
static const char* const compact_attrs[] = { ATTRS, "compact", NULL } ;
static const char* const dl_contents[] = { "dt", "dd", NULL } ;
static const char* const compact_attr[] = { "compact", NULL } ;
static const char* const label_attr[] = { "label", NULL } ;
static const char* const fieldset_contents[] = { FLOW, "legend" } ;
static const char* const font_attrs[] = { COREATTRS, I18N, "size", "color", "face" , NULL } ;
static const char* const form_contents[] = { HEADING, LIST, INLINE, "pre", "p", "div", "center", "noscript", "noframes", "blockquote", "isindex", "hr", "table", "fieldset", "address", NULL } ;
static const char* const form_attrs[] = { ATTRS, "method", "enctype", "accept", "name", "onsubmit", "onreset", "accept-charset", NULL } ;
static const char* const frame_attrs[] = { COREATTRS, "longdesc", "name", "src", "frameborder", "marginwidth", "marginheight", "noresize", "scrolling" , NULL } ;
static const char* const frameset_attrs[] = { COREATTRS, "rows", "cols", "onload", "onunload", NULL } ;
static const char* const frameset_contents[] = { "frameset", "frame", "noframes", NULL } ;
static const char* const head_attrs[] = { I18N, "profile", NULL } ;
static const char* const head_contents[] = { "title", "isindex", "base", "script", "style", "meta", "link", "object", NULL } ;
static const char* const hr_depr[] = { "align", "noshade", "size", "width", NULL } ;
static const char* const version_attr[] = { "version", NULL } ;
static const char* const html_content[] = { "head", "body", "frameset", NULL } ;
static const char* const iframe_attrs[] = { COREATTRS, "longdesc", "name", "src", "frameborder", "marginwidth", "marginheight", "scrolling", "align", "height", "width", NULL } ;
static const char* const img_attrs[] = { ATTRS, "longdesc", "name", "height", "width", "usemap", "ismap", NULL } ;
static const char* const embed_attrs[] = { COREATTRS, "align", "alt", "border", "code", "codebase", "frameborder", "height", "hidden", "hspace", "name", "palette", "pluginspace", "pluginurl", "src", "type", "units", "vspace", "width", NULL } ;
static const char* const input_attrs[] = { ATTRS, "type", "name", "value", "checked", "disabled", "readonly", "size", "maxlength", "src", "alt", "usemap", "ismap", "tabindex", "accesskey", "onfocus", "onblur", "onselect", "onchange", "accept", NULL } ;
static const char* const prompt_attrs[] = { COREATTRS, I18N, "prompt", NULL } ;
static const char* const label_attrs[] = { ATTRS, "for", "accesskey", "onfocus", "onblur", NULL } ;
static const char* const legend_attrs[] = { ATTRS, "accesskey", NULL } ;
static const char* const align_attr[] = { "align", NULL } ;
static const char* const link_attrs[] = { ATTRS, "charset", "href", "hreflang", "type", "rel", "rev", "media", NULL } ;
static const char* const map_contents[] = { BLOCK, "area", NULL } ;
static const char* const name_attr[] = { "name", NULL } ;
static const char* const action_attr[] = { "action", NULL } ;
static const char* const blockli_elt[] = { BLOCK, "li", NULL } ;
static const char* const meta_attrs[] = { I18N, "http-equiv", "name", "scheme", NULL } ;
static const char* const content_attr[] = { "content", NULL } ;
static const char* const type_attr[] = { "type", NULL } ;
static const char* const noframes_content[] = { "body", FLOW MODIFIER, NULL } ;
static const char* const object_contents[] = { FLOW, "param", NULL } ;
static const char* const object_attrs[] = { ATTRS, "declare", "classid", "codebase", "data", "type", "codetype", "archive", "standby", "height", "width", "usemap", "name", "tabindex", NULL } ;
static const char* const object_depr[] = { "align", "border", "hspace", "vspace", NULL } ;
static const char* const ol_attrs[] = { "type", "compact", "start", NULL} ;
static const char* const option_elt[] = { "option", NULL } ;
static const char* const optgroup_attrs[] = { ATTRS, "disabled", NULL } ;
static const char* const option_attrs[] = { ATTRS, "disabled", "label", "selected", "value", NULL } ;
static const char* const param_attrs[] = { "id", "value", "valuetype", "type", NULL } ;
static const char* const width_attr[] = { "width", NULL } ;
static const char* const pre_content[] = { PHRASE, "tt", "i", "b", "u", "s", "strike", "a", "br", "script", "map", "q", "span", "bdo", "iframe", NULL } ;
static const char* const script_attrs[] = { "charset", "src", "defer", "event", "for", NULL } ;
static const char* const language_attr[] = { "language", NULL } ;
static const char* const select_content[] = { "optgroup", "option", NULL } ;
static const char* const select_attrs[] = { ATTRS, "name", "size", "multiple", "disabled", "tabindex", "onfocus", "onblur", "onchange", NULL } ;
static const char* const style_attrs[] = { I18N, "media", "title", NULL } ;
static const char* const table_attrs[] = { ATTRS "summary", "width", "border", "frame", "rules", "cellspacing", "cellpadding", "datapagesize", NULL } ;
static const char* const table_depr[] = { "align", "bgcolor", NULL } ;
static const char* const table_contents[] = { "caption", "col", "colgroup", "thead", "tfoot", "tbody", "tr", NULL} ;
static const char* const tr_elt[] = { "tr", NULL } ;
static const char* const talign_attrs[] = { ATTRS, CELLHALIGN, CELLVALIGN, NULL} ;
static const char* const th_td_depr[] = { "nowrap", "bgcolor", "width", "height", NULL } ;
static const char* const th_td_attr[] = { ATTRS, "abbr", "axis", "headers", "scope", "rowspan", "colspan", CELLHALIGN, CELLVALIGN, NULL } ;
static const char* const textarea_attrs[] = { ATTRS, "name", "disabled", "readonly", "tabindex", "accesskey", "onfocus", "onblur", "onselect", "onchange", NULL } ;
static const char* const tr_contents[] = { "th", "td", NULL } ;
static const char* const bgcolor_attr[] = { "bgcolor", NULL } ;
static const char* const li_elt[] = { "li", NULL } ;
static const char* const ul_depr[] = { "type", "compact", NULL} ;
static const char* const dir_attr[] = { "dir", NULL} ;

#define DECL (const char**)

static const htmlElemDesc
html40ElementTable[] = {
{ "a",		0, 0, 0, 0, 0, 0, 1, "anchor ",
	DECL html_inline , NULL , DECL a_attrs , DECL target_attr, NULL
},
{ "abbr",	0, 0, 0, 0, 0, 0, 1, "abbreviated form",
	DECL html_inline , NULL , DECL html_attrs, NULL, NULL
},
{ "acronym",	0, 0, 0, 0, 0, 0, 1, "",
	DECL html_inline , NULL , DECL html_attrs, NULL, NULL
},
{ "address",	0, 0, 0, 0, 0, 0, 0, "information on author ",
	DECL inline_p  , NULL , DECL html_attrs, NULL, NULL
},
{ "applet",	0, 0, 0, 0, 1, 1, 2, "java applet ",
	DECL flow_param , NULL , NULL , DECL applet_attrs, NULL
},
{ "area",	0, 2, 2, 1, 0, 0, 0, "client-side image map area ",
	EMPTY ,  NULL , DECL area_attrs , DECL target_attr, DECL alt_attr
},
{ "b",		0, 3, 0, 0, 0, 0, 1, "bold text style",
	DECL html_inline , NULL , DECL html_attrs, NULL, NULL
},
{ "base",	0, 2, 2, 1, 0, 0, 0, "document base uri ",
	EMPTY , NULL , NULL , DECL target_attr, DECL href_attrs
},
{ "basefont",	0, 2, 2, 1, 1, 1, 1, "base font size " ,
	EMPTY , NULL , NULL, DECL basefont_attrs, NULL
},
{ "bdo",	0, 0, 0, 0, 0, 0, 1, "i18n bidi over-ride ",
	DECL html_inline , NULL , DECL core_i18n_attrs, NULL, DECL dir_attr
},
{ "big",	0, 3, 0, 0, 0, 0, 1, "large text style",
	DECL html_inline , NULL , DECL html_attrs, NULL, NULL
},
{ "blockquote",	0, 0, 0, 0, 0, 0, 0, "long quotation ",
	DECL html_flow , NULL , DECL quote_attrs , NULL, NULL
},
{ "body",	1, 1, 0, 0, 0, 0, 0, "document body ",
	DECL body_contents , "div" , DECL body_attrs, DECL body_depr, NULL
},
{ "br",		0, 2, 2, 1, 0, 0, 1, "forced line break ",
	EMPTY , NULL , DECL core_attrs, DECL clear_attrs , NULL
},
{ "button",	0, 0, 0, 0, 0, 0, 2, "push button ",
	DECL html_flow MODIFIER , NULL , DECL button_attrs, NULL, NULL
},
{ "caption",	0, 0, 0, 0, 0, 0, 0, "table caption ",
	DECL html_inline , NULL , DECL html_attrs, NULL, NULL
},
{ "center",	0, 3, 0, 0, 1, 1, 0, "shorthand for div align=center ",
	DECL html_flow , NULL , NULL, DECL html_attrs, NULL
},
{ "cite",	0, 0, 0, 0, 0, 0, 1, "citation",
	DECL html_inline , NULL , DECL html_attrs, NULL, NULL
},
{ "code",	0, 0, 0, 0, 0, 0, 1, "computer code fragment",
	DECL html_inline , NULL , DECL html_attrs, NULL, NULL
},
{ "col",	0, 2, 2, 1, 0, 0, 0, "table column ",