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.

	EMPTY , NULL , DECL col_attrs , NULL, NULL
},
{ "colgroup",	0, 1, 0, 0, 0, 0, 0, "table column group ",
	DECL col_elt , "col" , DECL col_attrs , NULL, NULL
},
{ "dd",		0, 1, 0, 0, 0, 0, 0, "definition description ",
	DECL html_flow , NULL , DECL html_attrs, NULL, NULL
},
{ "del",	0, 0, 0, 0, 0, 0, 2, "deleted text ",
	DECL html_flow , NULL , DECL edit_attrs , NULL, NULL
},
{ "dfn",	0, 0, 0, 0, 0, 0, 1, "instance definition",
	DECL html_inline , NULL , DECL html_attrs, NULL, NULL
},
{ "dir",	0, 0, 0, 0, 1, 1, 0, "directory list",
	DECL blockli_elt, "li" , NULL, DECL compact_attrs, NULL
},
{ "div",	0, 0, 0, 0, 0, 0, 0, "generic language/style container",
	DECL html_flow, NULL, DECL html_attrs, DECL align_attr, NULL
},
{ "dl",		0, 0, 0, 0, 0, 0, 0, "definition list ",
	DECL dl_contents , "dd" , DECL html_attrs, DECL compact_attr, NULL
},
{ "dt",		0, 1, 0, 0, 0, 0, 0, "definition term ",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "em",		0, 3, 0, 0, 0, 0, 1, "emphasis",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "embed",	0, 1, 2, 0, 1, 1, 1, "generic embedded object ",
	EMPTY, NULL, DECL embed_attrs, NULL, NULL
},
{ "fieldset",	0, 0, 0, 0, 0, 0, 0, "form control group ",
	DECL fieldset_contents , NULL, DECL html_attrs, NULL, NULL
},
{ "font",	0, 3, 0, 0, 1, 1, 1, "local change to font ",
	DECL html_inline, NULL, NULL, DECL font_attrs, NULL
},
{ "form",	0, 0, 0, 0, 0, 0, 0, "interactive form ",
	DECL form_contents, "fieldset", DECL form_attrs , DECL target_attr, DECL action_attr
},
{ "frame",	0, 2, 2, 1, 0, 2, 0, "subwindow " ,
	EMPTY, NULL, NULL, DECL frame_attrs, NULL
},
{ "frameset",	0, 0, 0, 0, 0, 2, 0, "window subdivision" ,
	DECL frameset_contents, "noframes" , NULL , DECL frameset_attrs, NULL
},
{ "h1",		0, 0, 0, 0, 0, 0, 0, "heading ",
	DECL html_inline, NULL, DECL html_attrs, DECL align_attr, NULL
},
{ "h2",		0, 0, 0, 0, 0, 0, 0, "heading ",
	DECL html_inline, NULL, DECL html_attrs, DECL align_attr, NULL
},
{ "h3",		0, 0, 0, 0, 0, 0, 0, "heading ",
	DECL html_inline, NULL, DECL html_attrs, DECL align_attr, NULL
},
{ "h4",		0, 0, 0, 0, 0, 0, 0, "heading ",
	DECL html_inline, NULL, DECL html_attrs, DECL align_attr, NULL
},
{ "h5",		0, 0, 0, 0, 0, 0, 0, "heading ",
	DECL html_inline, NULL, DECL html_attrs, DECL align_attr, NULL
},
{ "h6",		0, 0, 0, 0, 0, 0, 0, "heading ",
	DECL html_inline, NULL, DECL html_attrs, DECL align_attr, NULL
},
{ "head",	1, 1, 0, 0, 0, 0, 0, "document head ",
	DECL head_contents, NULL, DECL head_attrs, NULL, NULL
},
{ "hr",		0, 2, 2, 1, 0, 0, 0, "horizontal rule " ,
	EMPTY, NULL, DECL html_attrs, DECL hr_depr, NULL
},
{ "html",	1, 1, 0, 0, 0, 0, 0, "document root element ",
	DECL html_content , NULL , DECL i18n_attrs, DECL version_attr, NULL
},
{ "i",		0, 3, 0, 0, 0, 0, 1, "italic text style",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "iframe",	0, 0, 0, 0, 0, 1, 2, "inline subwindow ",
	DECL html_flow, NULL, NULL, DECL iframe_attrs, NULL
},
{ "img",	0, 2, 2, 1, 0, 0, 1, "embedded image ",
	EMPTY, NULL, DECL img_attrs, DECL align_attr, DECL src_alt_attrs
},
{ "input",	0, 2, 2, 1, 0, 0, 1, "form control ",
	EMPTY, NULL, DECL input_attrs , DECL align_attr, NULL
},
{ "ins",	0, 0, 0, 0, 0, 0, 2, "inserted text",
	DECL html_flow, NULL, DECL edit_attrs, NULL, NULL
},
{ "isindex",	0, 2, 2, 1, 1, 1, 0, "single line prompt ",
	EMPTY, NULL, NULL, DECL prompt_attrs, NULL
},
{ "kbd",	0, 0, 0, 0, 0, 0, 1, "text to be entered by the user",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "label",	0, 0, 0, 0, 0, 0, 1, "form field label text ",
	DECL html_inline MODIFIER, NULL, DECL label_attrs , NULL, NULL
},
{ "legend",	0, 0, 0, 0, 0, 0, 0, "fieldset legend ",
	DECL html_inline, NULL, DECL legend_attrs , DECL align_attr, NULL
},
{ "li",		0, 1, 1, 0, 0, 0, 0, "list item ",
	DECL html_flow, NULL, DECL html_attrs, NULL, NULL
},
{ "link",	0, 2, 2, 1, 0, 0, 0, "a media-independent link ",
	EMPTY, NULL, DECL link_attrs, DECL target_attr, NULL
},
{ "map",	0, 0, 0, 0, 0, 0, 2, "client-side image map ",
	DECL map_contents , NULL, DECL html_attrs , NULL, DECL name_attr
},
{ "menu",	0, 0, 0, 0, 1, 1, 0, "menu list ",
	DECL blockli_elt , NULL, NULL, DECL compact_attrs, NULL
},
{ "meta",	0, 2, 2, 1, 0, 0, 0, "generic metainformation ",
	EMPTY, NULL, DECL meta_attrs , NULL , DECL content_attr
},
{ "noframes",	0, 0, 0, 0, 0, 2, 0, "alternate content container for non frame-based rendering ",
	DECL noframes_content, "body" , DECL html_attrs, NULL, NULL
},
{ "noscript",	0, 0, 0, 0, 0, 0, 0, "alternate content container for non script-based rendering ",
	DECL html_flow, "div", DECL html_attrs, NULL, NULL
},
{ "object",	0, 0, 0, 0, 0, 0, 2, "generic embedded object ",
	DECL object_contents , "div" , DECL object_attrs, DECL object_depr, NULL
},
{ "ol",		0, 0, 0, 0, 0, 0, 0, "ordered list ",
	DECL li_elt , "li" , DECL html_attrs, DECL ol_attrs, NULL
},
{ "optgroup",	0, 0, 0, 0, 0, 0, 0, "option group ",
	DECL option_elt , "option", DECL optgroup_attrs, NULL, DECL label_attr
},
{ "option",	0, 1, 0, 0, 0, 0, 0, "selectable choice " ,
	DECL html_pcdata, NULL, DECL option_attrs, NULL, NULL
},
{ "p",		0, 1, 0, 0, 0, 0, 0, "paragraph ",
	DECL html_inline, NULL, DECL html_attrs, DECL align_attr, NULL
},
{ "param",	0, 2, 2, 1, 0, 0, 0, "named property value ",
	EMPTY, NULL, DECL param_attrs, NULL, DECL name_attr
},
{ "pre",	0, 0, 0, 0, 0, 0, 0, "preformatted text ",
	DECL pre_content, NULL, DECL html_attrs, DECL width_attr, NULL
},
{ "q",		0, 0, 0, 0, 0, 0, 1, "short inline quotation ",
	DECL html_inline, NULL, DECL quote_attrs, NULL, NULL
},
{ "s",		0, 3, 0, 0, 1, 1, 1, "strike-through text style",
	DECL html_inline, NULL, NULL, DECL html_attrs, NULL
},
{ "samp",	0, 0, 0, 0, 0, 0, 1, "sample program output, scripts, etc.",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "script",	0, 0, 0, 0, 0, 0, 2, "script statements ",
	DECL html_cdata, NULL, DECL script_attrs, DECL language_attr, DECL type_attr
},
{ "select",	0, 0, 0, 0, 0, 0, 1, "option selector ",
	DECL select_content, NULL, DECL select_attrs, NULL, NULL
},
{ "small",	0, 3, 0, 0, 0, 0, 1, "small text style",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "span",	0, 0, 0, 0, 0, 0, 1, "generic language/style container ",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "strike",	0, 3, 0, 0, 1, 1, 1, "strike-through text",
	DECL html_inline, NULL, NULL, DECL html_attrs, NULL
},
{ "strong",	0, 3, 0, 0, 0, 0, 1, "strong emphasis",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "style",	0, 0, 0, 0, 0, 0, 0, "style info ",
	DECL html_cdata, NULL, DECL style_attrs, NULL, DECL type_attr
},
{ "sub",	0, 3, 0, 0, 0, 0, 1, "subscript",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "sup",	0, 3, 0, 0, 0, 0, 1, "superscript ",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "table",	0, 0, 0, 0, 0, 0, 0, "",
	DECL table_contents , "tr" , DECL table_attrs , DECL table_depr, NULL
},
{ "tbody",	1, 0, 0, 0, 0, 0, 0, "table body ",
	DECL tr_elt , "tr" , DECL talign_attrs, NULL, NULL
},
{ "td",		0, 0, 0, 0, 0, 0, 0, "table data cell",
	DECL html_flow, NULL, DECL th_td_attr, DECL th_td_depr, NULL
},
{ "textarea",	0, 0, 0, 0, 0, 0, 1, "multi-line text field ",
	DECL html_pcdata, NULL, DECL textarea_attrs, NULL, DECL rows_cols_attr
},
{ "tfoot",	0, 1, 0, 0, 0, 0, 0, "table footer ",
	DECL tr_elt , "tr" , DECL talign_attrs, NULL, NULL
},
{ "th",		0, 1, 0, 0, 0, 0, 0, "table header cell",
	DECL html_flow, NULL, DECL th_td_attr, DECL th_td_depr, NULL
},
{ "thead",	0, 1, 0, 0, 0, 0, 0, "table header ",
	DECL tr_elt , "tr" , DECL talign_attrs, NULL, NULL
},
{ "title",	0, 0, 0, 0, 0, 0, 0, "document title ",
	DECL html_pcdata, NULL, DECL i18n_attrs, NULL, NULL
},
{ "tr",		0, 0, 0, 0, 0, 0, 0, "table row ",
	DECL tr_contents , "td" , DECL talign_attrs, DECL bgcolor_attr, NULL
},
{ "tt",		0, 3, 0, 0, 0, 0, 1, "teletype or monospaced text style",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
},
{ "u",		0, 3, 0, 0, 1, 1, 1, "underlined text style",
	DECL html_inline, NULL, NULL, DECL html_attrs, NULL
},
{ "ul",		0, 0, 0, 0, 0, 0, 0, "unordered list ",
	DECL li_elt , "li" , DECL html_attrs, DECL ul_depr, NULL
},
{ "var",	0, 0, 0, 0, 0, 0, 1, "instance of a variable or program argument",
	DECL html_inline, NULL, DECL html_attrs, NULL, NULL
}
};

/*
 * start tags that imply the end of current element
 */
static const char * const htmlStartClose[] = {
"form",		"form", "p", "hr", "h1", "h2", "h3", "h4", "h5", "h6",
		"dl", "ul", "ol", "menu", "dir", "address", "pre",
		"listing", "xmp", "head", NULL,
"head",		"p", NULL,
"title",	"p", NULL,
"body",		"head", "style", "link", "title", "p", NULL,
"frameset",	"head", "style", "link", "title", "p", NULL,
"li",		"p", "h1", "h2", "h3", "h4", "h5", "h6", "dl", "address",
		"pre", "listing", "xmp", "head", "li", NULL,
"hr",		"p", "head", NULL,
"h1",		"p", "head", NULL,
"h2",		"p", "head", NULL,
"h3",		"p", "head", NULL,
"h4",		"p", "head", NULL,
"h5",		"p", "head", NULL,
"h6",		"p", "head", NULL,
"dir",		"p", "head", NULL,
"address",	"p", "head", "ul", NULL,
"pre",		"p", "head", "ul", NULL,
"listing",	"p", "head", NULL,
"xmp",		"p", "head", NULL,
"blockquote",	"p", "head", NULL,
"dl",		"p", "dt", "menu", "dir", "address", "pre", "listing",
		"xmp", "head", NULL,
"dt",		"p", "menu", "dir", "address", "pre", "listing", "xmp",
                "head", "dd", NULL,
"dd",		"p", "menu", "dir", "address", "pre", "listing", "xmp",
                "head", "dt", NULL,
"ul",		"p", "head", "ol", "menu", "dir", "address", "pre",
		"listing", "xmp", NULL,
"ol",		"p", "head", "ul", NULL,
"menu",		"p", "head", "ul", NULL,
"p",		"p", "head", "h1", "h2", "h3", "h4", "h5", "h6", NULL,
"div",		"p", "head", NULL,
"noscript",	"p", "head", NULL,
"center",	"font", "b", "i", "p", "head", NULL,
"a",		"a", NULL,
"caption",	"p", NULL,
"colgroup",	"caption", "colgroup", "col", "p", NULL,
"col",		"caption", "col", "p", NULL,
"table",	"p", "head", "h1", "h2", "h3", "h4", "h5", "h6", "pre",
		"listing", "xmp", "a", NULL,
"th",		"th", "td", "p", "span", "font", "a", "b", "i", "u", NULL,
"td",		"th", "td", "p", "span", "font", "a", "b", "i", "u", NULL,      
"tr",		"th", "td", "tr", "caption", "col", "colgroup", "p", NULL,
"thead",	"caption", "col", "colgroup", NULL,
"tfoot",	"th", "td", "tr", "caption", "col", "colgroup", "thead",
		"tbody", "p", NULL,
"tbody",	"th", "td", "tr", "caption", "col", "colgroup", "thead",
		"tfoot", "tbody", "p", NULL,
"optgroup",	"option", NULL,
"option",	"option", NULL,
"fieldset",	"legend", "p", "head", "h1", "h2", "h3", "h4", "h5", "h6",
		"pre", "listing", "xmp", "a", NULL,
NULL
};

/*
 * The list of HTML elements which are supposed not to have
 * CDATA content and where a p element will be implied
 *
 * TODO: extend that list by reading the HTML SGML DTD on
 *       implied paragraph
 */
static const char *const htmlNoContentElements[] = {
    "html",
    "head",
    NULL
};

/*
 * The list of HTML attributes which are of content %Script;
 * NOTE: when adding ones, check htmlIsScriptAttribute() since
 *       it assumes the name starts with 'on'
 */
static const char *const htmlScriptAttributes[] = {
    "onclick",
    "ondblclick",
    "onmousedown",
    "onmouseup",
    "onmouseover",
    "onmousemove",
    "onmouseout",
    "onkeypress",
    "onkeydown",
    "onkeyup",
    "onload",
    "onunload",
    "onfocus",
    "onblur",
    "onsubmit",
    "onrest",
    "onchange",
    "onselect"
};

/*
 * This table is used by the htmlparser to know what to do with
 * broken html pages. By assigning different priorities to different
 * elements the parser can decide how to handle extra endtags.
 * Endtags are only allowed to close elements with lower or equal
 * priority.
 */ 

typedef struct {
    const char *name;
    int priority;
} elementPriority;

static const elementPriority htmlEndPriority[] = {
    {"div",   150},
    {"td",    160},
    {"th",    160},
    {"tr",    170},
    {"thead", 180},
    {"tbody", 180},
    {"tfoot", 180},
    {"table", 190},