Attachment #5525 for bug #10515

View | Details | Raw Unified | Return to bug 10515
Collapse All | Expand All

Lines 653-658 Link Here

(-)/cise/tmp/ppadala/tidy/include/html.h (+4 lines)
653	Lexer lexer, Node node);	653	Lexer lexer, Node node);
654	void PPrintXMLTree(Out *fout, uint mode, uint indent,	654	void PPrintXMLTree(Out *fout, uint mode, uint indent,
655	Lexer lexer, Node node);	655	Lexer lexer, Node node);
		656	void PrintSgml(Out *fout, uint mode, uint indent,
		657	Lexer lexer, Node node);
656	void PFlushLine(Out *out, uint indent);	658	void PFlushLine(Out *out, uint indent);
657	void PCondFlushLine(Out *out, uint indent);	659	void PCondFlushLine(Out *out, uint indent);
658	void PrintBody(Out fout, Lexer lexer, Node root); / Feature request #434940 - fix by Dave Raggett/Ignacio Vazquez-Abrams 21 Jun 01 */	660	void PrintBody(Out fout, Lexer lexer, Node root); / Feature request #434940 - fix by Dave Raggett/Ignacio Vazquez-Abrams 21 Jun 01 */
Lines 908-913 Link Here
908	extern Bool XmlOut;	910	extern Bool XmlOut;
909	extern Bool xHTML;	911	extern Bool xHTML;
910	extern Bool HtmlOut; /* Yes means set explicitly. */	912	extern Bool HtmlOut; /* Yes means set explicitly. */
		913	extern Bool DbSgml;
		914	extern Bool DbXml;
911	extern Bool XmlPi; /* add <?xml?> */	915	extern Bool XmlPi; /* add <?xml?> */
912	extern Bool XmlPIs; /* assume PIs end with ?> as per XML */	916	extern Bool XmlPIs; /* assume PIs end with ?> as per XML */
913	extern Bool XmlSpace;	917	extern Bool XmlSpace;

Lines 81-86 Link Here

(-)/cise/tmp/ppadala/tidy/src/config.c (+2 lines)
81	Bool XmlOut = no; /* create output as XML */	81	Bool XmlOut = no; /* create output as XML */
82	Bool xHTML = no; /* output extensible HTML */	82	Bool xHTML = no; /* output extensible HTML */
83	Bool HtmlOut = no; /* output plain-old HTML, even for XHTML input. Yes means set explicitly. */	83	Bool HtmlOut = no; /* output plain-old HTML, even for XHTML input. Yes means set explicitly. */
		84	Bool DbSgml = no; /* output docbook SGML */
		85	Bool DbXml = no; /* output docbook XML */
84	Bool XmlPi = no; /* add <?xml?> for XML docs */	86	Bool XmlPi = no; /* add <?xml?> for XML docs */
85	Bool RawOut = no; /* avoid mapping values > 127 to entities: not used for anything yet */	87	Bool RawOut = no; /* avoid mapping values > 127 to entities: not used for anything yet */
86	Bool UpperCaseTags = no; /* output tags in upper not lower case */	88	Bool UpperCaseTags = no; /* output tags in upper not lower case */




    return doctype;
}

Bool SetSgmlDocType(Lexer *lexer, Node *root)
{   char *fpi, *sysid;
    Node *doctype;

    if (doctype_mode == doctype_user && doctype_str)
    {
        fpi = doctype_str;
        sysid = "";
    }
    doctype = FindDocType(root);
    if(!doctype) /* The html file doesn't contain doctype */
        if ( !(doctype = NewXhtmlDocTypeNode( root )) )
            return no;
    
    lexer->txtstart = lexer->txtend = lexer->lexsize;

    /* add public identifier */
    AddStringLiteral(lexer, fpi);
    /* add system identifier */
    AddStringLiteral(lexer, sysid);
    
    lexer->txtend = lexer->lexsize;

    doctype->start = lexer->txtstart;
    doctype->end = lexer->txtend;

    return no; 
}

Bool SetXHTMLDocType(Lexer *lexer, Node *root)
{
    char *fpi = "", *sysid = "", *dtdsub, *name_space = XHTML_NAMESPACE; /* #578005 - fix by Anonymous 05 Jul 02 */

Lines 1054-1059 Link Here

(-)/cise/tmp/ppadala/tidy/src/localize.c (+2 lines)
1054	tidy_out(out, " -asxml to convert HTML to well formed XHTML\n");	1054	tidy_out(out, " -asxml to convert HTML to well formed XHTML\n");
1055	tidy_out(out, " -asxhtml to convert HTML to well formed XHTML\n");	1055	tidy_out(out, " -asxhtml to convert HTML to well formed XHTML\n");
1056	tidy_out(out, " -ashtml to force XHTML to well formed HTML\n");	1056	tidy_out(out, " -ashtml to force XHTML to well formed HTML\n");
		1057	tidy_out(out, " -dbsgml to convert HTML to Docbook SGML\n");
		1058	tidy_out(out, " -dbxml to convert HTML to Docbook XML\n");
1057	tidy_out(out, " -slides to burst into slides on H2 elements\n");	1059	tidy_out(out, " -slides to burst into slides on H2 elements\n");
1058		1060
1059	/* TRT */	1061	/* TRT */

Lines 501-507 Link Here

(-)/cise/tmp/ppadala/tidy/src/parser.c (-1 / +1 lines)
501	TrimTrailingSpace(lexer, element, text);	501	TrimTrailingSpace(lexer, element, text);
502	}	502	}
503		503
504	static Bool DescendantOf(Node element, Dict tag)	504	Bool DescendantOf(Node element, Dict tag)
505	{	505	{
506	Node *parent;	506	Node *parent;
507		507

Lines 6-14 Link Here

(-)/cise/tmp/ppadala/tidy/src/pprint.c (-11 / +641 lines)
6		6
7	CVS Info :	7	CVS Info :
8		8
9	$Author: creitzel $	9	$Author: terry_teague $
10	$Date: 2002/07/18 18:21:27 $	10	$Date: 2002/07/08 03:28:41 $
11	$Revision: 1.48 $	11	$Revision: 1.46 $
12		12
13	*/	13	*/
14		14
Lines 36-41 Link Here
36	static void PPrintPhp(Out *fout, uint indent,	36	static void PPrintPhp(Out *fout, uint indent,
37	Lexer lexer, Node node);	37	Lexer lexer, Node node);
38		38
		39	/* Tag types to distinguish printing */
		40	typedef enum {
		41	SgmlTagStart,
		42	SgmlTagEnd
		43	}SgmlTagType;
		44
		45	extern Bool DescendantOf(Node element, Dict tag);
39		46
40	#define NORMAL 0	47	#define NORMAL 0
41	#define PREFORMATTED 1	48	#define PREFORMATTED 1
Lines 1769-1774 Link Here
1769	}	1776	}
1770	}	1777	}
1771		1778
		1779	void PrintSgmlDefault(Out *fout)
		1780	{
		1781	char *str = "SGML cannot contain these elements";
		1782
		1783	fprintf(stderr, str);
		1784	}
		1785
		1786	void PrintSgmlBodyStart(Out *fout, uint indent)
		1787	{
		1788	char *str = "<article>";
		1789	PPrintString(fout, indent, str);
		1790	}
		1791
		1792	#define DIGIT(c) (c - 48)
		1793	#define TOTAL_H 6
		1794	static Bool seen_h[TOTAL_H] = {no, no, no, no, no, no};
		1795
		1796	/* Yuck ugly. FIXME */
		1797	#define SECT(i) (i - startsect)
		1798	static startsect = 0; /* We are at level 0(H1) initially */
		1799
		1800	void PrintSgmlBodyEnd(Out *fout, uint indent)
		1801	{ int i = TOTAL_H - 1;
		1802	char str[10];
		1803
		1804	while(i >= 0) {
		1805	if(seen_h[i] == yes) {
		1806	if(i == 5)
		1807	sprintf(str, "</simpleect>");
		1808	else
		1809	sprintf(str, "</sect%d>", SECT(i) + 1);
		1810	PPrintString(fout, indent, str);
		1811	seen_h[i] = no;
		1812	}
		1813	--i;
		1814	}
		1815
		1816	sprintf(str, "</article>");
		1817	PPrintString(fout, indent, str);
		1818	}
		1819
		1820	char GetContent(Lexer lexer, Node *node)
		1821	{ Node content, temp_node;
		1822	char str, temp, c;
		1823	Bool flag = no;
		1824	int i;
		1825
		1826	content = node->content;
		1827
		1828	/* Find the <a> tag */
		1829	for (temp_node = content;
		1830	temp_node && temp_node->tag != tag_a;
		1831	temp_node = temp_node->next)
		1832	;
		1833
		1834	if(temp_node == NULL) { /* There is no <a> .. </a> tag */
		1835	/* Discard all elements which are not text nodes */
		1836	temp_node = content;
		1837	for (temp_node = content;
		1838	temp_node && temp_node->type != TextNode;
		1839	temp_node = temp_node->next)
		1840	;
		1841	if(temp_node == NULL) { /* There's no TextNode either */
		1842	str = MemAlloc(1);
1843	str[0] = '\0';
1844	return str;
1845	}
1846	}
1847	content = temp_node;
1848
1849	if(content->type == TextNode) {
1850	int size = content->end - content->start;
1851
1852	str = MemAlloc(size + 1);
1853	str[size] = '\0';
1854	wstrncpy(str, lexer->lexbuf + content->start, size);
1855	}
1856	else if(content->tag == tag_a){
1857	AttVal *name;
1858	int size;
1859
1860	name = GetAttrByName(content, "name");
1861	if(name == NULL)
1862	name = GetAttrByName(content, "href");
1863
1864	if(name == NULL) { /* No href or name, let's take empty id */
1865	size = 0;
1866	str = MemAlloc(size + 1);
1867	str[size] = '\0';
1868	}
1869	else {
1870	size = wstrlen(name->value);
1871	str = MemAlloc(size + 1);
1872	str[size] = '\0';
1873	wstrncpy(str, name->value, size);
1874	}
1875	}
1876
1877	temp = str;
1878	if(str[0] == '#')
1879	flag = yes;
1880
1881	#define SGML_NAMELEN 44 /* Maximum id namelength */
1882
1883	i = 0;
1884
1885	while(*temp && i < SGML_NAMELEN) {
1886	if(flag)
1887	temp = (temp + 1);
1888	if(*temp == ' ')
1889	*temp = '_';
1890	++temp;
1891	++i;
1892	}
1893	*temp = '\0';
1894	return str;
1895	}
1896
1897	void PrintSectTag( Out fout, uint indent, Lexer lexer, Node *node,
1898	uint startsect)
1899	{ char sectnum = node->element[1];
1900	char str[100];
1901
1902	char *id = GetContent(lexer, node);
1903
1904	if(sectnum == '6') /* there's no sect6. We can do variety of
1905	things here. may be <section> .. */
1906	sprintf(str, "<simplesect id=\"%s\"><title>", id);
1907	else
1908	sprintf(str, "<sect%c id=\"%s\"><title>", SECT(sectnum), id);
1909	PPrintString(fout, indent, str);
1910	MemFree(id);
1911	}
1912
1913	Bool ImmediateDescendantOfHTags(Node *element)
1914	{ Node *parent = element->parent;
1915
1916	if (strlen(parent->element) == 2 &&
1917	parent->element[0] == 'h' &&
1918	IsDigit(parent->element[1]))
1919	return yes;
1920	return no;
1921	}
1922
1923	void PrintSgmlLink(Out fout, uint indent, Node node)
1924	{ AttVal *addr;
1925	char str[500]; /* FIXME allocate dynamically later */
1926
1927	addr = GetAttrByName(node, "name");
1928	if(addr == NULL) {
1929	addr = GetAttrByName(node, "href");
1930	if(!ImmediateDescendantOfHTags(node)) {
1931	if(addr->value[0] == '#')
1932	sprintf(str, "<link linkend=\"%s\">", addr->value + 1);
1933	else
1934	sprintf(str, "<ulink url=\"%s\">", addr->value);
1935	if( !DescendantOf(node, tag_p) &&
1936	node->prev && node->prev->type == TextNode)
1937	PPrintString(fout, indent, "<para>");
1938	PPrintString(fout, indent, str);
1939	}
1940	}
1941	else {
1942	if(!ImmediateDescendantOfHTags(node)) {
1943	if(!DescendantOf(node, tag_p))
1944	sprintf(str, "<para id=\"%s\">", addr->value);
1945	else /* We cannnot have a <para> inside another <para> */
1946	sprintf(str, "<anchor id=\"%s\"/>", addr->value);
1947	PPrintString(fout, indent, str);
1948	}
1949	}
1950	}
1951
1952	void PrintSgmlLinkEnd(Out fout, uint indent, Node node)
1953	{ AttVal *addr;
1954
1955	addr = GetAttrByName(node, "name");
1956	if(addr == NULL) {
1957	addr = GetAttrByName(node, "href");
1958	if(!ImmediateDescendantOfHTags(node)) {
1959	if(addr->value[0] == '#')
1960	PPrintString(fout, indent, "</link>");
1961	else
1962	PPrintString(fout, indent, "</ulink>");
1963	if( !DescendantOf(node, tag_p) &&
1964	node->prev && node->prev->type == TextNode)
1965	PPrintString(fout, indent, "</para>");
1966	}
1967	}
1968	else {
1969	if(!ImmediateDescendantOfHTags(node)) {
1970	if(!DescendantOf(node, tag_p))
1971	PPrintString(fout, indent, "</para>");
1972	/* else
1973	<anchor .. /> has already been placed. no need to
1974	do any thing */
1975	}
1976	}
1977	}
1978
1979
1980	void PrintSgmlTagString(Out *fout, uint mode, uint indent,
1981	SgmlTagType sgmltag_type, char *str)
1982	{ PPrintChar(str[0], mode \| CDATA);
1983	if(sgmltag_type == SgmlTagEnd)
1984	PPrintChar('/', mode);
1985	PPrintString(fout, indent, str + 1);
1986	}
1987
1988	void PrintSgmlList(Lexer lexer, Out fout,
1989	uint mode, uint indent,
1990	Node *node)
1991	{ if(node->tag == tag_ul)
1992	PPrintString(fout, indent, "<itemizedlist>");
1993	else if(node->tag == tag_ol)
1994	PPrintString(fout, indent, "<orderedlist>");
1995	else if(node->tag == tag_dl)
1996	PPrintString(fout, indent, "<variablelist>");
1997	}
1998
1999	void PrintSgmlListEnd(Lexer lexer, Out fout,
2000	uint mode, uint indent,
2001	Node *node)
2002	{ if(node->tag == tag_ul)
2003	PPrintString(fout, indent, "</itemizedlist>");
2004	else if(node->tag == tag_ol)
2005	PPrintString(fout, indent, "</orderedlist>");
2006	else if(node->tag == tag_dl)
2007	PPrintString(fout, indent, "</variablelist>");
2008	}
2009
2010	void PrintSgmlListItem(Out fout, uint indent, Node node)
2011	{ if(node->tag == tag_li)
2012	PPrintString(fout, indent, "<listitem>");
2013	else if(node->tag == tag_dd)
2014	PPrintString(fout, indent, "<listitem>");
2015	}
2016
2017	void PrintSgmlListItemEnd(Out fout, uint indent, Node node)
2018	{ if(node->tag == tag_li)
2019	PPrintString(fout, indent, "</listitem>");
2020	else if(node->tag == tag_dd)
2021	PPrintString(fout, indent, "</listitem></varlistentry>");
2022	}
2023
2024	void PrintSgmlImage(Out fout, uint indent, Node node)
2025	{ AttVal *addr;
2026	char str[100];
2027
2028	addr = GetAttrByName(node, "src");
2029	/* We can get other attributes like width, height etc.. */
2030	if(addr != NULL) {
2031	PPrintString(fout, indent, "<inlinemediaobject><imageobject>");
2032	PCondFlushLine(fout, indent);
2033	sprintf(str, "<imagedata fileref=\"%s\">", addr->value);
2034	PPrintString(fout, indent, str);
2035	PCondFlushLine(fout, indent);
2036	PPrintString(fout, indent, "</imageobject></inlinemediaobject>");
2037	PCondFlushLine(fout, indent);
2038	}
2039	}
2040
2041	int CountColumns(Node *node)
2042	{ Node temp, row_content;
2043	int ncols = 0;
2044
2045	temp = node->content;
2046
2047	/* FIXME */
2048	/* Perhaps this is not needed, check with HTML standard later */
2049	while(temp->tag != tag_tr)
2050	temp = temp->next;
2051
2052	/* This can contain th or td's */
2053	row_content = temp->content;
2054	while(row_content) {
2055	if(row_content->tag == tag_th \|\| row_content->tag == tag_td) {
2056	AttVal *colspan;
2057
2058	colspan = GetAttrByName(row_content, "colspan");
2059	if(colspan)
2060	ncols += atoi(colspan->value);
2061	else
2062	++ncols;
2063	}
2064	else
2065	fprintf(stderr, "PrintSgml: error in table processing\n");
2066	row_content = row_content->next;
2067	}
2068	return ncols;
2069	}
2070
2071	void PrintSgmlTable(Out fout, uint indent, Node node)
2072	{ int ncols;
2073	char str[100];
2074
2075	PPrintString(fout, indent, "<informaltable>");
2076	ncols = CountColumns(node);
2077	sprintf(str, "<tgroup cols=\"%d\"><tbody>", ncols);
2078	PPrintString(fout, indent, str);
2079	}
2080
2081	void PrintSgmlTableEnd(Out fout, uint indent, Node node)
2082	{
2083	PPrintString(fout, indent, "</tbody></tgroup></informaltable>");
2084	}
2085
2086	Bool DescendantOfAddress(Node *element)
2087	{
2088	Node *parent;
2089
2090	for (parent = element->parent;
2091	parent != null; parent = parent->parent)
2092	{ if (parent->element && wstrcasecmp(parent->element, "address") == 0)
2093	return yes;
2094	}
2095
2096	return no;
2097	}
2098
2099	void PrintSgmlTag( Out fout, uint mode, uint indent, Lexer lexer, Node *node,
2100	SgmlTagType sgmltag_type)
2101	{ static level = -1;
2102
2103	if(node->tag == tag_html) {
2104	if(sgmltag_type == SgmlTagStart)
2105	PrintSgmlBodyStart(fout, indent);
2106	else if(sgmltag_type == SgmlTagEnd)
2107	PrintSgmlBodyEnd(fout, indent);
2108	}
2109	else if(node->tag == tag_head)
2110	PrintSgmlTagString(fout, mode, indent, sgmltag_type,"<articleinfo>");
2111	else if(node->tag == tag_title)
2112	PrintSgmlTagString(fout, mode, indent, sgmltag_type,"<title>");
2113	/* May be we can replace with node->model & CM_LIST */
2114	else if(node->tag == tag_ul \|\| node->tag == tag_ol \|\|
2115	node->tag == tag_dl) {
2116	if(sgmltag_type == SgmlTagStart)
2117	PrintSgmlList(lexer, fout, mode, indent, node);
2118	else if(sgmltag_type == SgmlTagEnd)
2119	PrintSgmlListEnd(lexer, fout, mode, indent, node);
2120	}
2121	else if(node->tag == tag_dt) {
2122	if(sgmltag_type == SgmlTagStart)
2123	PPrintString(fout, indent, "<varlistentry><term>");
2124	else if(sgmltag_type == SgmlTagEnd)
2125	PPrintString(fout, indent, "</term>");
2126	}
2127	else if(node->tag == tag_li \|\| node->tag == tag_dd) {
2128	if(sgmltag_type == SgmlTagStart)
2129	PrintSgmlListItem(fout, indent, node);
2130	else if(sgmltag_type == SgmlTagEnd)
2131	PrintSgmlListItemEnd(fout, indent, node);
2132	}
2133	/* Later we should clean this before coming to PrintSgml */
2134	else if(node->tag == tag_p &&
2135	/* Table <entry> processing */
2136	!DescendantOf(node, tag_th) && !DescendantOf(node, tag_td) &&
2137	!DescendantOfAddress(node))
2138	PrintSgmlTagString(fout, mode, indent, sgmltag_type, "<para>");
2139	else if(node->tag == tag_blockquote)
2140	PrintSgmlTagString(fout, mode, indent, sgmltag_type, "<blockquote>");
2141	else if(node->tag == tag_pre &&
2142	/* Table <entry> processing */
2143	!DescendantOf(node, tag_th) && !DescendantOf(node, tag_td))
2144	PrintSgmlTagString(fout, mode, indent, sgmltag_type,
2145	"<programlisting>");
2146	else if(node->tag == tag_a) {
2147	if(sgmltag_type == SgmlTagStart)
2148	PrintSgmlLink(fout, indent, node);
2149	else if(sgmltag_type == SgmlTagEnd)
2150	PrintSgmlLinkEnd(fout, indent, node);
2151	}
2152	/* Table would require more processing */
2153	else if(node->tag == tag_table) {
2154	if(sgmltag_type == SgmlTagStart)
2155	PrintSgmlTable(fout, indent, node);
2156	else if(sgmltag_type == SgmlTagEnd)
2157	PrintSgmlTableEnd(fout, indent, node);
2158	}
2159	else if(node->tag == tag_tr)
2160	PrintSgmlTagString(fout, mode, indent, sgmltag_type, "<row>");
2161	else if(node->tag == tag_td \|\| node->tag == tag_th)
2162	PrintSgmlTagString(fout, mode, indent, sgmltag_type, "<entry>");
2163	else if(node->tag == tag_img) { /* This is a StartEndTag */
2164	if(sgmltag_type == SgmlTagStart)
2165	PrintSgmlImage(fout, indent, node);
2166	}
2167
2168	else if(wstrcasecmp(node->element, "cite") == 0)
2169	PrintSgmlTagString(fout, mode, indent, sgmltag_type,
2170	"<citation>");
2171	/* We should distinguish tag_strong and tag_em later
2172	haven't found proper docbook tag for <strong> */
2173	else if(node->tag == tag_em \|\| node->tag == tag_strong \|\|
2174	wstrcasecmp(node->element, "address") == 0) {
2175	if(sgmltag_type == SgmlTagStart) {
2176	if(DescendantOf(node, tag_p) \|\| DescendantOf(node, tag_pre))
2177	PPrintString(fout, indent, "<emphasis>");
2178	else
2179	PPrintString(fout, indent, "<para><emphasis>");
2180	}
2181	else if(sgmltag_type == SgmlTagEnd) {
2182	if(DescendantOf(node, tag_p) \|\| DescendantOf(node, tag_pre))
2183	PPrintString(fout, indent, "</emphasis>");
2184	else
2185	PPrintString(fout, indent, "</emphasis></para>");
2186	}
2187	}
2188	else {
2189	if(wstrcasecmp(node->element, "code") == 0 &&
2190	!(node->parent->tag == tag_dd \|\|
2191	node->parent->tag == tag_li))
2192	PrintSgmlTagString(fout, mode, indent,
2193	sgmltag_type, "<literal>");
2194	else if(strlen(node->element) == 2 &&
2195	node->element[0] == 'h' &&
2196	IsDigit(node->element[1])) {
2197	if(sgmltag_type == SgmlTagStart) {
2198	int sectnum = DIGIT(node->element[1]) - 1;
2199	char str[10];
2200	if(seen_h[sectnum] == no)
2201	seen_h[sectnum] = yes;
2202	else {
2203	int i = level;
2204	while(i > sectnum && seen_h[i] == yes) {
2205	if(i == 5)
2206	sprintf(str, "</simplesect>");
2207	else
2208	sprintf(str, "</sect%d>", SECT(i) + 1);
2209	PPrintString(fout, indent, str);
2210	seen_h[i] = no;
2211	--i;
2212	}
2213	if(sectnum == 5)
2214	sprintf(str, "</simplesect>");
2215	else
2216	sprintf(str, "</sect%d>", SECT(sectnum) + 1);
2217	PPrintString(fout, indent, str);
2218	}
2219	/* H1 is not the first level
2220	like the curses man2html pages */
2221	if(level == -1 && sectnum > 0)
2222	startsect = sectnum;
2223
2224	PrintSectTag(fout, indent, lexer, node, startsect);
2225	level = sectnum;
2226	}
2227	else
2228	PPrintString(fout, indent, "</title>");
2229	}
2230	}
2231	}
2232
2233	void PrintSgml( Out *fout, uint mode, uint indent,
2234	Lexer lexer, Node node)
2235	{ Node *content;
2236
2237	if (node == null)
2238	return;
2239
2240	if (node->type == TextNode) {
2241	if(DescendantOf(node, tag_dd) && !DescendantOf(node, tag_a) &&
2242	!DescendantOf(node, tag_p) &&
2243	/* We have to descide on this table stuff later
2244	* <entry> processing is complex */
2245	!DescendantOf(node, tag_td) && !DescendantOf(node, tag_th))
2246	/* && wstrcasecmp(node->parent->element, "code") != 0)
2247	above line may be needed later to properly convert <code> stuff */
2248	{
2249	PPrintString(fout, indent, "<para>");
2250	PPrintText(fout, mode, indent, lexer, node->start, node->end);
2251	PPrintString(fout, indent, "</para>");
2252	}
2253	else {
2254	if(DescendantOf(node, tag_style))
2255	fprintf(stderr, "PrintSgml: skipping style elements\n\n");
2256	else
2257	PPrintText(fout, mode, indent, lexer, node->start, node->end);
2258	}
2259	}
2260	else if(node->type == CDATATag && EscapeCdata)
2261	PPrintText(fout, mode, indent, lexer, node->start, node->end);
2262	else if (node->type == CommentTag)
2263	PPrintComment(fout, indent, lexer, node);
2264	else if (node->type == RootNode)
2265	{
2266	for (content = node->content;
2267	content != null;
2268	content = content->next)
2269	PrintSgml(fout, mode, indent, lexer, content);
2270	}
2271	else if (node->type == DocTypeTag)
2272	PPrintDocType(fout, indent, lexer, node);
2273	else if (node->type == CDATATag)
2274	PPrintCDATA(fout, indent, lexer, node);
2275	else if (node->type == SectionTag)
2276	PPrintSection(fout, indent, lexer, node);
2277	else if (node->type == AspTag \|\|
2278	node->type == JsteTag \|\|
2279	node->type == PhpTag )
2280	PrintSgmlDefault(fout);
2281	else if (node->type == ProcInsTag)
2282	PPrintPI(fout, indent, lexer, node);
2283	else if (node->type == XmlDecl)// && DbXml May be this is needed
2284	PPrintXmlDecl(fout, indent, lexer, node);
2285	else if (node->tag->model & CM_EMPTY \|\|
2286	(node->type == StartEndTag && !xHTML))
2287	{
2288	if (!(node->tag->model & CM_INLINE))
2289	PCondFlushLine(fout, indent);
2290
2291	if (MakeClean && node->tag == tag_wbr)
2292	PPrintString(fout, indent, " ");
2293	else
2294	PrintSgmlTag(fout, mode, indent, lexer, node, SgmlTagStart);
2295	}
2296	else {
2297	if (node->type == StartEndTag)
2298	node->type = StartTag;
2299
2300	if (node->tag && node->tag->parser == ParsePre)
2301	{
2302	PCondFlushLine(fout, indent);
2303
2304	indent = 0;
2305	PCondFlushLine(fout, indent);
2306
2307	PrintSgmlTag(fout, mode, indent, lexer, node, SgmlTagStart);
2308	PFlushLine(fout, indent);
2309
2310	for (content = node->content;
2311	content != null;
2312	content = content->next)
2313	PrintSgml(fout, (mode \| PREFORMATTED \| NOWRAP),
2314	indent, lexer, content);
2315
2316	PCondFlushLine(fout, indent);
2317	PrintSgmlTag(fout, mode, indent, lexer, node, SgmlTagEnd);
2318	PFlushLine(fout, indent);
2319
2320	if (IndentContent == no && node->next != null)
2321	PFlushLine(fout, indent);
2322	}
2323	else if (node->tag->model & CM_INLINE)
2324	{ PrintSgmlTag(fout, mode, indent, lexer, node, SgmlTagStart);
2325
2326	if (ShouldIndent(node))
2327	{
2328	PCondFlushLine(fout, indent);
2329	indent += spaces;
2330
2331	for (content = node->content;
2332	content != null;
2333	content = content->next)
2334	PrintSgml(fout, mode, indent, lexer, content);
2335
2336	PCondFlushLine(fout, indent);
2337	indent -= spaces;
2338	PCondFlushLine(fout, indent);
2339	}
2340	else
2341	{
2342
2343	for (content = node->content;
2344	content != null;
2345	content = content->next)
2346	PrintSgml(fout, mode, indent, lexer, content);
2347	}
2348
2349	PrintSgmlTag(fout, mode, indent, lexer, node, SgmlTagEnd);
2350	}
2351	else
2352	{ PCondFlushLine(fout, indent);
2353	if (SmartIndent && node->prev != null)
2354	PFlushLine(fout, indent);
2355
2356	PrintSgmlTag(fout, mode ,indent, lexer, node, SgmlTagStart);
2357	if (ShouldIndent(node))
2358	PCondFlushLine(fout, indent);
2359	else if (node->tag->model & CM_HTML \|\|
2360	node->tag == tag_noframes \|\|
2361	(node->tag->model & CM_HEAD && !(node->tag == tag_title)))
2362	PFlushLine(fout, indent);
2363
2364	if (ShouldIndent(node))
2365	{ PCondFlushLine(fout, indent);
2366	indent += spaces;
2367
2368	for (content = node->content;
2369	content != null;
2370	content = content->next)
2371	PrintSgml(fout, mode, indent, lexer, content);
2372	PCondFlushLine(fout, indent);
2373	indent -= spaces;
2374	PCondFlushLine(fout, indent);
2375	}
2376	else
2377	{ Node *last;
2378	last = null;
2379	for (content = node->content;
2380	content != null;
2381	content = content->next) {
2382	/* kludge for naked text before block level tag */
2383	if (last && !IndentContent && last->type == TextNode &&
2384	content->tag && !(content->tag->model & CM_INLINE) )
2385	{
2386	/* PFlushLine(fout, indent); */
2387	PFlushLine(fout, indent);
2388	}
2389
2390	PrintSgml(fout, mode,
2391	(ShouldIndent(node) ? indent+spaces : indent),
2392	lexer, content);
2393	last = content;
2394	}
2395	}
2396	PrintSgmlTag(fout, mode, indent, lexer, node, SgmlTagEnd);
2397	PFlushLine(fout, indent);
2398	if (IndentContent == no &&
2399	node->next != null &&
2400	HideEndTags == no &&
2401	(node->tag->model & (CM_BLOCK\|CM_LIST\|CM_DEFLIST\|CM_TABLE)))
2402	PFlushLine(fout, indent);
2403	}
2404	}
2405	}
2406
1772	void PPrintTree(Out *fout, uint mode, uint indent,	2407	void PPrintTree(Out *fout, uint mode, uint indent,
1773	Lexer lexer, Node node)	2408	Lexer lexer, Node node)
1774	{	2409	{
Lines 2034-2050 Link Here
2034	PPrintJste(fout, indent, lexer, node);	2669	PPrintJste(fout, indent, lexer, node);
2035	else if (node->type == PhpTag)	2670	else if (node->type == PhpTag)
2036	PPrintPhp(fout, indent, lexer, node);	2671	PPrintPhp(fout, indent, lexer, node);
2037	else if ( node->tag->model & CM_EMPTY	2672	else if (node->tag->model & CM_EMPTY \|\| (node->type == StartEndTag && !xHTML))
2038	\|\| (node->type == StartEndTag && !xHTML) )
2039	{	2673	{
2040	PCondFlushLine(fout, indent);	2674	PCondFlushLine(fout, indent);
2041	PPrintTag(lexer, fout, mode, indent, node);	2675	PPrintTag(lexer, fout, mode, indent, node);
2042	PFlushLine(fout, indent);	2676	PFlushLine(fout, indent);
2043		2677
2044	/* CPR: folks don't want so much vertical spacing in XML
2045	if (node->next)	2678	if (node->next)
2046	PFlushLine(fout, indent);	2679	PFlushLine(fout, indent);
2047	*/
2048	}	2680	}
2049	else /* some kind of container element */	2681	else /* some kind of container element */
2050	{	2682	{
Lines 2076-2082 Link Here
2076		2708
2077	PPrintTag(lexer, fout, mode, indent, node);	2709	PPrintTag(lexer, fout, mode, indent, node);
2078		2710
2079	if ( !mixed && node->content )	2711	if (!mixed)
2080	PFlushLine(fout, indent);	2712	PFlushLine(fout, indent);
2081		2713
2082	for (content = node->content;	2714	for (content = node->content;
Lines 2084-2099 Link Here
2084	content = content->next)	2716	content = content->next)
2085	PPrintXMLTree(fout, mode, cindent, lexer, content);	2717	PPrintXMLTree(fout, mode, cindent, lexer, content);
2086		2718
2087	if ( !mixed && node->content )	2719	if (!mixed)
2088	PCondFlushLine(fout, cindent);	2720	PCondFlushLine(fout, cindent);
2089		2721
2090	PPrintEndTag(fout, mode, indent, node);	2722	PPrintEndTag(fout, mode, indent, node);
2091	PCondFlushLine(fout, indent);	2723	PCondFlushLine(fout, indent);
2092		2724
2093	/* CPR: folks don't want so much vertical spacing in XML
2094	if (node->next)	2725	if (node->next)
2095	PFlushLine(fout, indent);	2726	PFlushLine(fout, indent);
2096	*/
2097	}	2727	}
2098	}	2728	}
2099		2729

Lines 2-8 Link Here

(-)/cise/tmp/ppadala/tidy/src/tab2space.c (-1 / +1 lines)
2	#include <stdlib.h>	2	#include <stdlib.h>
3	#include <string.h>	3	#include <string.h>
4		4
5	#ifndef __BEOS__	5	#if !(defined(__BEOS__) \|\| defined(linux))
6	typedef unsigned int uint;	6	typedef unsigned int uint;
7	#endif	7	#endif
8	typedef unsigned char byte;	8	typedef unsigned char byte;

Lines 1853-1858 Link Here

(-)/cise/tmp/ppadala/tidy/src/tidy.c (-4 / +37 lines)
1853	IndentContent = yes;	1853	IndentContent = yes;
1854	SmartIndent = yes;	1854	SmartIndent = yes;
1855	}	1855	}
		1856	else if (wstrcasecmp(arg, "dbsgml") == 0)
		1857	DbSgml = yes;
		1858	else if(wstrcasecmp(arg, "dbxml") == 0)
		1859	DbXml = yes;
1856	else if (wstrcasecmp(arg, "omit") == 0)	1860	else if (wstrcasecmp(arg, "omit") == 0)
1857	HideEndTags = yes;	1861	HideEndTags = yes;
1858	else if (wstrcasecmp(arg, "upper") == 0)	1862	else if (wstrcasecmp(arg, "upper") == 0)
Lines 2180-2185 Link Here
2180	else	2184	else
2181	{	2185	{
2182	lexer->warnings = 0;	2186	lexer->warnings = 0;
		2187
		2188	if (DbSgml \|\| DbXml) {
		2189	char *str;
		2190
		2191	if(DbSgml)
		2192	str = "article PUBLIC \"-//OASIS//DTD DocBook V4.1//EN\"";
		2193	else
		2194	str = "article PUBLIC \"-//OASIS//DTD DocBk XML V4.1.2 //EN\"";
		2195
		2196	EncloseBodyText = yes; /* We want those <p>s */
		2197	EncloseBlockText = yes;
		2198	LogicalEmphasis = yes;
		2199	DropFontTags = yes; /* <font> .. </font> are not needed */
		2200
		2201	/* May be this should be decided by user */
		2202	QuoteMarks = yes;
		2203
		2204	doctype_mode = doctype_user;
		2205	/* TidyDeInit does MemFree(doctype_str) if it's != NULL */
		2206	doctype_str = MemAlloc(wstrlen(str));
		2207	wstrcpy(doctype_str, str);
		2208	}
2183		2209
2184	document = ParseDocument(lexer);	2210	document = ParseDocument(lexer);
2185		2211
Lines 2226-2231 Link Here
2226	{	2252	{
2227	if (xHTML)	2253	if (xHTML)
2228	SetXHTMLDocType(lexer, document);	2254	SetXHTMLDocType(lexer, document);
		2255	else if(DbSgml)
		2256	SetSgmlDocType(lexer, document);
		2257	else if(DbXml)
		2258	SetSgmlDocType(lexer, document);
2229	else	2259	else
2230	FixDocType(lexer, document);	2260	FixDocType(lexer, document);
2231		2261
Lines 2247-2253 Link Here
2247	}	2277	}
2248		2278
2249	/* ensure presence of initial <?XML version="1.0"?> */	2279	/* ensure presence of initial <?XML version="1.0"?> */
2250	if (XmlOut && XmlPi)	2280	if ((XmlOut && XmlPi) \|\| DbXml)
2251	FixXmlDecl(lexer, document);	2281	FixXmlDecl(lexer, document);
2252		2282
2253	/*	2283	/*
Lines 2381-2389 Link Here
2381	/* Feature request #434940 - fix by Dave Raggett/Ignacio Vazquez-Abrams 21 Jun 01 */	2411	/* Feature request #434940 - fix by Dave Raggett/Ignacio Vazquez-Abrams 21 Jun 01 */
2382	else if (BodyOnly)	2412	else if (BodyOnly)
2383	PrintBody(&out, lexer, document);	2413	PrintBody(&out, lexer, document);
2384	else	2414	else {
2385	PPrintTree(&out, null, 0, lexer, document);	2415	if(DbSgml \|\| DbXml)
2386		2416	PrintSgml(&out, null, 0, lexer, document);
		2417	else
		2418	PPrintTree(&out, null, 0, lexer, document);
		2419	}
2387	PFlushLine(&out, 0);	2420	PFlushLine(&out, 0);
2388	}	2421	}
2389		2422

Return to bug 10515

Lines 1674-1679 Link Here

(-)/cise/tmp/ppadala/tidy/src/lexer.c (+29 lines)
1674	return doctype;	1674	return doctype;
1675	}	1675	}
1676		1676
		1677	Bool SetSgmlDocType(Lexer lexer, Node root)
		1678	{ char fpi, sysid;
		1679	Node *doctype;
		1680
		1681	if (doctype_mode == doctype_user && doctype_str)
		1682	{
		1683	fpi = doctype_str;
		1684	sysid = "";
		1685	}
		1686	doctype = FindDocType(root);
		1687	if(!doctype) /* The html file doesn't contain doctype */
		1688	if ( !(doctype = NewXhtmlDocTypeNode( root )) )
		1689	return no;
		1690
		1691	lexer->txtstart = lexer->txtend = lexer->lexsize;
		1692
		1693	/* add public identifier */
		1694	AddStringLiteral(lexer, fpi);
		1695	/* add system identifier */
		1696	AddStringLiteral(lexer, sysid);
		1697
		1698	lexer->txtend = lexer->lexsize;
		1699
		1700	doctype->start = lexer->txtstart;
		1701	doctype->end = lexer->txtend;
		1702
		1703	return no;
		1704	}
		1705
1677	Bool SetXHTMLDocType(Lexer lexer, Node root)	1706	Bool SetXHTMLDocType(Lexer lexer, Node root)
1678	{	1707	{
1679	char fpi = "", sysid = "", dtdsub, name_space = XHTML_NAMESPACE; /* #578005 - fix by Anonymous 05 Jul 02 */	1708	char fpi = "", sysid = "", dtdsub, name_space = XHTML_NAMESPACE; /* #578005 - fix by Anonymous 05 Jul 02 */