WebSVN – Kolibri OS – Diff – /programs/cmm/browser/TWB/parse_tag.h


struct _tag
{
	char name[32];
	char prior[32];
	bool opened;
	collection attributes;
	collection values;
	dword value;
	bool is();
	bool parse();
	void debug_tag();
	dword get_next_param();
	dword get_value_of();
} tag=0;
 
bool _tag::is(dword _text) 
{ 
	if ( !strcmp(#name, _text) ) {
		return true;
	} else {
		return false; 
	}
}
 
bool _tag::parse(dword _bufpos, bufend)
{
	bool retok = true;
	dword bufpos = ESDWORD[_bufpos];
	dword params, paramsend;
 
	dword closepos;
	dword whitepos;
	dword openpos;
 
	if (name) strcpy(#prior, #name); else prior = '\0';
	name = '\0';
	attributes.drop();
	values.drop();		
 
	if (!strncmp(bufpos,"!--",3))
	{
		bufpos+=3;
		//STRSTR
		while (strncmp(bufpos,"-->",3)!=0) && (bufpos < bufend)
		{
			bufpos++;
		}
		bufpos+=2;
		retok = false;
		goto _RET;
	}
 
	if (ESBYTE[bufpos] == '/') {
		opened = false;
		bufpos++;
	} else {
		opened = true;
	}
 
	closepos = strchr(bufpos, '>');
	whitepos = strchrw(bufpos, bufend-bufpos);
 
	if (debug_mode) {
		if (!closepos) debugln("null closepos");
		if (!whitepos) debugln("null whitepos");
	}
 
	if (!whitepos) || (whitepos > closepos) {
		//no param
		strncpy(#name, bufpos, math.min(closepos - bufpos, sizeof(tag.name)));
		debug_tag();
		bufpos = closepos;
	} else {
		//we have param
		while (chrlnum(whitepos, '\"', closepos - whitepos)%2) { //alt="Next>>"
			/*
			openpos = strchr(closepos+1, '<');
			closepos = strchr(closepos+1, '>');
			if (openpos) && (openpos < closepos) {
				closepos = openpos - 1;
				break;
			}
			*/
			if (!openpos = strchr(closepos+1, '<')) break;
			if (openpos < strchr(closepos+1, '>')) break;
			if (!closepos = EAX) {closepos = bufend;break;}
		}
		strncpy(#name, bufpos, math.min(whitepos - bufpos, sizeof(tag.name)));
		debug_tag();
		bufpos = closepos;
 
		params = malloc(closepos - whitepos + 1);
		strncpy(params, whitepos, closepos - whitepos);
		if (debug_mode) { debug("params: "); debugln(params+1); }
		paramsend = params + closepos - whitepos;
		while (paramsend = get_next_param(params, paramsend-1));
		free(params);
	}
 
	if (name) {
		strlwr(#name);
		// ignore text inside the next tags
		if (is("script")) || (is("style")) || (is("binary")) || (is("select")) { 
			strcpy(#prior, #name);
			sprintf(#name, "", #prior);
			if (strstri(bufpos, #name)) bufpos = EAX-1;
			retok = false;
		} else {
			if (name[strlen(#name)-1]=='/') name[strlen(#name)-1]=NULL; //for 
		}
	} else {
		retok = false;
	}
 
_RET:
	ESDWORD[_bufpos] = bufpos;
	return retok;
}
 
void _tag::debug_tag()
{
	if (debug_mode) { 
		debugch('<'); 
		if (!opened) debugch('/');
		debug(#name);
		debugln(">");
	}
}
 
dword _tag::get_next_param(dword ps, pe)
{
	// "ps" - param start
	// "pe" - param end
	// "q"  - quote char
	char q = NULL;
	dword fixeq;
	dword val;
	dword attr;
	
	if (ESBYTE[pe] == '/') pe--;
	while (pe>ps) && (__isWhite(ESBYTE[pe])) pe--;
 
	if (ESBYTE[pe] == '"') || (ESBYTE[pe] == '\'')
	{
		//remove quote
		q = ESBYTE[pe];
		ESBYTE[pe] = '\0';
		pe--;
 
		//find VAL start and copy
		pe = strrchr(ps, q) + ps;
		val = pe;
		pe--;
		ESBYTE[pe] = '\0'; 
 
		//find ATTR end
		while (pe > ps) && (ESBYTE[pe] != '=') pe--;
		ESBYTE[pe] = '\0';
	}
	else
	{
		//find VAL start and copy
		while (pe > ps) && (ESBYTE[pe] != '=') pe--;
		val = pe+1;
		ESBYTE[pe] = '\0';
		//already have ATTR end
	}
 
	//find ATTR start and copy
	while (pe>ps) && (!__isWhite(ESBYTE[pe])) pe--;
	attr = pe + 1;
	ESBYTE[pe] = '\0';
 
	// Fix case: src=./images/logo?sid=e8ece8b38b
	// Exchange '=' and '\0' position.
	// attr: src=./images/logo?sid   =>   src
	// val:  e8ece8b38b              =>   ./images/logo?sid=e8ece8b38b
	fixeq = strchr(attr,'=');
	if (!q) && (fixeq) {
		ESBYTE[val-1] >< ESBYTE[fixeq];
		val = fixeq+1;
	}
	strlwr(attr);
	strrtrim(val);
 
	attributes.add(attr);
	values.add(val);
 
	if (debug_mode) {
		debug("atr: "); debugln(attr);
		debug("val: "); debugln(val);
		debugch('\n');
	}
 
	if (pe==ps) return NULL;
	return pe;
}
 
dword _tag::get_value_of(dword _attr_name)
{
	int pos = attributes.get_pos_by_name(_attr_name);
	if (pos == -1) {
		value = 0;
	} else {
		value = values.get(pos);
	}
	return value;
}
>

Rev 8445	Rev 8490
1	struct _tag	1	struct _tag
2	{	2	{
3	char name[32];	3	char name[32];
4	char prior[32];	4	char prior[32];
5	bool opened;	5	bool opened;
6	collection attributes;	6	collection attributes;
7	collection values;	7	collection values;
8	dword value;	8	dword value;
9	bool is();	9	bool is();
10	bool parse();	10	bool parse();
11	void debug_tag();	11	void debug_tag();
12	dword get_next_param();	12	dword get_next_param();
13	dword get_value_of();	13	dword get_value_of();
14	} tag=0;	14	} tag=0;
15		15
16	bool _tag::is(dword _text)	16	bool _tag::is(dword _text)
17	{	17	{
18	if ( !strcmp(#name, _text) ) {	18	if ( !strcmp(#name, _text) ) {
19	return true;	19	return true;
20	} else {	20	} else {
21	return false;	21	return false;
22	}	22	}
23	}	23	}
24		24
25	bool _tag::parse(dword _bufpos, bufend)	25	bool _tag::parse(dword _bufpos, bufend)
26	{	26	{
27	bool retok = true;	27	bool retok = true;
28	dword bufpos = ESDWORD[_bufpos];	28	dword bufpos = ESDWORD[_bufpos];
29	dword params, paramsend;	29	dword params, paramsend;
30		30
31	dword closepos;	31	dword closepos;
32	dword whitepos;	32	dword whitepos;
33		33	dword openpos;
-		34
34	if (name) strcpy(#prior, #name); else prior = '\0';	35	if (name) strcpy(#prior, #name); else prior = '\0';
35	name = '\0';	36	name = '\0';
36	attributes.drop();	37	attributes.drop();
37	values.drop();	38	values.drop();
38		39
39	if (!strncmp(bufpos,"!--",3))	40	if (!strncmp(bufpos,"!--",3))
40	{	41	{
41	bufpos+=3;	42	bufpos+=3;
42	//STRSTR	43	//STRSTR
43	while (strncmp(bufpos,"-->",3)!=0) && (bufpos < bufend)	44	while (strncmp(bufpos,"-->",3)!=0) && (bufpos < bufend)
44	{	45	{
45	bufpos++;	46	bufpos++;
46	}	47	}
47	bufpos+=2;	48	bufpos+=2;
48	retok = false;	49	retok = false;
49	goto _RET;	50	goto _RET;
50	}	51	}
51		52
52	if (ESBYTE[bufpos] == '/') {	53	if (ESBYTE[bufpos] == '/') {
53	opened = false;	54	opened = false;
54	bufpos++;	55	bufpos++;
55	} else {	56	} else {
56	opened = true;	57	opened = true;
57	}	58	}
58		59
59	closepos = strchr(bufpos, '>');	60	closepos = strchr(bufpos, '>');
60	whitepos = strchrw(bufpos, bufend-bufpos);	61	whitepos = strchrw(bufpos, bufend-bufpos);
61		62
62	if (debug_mode) {	63	if (debug_mode) {
63	if (!closepos) debugln("null closepos");	64	if (!closepos) debugln("null closepos");
64	if (!whitepos) debugln("null whitepos");	65	if (!whitepos) debugln("null whitepos");
65	}	66	}
66		67
67	if (!whitepos) \|\| (whitepos > closepos) {	68	if (!whitepos) \|\| (whitepos > closepos) {
68	//no param	69	//no param
69	strncpy(#name, bufpos, math.min(closepos - bufpos, sizeof(tag.name)));	70	strncpy(#name, bufpos, math.min(closepos - bufpos, sizeof(tag.name)));
70	debug_tag();	71	debug_tag();
71	bufpos = closepos;	72	bufpos = closepos;
72	} else {	73	} else {
73	//we have param	74	//we have param
74	strncpy(#name, bufpos, math.min(whitepos - bufpos, sizeof(tag.name)));	75	while (chrlnum(whitepos, '\"', closepos - whitepos)%2) { //alt="Next>>"
-		76	/*
-		77	openpos = strchr(closepos+1, '<');
-		78	closepos = strchr(closepos+1, '>');
-		79	if (openpos) && (openpos < closepos) {
-		80	closepos = openpos - 1;
-		81	break;
-		82	}
-		83	*/
-		84	if (!openpos = strchr(closepos+1, '<')) break;
-		85	if (openpos < strchr(closepos+1, '>')) break;
-		86	if (!closepos = EAX) {closepos = bufend;break;}
-		87	}
-		88	strncpy(#name, bufpos, math.min(whitepos - bufpos, sizeof(tag.name)));
75	debug_tag();	89	debug_tag();
76	bufpos = closepos;	90	bufpos = closepos;
77		91
78	params = malloc(closepos - whitepos + 1);	92	params = malloc(closepos - whitepos + 1);
79	strncpy(params, whitepos, closepos - whitepos);	93	strncpy(params, whitepos, closepos - whitepos);
80	if (debug_mode) { debug("params: "); debugln(params+1); }	94	if (debug_mode) { debug("params: "); debugln(params+1); }
81	paramsend = params + closepos - whitepos;	95	paramsend = params + closepos - whitepos;
82	while (paramsend = get_next_param(params, paramsend-1));	96	while (paramsend = get_next_param(params, paramsend-1));
83	free(params);	97	free(params);
84	}	98	}
85		99
86	if (name) {	100	if (name) {
87	strlwr(#name);	101	strlwr(#name);
88	// ignore text inside the next tags	102	// ignore text inside the next tags
89	if (is("script")) \|\| (is("style")) \|\| (is("binary")) \|\| (is("select")) {	103	if (is("script")) \|\| (is("style")) \|\| (is("binary")) \|\| (is("select")) {
90	strcpy(#prior, #name);	104	strcpy(#prior, #name);
91	sprintf(#name, "", #prior);	105	sprintf(#name, "", #prior);
92	if (strstri(bufpos, #name)) bufpos = EAX-1;	106	if (strstri(bufpos, #name)) bufpos = EAX-1;
93	retok = false;	107	retok = false;
94	} else {	108	} else {
95	if (name[strlen(#name)-1]=='/') name[strlen(#name)-1]=NULL; //for	109	if (name[strlen(#name)-1]=='/') name[strlen(#name)-1]=NULL; //for
96	}	110	}
97	} else {	111	} else {
98	retok = false;	112	retok = false;
99	}	113	}
100		114
101	_RET:	115	_RET:
102	ESDWORD[_bufpos] = bufpos;	116	ESDWORD[_bufpos] = bufpos;
103	return retok;	117	return retok;
104	}	118	}
105		119
106	void _tag::debug_tag()	120	void _tag::debug_tag()
107	{	121	{
108	if (debug_mode) {	122	if (debug_mode) {
109	debugch('<');	123	debugch('<');
110	if (!opened) debugch('/');	124	if (!opened) debugch('/');
111	debug(#name);	125	debug(#name);
112	debugln(">");	126	debugln(">");
113	}	127	}
114	}	128	}
115		129
116	dword _tag::get_next_param(dword ps, pe)	130	dword _tag::get_next_param(dword ps, pe)
117	{	131	{
118	// "ps" - param start	132	// "ps" - param start
119	// "pe" - param end	133	// "pe" - param end
120	// "q" - quote char	134	// "q" - quote char
121	char q = NULL;	135	char q = NULL;
122	dword fixeq;	136	dword fixeq;
123	dword val;	137	dword val;
124	dword attr;	138	dword attr;
125		139
126	if (ESBYTE[pe] == '/') pe--;	140	if (ESBYTE[pe] == '/') pe--;
127	while (pe>ps) && (__isWhite(ESBYTE[pe])) pe--;	141	while (pe>ps) && (__isWhite(ESBYTE[pe])) pe--;
128		142
129	if (ESBYTE[pe] == '"') \|\| (ESBYTE[pe] == '\'')	143	if (ESBYTE[pe] == '"') \|\| (ESBYTE[pe] == '\'')
130	{	144	{
131	//remove quote	145	//remove quote
132	q = ESBYTE[pe];	146	q = ESBYTE[pe];
133	ESBYTE[pe] = '\0';	147	ESBYTE[pe] = '\0';
134	pe--;	148	pe--;
135		149
136	//find VAL start and copy	150	//find VAL start and copy
137	pe = strrchr(ps, q) + ps;	151	pe = strrchr(ps, q) + ps;
138	val = pe;	152	val = pe;
139	pe--;	153	pe--;
140	ESBYTE[pe] = '\0';	154	ESBYTE[pe] = '\0';
141		155
142	//find ATTR end	156	//find ATTR end
143	while (pe > ps) && (ESBYTE[pe] != '=') pe--;	157	while (pe > ps) && (ESBYTE[pe] != '=') pe--;
144	ESBYTE[pe] = '\0';	158	ESBYTE[pe] = '\0';
145	}	159	}
146	else	160	else
147	{	161	{
148	//find VAL start and copy	162	//find VAL start and copy
149	while (pe > ps) && (ESBYTE[pe] != '=') pe--;	163	while (pe > ps) && (ESBYTE[pe] != '=') pe--;
150	val = pe+1;	164	val = pe+1;
151	ESBYTE[pe] = '\0';	165	ESBYTE[pe] = '\0';
152	//already have ATTR end	166	//already have ATTR end
153	}	167	}
154		168
155	//find ATTR start and copy	169	//find ATTR start and copy
156	while (pe>ps) && (!__isWhite(ESBYTE[pe])) pe--;	170	while (pe>ps) && (!__isWhite(ESBYTE[pe])) pe--;
157	attr = pe + 1;	171	attr = pe + 1;
158	ESBYTE[pe] = '\0';	172	ESBYTE[pe] = '\0';
159		173
160	// Fix case: src=./images/logo?sid=e8ece8b38b	174	// Fix case: src=./images/logo?sid=e8ece8b38b
161	// Exchange '=' and '\0' position.	175	// Exchange '=' and '\0' position.
162	// attr: src=./images/logo?sid => src	176	// attr: src=./images/logo?sid => src
163	// val: e8ece8b38b => ./images/logo?sid=e8ece8b38b	177	// val: e8ece8b38b => ./images/logo?sid=e8ece8b38b
164	fixeq = strchr(attr,'=');	178	fixeq = strchr(attr,'=');
165	if (!q) && (fixeq) {	179	if (!q) && (fixeq) {
166	ESBYTE[val-1] >< ESBYTE[fixeq];	180	ESBYTE[val-1] >< ESBYTE[fixeq];
167	val = fixeq+1;	181	val = fixeq+1;
168	}	182	}
169	strlwr(attr);	183	strlwr(attr);
170	strrtrim(val);	184	strrtrim(val);
171		185
172	attributes.add(attr);	186	attributes.add(attr);
173	values.add(val);	187	values.add(val);
174		188
175	if (debug_mode) {	189	if (debug_mode) {
176	debug("atr: "); debugln(attr);	190	debug("atr: "); debugln(attr);
177	debug("val: "); debugln(val);	191	debug("val: "); debugln(val);
178	debugch('\n');	192	debugch('\n');
179	}	193	}
180		194
181	if (pe==ps) return NULL;	195	if (pe==ps) return NULL;
182	return pe;	196	return pe;
183	}	197	}
184		198
185	dword _tag::get_value_of(dword _attr_name)	199	dword _tag::get_value_of(dword _attr_name)
186	{	200	{
187	int pos = attributes.get_pos_by_name(_attr_name);	201	int pos = attributes.get_pos_by_name(_attr_name);
188	if (pos == -1) {	202	if (pos == -1) {
189	value = 0;	203	value = 0;
190	} else {	204	} else {
191	value = values.get(pos);	205	value = values.get(pos);
192	}	206	}
193	return value;	207	return value;
194	}	208	}
-		209	>

Subversion Repositories Kolibri OS

(root)/programs/cmm/browser/TWB/parse_tag.h – Rev 8445 → 8490