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


/*
struct _DOM {
	collection tag;
	collection tagparam;
	collection_int parent;
} dom;
*/
 
struct _tag
{
	char name[32];
	char prior[32];
	bool opened;
	collection attributes;
	collection values;
	dword value;
	dword number;
	bool is();
	bool parse();
	dword get_next_param();
	dword get_value_of();
	signed get_number_of();
} tag=0;
 
bool _tag::is(dword _text) 
{ 
	return streq(#name, _text);
}
 
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 (!whitepos) || (whitepos > closepos) {
		//no param
		strncpy(#name, bufpos, math.min(closepos - bufpos, sizeof(tag.name)));
		bufpos = closepos;
	} else {
		//we have param
		while (chrlnum(whitepos, '\"', closepos - whitepos)%2) { //alt="Next>>"
			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)));
		bufpos = closepos;
 
		params = malloc(closepos - whitepos + 1);
		strncpy(params, whitepos, closepos - whitepos);
		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;
}
 
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;
		if (pe>ps) 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 (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;
}
 
signed _tag::get_number_of(dword _attr_name)
{
	if (get_value_of(_attr_name)) {
		number = atoi(value);
	} else {
		number = 0;
	}
	return number;
}

Rev 9089	Rev 9103
-		1	/*
-		2	struct _DOM {
-		3	collection tag;
-		4	collection tagparam;
-		5	collection_int parent;
-		6	} dom;
-		7	*/
-		8
1	struct _tag	9	struct _tag
2	{	10	{
3	char name[32];	11	char name[32];
4	char prior[32];	12	char prior[32];
5	bool opened;	13	bool opened;
6	collection attributes;	14	collection attributes;
7	collection values;	15	collection values;
8	dword value;	16	dword value;
9	dword number;	17	dword number;
10	bool is();	18	bool is();
11	bool parse();	19	bool parse();
12	dword get_next_param();	20	dword get_next_param();
13	dword get_value_of();	21	dword get_value_of();
14	signed get_number_of();	22	signed get_number_of();
15	} tag=0;	23	} tag=0;
16		24
17	bool _tag::is(dword _text)	25	bool _tag::is(dword _text)
18	{	26	{
19	return streq(#name, _text);	27	return streq(#name, _text);
20	}	28	}
21		29
22	bool _tag::parse(dword _bufpos, bufend)	30	bool _tag::parse(dword _bufpos, bufend)
23	{	31	{
24	bool retok = true;	32	bool retok = true;
25	dword bufpos = ESDWORD[_bufpos];	33	dword bufpos = ESDWORD[_bufpos];
26	dword params, paramsend;	34	dword params, paramsend;
27		35
28	dword closepos;	36	dword closepos;
29	dword whitepos;	37	dword whitepos;
30	dword openpos;	38	dword openpos;
31		39
32	if (name) strcpy(#prior, #name); else prior = '\0';	40	if (name) strcpy(#prior, #name); else prior = '\0';
33	name = '\0';	41	name = '\0';
34	attributes.drop();	42	attributes.drop();
35	values.drop();	43	values.drop();
36		44
37	if (!strncmp(bufpos,"!--",3))	45	if (!strncmp(bufpos,"!--",3))
38	{	46	{
39	bufpos+=3;	47	bufpos+=3;
40	//STRSTR	48	//STRSTR
41	while (strncmp(bufpos,"-->",3)!=0) && (bufpos < bufend)	49	while (strncmp(bufpos,"-->",3)!=0) && (bufpos < bufend)
42	{	50	{
43	bufpos++;	51	bufpos++;
44	}	52	}
45	bufpos+=2;	53	bufpos+=2;
46	retok = false;	54	retok = false;
47	goto _RET;	55	goto _RET;
48	}	56	}
49		57
50	if (ESBYTE[bufpos] == '/') {	58	if (ESBYTE[bufpos] == '/') {
51	opened = false;	59	opened = false;
52	bufpos++;	60	bufpos++;
53	} else {	61	} else {
54	opened = true;	62	opened = true;
55	}	63	}
56		64
57	closepos = strchr(bufpos, '>');	65	closepos = strchr(bufpos, '>');
58	whitepos = strchrw(bufpos, bufend-bufpos);	66	whitepos = strchrw(bufpos, bufend-bufpos);
59		67
60	if (!whitepos) \|\| (whitepos > closepos) {	68	if (!whitepos) \|\| (whitepos > closepos) {
61	//no param	69	//no param
62	strncpy(#name, bufpos, math.min(closepos - bufpos, sizeof(tag.name)));	70	strncpy(#name, bufpos, math.min(closepos - bufpos, sizeof(tag.name)));
63	bufpos = closepos;	71	bufpos = closepos;
64	} else {	72	} else {
65	//we have param	73	//we have param
66	while (chrlnum(whitepos, '\"', closepos - whitepos)%2) { //alt="Next>>"	74	while (chrlnum(whitepos, '\"', closepos - whitepos)%2) { //alt="Next>>"
67	if (!openpos = strchr(closepos+1, '<')) break;	75	if (!openpos = strchr(closepos+1, '<')) break;
68	if (openpos < strchr(closepos+1, '>')) break;	76	if (openpos < strchr(closepos+1, '>')) break;
69	if (!closepos = EAX) {closepos = bufend;break;}	77	if (!closepos = EAX) {closepos = bufend;break;}
70	}	78	}
71	strncpy(#name, bufpos, math.min(whitepos - bufpos, sizeof(tag.name)));	79	strncpy(#name, bufpos, math.min(whitepos - bufpos, sizeof(tag.name)));
72	bufpos = closepos;	80	bufpos = closepos;
73		81
74	params = malloc(closepos - whitepos + 1);	82	params = malloc(closepos - whitepos + 1);
75	strncpy(params, whitepos, closepos - whitepos);	83	strncpy(params, whitepos, closepos - whitepos);
76	paramsend = params + closepos - whitepos;	84	paramsend = params + closepos - whitepos;
77	while (paramsend = get_next_param(params, paramsend-1));	85	while (paramsend = get_next_param(params, paramsend-1));
78	free(params);	86	free(params);
79	}	87	}
80		88
81	if (name) {	89	if (name) {
82	strlwr(#name);	90	strlwr(#name);
83	// ignore text inside the next tags	91	// ignore text inside the next tags
84	if (is("script")) \|\| (is("style")) \|\| (is("binary")) \|\| (is("select")) {	92	if (is("script")) \|\| (is("style")) \|\| (is("binary")) \|\| (is("select")) {
85	strcpy(#prior, #name);	93	strcpy(#prior, #name);
86	sprintf(#name, "", #prior);	94	sprintf(#name, "", #prior);
87	if (strstri(bufpos, #name)) bufpos = EAX-1;	95	if (strstri(bufpos, #name)) bufpos = EAX-1;
88	retok = false;	96	retok = false;
89	} else {	97	} else {
90	if (name[strlen(#name)-1]=='/') name[strlen(#name)-1]=NULL; //for	98	if (name[strlen(#name)-1]=='/') name[strlen(#name)-1]=NULL; //for
91	}	99	}
92	} else {	100	} else {
93	retok = false;	101	retok = false;
94	}	102	}
95		103
96	_RET:	104	_RET:
97	ESDWORD[_bufpos] = bufpos;	105	ESDWORD[_bufpos] = bufpos;
98	return retok;	106	return retok;
99	}	107	}
100		108
101	dword _tag::get_next_param(dword ps, pe)	109	dword _tag::get_next_param(dword ps, pe)
102	{	110	{
103	// "ps" - param start	111	// "ps" - param start
104	// "pe" - param end	112	// "pe" - param end
105	// "q" - quote char	113	// "q" - quote char
106	char q = NULL;	114	char q = NULL;
107	dword fixeq;	115	dword fixeq;
108	dword val;	116	dword val;
109	dword attr;	117	dword attr;
110		118
111	if (ESBYTE[pe] == '/') pe--;	119	if (ESBYTE[pe] == '/') pe--;
112	while (pe>ps) && (__isWhite(ESBYTE[pe])) pe--;	120	while (pe>ps) && (__isWhite(ESBYTE[pe])) pe--;
113		121
114	if (ESBYTE[pe] == '"') \|\| (ESBYTE[pe] == '\'')	122	if (ESBYTE[pe] == '"') \|\| (ESBYTE[pe] == '\'')
115	{	123	{
116	//remove quote	124	//remove quote
117	q = ESBYTE[pe];	125	q = ESBYTE[pe];
118	ESBYTE[pe] = '\0';	126	ESBYTE[pe] = '\0';
119	pe--;	127	pe--;
120		128
121	//find VAL start and copy	129	//find VAL start and copy
122	pe = strrchr(ps, q) + ps;	130	pe = strrchr(ps, q) + ps;
123	val = pe;	131	val = pe;
124	pe--;	132	if (pe>ps) pe--;
125	ESBYTE[pe] = '\0';	133	ESBYTE[pe] = '\0';
126		134
127	//find ATTR end	135	//find ATTR end
128	while (pe > ps) && (ESBYTE[pe] != '=') pe--;	136	while (pe > ps) && (ESBYTE[pe] != '=') pe--;
129	ESBYTE[pe] = '\0';	137	ESBYTE[pe] = '\0';
130	}	138	}
131	else	139	else
132	{	140	{
133	//find VAL start and copy	141	//find VAL start and copy
134	while (pe > ps) && (ESBYTE[pe] != '=') pe--;	142	while (pe > ps) && (ESBYTE[pe] != '=') pe--;
135	val = pe+1;	143	val = pe+1;
136	ESBYTE[pe] = '\0';	144	ESBYTE[pe] = '\0';
137	//already have ATTR end	145	//already have ATTR end
138	}	146	}
139		147
140	//find ATTR start and copy	148	//find ATTR start and copy
141	while (pe>ps) && (!__isWhite(ESBYTE[pe])) pe--;	149	while (pe>ps) && (!__isWhite(ESBYTE[pe])) pe--;
142	attr = pe + 1;	150	attr = pe + 1;
143	ESBYTE[pe] = '\0';	151	ESBYTE[pe] = '\0';
144		152
145	// Fix case: src=./images/logo?sid=e8ece8b38b	153	// Fix case: src=./images/logo?sid=e8ece8b38b
146	// Exchange '=' and '\0' position.	154	// Exchange '=' and '\0' position.
147	// attr: src=./images/logo?sid => src	155	// attr: src=./images/logo?sid => src
148	// val: e8ece8b38b => ./images/logo?sid=e8ece8b38b	156	// val: e8ece8b38b => ./images/logo?sid=e8ece8b38b
149	fixeq = strchr(attr,'=');	157	fixeq = strchr(attr,'=');
150	if (!q) && (fixeq) {	158	if (!q) && (fixeq) {
151	ESBYTE[val-1] >< ESBYTE[fixeq];	159	ESBYTE[val-1] >< ESBYTE[fixeq];
152	val = fixeq+1;	160	val = fixeq+1;
153	}	161	}
154	strlwr(attr);	162	strlwr(attr);
155	strrtrim(val);	163	strrtrim(val);
156		164
157	attributes.add(attr);	165	attributes.add(attr);
158	values.add(val);	166	values.add(val);
159		167
160	if (pe==ps) return NULL;	168	if (pe==ps) return NULL;
161	return pe;	169	return pe;
162	}	170	}
163		171
164	dword _tag::get_value_of(dword _attr_name)	172	dword _tag::get_value_of(dword _attr_name)
165	{	173	{
166	int pos = attributes.get_pos_by_name(_attr_name);	174	int pos = attributes.get_pos_by_name(_attr_name);
167	if (pos == -1) {	175	if (pos == -1) {
168	value = 0;	176	value = 0;
169	} else {	177	} else {
170	value = values.get(pos);	178	value = values.get(pos);
171	}	179	}
172	return value;	180	return value;
173	}	181	}
174		182
175	signed _tag::get_number_of(dword _attr_name)	183	signed _tag::get_number_of(dword _attr_name)
176	{	184	{
177	if (get_value_of(_attr_name)) {	185	if (get_value_of(_attr_name)) {
178	number = atoi(value);	186	number = atoi(value);
179	} else {	187	} else {
180	number = 0;	188	number = 0;
181	}	189	}
182	return number;	190	return number;
183	}	191	}

Subversion Repositories Kolibri OS

(root)/programs/cmm/browser/TWB/parse_tag.h – Rev 9089 → 9103