[ia32rtools.git] / tools / protoparse.h


struct parsed_proto {
	struct {
		char *reg;
		const char *type;
		void *datap;
	} arg[16];
	const char *ret_type;
	int is_stdcall;
	int argc;
	int argc_stack;
	int argc_reg;
};

static const char *hdrfn;
static int hdrfline = 0;

static int find_protostr(char *dst, size_t dlen, FILE *fhdr, const char *sym)
{
	int line = 0;
	char *p;

	rewind(fhdr);

	while (fgets(dst, dlen, fhdr))
	{
		line++;
		if (strstr(dst, sym) != NULL)
			break;
	}
	hdrfline = line;

	if (feof(fhdr))
		return -1;

	p = dst + strlen(dst);
	for (p--; p > dst && my_isblank(*p); --p)
		*p = 0;

	return 0;
}

static int get_regparm(char *dst, size_t dlen, char *p)
{
	int i, o;

	if (*p != '<')
		return 0;

	for (o = 0, i = 1; o < dlen; i++) {
		if (p[i] == 0)
			return 0;
		if (p[i] == '>')
			break;
		dst[o++] = p[i];
	}
	dst[o] = 0;
	return i + 1;
}

// hmh..
static const char *known_types[] = {
	"const void *",
	"void *",
	"char *",
	"FILE *",
	"unsigned __int8",
	"unsigned __int16",
	"unsigned int",
	"signed int",
	"char",
	"__int8",
	"__int16",
	"int",
	"bool",
	"void",
	"BYTE",
	"WORD",
	"DWORD",
	"_DWORD",
	"HMODULE",
	"HANDLE",
	"HWND",
	"LPCSTR",
	"size_t",
};

static const char *check_type(const char *name)
{
	int i, l;

	for (i = 0; i < ARRAY_SIZE(known_types); i++) {
		l = strlen(known_types[i]);
		if (strncmp(known_types[i], name, l) == 0)
			return known_types[i];
	}

	return NULL;
}

/* args are always expanded to 32bit */
static const char *map_reg(const char *reg)
{
	const char *regs_f[] = { "eax", "ebx", "ecx", "edx", "esi", "edi" };
	const char *regs_w[] = { "ax",  "bx",  "cx",  "dx",  "si",  "di" };
	const char *regs_b[] = { "al",  "bl",  "cl",  "dl" };
	int i;

	for (i = 0; i < ARRAY_SIZE(regs_w); i++)
		if (IS(reg, regs_w[i]))
			return regs_f[i];

	for (i = 0; i < ARRAY_SIZE(regs_b); i++)
		if (IS(reg, regs_b[i]))
			return regs_f[i];

	return reg;
}

static int parse_protostr(char *protostr, struct parsed_proto *pp)
{
	char regparm[16];
	char buf[256];
	char cconv[32];
	const char *kt;
	int xarg = 0;
	int ret;
	char *p;
	int i;

	p = protostr;
	if (p[0] == '/' && p[1] == '/') {
		//printf("warning: decl for sym '%s' is commented out\n", sym);
		p = sskip(p + 2);
	}

	kt = check_type(p);
	if (kt == NULL) {
		printf("%s:%d:%ld: unhandled return in '%s'\n",
			hdrfn, hdrfline, (p - protostr) + 1, protostr);
		return 1;
	}
	pp->ret_type = kt;
	p += strlen(kt);
	p = sskip(p);

	p = next_word(cconv, sizeof(cconv), p);
	p = sskip(p);
	if (cconv[0] == 0) {
		printf("%s:%d:%ld: cconv missing\n",
			hdrfn, hdrfline, (p - protostr) + 1);
		return 1;
	}
	if      (IS(cconv, "__cdecl"))
		pp->is_stdcall = 0;
	else if (IS(cconv, "__stdcall"))
		pp->is_stdcall = 1;
	else if (IS(cconv, "__fastcall"))
		pp->is_stdcall = 1;
	else if (IS(cconv, "__thiscall"))
		pp->is_stdcall = 1;
	else if (IS(cconv, "__userpurge"))
		pp->is_stdcall = 1; // in all cases seen..
	else if (IS(cconv, "__usercall"))
		pp->is_stdcall = 0; // ..or is it?
	else {
		printf("%s:%d:%ld: unhandled cconv: '%s'\n",
			hdrfn, hdrfline, (p - protostr) + 1, cconv);
		return 1;
	}

	p = next_idt(buf, sizeof(buf), p);
	p = sskip(p);
	if (buf[0] == 0) {
		printf("%s:%d:%ld: func name missing\n",
				hdrfn, hdrfline, (p - protostr) + 1);
		return 1;
	}

	ret = get_regparm(regparm, sizeof(regparm), p);
	if (ret > 0) {
		if (!IS(regparm, "eax") && !IS(regparm, "ax")
		 && !IS(regparm, "al"))
		{
			printf("%s:%d:%ld: bad regparm: %s\n",
				hdrfn, hdrfline, (p - protostr) + 1, regparm);
			return 1;
		}
		p += ret;
		p = sskip(p);
	}

	if (*p != '(') {
		printf("%s:%d:%ld: '(' expected, got '%c'\n",
				hdrfn, hdrfline, (p - protostr) + 1, *p);
		return 1;
	}
	p++;

	while (1) {
		p = sskip(p);
		if (*p == ')')
			break;
		if (*p == ',')
			p = sskip(p + 1);

		xarg++;

		kt = check_type(p);
		if (kt == NULL) {
			printf("%s:%d:%ld: unhandled type for arg%d\n",
				hdrfn, hdrfline, (p - protostr) + 1, xarg);
			return 1;
		}
		pp->arg[xarg - 1].type = kt;
		p += strlen(kt);
		p = sskip(p);

		p = next_idt(buf, sizeof(buf), p);
		p = sskip(p);
#if 0
		if (buf[0] == 0) {
			printf("%s:%d:%ld: idt missing for arg%d\n",
				hdrfn, hdrfline, (p - protostr) + 1, xarg);
			return 1;
		}
#endif
		pp->arg[xarg - 1].reg = NULL;

		ret = get_regparm(regparm, sizeof(regparm), p);
		if (ret > 0) {
			p += ret;
			p = sskip(p);

			pp->arg[xarg - 1].reg = strdup(map_reg(regparm));
		}
	}

	if (xarg > 0 && (IS(cconv, "__fastcall") || IS(cconv, "__thiscall"))) {
		if (pp->arg[0].reg != NULL) {
			printf("%s:%d: %s with arg1 spec %s?\n",
				hdrfn, hdrfline, cconv, pp->arg[0].reg);
		}
		pp->arg[0].reg = strdup("ecx");
	}

	if (xarg > 1 && IS(cconv, "__fastcall")) {
		if (pp->arg[1].reg != NULL) {
			printf("%s:%d: %s with arg2 spec %s?\n",
				hdrfn, hdrfline, cconv, pp->arg[1].reg);
		}
		pp->arg[1].reg = strdup("edx");
	}

	pp->argc = xarg;

	for (i = 0; i < pp->argc; i++) {
		if (pp->arg[i].reg == NULL)
			pp->argc_stack++;
		else
			pp->argc_reg++;
	}

	return 0;
}

static int proto_parse(FILE *fhdr, const char *sym, struct parsed_proto *pp)
{
	char protostr[256];
	int ret;

	memset(pp, 0, sizeof(*pp));

	ret = find_protostr(protostr, sizeof(protostr), fhdr, sym);
	if (ret != 0) {
		printf("%s: sym '%s' is missing\n", hdrfn, sym);
		return ret;
	}

	return parse_protostr(protostr, pp);
}

static void proto_release(struct parsed_proto *pp)
{
	int i;

	for (i = 0; i < pp->argc; i++) {
		if (pp->arg[i].reg == NULL)
			free(pp->arg[i].reg);
	}
}
Commit	Line	Data
c36e914d	1
	2	struct parsed_proto {
	3	struct {
	4	char *reg;
	5	const char *type;
91977a1c	6	void *datap;
c36e914d	7	} arg[16];
	8	const char *ret_type;
	9	int is_stdcall;
	10	int argc;
	11	int argc_stack;
	12	int argc_reg;
	13	};
	14
	15	static const char *hdrfn;
	16	static int hdrfline = 0;
	17
	18	static int find_protostr(char dst, size_t dlen, FILE fhdr, const char *sym)
	19	{
	20	int line = 0;
	21	char *p;
	22
	23	rewind(fhdr);
	24
	25	while (fgets(dst, dlen, fhdr))
	26	{
	27	line++;
	28	if (strstr(dst, sym) != NULL)
	29	break;
	30	}
	31	hdrfline = line;
	32
	33	if (feof(fhdr))
	34	return -1;
	35
	36	p = dst + strlen(dst);
	37	for (p--; p > dst && my_isblank(*p); --p)
	38	*p = 0;
	39
	40	return 0;
	41	}
	42
	43	static int get_regparm(char dst, size_t dlen, char p)
	44	{
	45	int i, o;
	46
	47	if (*p != '<')
	48	return 0;
	49
	50	for (o = 0, i = 1; o < dlen; i++) {
	51	if (p[i] == 0)
	52	return 0;
	53	if (p[i] == '>')
	54	break;
	55	dst[o++] = p[i];
	56	}
	57	dst[o] = 0;
	58	return i + 1;
	59	}
	60
	61	// hmh..
	62	static const char *known_types[] = {
	63	"const void *",
	64	"void *",
	65	"char *",
	66	"FILE *",
	67	"unsigned __int8",
	68	"unsigned __int16",
	69	"unsigned int",
	70	"signed int",
71	"char",
72	"__int8",
73	"__int16",
74	"int",
75	"bool",
76	"void",
77	"BYTE",
78	"WORD",
79	"DWORD",
91977a1c	80	"_DWORD",
c36e914d	81	"HMODULE",
	82	"HANDLE",
	83	"HWND",
	84	"LPCSTR",
	85	"size_t",
	86	};
	87
	88	static const char check_type(const char name)
	89	{
	90	int i, l;
	91
	92	for (i = 0; i < ARRAY_SIZE(known_types); i++) {
	93	l = strlen(known_types[i]);
	94	if (strncmp(known_types[i], name, l) == 0)
	95	return known_types[i];
	96	}
	97
	98	return NULL;
	99	}
	100
	101	/* args are always expanded to 32bit */
	102	static const char map_reg(const char reg)
	103	{
	104	const char *regs_f[] = { "eax", "ebx", "ecx", "edx", "esi", "edi" };
	105	const char *regs_w[] = { "ax", "bx", "cx", "dx", "si", "di" };
	106	const char *regs_b[] = { "al", "bl", "cl", "dl" };
	107	int i;
	108
	109	for (i = 0; i < ARRAY_SIZE(regs_w); i++)
	110	if (IS(reg, regs_w[i]))
	111	return regs_f[i];
	112
	113	for (i = 0; i < ARRAY_SIZE(regs_b); i++)
	114	if (IS(reg, regs_b[i]))
	115	return regs_f[i];
	116
	117	return reg;
	118	}
	119
	120	static int parse_protostr(char protostr, struct parsed_proto pp)
	121	{
	122	char regparm[16];
	123	char buf[256];
	124	char cconv[32];
	125	const char *kt;
	126	int xarg = 0;
	127	int ret;
	128	char *p;
	129	int i;
	130
	131	p = protostr;
	132	if (p[0] == '/' && p[1] == '/') {
	133	//printf("warning: decl for sym '%s' is commented out\n", sym);
	134	p = sskip(p + 2);
	135	}
	136
	137	kt = check_type(p);
	138	if (kt == NULL) {
	139	printf("%s:%d:%ld: unhandled return in '%s'\n",
	140	hdrfn, hdrfline, (p - protostr) + 1, protostr);
	141	return 1;
	142	}
	143	pp->ret_type = kt;
	144	p += strlen(kt);
145	p = sskip(p);
146
147	p = next_word(cconv, sizeof(cconv), p);
148	p = sskip(p);
149	if (cconv[0] == 0) {
150	printf("%s:%d:%ld: cconv missing\n",
151	hdrfn, hdrfline, (p - protostr) + 1);
152	return 1;
153	}
154	if (IS(cconv, "__cdecl"))
155	pp->is_stdcall = 0;
156	else if (IS(cconv, "__stdcall"))
157	pp->is_stdcall = 1;
158	else if (IS(cconv, "__fastcall"))
159	pp->is_stdcall = 1;
160	else if (IS(cconv, "__thiscall"))
161	pp->is_stdcall = 1;
162	else if (IS(cconv, "__userpurge"))
163	pp->is_stdcall = 1; // in all cases seen..
164	else if (IS(cconv, "__usercall"))
165	pp->is_stdcall = 0; // ..or is it?
166	else {
167	printf("%s:%d:%ld: unhandled cconv: '%s'\n",
168	hdrfn, hdrfline, (p - protostr) + 1, cconv);
169	return 1;
170	}
171
172	p = next_idt(buf, sizeof(buf), p);
173	p = sskip(p);
174	if (buf[0] == 0) {
175	printf("%s:%d:%ld: func name missing\n",
176	hdrfn, hdrfline, (p - protostr) + 1);
177	return 1;
178	}
179
180	ret = get_regparm(regparm, sizeof(regparm), p);
181	if (ret > 0) {
182	if (!IS(regparm, "eax") && !IS(regparm, "ax")
183	&& !IS(regparm, "al"))
184	{
185	printf("%s:%d:%ld: bad regparm: %s\n",
186	hdrfn, hdrfline, (p - protostr) + 1, regparm);
187	return 1;
188	}
189	p += ret;
190	p = sskip(p);
191	}
192
193	if (*p != '(') {
194	printf("%s:%d:%ld: '(' expected, got '%c'\n",
195	hdrfn, hdrfline, (p - protostr) + 1, *p);
196	return 1;
197	}
198	p++;
199
200	while (1) {
201	p = sskip(p);
202	if (*p == ')')
203	break;
204	if (*p == ',')
205	p = sskip(p + 1);
206
207	xarg++;
208
209	kt = check_type(p);
210	if (kt == NULL) {
211	printf("%s:%d:%ld: unhandled type for arg%d\n",
212	hdrfn, hdrfline, (p - protostr) + 1, xarg);
213	return 1;
214	}
215	pp->arg[xarg - 1].type = kt;
216	p += strlen(kt);
217	p = sskip(p);
218
219	p = next_idt(buf, sizeof(buf), p);
220	p = sskip(p);
221	#if 0
222	if (buf[0] == 0) {
223	printf("%s:%d:%ld: idt missing for arg%d\n",
224	hdrfn, hdrfline, (p - protostr) + 1, xarg);
225	return 1;
226	}
227	#endif
228	pp->arg[xarg - 1].reg = NULL;
229
230	ret = get_regparm(regparm, sizeof(regparm), p);
231	if (ret > 0) {
232	p += ret;
233	p = sskip(p);
234
235	pp->arg[xarg - 1].reg = strdup(map_reg(regparm));
236	}
237	}
238
239	if (xarg > 0 && (IS(cconv, "__fastcall") \|\| IS(cconv, "__thiscall"))) {
240	if (pp->arg[0].reg != NULL) {
241	printf("%s:%d: %s with arg1 spec %s?\n",
242	hdrfn, hdrfline, cconv, pp->arg[0].reg);
243	}
244	pp->arg[0].reg = strdup("ecx");
245	}
246
247	if (xarg > 1 && IS(cconv, "__fastcall")) {
248	if (pp->arg[1].reg != NULL) {
249	printf("%s:%d: %s with arg2 spec %s?\n",
250	hdrfn, hdrfline, cconv, pp->arg[1].reg);
251	}
252	pp->arg[1].reg = strdup("edx");
253	}
254
255	pp->argc = xarg;
256
257	for (i = 0; i < pp->argc; i++) {
258	if (pp->arg[i].reg == NULL)
259	pp->argc_stack++;
260	else
261	pp->argc_reg++;
262	}
263
264	return 0;
265	}
266
267	static int proto_parse(FILE fhdr, const char sym, struct parsed_proto *pp)
268	{
269	char protostr[256];
270	int ret;
271
272	memset(pp, 0, sizeof(*pp));
273
274	ret = find_protostr(protostr, sizeof(protostr), fhdr, sym);
275	if (ret != 0) {
276	printf("%s: sym '%s' is missing\n", hdrfn, sym);
277	return ret;
278	}
279
280	return parse_protostr(protostr, pp);
281	}
282
283	static void proto_release(struct parsed_proto *pp)
284	{
285	int i;
286
287	for (i = 0; i < pp->argc; i++) {
288	if (pp->arg[i].reg == NULL)
289	free(pp->arg[i].reg);
290	}
291	}