[ia32rtools.git] / tools / protoparse.h


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