X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=lexer.c;h=82bccd05a960748a17a9fdc78f94bb8bcfe34b18;hb=a223a2177b4add7bc2f341675ec1932943553245;hp=73dd134d4ae2984a20c719cd10c02b3e70ac0bd6;hpb=81d6db8265316b7f0d642dc4306c93c633f0b548;p=cparser

diff --git a/lexer.c b/lexer.c
index 73dd134..82bccd0 100644
--- a/lexer.c
+++ b/lexer.c
@@ -7,6 +7,7 @@
 #include "adt/strset.h"
 #include "adt/util.h"
 #include "type_t.h"
+#include "target_architecture.h"
 
 #include <assert.h>
 #include <errno.h>
@@ -17,6 +18,19 @@
 //#define DEBUG_CHARS
 #define MAX_PUTBACK 3
 
+#ifdef _WIN32
+/* No strtold on windows and no replacement yet */
+#define strtold(s, e) strtod(s, e)
+#endif
+
+#if defined HAS_SIGNED_CHAR
+typedef signed char char_type;
+#elif defined HAS_UNSIGNED_CHAR
+typedef unsigned char char_type;
+#else
+#	error signedness of char not determined
+#endif
+
 static int         c;
 token_t            lexer_token;
 symbol_t          *symbol_L;
@@ -75,7 +89,7 @@ static inline void put_back(int pc)
 	//assert(bufpos < buf+MAX_PUTBACK || *bufpos == pc);
 
 	char *p = buf + (bufpos - buf);
-	*p = pc;
+	*p = (char) pc;
 
 	/* going backwards in the buffer is legal as long as it's not more often
 	 * than MAX_PUTBACK */
@@ -234,14 +248,14 @@ static void parse_symbol(void)
 	symbol_t *symbol;
 	char     *string;
 
-	obstack_1grow(&symbol_obstack, c);
+	obstack_1grow(&symbol_obstack, (char) c);
 	next_char();
 
 	while(1) {
 		switch(c) {
 		DIGITS
 		SYMBOL_CHARS
-			obstack_1grow(&symbol_obstack, c);
+			obstack_1grow(&symbol_obstack, (char) c);
 			next_char();
 			break;
 
@@ -264,39 +278,78 @@ end_symbol:
 	}
 }
 
-static void parse_integer_suffix(void)
+static void parse_integer_suffix(bool is_oct_hex)
 {
+	bool is_unsigned  = false;
+	bool min_long     = false;
+	bool min_longlong = false;
+
 	if(c == 'U' || c == 'u') {
+		is_unsigned = true;
 		next_char();
 		if(c == 'L' || c == 'l') {
+			min_long = true;
 			next_char();
 			if(c == 'L' || c == 'l') {
+				min_longlong = true;
 				next_char();
-				lexer_token.datatype = type_ulonglong;
-			} else {
-				lexer_token.datatype = type_ulong;
 			}
-		} else {
-			lexer_token.datatype = type_uint;
 		}
 	} else if(c == 'l' || c == 'L') {
+		min_long = true;
 		next_char();
 		if(c == 'l' || c == 'L') {
+			min_longlong = true;
 			next_char();
 			if(c == 'u' || c == 'U') {
+				is_unsigned = true;
 				next_char();
-				lexer_token.datatype = type_ulonglong;
-			} else {
-				lexer_token.datatype = type_longlong;
 			}
 		} else if(c == 'u' || c == 'U') {
+			is_unsigned = true;
 			next_char();
 			lexer_token.datatype = type_ulong;
-		} else {
-			lexer_token.datatype = type_int;
 		}
+	}
+
+	if(!is_unsigned) {
+		long long v = lexer_token.v.intvalue;
+		if(!min_long) {
+			if(v >= TARGET_INT_MIN && v <= TARGET_INT_MAX) {
+				lexer_token.datatype = type_int;
+				return;
+			} else if(is_oct_hex && v >= 0 && v <= TARGET_UINT_MAX) {
+				lexer_token.datatype = type_uint;
+				return;
+			}
+		}
+		if(!min_longlong) {
+			if(v >= TARGET_LONG_MIN && v <= TARGET_LONG_MAX) {
+				lexer_token.datatype = type_long;
+				return;
+			} else if(is_oct_hex && v >= 0 && v <= TARGET_ULONG_MAX) {
+				lexer_token.datatype = type_ulong;
+				return;
+			}
+		}
+		unsigned long long uv = (unsigned long long) v;
+		if(is_oct_hex && uv > (unsigned long long) TARGET_LONGLONG_MAX) {
+			lexer_token.datatype = type_ulonglong;
+			return;
+		}
+
+		lexer_token.datatype = type_longlong;
 	} else {
-		lexer_token.datatype = type_int;
+		unsigned long long v = (unsigned long long) lexer_token.v.intvalue;
+		if(!min_long && v <= TARGET_UINT_MAX) {
+			lexer_token.datatype = type_uint;
+			return;
+		}
+		if(!min_longlong && v <= TARGET_ULONG_MAX) {
+			lexer_token.datatype = type_ulong;
+			return;
+		}
+		lexer_token.datatype = type_ulonglong;
 	}
 }
 
@@ -320,11 +373,88 @@ static void parse_floating_suffix(void)
 	}
 }
 
-static inline bool is_hex_digit(int c)
-{
-	return (c >= '0' && c <= '9')
-			|| (c >= 'a' && c <= 'f')
-			|| (c >= 'A' && c <= 'F');
+/**
+ * A replacement for strtoull. Only those parts needed for
+ * our parser are implemented.
+ */
+static unsigned long long parse_int_string(const char *s, const char **endptr, int base) {
+	unsigned long long v = 0;
+
+	switch (base) {
+	case 16:
+		for (;; ++s) {
+			/* check for overrun */
+			if (v >= 0x1000000000000000ULL)
+				break;
+			switch (tolower(*s)) {
+			case '0': v <<= 4; break;
+			case '1': v <<= 4; v |= 0x1; break;
+			case '2': v <<= 4; v |= 0x2; break;
+			case '3': v <<= 4; v |= 0x3; break;
+			case '4': v <<= 4; v |= 0x4; break;
+			case '5': v <<= 4; v |= 0x5; break;
+			case '6': v <<= 4; v |= 0x6; break;
+			case '7': v <<= 4; v |= 0x7; break;
+			case '8': v <<= 4; v |= 0x8; break;
+			case '9': v <<= 4; v |= 0x9; break;
+			case 'a': v <<= 4; v |= 0xa; break;
+			case 'b': v <<= 4; v |= 0xb; break;
+			case 'c': v <<= 4; v |= 0xc; break;
+			case 'd': v <<= 4; v |= 0xd; break;
+			case 'e': v <<= 4; v |= 0xe; break;
+			case 'f': v <<= 4; v |= 0xf; break;
+			default:
+				goto end;
+			}
+		}
+		break;
+	case 8:
+		for (;; ++s) {
+			/* check for overrun */
+			if (v >= 0x2000000000000000ULL)
+				break;
+			switch (tolower(*s)) {
+			case '0': v <<= 3; break;
+			case '1': v <<= 3; v |= 1; break;
+			case '2': v <<= 3; v |= 2; break;
+			case '3': v <<= 3; v |= 3; break;
+			case '4': v <<= 3; v |= 4; break;
+			case '5': v <<= 3; v |= 5; break;
+			case '6': v <<= 3; v |= 6; break;
+			case '7': v <<= 3; v |= 7; break;
+			default:
+				goto end;
+			}
+		}
+		break;
+	case 10:
+		for (;; ++s) {
+			/* check for overrun */
+			if (v > 0x1999999999999999ULL)
+				break;
+			switch (tolower(*s)) {
+			case '0': v *= 10; break;
+			case '1': v *= 10; v += 1; break;
+			case '2': v *= 10; v += 2; break;
+			case '3': v *= 10; v += 3; break;
+			case '4': v *= 10; v += 4; break;
+			case '5': v *= 10; v += 5; break;
+			case '6': v *= 10; v += 6; break;
+			case '7': v *= 10; v += 7; break;
+			case '8': v *= 10; v += 8; break;
+			case '9': v *= 10; v += 9; break;
+			default:
+				goto end;
+			}
+		}
+		break;
+	default:
+		assert(0);
+		break;
+	}
+end:
+	*endptr = s;
+	return v;
 }
 
 static void parse_number_hex(void)
@@ -332,8 +462,8 @@ static void parse_number_hex(void)
 	assert(c == 'x' || c == 'X');
 	next_char();
 
-	while(is_hex_digit(c)) {
-		obstack_1grow(&symbol_obstack, c);
+	while(isxdigit(c)) {
+		obstack_1grow(&symbol_obstack, (char) c);
 		next_char();
 	}
 	obstack_1grow(&symbol_obstack, '\0');
@@ -348,15 +478,15 @@ static void parse_number_hex(void)
 		lexer_token.type = T_ERROR;
 	}
 
-	char *endptr;
+	const char *endptr;
 	lexer_token.type       = T_INTEGER;
-	lexer_token.v.intvalue = strtoull(string, &endptr, 16);
+	lexer_token.v.intvalue = parse_int_string(string, &endptr, 16);
 	if(*endptr != '\0') {
 		parse_error("hex number literal too long");
 	}
 
 	obstack_free(&symbol_obstack, string);
-	parse_integer_suffix();
+	parse_integer_suffix(true);
 }
 
 static inline bool is_octal_digit(int chr)
@@ -367,28 +497,28 @@ static inline bool is_octal_digit(int chr)
 static void parse_number_oct(void)
 {
 	while(is_octal_digit(c)) {
-		obstack_1grow(&symbol_obstack, c);
+		obstack_1grow(&symbol_obstack, (char) c);
 		next_char();
 	}
 	obstack_1grow(&symbol_obstack, '\0');
 	char *string = obstack_finish(&symbol_obstack);
 
-	char *endptr;
+	const char *endptr;
 	lexer_token.type       = T_INTEGER;
-	lexer_token.v.intvalue = strtoull(string, &endptr, 8);
+	lexer_token.v.intvalue = parse_int_string(string, &endptr, 8);
 	if(*endptr != '\0') {
 		parse_error("octal number literal too long");
 	}
 
 	obstack_free(&symbol_obstack, string);
-	parse_integer_suffix();
+	parse_integer_suffix(true);
 }
 
 static void parse_number_dec(void)
 {
 	bool is_float = false;
 	while(isdigit(c)) {
-		obstack_1grow(&symbol_obstack, c);
+		obstack_1grow(&symbol_obstack, (char) c);
 		next_char();
 	}
 
@@ -397,7 +527,7 @@ static void parse_number_dec(void)
 		next_char();
 
 		while(isdigit(c)) {
-			obstack_1grow(&symbol_obstack, c);
+			obstack_1grow(&symbol_obstack, (char) c);
 			next_char();
 		}
 		is_float = true;
@@ -407,12 +537,12 @@ static void parse_number_dec(void)
 		next_char();
 
 		if(c == '-' || c == '+') {
-			obstack_1grow(&symbol_obstack, c);
+			obstack_1grow(&symbol_obstack, (char) c);
 			next_char();
 		}
 
 		while(isdigit(c)) {
-			obstack_1grow(&symbol_obstack, c);
+			obstack_1grow(&symbol_obstack, (char) c);
 			next_char();
 		}
 		is_float = true;
@@ -421,8 +551,8 @@ static void parse_number_dec(void)
 	obstack_1grow(&symbol_obstack, '\0');
 	char *string = obstack_finish(&symbol_obstack);
 
-	char *endptr;
 	if(is_float) {
+		char *endptr;
 		lexer_token.type         = T_FLOATINGPOINT;
 		lexer_token.v.floatvalue = strtold(string, &endptr);
 
@@ -432,14 +562,15 @@ static void parse_number_dec(void)
 
 		parse_floating_suffix();
 	} else {
+		const char *endptr;
 		lexer_token.type       = T_INTEGER;
-		lexer_token.v.intvalue = strtoull(string, &endptr, 10);
+		lexer_token.v.intvalue = parse_int_string(string, &endptr, 10);
 
 		if(*endptr != '\0') {
 			parse_error("invalid number literal");
 		}
 
-		parse_integer_suffix();
+		parse_integer_suffix(false);
 	}
 	obstack_free(&symbol_obstack, string);
 }
@@ -492,7 +623,7 @@ static int parse_octal_sequence(const int first_digit)
 	if (!is_octal_digit(c)) return value;
 	value = 8 * value + c - '0';
 	next_char();
-	return value;
+	return (char_type)value;
 }
 
 static int parse_hex_sequence(void)
@@ -511,7 +642,7 @@ static int parse_hex_sequence(void)
 		next_char();
 	}
 
-	return value;
+	return (char_type)value;
 }
 
 static int parse_escape_sequence(void)
@@ -584,7 +715,7 @@ static void parse_string_literal(void)
 		switch(c) {
 		case '\\':
 			tc = parse_escape_sequence();
-			obstack_1grow(&symbol_obstack, tc);
+			obstack_1grow(&symbol_obstack, (char) tc);
 			break;
 
 		case EOF:
@@ -599,7 +730,7 @@ static void parse_string_literal(void)
 			goto end_of_string;
 
 		default:
-			obstack_1grow(&symbol_obstack, c);
+			obstack_1grow(&symbol_obstack, (char) c);
 			next_char();
 			break;
 		}
@@ -623,6 +754,66 @@ end_of_string:
 	lexer_token.v.string = result;
 }
 
+static void parse_wide_string_literal(void)
+{
+	const unsigned start_linenr = lexer_token.source_position.linenr;
+
+	assert(c == '"');
+	next_char();
+
+	while(1) {
+		switch(c) {
+			case '\\': {
+				wchar_rep_t tc = parse_escape_sequence();
+				obstack_grow(&symbol_obstack, &tc, sizeof(tc));
+				break;
+			}
+
+			case EOF:
+				error_prefix_at(lexer_token.source_position.input_name,
+				                start_linenr);
+				fprintf(stderr, "string has no end\n");
+				lexer_token.type = T_ERROR;
+				return;
+
+			case '"':
+				next_char();
+				goto end_of_string;
+
+			default: {
+				wchar_rep_t tc = c;
+				obstack_grow(&symbol_obstack, &tc, sizeof(tc));
+				next_char();
+				break;
+			}
+		}
+	}
+
+end_of_string:;
+
+	/* TODO: concatenate multiple strings separated by whitespace... */
+
+	/* add finishing 0 to the string */
+	wchar_rep_t nul = L'\0';
+	obstack_grow(&symbol_obstack, &nul, sizeof(nul));
+	const size_t             size   = (size_t)obstack_object_size(&symbol_obstack) / sizeof(wchar_rep_t);
+	const wchar_rep_t *const string = obstack_finish(&symbol_obstack);
+
+#if 0 /* TODO hash */
+	/* check if there is already a copy of the string */
+	const wchar_rep_t *const result = strset_insert(&stringset, string);
+	if(result != string) {
+		obstack_free(&symbol_obstack, string);
+	}
+#else
+	const wchar_rep_t *const result = string;
+#endif
+
+	lexer_token.type                = T_WIDE_STRING_LITERAL;
+	lexer_token.v.wide_string.begin = result;
+	lexer_token.v.wide_string.size  = size;
+}
+
 static void parse_character_constant(void)
 {
 	eat('\'');
@@ -763,7 +954,7 @@ static void parse_line_directive(void)
 	if(pp_token.type != T_INTEGER) {
 		parse_error("expected integer");
 	} else {
-		lexer_token.source_position.linenr = pp_token.v.intvalue - 1;
+		lexer_token.source_position.linenr = (unsigned int)(pp_token.v.intvalue - 1);
 		next_pp_token();
 	}
 	if(pp_token.type == T_STRING_LITERAL) {
@@ -872,8 +1063,7 @@ void lexer_next_preprocessing_token(void)
 			/* might be a wide string ( L"string" ) */
 			if(c == '"' && (lexer_token.type == T_IDENTIFIER &&
 			   lexer_token.v.symbol == symbol_L)) {
-			   	parse_string_literal();
-			   	return;
+				parse_wide_string_literal();
 			}
 			return;
 
@@ -1046,14 +1236,14 @@ void init_lexer(void)
 {
 	strset_init(&stringset);
 
-	type_int       = make_atomic_type(ATOMIC_TYPE_INT, TYPE_QUALIFIER_CONST);
-	type_uint      = make_atomic_type(ATOMIC_TYPE_UINT, TYPE_QUALIFIER_CONST);
-	type_long      = make_atomic_type(ATOMIC_TYPE_LONG, TYPE_QUALIFIER_CONST);
-	type_ulong     = make_atomic_type(ATOMIC_TYPE_ULONG, TYPE_QUALIFIER_CONST);
+	type_int       = make_atomic_type(ATOMIC_TYPE_INT, TYPE_QUALIFIER_NONE);
+	type_uint      = make_atomic_type(ATOMIC_TYPE_UINT, TYPE_QUALIFIER_NONE);
+	type_long      = make_atomic_type(ATOMIC_TYPE_LONG, TYPE_QUALIFIER_NONE);
+	type_ulong     = make_atomic_type(ATOMIC_TYPE_ULONG, TYPE_QUALIFIER_NONE);
 	type_longlong  = make_atomic_type(ATOMIC_TYPE_LONGLONG,
-	                                  TYPE_QUALIFIER_CONST);
+	                                  TYPE_QUALIFIER_NONE);
 	type_ulonglong = make_atomic_type(ATOMIC_TYPE_ULONGLONG,
-	                                  TYPE_QUALIFIER_CONST);
+	                                  TYPE_QUALIFIER_NONE);
 
 	type_float      = make_atomic_type(ATOMIC_TYPE_FLOAT, TYPE_QUALIFIER_CONST);
 	type_double     = make_atomic_type(ATOMIC_TYPE_DOUBLE,
@@ -1083,7 +1273,7 @@ void exit_lexer(void)
 static __attribute__((unused))
 void dbg_pos(const source_position_t source_position)
 {
-	fprintf(stdout, "%s:%d\n", source_position.input_name,
+	fprintf(stdout, "%s:%u\n", source_position.input_name,
 	        source_position.linenr);
 	fflush(stdout);
 }