lexical analysis almost done

3 files changed, 149 insertions, 0 deletions

diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..6deae78
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,11 @@
+all: cibic
+
+run: 
+	./cibic
+
+cibic: lex.yy.c cibic.tab.c
+	gcc -o cibic lex.yy.c cibic.tab.c
+lex.yy.c: cibic.l
+	flex cibic.l
+cibic.tab.c: cibic.y
+	bison -d cibic.y
diff --git a/cibic.l b/cibic.l
new file mode 100644
index 0000000..37de015
--- /dev/null
+++ b/cibic.l
@@ -0,0 +1,95 @@
+%{
+#include "cibic.tab.h"
+%}
+
+letter [a-zA-Z_$]
+digit [0-9]
+%s IN_BLOCK_COMMENT IN_INLINE_COMMENT
+%%
+
+<INITIAL>{
+"/*"    BEGIN(IN_BLOCK_COMMENT);
+}
+<IN_BLOCK_COMMENT>{
+"*/"    BEGIN(INITIAL);
+[^*\n]+   // eat comment in chunks
+"*"       // eat the lone star
+\n
+}
+
+<INITIAL>{
+"//"    BEGIN(IN_INLINE_COMMENT);
+}
+<IN_INLINE_COMMENT>{
+\n      BEGIN(INITIAL);
+[^\n]+ 
+}
+
+"void"                                      { return KW_VOID; }
+"char"                                      { return KW_CHAR; }
+"int"                                       { return KW_INT; }
+"struct"                                    { return KW_STRUCT; }
+"union"                                     { return KW_UNION; }
+"if"                                        { return KW_IF; }
+"else"                                      { return KW_ELSE; }
+"while"                                     { return KW_WHILE; }
+"for"                                       { return KW_FOR; }
+"continue"                                  { return KW_CONT; }
+"break"                                     { return KW_BREAK; }
+"return"                                    { return KW_RET; }
+"sizeof"                                    { return KW_SIZEOF; }
+
+{letter}({letter}|{digit})*                 { 
+    yylval.strval = strdup(yytext); 
+    return IDENTIFIER; 
+}
+
+({digit}+)|(0[xX][0-9a-fA-F]+)              { 
+    if (*yytext == '0')
+    {
+        if (*(yytext + 1) == 'x' || *(yytext + 1) == 'X')
+            sscanf(yytext, "%x", &yylval.intval);
+        else    // FIXME: error report if it is not a valid octal
+            sscanf(yytext, "%o", &yylval.intval);
+    }
+    else yylval.intval = atoi(yytext);
+    return INT_CONST; 
+}
+
+'{letter}'                                  { 
+    yylval.intval = *(yytext);
+    return CHAR_CONST; 
+}
+
+\"[^\n\"]*\"                                {
+    yylval.strval = strndup(yytext + 1, strlen(yytext) - 2);
+    return STR_CONST; 
+}
+
+"||"                                        { return OPT_OR; }
+"&&"                                        { return OPT_AND; }
+"=="                                        { return OPT_EQ; }
+"!="                                        { return OPT_NE; }
+"<="                                        { return OPT_LE; }
+"<<"                                        { return OPT_SHL; }
+">>"                                        { return OPT_SHR; }
+"++"                                        { return OPT_INC; }
+"--"                                        { return OPT_DEC; }
+"->"                                        { return OPT_PTR; }
+
+"*="                                        { return ASS_MUL; }
+"/="                                        { return ASS_DIV; }
+"%="                                        { return ASS_MOD; }
+"+="                                        { return ASS_ADD; }
+"-="                                        { return ASS_SUB; }
+"<<="                                       { return ASS_SHL; }
+">>="                                       { return ASS_SHR; }
+"&="                                        { return ASS_AND; }
+"^="                                        { return ASS_XOR; }
+"|="                                        { return ASS_OR; }
+
+[();,={}\[\]*|\^&<>+\-*//%~!.]              { return *yytext; }
+
+[ \t\n\r]                                   /* skip whitespaces */
+.                                           { return UNKNOWN; }
+%%
diff --git a/cibic.y b/cibic.y
new file mode 100644
index 0000000..92f0cfe
--- /dev/null
+++ b/cibic.y
@@ -0,0 +1,43 @@
+%{
+#include <stdio.h>
+%}
+%token IDENTIFIER INT_CONST CHAR_CONST STR_CONST
+%token KW_VOID KW_CHAR KW_INT KW_STRUCT KW_UNION KW_IF KW_ELSE KW_WHILE
+%token KW_FOR KW_CONT KW_BREAK KW_RET KW_SIZEOF
+%token OPT_OR OPT_AND OPT_EQ OPT_NE OPT_LE OPT_GE OPT_SHL OPT_SHR OPT_INC OPT_DEC OPT_PTR
+%token ASS_MUL ASS_DIV ASS_MOD ASS_ADD ASS_SUB ASS_SHL ASS_SHR ASS_AND ASS_XOR ASS_OR 
+%token UNKNOWN
+%union {
+    int intval;
+    char *strval;
+}
+%%
+program
+    : body { 
+        printf("\n")}
+body
+    : IDENTIFIER
+%%
+int yywrap() { 
+    return 1; 
+}
+
+int yyerror(char *s) {
+}
+
+extern FILE *yyin;
+int main() {
+    int ret;
+    //yyin = fopen("in", "r");
+    while (ret = yylex()) 
+    {
+        printf("%d\n", ret);    
+        if (ret == IDENTIFIER)
+            printf("id: %s\n", yylval.strval);
+        else if (ret == INT_CONST)
+            printf("int: %d\n", yylval.intval);
+        else if (ret == STR_CONST)
+            printf("str: %s\n", yylval.strval);
+    }
+    return 0;
+}