lexical analysis almost done

1 files changed, 95 insertions, 0 deletions

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; }
+%%