mah lexerHEADmaster

1 files changed, 123 insertions, 0 deletions

diff --git a/lexer.l b/lexer.l
new file mode 100644
index 0000000..dc499e6
--- /dev/null
+++ b/lexer.l
@@ -0,0 +1,123 @@
+%{
+#include <stdlib.h>
+#include "token.h"
+
+extern int yycolumn, yylength, yyline, yylval;
+
+extern char *stringTable;
+extern int stringTableIndices[];
+extern int nStringsStored;
+extern int nCharsStored;
+
+/* functions */
+void ReportError(char *, int, int);
+int getStrTableIndex(char *, int, int);
+void RemoveQuotes(char *);
+
+%}
+
+/* regular definitions */
+
+id            ([a-zA-Z])+
+delim         [ \t]
+ws            {delim}+
+newline       [\n]
+
+%%
+{newline}                                                           {yyline++; yycolumn=0;}
+{ws}                                                                {yycolumn+=yyleng;}
+"/*"([^*]|[ \t\n]|(\*+([^*/]|[ \t\n])))*                            {ReportError("comment doesn't terminate",yyline,yycolumn);}
+"/*"([^*]|[ \t\n]|(\*+([^*/]|[ \t\n])))*"*"+"/"                     {yycolumn+=yyleng;};
+
+"&&"                                                                {yycolumn+=yyleng; return (ANDnum);}
+":="                                                                {yycolumn+=yyleng; return (ASSGNnum);}
+[dD][eE][cC][lL][aA][rR][aA][tT][iI][oO][nN][sS]                    {yycolumn+=yyleng; return (DECLARATIONSnum);}
+"."                                                                 {yycolumn+=yyleng; return (DOTnum);}
+[eE][nN][dD][dD][eE][cC][lL][aA][rR][aA][tT][iI][oO][nN][sS]        {yycolumn+=yyleng; return (ENDDECLARATIONSnum);}
+"="                                                                 {yycolumn+=yyleng; return (EQUALnum);}
+">"                               {yycolumn+=yyleng; return (GTnum);}
+[iI][nN][tT]                      {yycolumn+=yyleng; return (INTnum);}
+"["                               {yycolumn+=yyleng; return (LBRACnum);}
+"("                               {yycolumn+=yyleng; return (LPARENnum);}
+[mM][eE][tT][hH][oO][dD]          {yycolumn+=yyleng; return (METHODnum);}
+"!="                              {yycolumn+=yyleng; return (NEnum);}
+"||"                              {yycolumn+=yyleng; return (ORnum);}
+[pP][rR][oO][gG][rR][aA][mM]      {yycolumn+=yyleng; return (PROGRAMnum);}
+"]"                               {yycolumn+=yyleng; return (RBRACnum);}
+")"                               {yycolumn+=yyleng; return (RPARENnum);}
+";"                               {yycolumn+=yyleng; return (SEMInum);}
+[vV][aA][lL]                      {yycolumn+=yyleng; return (VALnum);}
+[wW][hH][iI][lL][eE]              {yycolumn+=yyleng; return (WHILEnum);}
+[cC][lL][aA][sS][sS]              {yycolumn+=yyleng; return (CLASSnum);}
+","                               {yycolumn+=yyleng; return (COMMAnum);}
+"/"                               {yycolumn+=yyleng; return (DIVIDEnum);}
+[eE][lL][sS][eE]                  {yycolumn+=yyleng; return (ELSEnum);}
+"=="                              {yycolumn+=yyleng; return (EQnum);}
+">="                              {yycolumn+=yyleng; return (GEnum);}
+[iI][fF]                          {yycolumn+=yyleng; return (IFnum);}
+"{"                               {yycolumn+=yyleng; return (LBRACEnum);}
+"<="                              {yycolumn+=yyleng; return (LEnum);}
+"<"                               {yycolumn+=yyleng; return (LTnum);}
+"-"                               {yycolumn+=yyleng; return (MINUSnum);}
+"!"                               {yycolumn+=yyleng; return (NOTnum);}
+"+"                               {yycolumn+=yyleng; return (PLUSnum);}
+"}"                               {yycolumn+=yyleng; return (RBRACEnum);}
+[rR][eE][tT][uU][rR][nN]          {yycolumn+=yyleng; return (RETURNnum);}
+"*"                               {yycolumn+=yyleng; return (TIMESnum);}
+[vV][oO][iI][dD]                  {yycolumn+=yyleng; return (VOIDnum);}
+
+'(.|[ \t])*'                      {RemoveQuotes(yytext); yylval = getStrTableIndex(yytext,yyline,yycolumn); yycolumn+=yyleng; return (SCONSTnum);}
+'(.|[ \t\n])*'                    {ReportError("string constant crosses line boundary",yyline,yycolumn); yyline++; yycolumn=0;}
+[1-9][0-9]*                       {yylval = atoi(yytext); yycolumn+=yyleng; return (ICONSTnum);}
+[0-9][a-zA-Z0-9]*                 {ReportError("identifier starting with a digit",yyline,yycolumn); yycolumn+=yyleng;}
+[a-zA-Z][a-zA-Z0-9]*              {yylval = getStrTableIndex(yytext,yyline,yycolumn); yycolumn+=yyleng; return (IDnum);}
+
+
+%%
+
+void ReportError(char *message,int line, int column) {
+    fprintf( stderr, "\nError at line %d, column %d (first char): %s\n\n", line, column, message);
+    return;
+}
+
+int getStrTableIndex(char *string, int line, int column) {
+    int i;
+    
+    /* use index array to search table for string*/
+    for (i=0; i < nStringsStored; i++) {
+        /* return index if found */
+        if (strcmp(string, stringTable + stringTableIndices[i]) == 0) {
+            return stringTableIndices[i];
+        }
+    }
+    
+    /* error if string table is full */
+    if (nStringsStored == LIMIT1) {
+        ReportError("can't add string to string table: max character count.",line, column);
+        return -1;
+    }
+    
+    /* error if index table is full */
+    if (nStringsStored == LIMIT2) {
+        ReportError("can't add string to string table: max number of unique strings",line, column);
+        return -1;
+    }
+    
+    /* add string to table if not full and return index */
+    stringTableIndices[i] = nCharsStored;
+    memcpy(stringTable + stringTableIndices[i],string,strlen(string)+1);
+    nCharsStored+=strlen(string)+1;
+    nStringsStored++;
+    
+    return stringTableIndices[i];
+}
+
+/* this lexer detects string constants by their quotes, so yytext includes them
+ * even though they aren't really part of the string 
+ * so this removes them */
+void RemoveQuotes(char *string) {
+    memmove(string, string+1, strlen(string));
+    string[strlen(string) - 1] = '\0';
+    
+    return;
+}