Merge pull request #10 from UB-CSE443/Sprint1-TokenizeID_MainFunction-FE-t#04

Merge #04

lexicalStructure.lex

@@ -3,56 +3,42 @@
 /* definitions */
 
 %option noyywrap
-
 %{
 #include "typedefs.h"
-%}
 int line_number = 1, column_number = 1;
+%}
 
+COM ([^*]|\*+[^)*])*
+ID [A-Za-z_][0-9A-Za-z_]*
 DIGIT [0-9]
 CHAR \\n|\\t|\\'|[^'\n\t\\]
 	/* char can be a newline, tab, an escaped quote, or anything but a single quote, an actual line break, an actual tab, or a backslash by itself (to prevent confusion from escaped quote */
 SCHAR \\n|\\t|\\\"|[^\"\n\\]  
 	/* similar to above, a string Char (SCHAR) is the same as a CHAR except we cannot have double quotes instead of single quotes. Double quotes need to be escaped in Flex unlike single quotes based on documentation */
-STARTCOM \(\*
-ENDCOM \*\)
-COMMENTCHAR [^\*]|\*[^\)]
-	/*Making the contents of a comment anything that is either not a * or not a * followed by ) to terminate comments at the first ENDCOM */
+
 %%
 
+"integer"       {return T_INTEGER;}
+"address"       {return T_ADDRESS;}
+"Boolean"       {return T_BOOLEAN;}
+"character"     {return T_CHARACTER;}
 
-\n      line_number++ column_number = 1;
-.       column_number++;
+{DIGIT}+        {return C_INTEGER;}
+"null"          {return C_NULL;}
 
-"integer"       {return T_INTEGER}
-"address"       {return T_ADDRESS}
-"Boolean"       {return T_BOOLEAN}
-"character"     {return T_CHARACTER}
+"while"         {return WHILE;}
+"if"            {return IF;}
+"then"          {return THEN;}
+"else"          {return ELSE;}
+"type"          {return TYPE;}
+"function"      {return FUNCTION;}
+"return"        {return RETURN;}
+"external"      {return EXTERNAL;}
+"as"            {return AS;}
 
-
-	/* rules */
-{DIGIT}+ {printf( "C_INTEGER: %s (%d)\n", yytext, atoi( yytext ) );}
-
-"null" {printf( "C_NULL: %s (%d)\n", yytext, atoi( yytext ) );}
-
-"while"         {return WHILE}
-"if"            {return IF}
-"then"          {return THEN}
-"else"          {return ELSE}
-"type"          {return TYPE}
-"function"      {return FUNCTION}
-"return"        {return RETURN}
-"external"      {return EXTERNAL}
-"as"            {return AS}
-
-'{CHAR}' {printf( "C_CHARACTER: %s (%d)\n", yytext, atoi( yytext ) );} /*using double \ per documentation to show escaped chars*/
-
-"true" {printf( "C_TRUE: %s (%d)\n", yytext, atoi( yytext ) );}
-
-
-"false" {printf( "C_FALSE: %s (%d)\n", yytext, atoi( yytext ) );}
-
-	/* OPERATORS */
+'{CHAR}'        {return C_CHARACTER;}
+"true"          {return C_TRUE;}
+"false"         {return C_FALSE;}
 
 "+"		        {return ADD;}
 "-"		        {return SUB_OR_NEG;}
@@ -66,15 +52,21 @@ COMMENTCHAR [^\*]|\*[^\)]
 "&"		        {return AND;}
 "|"		        {return OR;}
 "."		        {return DOT;}
+
+";"             {return SEMI_COLON;}
+":"             {return COLON;}
+","             {return COMMA;}
+"->"            {return ARROW;}
+
 "reserve"	    {return RESERVE;}
 "release" 	    {return RELEASE;}
 
+\"{SCHAR}*\"    {return C_STRING;}
+"(*"{COM}"*)"   {return COMMENT;}
 
-\"{SCHAR}*\" {printf( "C_STRING: %s (%d)\n", yytext, atoi( yytext ) );}
-
-\(\*{COMMENTCHAR}*\*\) {printf( "COMMENT: %s (%d)\n", yytext, atoi( yytext ) );}
-
-.|\n  
+{ID}            {return ID;}
 
+\n              {line_number++; column_number = 1;}
+.               {column_number++;}
 
 %%

tests/test_comments.alpha

@@ -0,0 +1,9 @@
+(* hello *)
+(* hello *)
+(* I'd think this is a legal "string" that contains several \n \t 
+escaped characters, isn't it? *)
+(* \ *)
+(* *)
+(*{COMMENT}+ *)
+(* * *)
+(* (hello) *)

tests/test_otherpunc.alpha

@@ -0,0 +1,6 @@
+;
+:
+,
+->
+->>
+-->

tests/test_types.alpha

@@ -1,10 +0,0 @@
-integer
-Integer
-address
-Address
-Boolean
-boolean
-character
-Character
-string
-String

tests/test_variables.alpha

@@ -0,0 +1,10 @@
+valid1
+Valid2
+_valid3
+_valid_name_4
+VALID
+0Invalid
+1invalid
+"invalid
+invalid=
+String