Merge pull request #12 from UB-CSE443/Sprint1-TokenLocationLogic-NoTask

Sprint1 token location logic no task

Makefile

@@ -1,7 +1,7 @@
 CC := gcc
 FLEX := flex
 LEX := lexicalStructure.lex
-EXE := lexicalStructure
+EXE := runner
 CFLAGS := -std=c99 -Wall
 CPPFLAGS :=
 
@@ -16,7 +16,10 @@ flex.o: lex.yy.c typedefs.h
 
 lex.yy.c: lexicalStructure.lex
 	$(FLEX) -o lex.yy.c $(LEX)
-  
+
+debug: CFLAGS += -DDEBUG=1
+debug: clean runner
+
 test:
 	./$(EXE) ./tests/test_comments.alpha
 	./$(EXE) ./tests/test_generalTokenTest.alpha
@@ -25,9 +28,16 @@ test:
 	./$(EXE) ./tests/test_otherpunc.alpha
 	./$(EXE) ./tests/test_simpleIntTest.alpha
 	./$(EXE) ./tests/test_simpleLiterals.alpha
-
+	./$(EXE) -tok ./tests/test_comments.alpha
+	./$(EXE) -tok ./tests/test_generalTokenTest.alpha
+	./$(EXE) -tok ./tests/test_keywords.alpha
+	./$(EXE) -tok ./tests/test_operators.alpha
+	./$(EXE) -tok ./tests/test_otherpunc.alpha
+	./$(EXE) -tok ./tests/test_simpleIntTest.alpha
+	./$(EXE) -tok ./tests/test_simpleLiterals.alpha
 clean:
 	rm -f *.o
 	rm -f lex.yy.c
 	rm -f $(EXE)
 	rm -f flex.h
+	rm -f *.tok

Makefile.save

@@ -0,0 +1,43 @@
+CC := gcc
+FLEX := flex
+LEX := lexicalStructure.lex
+EXE := runner
+CFLAGS := -std=c99 -Wall
+CPPFLAGS :=
+
+runner: flex.o runner.o
+	$(CC) -o runner runner.o flex.o
+
+debug: CFLAGS += -DDEBUG=true
+debug: clean runner
+
+runner.o: runner.c runner.h flex.h
+	$(CC) $(CFLAGS) -o runner.o -c runner.c
+
+flex.o: lex.yy.c typedefs.h
+	$(CC) $(CFLAGS) -o flex.o -c lex.yy.c
+
+lex.yy.c: lexicalStructure.lex
+	$(FLEX) -o lex.yy.c $(LEX)
+  
+test:
+	./$(EXE) ./tests/test_comments.alpha
+	./$(EXE) ./tests/test_generalTokenTest.alpha
+	./$(EXE) ./tests/test_keywords.alpha
+	./$(EXE) ./tests/test_operators.alpha
+	./$(EXE) ./tests/test_otherpunc.alpha
+	./$(EXE) ./tests/test_simpleIntTest.alpha
+	./$(EXE) ./tests/test_simpleLiterals.alpha
+	./$(EXE) -tok ./tests/test_comments.alpha
+	./$(EXE) -tok ./tests/test_generalTokenTest.alpha
+	./$(EXE) -tok ./tests/test_keywords.alpha
+	./$(EXE) -tok ./tests/test_operators.alpha
+	./$(EXE) -tok ./tests/test_otherpunc.alpha
+	./$(EXE) -tok ./tests/test_simpleIntTest.alpha
+	./$(EXE) -tok ./tests/test_simpleLiterals.alpha
+clean:
+	rm -f *.o
+	rm -f lex.yy.c
+	rm -f $(EXE)
+	rm -f flex.h
+	rm -f *.tok

lexicalStructure.lex

@@ -8,13 +8,15 @@
 	#include <stdbool.h>
 	#include "typedefs.h"
 	int line_number = 1, column_number = 1;
-	bool DEBUG = true;
+	#ifndef DEBUG
+		#define DEBUG 0
+	#endif
 %}
 
 COM ([^*]|\*+[^)*])*
 ID [A-Za-z_][0-9A-Za-z_]*
 DIGIT [0-9]
-CHAR \\n|\\t|\\'|[^'\n\t\\]
+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 */

runner.c

@@ -26,8 +26,25 @@ int main(int argc, char *argv[]) {
         if (arg == TOK_ARG) {
             fprintf(output, "%d %d %3d \"%s\"\n", line_number, column_number, token, yytext);
         }
+	if(token == COMMENT){
+		for (int i = 0; i < yyleng; i++) {
+			if(yytext[i] == '\n'){
+				line_number++;
+				column_number = 0;
+			}
+			column_number++;
+		}
+		continue;
+	}
+	column_number += yyleng;
     }
 
+    if (yyin != NULL) {
+        fclose(yyin);
+    }
+    if (output != NULL && arg == TOK_ARG) {
+        fclose(output);
+    }
     return 0;
 }
 
@@ -40,21 +57,20 @@ char *is_tok(int argc, char *argv[]) {
             return INVALID_ARG;
         }
 
-        int ignore_path = 0;
-        int count_since_slash = 0;
-        for (int i = 0; i < file_len; i++) {
-            count_since_slash++;
-            if (input_prog[i] == '/') {
-                ignore_path += count_since_slash;
-                count_since_slash = 0;
-            }
+        const char *basename = input_prog;
+        const char *slash = strrchr(input_prog, '/');
+        if (slash != NULL) {
+            basename = slash + 1;
         }
+
+        // Calculate lengths
+        int basename_len = strlen(basename);
+        char* FILE_tok = calloc(basename_len - ALPHA_OFFSET + TOK_LEN + 1, sizeof(char));
+        
+        // Copy filename and add .tok extension
+        strncpy(FILE_tok, basename, basename_len - ALPHA_OFFSET);
+        strcat(FILE_tok, ".tok");
         
-        file_len -= ignore_path;
-        input_prog += ignore_path;
-        char* FILE_tok = calloc(sizeof(char), file_len - ALPHA_OFFSET + TOK_LEN);
-        strncpy(FILE_tok, input_prog, file_len - ALPHA_OFFSET); //copy name of prog before .alpha
-        strcpy(FILE_tok + sizeof(char) * (file_len - ALPHA_OFFSET), ".tok"); //add .tok to end of file name
         return FILE_tok;
     }
     return CHECK_OTHER;

runner.h

@@ -14,6 +14,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include "flex.h"
+#include "typedefs.h"
 
 extern int line_number, column_number;
 extern char *yytext;
@@ -22,4 +23,4 @@ int arg;
 
 int main(int argc, char* argv[]);
 char *is_tok(int argc, char* argv[]);
-int is_alpha_file(char *file, int file_len);
+int is_alpha_file(char *file, int file_len);

tests/test_simpleIntTest.alpha

@@ -1,4 +1,5 @@
 45
 123
 8392
-
+40 40 
+ 200 50 21783

tests/test_simpleLiterals.alpha

@@ -4,7 +4,7 @@
 12893 "this is not a string (*one valid token before this*)
 (* spacey comment here     
 over multiple lines
-will it work? *)
+will it work? *) false
 "
 '''
 '\'