Merge pull request #2 from Dayof/dev

Second parser version before deadline

README.md

@@ -1,18 +1,18 @@
-# CPPython
+# CPPython - Lexer/Parser
 
 ## Requirements
 
 - Flex 2.6.4
+- Bison 3.5.1
 - GCC 9.3.0
+- Make 4.2.1
 
 ## Usage
 
 ```bash
-$ cd src
-$ chmod +x build.sh
-$ ./build.sh
-$ ./cppython tests/valid_1.ppy # for valid test
-$ ./cppython tests/incorrect_1.ppy # for invalid test
+$ make
+$ ./cppython tests/parser/valid_1.ppy # for valid test
+$ ./cppython tests/parser/invalid_1.ppy # for invalid test
 ```
 
 ## Author

src/Makefile

@@ -1,15 +1,15 @@
-FILES = parser.c lexer.c ast.c sym_tab.c main.c 
+FILES = parser/parser.c lexer/lexer.c core/ast.c core/sym_tab.c core/main.c 
 CFLAGS = -g -Wall -pedantic -x c
 CC = gcc
 
-cppython: $(FILES) ast.h sym_tab.h
- $(CC) $(CFLAGS) $(FILES) -o cppython -lfl
+cppython: $(FILES) core/ast.h core/sym_tab.h
+ $(CC) $(CFLAGS) $(FILES) -I core -I lexer -I parser -o cppython -lfl
 
-lexer.c: cppython.lex
- flex cppython.lex
+lexer/lexer.c: lexer/cppython.lex
+  flex lexer/cppython.lex
 
-parser.c: cppython.y
- bison -d -v cppython.y
+parser/parser.c: parser/cppython.y
+  bison -d -v parser/cppython.y
 
 clean:
- rm -f *.o *~ lexer.c lexer.h parser.c parser.h parser.output cppython
+ rm -f *.o *~ lexer/lexer.c lexer/lexer.h parser/parser.c parser/parser.h parser/parser.output cppython

src/README.md

@@ -1,20 +1,6 @@
-# CPPython - Lexer
+# CPPython - Lexer/Parser
 
-## Requirements
-
-- Flex 2.6.4
-- Bison 3.5.1
-- GCC 9.3.0
-
-## Usage
-
-```bash
-$ make
-$ ./cppython tests/parser/valid_1.ppy # for valid test
-$ ./cppython tests/parser/invalid_1.ppy # for invalid test
-```
-
-## Output examples
+## Examples
 
 - Valid input 1:
  - Commands: ./cppython tests/parser/valid_1.ppy
@@ -58,8 +44,3 @@ $ ./cppython tests/parser/invalid_1.ppy # for invalid test
 
  LexerError: line 1, column 2, token '^' is not recognized
  ```
-
-## Author
-
-Name: Dayanne Fernandes da Cunha 
-University ID: 130107191

src/core/ast.c

@@ -0,0 +1,75 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include "ast.h"
+
+ast_node* create_bin_expr(char *operator, ast_node* left, ast_node* right) {
+ if (PARSER_VERBOSE) printf("\nCreating binary expression node: .%d. .%s. .%d.\n",
+ left->op.integer_expr, operator, right->op.integer_expr);
+ ast_node* expr = (ast_node*) malloc(sizeof(ast_node));
+ expr->tag = BINARY_TYPE;
+ expr->op.binary_expr.operator = operator;
+ expr->op.binary_expr.left = left;
+ expr->op.binary_expr.right = right;
+ return expr;
+}
+
+ast_node* create_int_expr(int value) {
+ if (PARSER_VERBOSE) printf("\n\nCreating integer expression node: %d\n", value);
+ ast_node* expr = (ast_node*) malloc(sizeof(ast_node));
+ expr->tag = INTEGER_TYPE;
+ expr->op.integer_expr = value;
+ return expr;
+}
+
+void create_ast(ast_node* expression) {
+ if (PARSER_VERBOSE) printf("\nCreating AST.\n");
+ root = expression;
+ return;
+}
+
+void create_empy_ast() {
+ if (PARSER_VERBOSE) printf("\nCreating empty AST.\n");
+ root = NULL;
+ return;
+}
+
+ast_node* print_exp(ast_node* node) {
+ if (PARSER_VERBOSE) {
+ if (node == NULL) {
+ printf("Empty expression.\n");
+ } else {
+ printf("TAG: %d\n", node->tag);
+ }
+ }
+ return node;
+}
+
+void print_ast(ast_node* node) {
+ if (node == NULL) {
+ printf("Empty AST.\n");
+ return;
+ }
+
+ if (PARSER_VERBOSE) printf("TAG: %d\n", node->tag);
+
+ for (int i=0; i < AST_LVL; ++i) printf(" ");
+
+ // terminal leaf
+ if (node->tag == INTEGER_TYPE) {
+ printf("INT: %d\n", node->op.integer_expr);
+ return;
+ // terminal leaf
+ } else if (node->tag == VAR_TYPE) {
+ printf("VAR: %s\n", node->op.variable_expr);
+ return;
+ // non terminal node
+ } else if (node->tag == BINARY_TYPE) {
+ printf("OP: %s\n", node->op.binary_expr.operator);
+ AST_LVL += 1;
+ print_ast(node->op.binary_expr.right);
+ print_ast(node->op.binary_expr.left);
+ } else {
+ printf("Print AST unknown error.\n");
+ return;
+ }
+}

src/ast.h → src/core/ast.h

@@ -1,28 +1,35 @@
 #ifndef __AST_H__
 #define __AST_H__
 
+int LEX_VERBOSE, PARSER_VERBOSE, MAIN_VERBOSE, AST_LVL;
 int line, column;
 
+enum TAG {
+ INTEGER_TYPE=0, 
+ VAR_TYPE, 
+ BINARY_TYPE
+};
+
 typedef struct exp {
+ int tag;
  union { 
  int integer_expr;
- char *string_expr;
  char *variable_expr;
  struct { 
  char *operator;
  struct exp* left;
  struct exp* right;
  } binary_expr;
-  } op;
+ } op;
 } ast_node;
 
-typedef struct expr_list { 
- ast_node* head;
- struct expr_list* next;
-} ast;
+ast_node* root;
+
+void print_ast(ast_node* node);
+ast_node* print_exp(ast_node* node);
 
+void create_empy_ast();
 void create_ast(ast_node* expression);
-ast_node* show(ast_node* expression);
 ast_node* create_int_expr(int value);
 ast_node* create_bin_expr(char *operator, ast_node* left, ast_node* right);
 

src/main.c → src/core/main.c

@@ -5,18 +5,22 @@
 
 int main (int argc, char *argv[]) {
  printf("Welcome to CPPython interpreter:\n");
+ MAIN_VERBOSE = LEX_VERBOSE = PARSER_VERBOSE = AST_LVL = 0;
+ root = NULL;
 
  // init lexer and parser
  printf("Lexer/parser:\n");
  line = column = 1;
  yyin = fopen(argv[1], "r");
- printf("\nline %d. ", line);
+ if (MAIN_VERBOSE) printf("\nline %d. ", line);
  do {
  yyparse();
  } while (!feof(yyin));
  fclose(yyin);
- printf("\n");
- printf("Lexer and parser finished.\n\n");
+ printf("\nLexer and parser finished.\n\n");
+
+ printf("Abstract Syntax Tree:\n");
+ print_ast(root);
 
  return 0;
 }

src/cppython.lex → src/lexer/cppython.lex

@@ -6,7 +6,7 @@
  #include "ast.h"
  #include "parser.h"
 
- enum TOKENS{
+ enum TOKENS {
  ERROR_TOK=1,
  NEWLINE_TOK,
  WHITESPACE_TOK,
@@ -26,7 +26,7 @@
  };
 %}
 
-%option outfile="lexer.c" header-file="lexer.h"
+%option outfile="lexer/lexer.c" header-file="lexer/lexer.h"
 %option nounput
 %option noinput
 
@@ -96,33 +96,33 @@ NUMBER ({INTEGER}|{FLOAT})
 void handle_token(int token) {
  switch (token) {
  case INTEGER_TOK:
- printf("Token: <integer, '%s'>", yytext);
+ if (LEX_VERBOSE) printf("Token: <integer, '%s'>", yytext);
  yylval.value = atoi(yytext); 
  break;
  case SUB_TOK:
- printf("Token: <sub, '%s'>", yytext);
+ if (LEX_VERBOSE) printf("Token: <sub, '%s'>", yytext);
  yylval.op = yytext;
  break;
  case ADD_TOK:
- printf("Token: <add, '%s'>", yytext);
+ if (LEX_VERBOSE) printf("Token: <add, '%s'>", yytext);
  yylval.op = yytext;
  break;
  case MULT_TOK:
- printf("Token: <mult, '%s'>", yytext);
+ if (LEX_VERBOSE) printf("Token: <mult, '%s'>", yytext);
  yylval.op = yytext;
  break;
  case DIV_TOK:
- printf("Token: <div, '%s'>", yytext);
+ if (LEX_VERBOSE) printf("Token: <div, '%s'>", yytext);
  yylval.op = yytext;
  break;
  case NEWLINE_TOK:
  line += 1;
  column = 0; // reset column index 
- printf("\nline %d. ", line);
+ if (LEX_VERBOSE) printf("\nline %d. ", line);
  break;
  case ERROR_TOK:
- printf("\nLexerError: line %d, column %d, token '%s' is not recognized\n",
- line, column, yytext);
+ if (LEX_VERBOSE) printf("\nLexerError: line %d, column %d, token '%s' is not recognized\n",
+  line, column, yytext);
  exit(1);
  default:
  break; // ignore