Merge pull request #105 from udem-dlteam/laurent/random-improvements

A few small changes

debug.c

@@ -1,5 +1,3 @@
-// Shell backend already defines a print_string_char
-#ifndef sh
 void print_string_char(int c) {
   if (c == 7)       putstr("\\a");
   else if (c == 8)  putstr("\\b");
@@ -12,12 +10,9 @@ void print_string_char(int c) {
   else if (c < 32 || c > 126) { putchar('\\'); putint(c >> 6); putint((c >> 3) & 7); putint(c & 7); }
   else putchar(c);
 }
-#else
-void print_string_char(int c);
-#endif
 
+int print_tok_indent = 0;
 void print_tok(int tok, int val) {
-
   int i;
 
   if      (tok == AUTO_KW)      putstr("auto");
@@ -101,15 +96,20 @@ void print_tok(int tok, int val) {
     putchar('"');
   } else if (tok == MACRO_ARG) {
     putstr("ARG["); putint(val); putstr("]");
+  } else if (tok == '{') {
+    putchar(tok);
+    print_tok_indent += 2;
+  } else if (tok == '}') {
+    print_tok_indent -= 2;
+    putchar(tok);
+  } else if (tok == '\n') {
+    putchar(tok);
+    for (i = 0; i < print_tok_indent; i++) putchar(' ');
   } else {
     putchar(tok);
   }
 
-  if (tok == ';') { // Simple heuristic to print newlines. This makes the output more readable.
-    putchar('\n');
-  } else {
-    putchar(' ');
-  }
+  if (tok != '\n') putchar(' ');
 }
 
 // Show the type of a token.
@@ -181,6 +181,7 @@ void print_tok_type(int tok) {
   else if (tok == MACRO)            putstr("macro");
   else if (tok == MACRO_ARG)        putstr("macro argument");
   else if (tok == EOF)              putstr("end of file");
+  else if (tok == '\n')             putstr("newline");
   else                              { putchar('\''); putchar(tok); putchar('\''); }
 }
 

pnut.c

@@ -76,8 +76,10 @@
 #endif
 
 // Options that turns Pnut into a C preprocessor or some variant of it
+// DEBUG_GETCHAR: Read and print the input character by character.
 // DEBUG_CPP: Run preprocessor like gcc -E. This can be useful for debugging the preprocessor.
 // DEBUG_EXPAND_INCLUDES: Reads the input file and includes the contents of the included files.
+// DEBUG_PARSER: Runs the tokenizer on the input. Outputs nothing.
 
 typedef int bool;
 
@@ -948,19 +950,6 @@ int read_macro_tokens(int args) {
   return toks;
 }
 
-#ifdef DEBUG_CPP
-void print_macro_raw_tokens(int tokens) {
-  int i = 0;
-  while (tokens != 0) {
-    // print_tok(car(car(tokens)), cdr(car(tokens)));
-    putchar(car(car(tokens))); putchar('('); putint(car(car(tokens))); putchar(')');
-    tokens = cdr(tokens);
-    i += 1;
-  }
-  putstr("("); putint(i); putstr(" tokens)");
-}
-#endif
-
 // A few things that are different from the standard:
 // - We allow sequence of commas in the argument list
 // - Function-like macros with 0 arguments can be called either without parenthesis or with ().
@@ -1003,22 +992,6 @@ void handle_define() {
   // Accumulate tokens so they can be replayed when the macro is used
   heap[macro + 3] = cons(read_macro_tokens(args), args_count);
 
-#ifdef DEBUG_CPP
-  putstr("# ");
-  putstr(string_pool + heap[macro + 1]);
-  if (args_count != -1) putchar('('); // Function-like macro
-
-  while (args_count > 0) {
-    putstr(string_pool + heap[car(args) + 1]);
-    args = cdr(args);
-    args_count -= 1;
-    if (args_count > 0) putstr(", ");
-  }
-
-  if (args_count != -1) putstr(") ");
-  print_macro_raw_tokens(car(heap[macro + 3]));
-  putchar('\n');
-#endif
 }
 
 int eval_constant(ast expr, bool if_macro) {
@@ -1630,7 +1603,6 @@ void paste_tokens(int left_tok, int left_val) {
 
 void get_tok() {
 
-  bool first_time = true; // Used to simulate a do-while loop
 #ifdef SH_INCLUDE_C_CODE
   int prev_char_buf_ix = declaration_char_buf_ix;
   // Save the cursor in a local variable so we can restore it when the token is
@@ -1645,8 +1617,7 @@ void get_tok() {
 #endif
 
   // This outer loop is used to skip over tokens removed by #ifdef/#ifndef/#else
-  while (first_time || !if_macro_mask) {
-    first_time = false;
+  do {
 #ifdef SH_INCLUDE_C_CODE
     declaration_char_buf_ix = prev_char_buf_ix; // Skip over tokens that are masked off
 #endif
@@ -2020,7 +1991,8 @@ void get_tok() {
         }
       }
     }
-  }
+  } while (!if_macro_mask);
+
 #ifdef SH_INCLUDE_C_CODE
   last_tok_char_buf_ix = prev_last_tok_char_buf_ix - 1;
 #endif
@@ -3145,7 +3117,7 @@ ast parse_compound_statement() {
 
 // Select code generator
 
-#if !(defined DEBUG_CPP) && !(defined DEBUG_EXPAND_INCLUDES)
+#if !(defined DEBUG_CPP) && !(defined DEBUG_EXPAND_INCLUDES) && !(defined DEBUG_PARSER) && !(defined DEBUG_GETCHAR)
 #ifdef sh
 #include "sh.c"
 #endif
@@ -3217,36 +3189,35 @@ int main(int argc, char **argv) {
     fatal_error("no input file");
   }
 
-#if !(defined DEBUG_CPP) && !(defined DEBUG_EXPAND_INCLUDES)
-  codegen_begin();
-#endif
-
   ch = '\n';
+
+#if defined DEBUG_GETCHAR // Read input
+  while (ch != EOF) {
+    get_ch();
+  }
+#elif defined DEBUG_CPP || defined DEBUG_EXPAND_INCLUDES // Tokenize input, output tokens
   get_tok();
 
-#if defined DEBUG_EXPAND_INCLUDES
   while (tok != EOF) {
+    skip_newlines = false; // Don't skip newlines so print_tok knows where to break lines
+    print_tok(tok, val);
     get_tok();
   }
-#else
-  while (tok != EOF) {
-#ifdef DEBUG_CPP
-    print_tok(tok, val);
+#elif defined DEBUG_PARSER // Parse input, output nothing
     get_tok();
+  while (tok != EOF) {
+    decl = parse_definition(0);
+  }
 #else
-
+  codegen_begin();
+  get_tok();
+  while (tok != EOF) {
     decl = parse_definition(0);
 #ifdef SH_INCLUDE_C_CODE
     output_declaration_c_code(get_op(decl) == '=' | get_op(decl) == VAR_DECLS);
 #endif
     codegen_glo_decl(decl);
-
-#endif
   }
-#endif
-
-
-#if !(defined DEBUG_CPP) && !(defined DEBUG_EXPAND_INCLUDES)
   codegen_end();
 #endif
 

run-pnut-variant.sh

@@ -0,0 +1,51 @@
+#! /bin/sh
+
+set -e
+
+: ${PNUT_OPTIONS:=} # Default to empty options
+
+TEMP_DIR="build"
+PNUT_SH_OPTIONS="$PNUT_OPTIONS -DRT_NO_INIT_GLOBALS -Dsh"
+PNUT_SH_OPTIONS_FAST="$PNUT_SH_OPTIONS -DSH_SAVE_VARS_WITH_SET -DOPTIMIZE_CONSTANT_PARAM"
+PNUT_VARIANT_OPTIONS=""
+
+run_with_shell() {
+
+  echo "Running $variant with $1"
+
+  time $1 "$TEMP_DIR/pnut-$variant.sh" $PNUT_SH_OPTIONS "pnut.c" > "$TEMP_DIR/pnut-$variant.output"
+}
+
+# Parse the arguments
+shell="$SHELL" # Use current shell as the default. "all" to test all shells.
+variant=""
+
+while [ $# -gt 0 ]; do
+  case $1 in
+    --shell)      shell="$2";                                                    shift 2 ;;
+    --fast)       PNUT_SH_OPTIONS="$PNUT_SH_OPTIONS_FAST";                       shift 1 ;;
+    --reader)     PNUT_VARIANT_OPTIONS="-DDEBUG_GETCHAR -Ush";  variant=${1#--}; shift 1 ;;
+    --tokenizer)  PNUT_VARIANT_OPTIONS="-DDEBUG_CPP -Ush";      variant=${1#--}; shift 1 ;;
+    --parser)     PNUT_VARIANT_OPTIONS="-DDEBUG_PARSER -Ush";   variant=${1#--}; shift 1 ;;
+    *) echo "Unknown option: $1"; exit 1;;
+  esac
+done
+
+pnut_exec="$TEMP_DIR/pnut-$variant"
+
+if [ ! -d "$TEMP_DIR" ]; then mkdir "$TEMP_DIR"; fi
+
+gcc -o "$pnut_exec" $PNUT_SH_OPTIONS pnut.c
+$pnut_exec $PNUT_SH_OPTIONS $PNUT_VARIANT_OPTIONS "pnut.c" > "$TEMP_DIR/pnut-$variant.sh"
+
+if [ "$shell" = "all" ]; then
+  set +e # Don't exit on error because we want to test all shells.
+  run_with_shell "dash"
+  run_with_shell "ksh"
+  run_with_shell "bash"
+  run_with_shell "yash"
+  run_with_shell "mksh"
+  run_with_shell "zsh"
+else
+  run_with_shell "$shell"
+fi

sh.c

@@ -24,19 +24,6 @@ void handle_shell_include() {
   }
 }
 
-void print_string_char(int c) {
-  if (c == 7)       putstr("\\a");
-  else if (c == 8)  putstr("\\b");
-  else if (c == 12) putstr("\\f");
-  else if (c == 10) putstr("\\n");
-  else if (c == 13) putstr("\\r");
-  else if (c == 9)  putstr("\\t");
-  else if (c == 11) putstr("\\v");
-  else if (c == '\\' || c == '\'' || c == '\"') { putstr("\\"); putchar(c); }
-  else if (c < 32 || c > 126) { putstr("\\"); putint(c>>6); putint((c>>3)&7); putint(c&7); }
-  else putchar(c);
-}
-
 // codegen
 
 #define text int