1 files changed, 303 insertions, 47 deletions

diff --git a/awkgram.y b/awkgram.y
index beb85d5a..3b555a6e 100644
--- a/awkgram.y
+++ b/awkgram.y
@@ -83,10 +83,17 @@ static void check_funcs(void);
 
 static ssize_t read_one_line(int fd, void *buffer, size_t count);
 static int one_line_close(int fd);
+static void split_comment(void);
+static void check_comment(void);
 
 static bool at_seen = false;
 static bool want_source = false;
 static bool want_regexp = false;	/* lexical scanning kludge */
+static enum {
+	FUNC_HEADER,
+	FUNC_BODY,
+	DONT_CHECK
+} want_param_names = DONT_CHECK;	/* ditto */
 static char *in_function;		/* parsing kludge */
 static int rule = 0;
 
@@ -141,6 +148,15 @@ static INSTRUCTION *ip_atexit = NULL;
 static INSTRUCTION *ip_end;
 static INSTRUCTION *ip_endfile;
 static INSTRUCTION *ip_beginfile;
+INSTRUCTION *main_beginfile;
+
+static INSTRUCTION *comment = NULL;
+static INSTRUCTION *program_comment = NULL;
+static INSTRUCTION *function_comment = NULL;
+static INSTRUCTION *block_comment = NULL;
+
+static bool func_first = true;
+static bool first_rule = true;
 
 static inline INSTRUCTION *list_create(INSTRUCTION *x);
 static inline INSTRUCTION *list_append(INSTRUCTION *l, INSTRUCTION *x);
@@ -218,6 +234,7 @@ rule
 	: pattern action
 	  {
 		(void) append_rule($1, $2);
+		first_rule = false;
 	  }
 	| pattern statement_term
 	  {
@@ -234,6 +251,7 @@ rule
 	  {
 		in_function = NULL;
 		(void) mk_function($1, $2);
+		want_param_names = DONT_CHECK;
 		yyerrok;
 	  }
 	| '@' LEX_INCLUDE source statement_term
@@ -282,9 +300,24 @@ library
 
 pattern
 	: /* empty */
-	  {	$$ = NULL; rule = Rule; }
+	  {
+		rule = Rule;
+		if (comment != NULL) {
+			$$ = list_create(comment);
+			comment = NULL;
+		} else
+			$$ = NULL;
+	  }
 	| exp
-	  {	$$ = $1; rule = Rule; }
+	  {
+		rule = Rule;
+		if (comment != NULL) {
+			$$ = list_prepend($1, comment);
+			comment = NULL;
+		} else
+			$$ = $1;
+	  }
+		
 	| exp ',' opt_nls exp
 	  {
 		INSTRUCTION *tp;
@@ -308,41 +341,55 @@ pattern
 			($1->nexti + 1)->condpair_left = $1->lasti;
 			($1->nexti + 1)->condpair_right = $4->lasti;
 		}
-		$$ = list_append(list_merge($1, $4), tp);
+		if (comment != NULL) {
+			$$ = list_append(list_merge(list_prepend($1, comment), $4), tp);
+			comment = NULL;
+		} else
+			$$ = list_append(list_merge($1, $4), tp);
 		rule = Rule;
 	  }
 	| LEX_BEGIN
 	  {
 		static int begin_seen = 0;
+
+		func_first = false;
 		if (do_lint_old && ++begin_seen == 2)
 			warning_ln($1->source_line,
 				_("old awk does not support multiple `BEGIN' or `END' rules"));
 
 		$1->in_rule = rule = BEGIN;
 		$1->source_file = source;
+		check_comment();
 		$$ = $1;
 	  }
 	| LEX_END
 	  {
 		static int end_seen = 0;
+
+		func_first = false;
 		if (do_lint_old && ++end_seen == 2)
 			warning_ln($1->source_line,
 				_("old awk does not support multiple `BEGIN' or `END' rules"));
 
 		$1->in_rule = rule = END;
 		$1->source_file = source;
+		check_comment();
 		$$ = $1;
 	  }
 	| LEX_BEGINFILE
 	  {
+		func_first = false;
 		$1->in_rule = rule = BEGINFILE;
 		$1->source_file = source;
+		check_comment();
 		$$ = $1;
 	  }
 	| LEX_ENDFILE
 	  {
+		func_first = false;
 		$1->in_rule = rule = ENDFILE;
 		$1->source_file = source;
+		check_comment();
 		$$ = $1;
 	  }
 	;
@@ -350,10 +397,12 @@ pattern
 action
 	: l_brace statements r_brace opt_semi opt_nls
 	  {
+		INSTRUCTION *ip;
 		if ($2 == NULL)
-			$$ = list_create(instruction(Op_no_op));
+			ip = list_create(instruction(Op_no_op));
 		else
-			$$ = $2;
+			ip = $2;
+		$$ = ip;
 	  }
 	;
 
@@ -381,16 +430,33 @@ lex_builtin
 	;
 		
 function_prologue
-	: LEX_FUNCTION func_name '(' opt_param_list r_paren opt_nls
+	: LEX_FUNCTION func_name '(' { want_param_names = FUNC_HEADER; } opt_param_list r_paren opt_nls
 	  {
+		/*
+		 *  treat any comments between BOF and the first function
+		 *  definition (with no intervening BEGIN etc block) as
+		 *  program comments.  Special kludge: iff there are more
+		 *  than one such comments, treat the last as a function
+		 *  comment.
+		 */
+		if (comment != NULL && func_first
+		    && strstr(comment->memory->stptr, "\n\n") != NULL)
+			split_comment();
+		/* save any other pre-function comment as function comment  */
+		if (comment != NULL) {
+			function_comment = comment;
+			comment = NULL;
+		}
+		func_first = false;
 		$1->source_file = source;
-		if (install_function($2->lextok, $1, $4) < 0)
+		if (install_function($2->lextok, $1, $5) < 0)
 			YYABORT;
 		in_function = $2->lextok;
 		$2->lextok = NULL;
 		bcfree($2);
-		/* $4 already free'd in install_function */
+		/* $5 already free'd in install_function */
 		$$ = $1;
+		want_param_names = FUNC_BODY;
 	  }
 	;
 
@@ -440,19 +506,39 @@ a_slash
 
 statements
 	: /* empty */
-	  {	$$ = NULL; }
+	  {
+		if (comment != NULL) {
+			$$ = list_create(comment);
+			comment = NULL;
+		} else $$ = NULL;
+	  }
 	| statements statement
 	  {
-		if ($2 == NULL)
-			$$ = $1;
-		else {
+		if ($2 == NULL) {
+			if (comment == NULL)
+				$$ = $1;
+			else {
+				$$ = list_append($1, comment);
+				comment = NULL;
+			}
+		} else {
 			add_lint($2, LINT_no_effect);
-			if ($1 == NULL)
-				$$ = $2;
-			else
+			if ($1 == NULL) {
+				if (comment == NULL)
+					$$ = $2;
+				else {
+					$$ = list_append($2, comment);
+					comment = NULL;
+				}
+			} else {
+				if (comment != NULL) {
+					list_append($2, comment);
+					comment = NULL;
+				}
 				$$ = list_merge($1, $2);
+			}
 		}
-	    yyerrok;
+		yyerrok;
 	  }
 	| statements error
 	  {	$$ = NULL; }
@@ -496,7 +582,7 @@ statement
 		} /*  else
 				curr = NULL; */
 
-		for(; curr != NULL; curr = nextc) {
+		for (; curr != NULL; curr = nextc) {
 			INSTRUCTION *caseexp = curr->case_exp;
 			INSTRUCTION *casestmt = curr->case_stmt;
 
@@ -1184,7 +1270,7 @@ opt_param_list
 	: /* empty */
 	  { $$ = NULL; }
 	| param_list
-	  { $$ = $1 ; }
+	  { $$ = $1; }
 	;
 
 param_list
@@ -1364,7 +1450,7 @@ common_exp
 			n1 = force_string(n1);
 			n2 = force_string(n2);
 			nlen = n1->stlen + n2->stlen;
-			erealloc(n1->stptr, char *, nlen + 2, "constant fold");
+			erealloc(n1->stptr, char *, nlen + 1, "constant fold");
 			memcpy(n1->stptr + n1->stlen, n2->stptr, n2->stlen);
 			n1->stlen = nlen;
 			n1->stptr[nlen] = '\0';
@@ -1890,6 +1976,7 @@ static const struct token tokentab[] = {
 {"include",	Op_symbol,	 LEX_INCLUDE,	GAWKX,	0,	0},
 {"index",	Op_builtin,	 LEX_BUILTIN,	A(2),		do_index,	0},
 {"int",		Op_builtin,	 LEX_BUILTIN,	A(1),		do_int,	MPF(int)},
+{"intdiv",	Op_builtin,	 LEX_BUILTIN,	GAWKX|A(3),	do_intdiv,	MPF(intdiv)},
 {"isarray",	Op_builtin,	 LEX_BUILTIN,	GAWKX|A(1),	do_isarray,	0},
 {"length",	Op_builtin,	 LEX_LENGTH,	A(0)|A(1),	do_length,	0},
 {"load",  	Op_symbol,	 LEX_LOAD,	GAWKX,		0,	0},
@@ -2098,7 +2185,6 @@ yyerror(const char *m, ...)
 	char *buf;
 	int count;
 	static char end_of_file_line[] = "(END OF FILE)";
-	char save;
 
 	print_included_from();
 
@@ -2126,24 +2212,15 @@ yyerror(const char *m, ...)
 		bp = thisline + strlen(thisline);
 	}
 
-	/*
-	 * Saving and restoring *bp keeps valgrind happy,
-	 * since the guts of glibc uses strlen, even though
-	 * we're passing an explict precision. Sigh.
-	 *
-	 * 8/2003: We may not need this anymore.
-	 */
-	save = *bp;
-	*bp = '\0';
-
 	msg("%.*s", (int) (bp - thisline), thisline);
 
-	*bp = save;
 	va_start(args, m);
 	if (mesg == NULL)
 		mesg = m;
 
-	count = (bp - thisline) + strlen(mesg) + 2 + 1;
+	count = strlen(mesg) + 1;
+	if (lexptr != NULL)
+		count += (lexeme - thisline) + 2;
 	emalloc(buf, char *, count, "yyerror");
 
 	bp = buf;
@@ -2226,6 +2303,11 @@ mk_program()
 				cp = end_block;
 			else
 				cp = list_merge(begin_block, end_block);
+			if (program_comment != NULL) {
+				(void) list_prepend(cp, program_comment);
+			}  
+			if (comment != NULL)
+				(void) list_append(cp, comment);
 			(void) list_append(cp, ip_atexit);
 			(void) list_append(cp, instruction(Op_stop));
 
@@ -2258,6 +2340,12 @@ mk_program()
 	if (begin_block != NULL)
 		cp = list_merge(begin_block, cp);
 
+	if (program_comment != NULL) {
+		(void) list_prepend(cp, program_comment);
+	}  
+	if (comment != NULL) {
+		(void) list_append(cp, comment);
+	} 
 	(void) list_append(cp, ip_atexit);
 	(void) list_append(cp, instruction(Op_stop));
 
@@ -2265,6 +2353,10 @@ out:
 	/* delete the Op_list, not needed */
 	tmp = cp->nexti;
 	bcfree(cp);
+	/* these variables are not used again but zap them anyway.  */
+	comment = NULL;
+	function_comment = NULL;
+	program_comment = NULL;
 	return tmp;
 
 #undef begin_block
@@ -2291,7 +2383,7 @@ parse_program(INSTRUCTION **pcode)
 		ip_newfile = ip_rec = ip_atexit = ip_beginfile = ip_endfile = NULL;
 	else {
 		ip_endfile = instruction(Op_no_op);
-		ip_beginfile = instruction(Op_no_op);
+		main_beginfile = ip_beginfile = instruction(Op_no_op);
 		ip_rec = instruction(Op_get_record); /* target for `next', also ip_newfile */
 		ip_newfile = bcalloc(Op_newfile, 2, 0); /* target for `nextfile' */
 		ip_newfile->target_jmp = ip_end;
@@ -2762,7 +2854,7 @@ get_src_buf()
 		lexend = lexptr + n;
 		if (n == 0) {
 			static bool warned = false;
-			if (do_lint && newfile && ! warned){
+			if (do_lint && newfile && ! warned) {
 				warned = true;
 				sourceline = 0;
 				lintwarn(_("source file `%s' is empty"), source);
@@ -2853,7 +2945,7 @@ again:
 			mbstate_t tmp_state;
 			size_t mbclen;
 	
-			for (idx = 0 ; lexptr + idx < lexend ; idx++) {
+			for (idx = 0; lexptr + idx < lexend; idx++) {
 				tmp_state = cur_mbstate;
 				mbclen = mbrlen(lexptr, idx + 1, &tmp_state);
 
@@ -2910,6 +3002,105 @@ pushback(void)
 	(! lexeof && lexptr && lexptr > lexptr_begin ? lexptr-- : lexptr);
 }
 
+/* check_comment --- check for block comment */
+
+void
+check_comment(void)
+{
+	if (comment != NULL) {
+		if (first_rule) {
+			program_comment = comment;
+		} else 
+			block_comment = comment;
+		comment = NULL;
+	}
+	first_rule = false;
+}
+
+/*
+ * get_comment --- collect comment text.
+ * 	Flag = EOL_COMMENT for end-of-line comments.
+ * 	Flag = FULL_COMMENT for self-contained comments.
+ */
+
+int
+get_comment(int flag)
+{
+	int c;
+	int sl;
+	tok = tokstart;
+	tokadd('#');
+	sl = sourceline;
+
+	while (true) {
+		while ((c = nextc(false)) != '\n' && c != END_FILE) {
+			tokadd(c);
+		}
+		if (flag == EOL_COMMENT) {
+			/* comment at end of line.  */
+			if (c == '\n')
+				tokadd(c);
+			break;
+		}
+		if (c == '\n') {
+			tokadd(c);
+			sourceline++;
+			do {
+				c = nextc(false);
+				if (c == '\n') {
+					sourceline++;
+					tokadd(c);
+				}
+			} while (isspace(c) && c != END_FILE);
+			if (c == END_FILE)
+				break;
+			else if (c != '#') {
+				pushback();
+				sourceline--;
+				break;
+			} else
+				tokadd(c);
+		} else
+			break;
+	}
+	comment = bcalloc(Op_comment, 1, sl);
+	comment->source_file = source;
+	comment->memory = make_str_node(tokstart, tok - tokstart, 0);
+	comment->memory->comment_type = flag;
+
+	return c;
+}
+
+/* split_comment --- split initial comment text into program and function parts */
+
+static void
+split_comment(void)
+{
+	char *p;
+	int l;
+	NODE *n;
+
+	p = comment->memory->stptr;
+	l = comment->memory->stlen - 3;
+	/* have at least two comments so split at last blank line (\n\n)  */
+	while (l >= 0) {
+		if (p[l] == '\n' && p[l+1] == '\n') {
+			function_comment = comment;
+			n = function_comment->memory;
+			function_comment->memory = make_str_node(p + l + 2, n->stlen - l - 2, 0);
+			/* create program comment  */
+			program_comment = bcalloc(Op_comment, 1, sourceline);
+			program_comment->source_file = comment->source_file;
+			p[l + 2] = 0;
+			program_comment->memory = make_str_node(p, l + 2, 0);
+			comment = NULL;
+			freenode(n);
+			break;
+		}
+		else
+			l--;
+	}
+}
 
 /* allow_newline --- allow newline after &&, ||, ? and : */
 
@@ -2925,8 +3116,13 @@ allow_newline(void)
 			break;
 		}
 		if (c == '#') {
-			while ((c = nextc(false)) != '\n' && c != END_FILE)
-				continue;
+			if (do_pretty_print && ! do_profile) {
+			/* collect comment byte code iff doing pretty print but not profiling.  */
+				c = get_comment(EOL_COMMENT);
+			} else {
+				while ((c = nextc(false)) != '\n' && c != END_FILE)
+					continue;
+			}
 			if (c == END_FILE) {
 				pushback();
 				break;
@@ -2949,7 +3145,8 @@ allow_newline(void)
  * removes the warnings.
  */
 
-static int newline_eof()
+static int
+newline_eof()
 {
 	/* NB: a newline at end does not start a source line. */
 	if (lasttok != NEWLINE) {
@@ -3136,9 +3333,23 @@ retry:
 		return lasttok = NEWLINE;
 
 	case '#':		/* it's a comment */
-		while ((c = nextc(false)) != '\n') {
+		if (do_pretty_print && ! do_profile) {
+			/*
+			 * Collect comment byte code iff doing pretty print
+			 * but not profiling.
+			 */
+			if (lasttok == NEWLINE || lasttok == 0)
+				c = get_comment(FULL_COMMENT);
+			else
+				c = get_comment(EOL_COMMENT);
+
 			if (c == END_FILE)
 				return lasttok = NEWLINE_EOF;
+		} else {
+			while ((c = nextc(false)) != '\n') {
+				if (c == END_FILE)
+					return lasttok = NEWLINE_EOF;
+			}
 		}
 		sourceline++;
 		return lasttok = NEWLINE;
@@ -3150,7 +3361,7 @@ retry:
 	case '\\':
 #ifdef RELAXED_CONTINUATION
 		/*
-		 * This code puports to allow comments and/or whitespace
+		 * This code purports to allow comments and/or whitespace
 		 * after the `\' at the end of a line used for continuation.
 		 * Use it at your own risk. We think it's a bad idea, which
 		 * is why it's not on by default.
@@ -3167,9 +3378,13 @@ retry:
 					lintwarn(
 		_("use of `\\ #...' line continuation is not portable"));
 				}
-				while ((c = nextc(false)) != '\n')
-					if (c == END_FILE)
-						break;
+				if (do_pretty_print && ! do_profile)
+					c = get_comment(EOL_COMMENT);
+				else {
+					while ((c = nextc(false)) != '\n')
+						if (c == END_FILE)
+							break;
+				}
 			}
 			pushback();
 		}
@@ -3381,7 +3596,7 @@ retry:
 				lastline = sourceline;
 			return lasttok = c;
 		}
-		did_newline++;
+		did_newline = true;
 		--lexptr;	/* pick up } next time */
 		return lasttok = NEWLINE;
 
@@ -3679,6 +3894,27 @@ retry:
 				&& lasttok != '@')
 			goto out;
 
+		/* allow parameter names to shadow the names of gawk extension built-ins */
+		if ((tokentab[mid].flags & GAWKX) != 0) {
+			switch (want_param_names) {
+			case FUNC_HEADER:
+				/* in header, defining parameter names */
+				goto out;
+			case FUNC_BODY:
+				/* in body, name must be in symbol table for it to be a parameter */
+				if (lookup(tokstart) != NULL)
+					goto out;
+				/* else
+					fall through */
+			case DONT_CHECK:
+				/* regular code */
+				break;
+			default:
+				cant_happen();
+				break;
+			}
+		}
+
 		if (do_lint) {
 			if ((tokentab[mid].flags & GAWKX) != 0 && (warntab[mid] & GAWKX) == 0) {
 				lintwarn(_("`%s' is a gawk extension"),
@@ -3903,7 +4139,7 @@ snode(INSTRUCTION *subn, INSTRUCTION *r)
 	}
 
 #ifdef HAVE_MPFR
-	/* N.B.: There isn't any special processing for an alternate function below */
+	/* N.B.: If necessary, add special processing for alternate builtin, below */
 	if (do_mpfr && tokentab[idx].ptr2)
 		r->builtin =  tokentab[idx].ptr2;
 	else
@@ -3932,6 +4168,15 @@ snode(INSTRUCTION *subn, INSTRUCTION *r)
 		arg = subn->nexti;
 		if (arg->nexti == arg->lasti && arg->nexti->opcode == Op_push)
 			arg->nexti->opcode = Op_push_arg;	/* argument may be array */
+	} else if (r->builtin == do_intdiv
+#ifdef HAVE_MPFR
+		   || r->builtin == MPF(intdiv)
+#endif
+			) {
+		arg = subn->nexti->lasti->nexti->lasti->nexti;	/* 3rd arg list */
+		ip = arg->lasti;
+		if (ip->opcode == Op_push)
+			ip->opcode = Op_push_array;
 	} else if (r->builtin == do_match) {
 		static bool warned = false;
 
@@ -4227,6 +4472,14 @@ mk_function(INSTRUCTION *fi, INSTRUCTION *def)
 			(t + 1)->tail_call = true;
 	}
 
+	/* add any pre-function comment to start of action for profile.c  */
+
+	if (function_comment != NULL) {
+		function_comment->source_line = 0;
+		(void) list_prepend(def, function_comment);
+		function_comment = NULL;
+	}
+
 	/* add an implicit return at end;
 	 * also used by 'return' command in debugger
 	 */
@@ -4930,7 +5183,11 @@ append_rule(INSTRUCTION *pattern, INSTRUCTION *action)
 		(rp + 1)->lasti = action->lasti;
 		(rp + 2)->first_line = pattern->source_line;
 		(rp + 2)->last_line = lastline;
-		ip = list_prepend(action, rp);
+		if (block_comment != NULL) {
+			ip = list_prepend(list_prepend(action, block_comment), rp);
+			block_comment = NULL;
+		} else
+			ip = list_prepend(action, rp);
 
 	} else {
 		rp = bcalloc(Op_rule, 3, 0);
@@ -4972,7 +5229,6 @@ append_rule(INSTRUCTION *pattern, INSTRUCTION *action)
 						action),
 					tp);
 		}
-
 	}
 
 	list_append(rule_list, rp + 1);