X-Git-Url: https://ncurses.scripts.mit.edu/?p=ncurses.git;a=blobdiff_plain;f=ncurses%2Ftinfo%2Fcomp_scan.c;h=fe6e8e7631cee4c30aa09bb79c2dc08438ca7c15;hp=459113bf1b972bdd6ebbe3770fd23219beeb8a53;hb=63e3cafac136d773331dee15633f1c9ef08c5493;hpb=8b06e371ed1bce3dd6f37138e6becb5e1a562fe0

diff --git a/ncurses/tinfo/comp_scan.c b/ncurses/tinfo/comp_scan.c
index 459113bf..fe6e8e76 100644
--- a/ncurses/tinfo/comp_scan.c
+++ b/ncurses/tinfo/comp_scan.c
@@ -1,5 +1,5 @@
 /****************************************************************************
- * Copyright (c) 1998-2011,2012 Free Software Foundation, Inc.              *
+ * Copyright (c) 1998-2012,2013 Free Software Foundation, Inc.              *
  *                                                                          *
  * Permission is hereby granted, free of charge, to any person obtaining a  *
  * copy of this software and associated documentation files (the            *
@@ -50,7 +50,7 @@
 #include <ctype.h>
 #include <tic.h>
 
-MODULE_ID("$Id: comp_scan.c,v 1.96 2012/02/22 22:26:58 tom Exp $")
+MODULE_ID("$Id: comp_scan.c,v 1.102 2013/11/16 19:57:50 tom Exp $")
 
 /*
  * Maximum length of string capability we'll accept before raising an error.
@@ -128,14 +128,19 @@ _nc_reset_input(FILE *fp, char *buf)
  *	Returns the final nonblank character on the current input buffer
  */
 static int
-last_char(void)
+last_char(int from_end)
 {
     size_t len = strlen(bufptr);
+    int result = 0;
+
     while (len--) {
-	if (!isspace(UChar(bufptr[len])))
-	    return bufptr[len];
+	if (!isspace(UChar(bufptr[len]))) {
+	    if (from_end < (int) len)
+		result = bufptr[(int) len - from_end];
+	    break;
+	}
     }
-    return 0;
+    return result;
 }
 
 /*
@@ -196,7 +201,8 @@ next_char(void)
 		    result = typeRealloc(char, allocated, result);
 		    if (result == 0)
 			return (EOF);
-		    bufstart = result;
+		    if (bufstart)
+			bufstart = result;
 		}
 		if (used == 0)
 		    _nc_curr_file_pos = ftell(yyin);
@@ -340,13 +346,13 @@ _nc_get_token(bool silent)
 {
     static const char terminfo_punct[] = "@%&*!#";
 
-    char *after_list;
-    char *after_name;
+    char *after_name;		/* after primary name */
+    char *after_list;		/* after primary and alias list */
     char *numchk;
     char *tok_ptr;
     char *s;
     char numbuf[80];
-    int ch;
+    int ch, c0, c1;
     int dot_flag = FALSE;
     int type;
     long number;
@@ -374,7 +380,7 @@ _nc_get_token(bool silent)
 
     if (end_of_stream()) {
 	yyin = 0;
-	next_char();		/* frees its allocated memory */
+	(void) next_char();	/* frees its allocated memory */
 	if (tok_buf != 0) {
 	    if (_nc_curr_token.tk_name == tok_buf)
 		_nc_curr_token.tk_name = 0;
@@ -426,7 +432,7 @@ _nc_get_token(bool silent)
 #if NCURSES_EXT_FUNCS
 	    && !(ch == '.' && _nc_disable_period)
 #endif
-	    && !strchr(terminfo_punct, (char) ch)) {
+	    && ((strchr) (terminfo_punct, (char) ch) == 0)) {
 	    if (!silent)
 		_nc_warning("Illegal character (expected alphanumeric or %s) - '%s'",
 			    terminfo_punct, unctrl(UChar(ch)));
@@ -459,7 +465,7 @@ _nc_get_token(bool silent)
 		    after_list = tok_ptr;
 		    if (after_name == 0)
 			after_name = tok_ptr;
-		} else if (ch == ':' && last_char() != ',') {
+		} else if (ch == ':' && last_char(0) != ',') {
 		    _nc_syntax = SYN_TERMCAP;
 		    separator = ':';
 		    break;
@@ -473,12 +479,64 @@ _nc_get_token(bool silent)
 		    if (after_name == 0)
 			break;
 		    /*
-		     * If we see a comma, we assume this is terminfo unless we
-		     * subsequently run into a colon.  But we don't stop
-		     * looking for a colon until hitting a newline.  This
-		     * allows commas to be embedded in description fields of
-		     * either syntax.
+		     * We saw a comma, but are not entirely sure this is
+		     * terminfo format, since we can still be parsing the
+		     * description field (for either syntax).
+		     *
+		     * A properly formatted termcap line ends with either a
+		     * colon, or a backslash after a colon.  It is possible
+		     * to have a backslash in the middle of a capability, but
+		     * then there would be no leading whitespace on the next
+		     * line - something we want to discourage.
 		     */
+		    c0 = last_char(0);
+		    c1 = last_char(1);
+		    if (c1 != ':' && c0 != '\\' && c0 != ':') {
+			bool capability = FALSE;
+
+			/*
+			 * Since it is not termcap, assume the line is terminfo
+			 * format.  However, the comma can be embedded in a
+			 * description field.  It also can be a separator
+			 * between a description field and a capability.
+			 *
+			 * Improve the guess by checking if the next word after
+			 * the comma does not look like a capability.  In that
+			 * case, extend the description past the comma.
+			 */
+			for (s = bufptr; isspace(UChar(*s)); ++s) {
+			    ;
+			}
+			if (islower(UChar(*s))) {
+			    char *name = s;
+			    while (isalnum(UChar(*s))) {
+				++s;
+			    }
+			    if (*s == '#' || *s == '=' || *s == '@') {
+				/*
+				 * Checking solely with syntax allows us to
+				 * support extended capabilities with string
+				 * values.
+				 */
+				capability = TRUE;
+			    } else if (*s == ',') {
+				c0 = *s;
+				*s = '\0';
+				/*
+				 * Otherwise, we can handle predefined boolean
+				 * capabilities, still aided by syntax.
+				 */
+				if (_nc_find_entry(name,
+						   _nc_get_hash_table(FALSE))) {
+				    capability = TRUE;
+				}
+				*s = (char) c0;
+			    }
+			}
+			if (capability) {
+			    break;
+			}
+		    }
 		} else
 		    ch = eat_escaped_newline(ch);