ncurses 6.0 - patch 20170422
[ncurses.git] / ncurses / tinfo / comp_scan.c
index 28ae9e4e87135293679765a375ca1861bab363c4..40d7f6a8bc0dc59cd49014cd8dd90d10d2ec5917 100644 (file)
@@ -1,5 +1,5 @@
 /****************************************************************************
- * Copyright (c) 1998-2008,2010 Free Software Foundation, Inc.              *
+ * Copyright (c) 1998-2016,2017 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 1.86 2010/01/23 17:13:15 tom Exp $")
+MODULE_ID("$Id: comp_scan.c,v 1.106 2017/04/22 11:41:12 tom Exp $")
 
 /*
  * Maximum length of string capability we'll accept before raising an error.
@@ -61,6 +61,7 @@ MODULE_ID("$Id: comp_scan.c,v 1.86 2010/01/23 17:13:15 tom Exp $")
 #define iswhite(ch)    (ch == ' '  ||  ch == '\t')
 
 NCURSES_EXPORT_VAR (int) _nc_syntax = 0;         /* termcap or terminfo? */
+NCURSES_EXPORT_VAR (int) _nc_strict_bsd = 1;  /* ncurses extended termcap? */
 NCURSES_EXPORT_VAR (long) _nc_curr_file_pos = 0; /* file offset of current line */
 NCURSES_EXPORT_VAR (long) _nc_comment_start = 0; /* start of comment range before name */
 NCURSES_EXPORT_VAR (long) _nc_comment_end = 0;   /* end of comment range before name */
@@ -127,14 +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;
 }
 
 /*
@@ -183,19 +189,19 @@ next_char(void)
         * quite hard to get completely right.  Try it and see.  If you
         * succeed, don't forget to hack push_back() correspondingly.
         */
-       size_t used;
        size_t len;
 
        do {
+           size_t used = 0;
            bufstart = 0;
-           used = 0;
            do {
                if (used + (LEXBUFSIZ / 4) >= allocated) {
                    allocated += (allocated + LEXBUFSIZ);
                    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);
@@ -203,12 +209,16 @@ next_char(void)
                if (fgets(result + used, (int) (allocated - used), yyin) != 0) {
                    bufstart = result;
                    if (used == 0) {
+                       if (_nc_curr_line == 0
+                           && IS_TIC_MAGIC(result)) {
+                           _nc_err_abort("This is a compiled terminal description, not a source");
+                       }
                        _nc_curr_line++;
                        _nc_curr_col = 0;
                    }
                } else {
                    if (used != 0)
-                       strcat(result, "\n");
+                       _nc_STRCAT(result, "\n", allocated);
                }
                if ((bufptr = bufstart) != 0) {
                    used = strlen(bufptr);
@@ -252,12 +262,12 @@ next_char(void)
 }
 
 static void
-push_back(char c)
+push_back(int c)
 /* push a character back onto the input stream */
 {
     if (bufptr == bufstart)
        _nc_syserr_abort("Can't backspace off beginning of line");
-    *--bufptr = c;
+    *--bufptr = (char) c;
     _nc_curr_col--;
 }
 
@@ -295,6 +305,8 @@ eat_escaped_newline(int ch)
        *tok_ptr++ = (char) ch; \
        *tok_ptr = '\0'
 
+static char *tok_buf;
+
 /*
  *     int
  *     get_token()
@@ -332,15 +344,14 @@ NCURSES_EXPORT(int)
 _nc_get_token(bool silent)
 {
     static const char terminfo_punct[] = "@%&*!#";
-    static char *tok_buf;
 
-    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;
@@ -368,11 +379,10 @@ _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;
-           FreeAndNull(tok_buf);
        }
        return (EOF);
     }
@@ -386,6 +396,7 @@ _nc_get_token(bool silent)
     }
 
     ch = eat_escaped_newline(ch);
+    _nc_curr_token.tk_valstring = 0;
 
 #ifdef TRACE
     old_line = _nc_curr_line;
@@ -420,7 +431,7 @@ _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)));
@@ -453,7 +464,7 @@ _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;
@@ -467,19 +478,70 @@ _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);
 
                if (OkToAdd()) {
                    AddCh(ch);
                } else {
-                   ch = EOF;
                    break;
                }
            }
@@ -606,10 +668,18 @@ _nc_get_token(bool silent)
                    if (numchk == numbuf)
                        _nc_warning("no value given for `%s'", tok_buf);
                    if ((*numchk != '\0') || (ch != separator))
-                       _nc_warning("Missing separator");
+                       _nc_warning("Missing separator for `%s'", tok_buf);
+                   if (number < 0)
+                       _nc_warning("value of `%s' cannot be negative", tok_buf);
+                   if (number > MAX_OF_TYPE(NCURSES_INT2)) {
+                       _nc_warning("limiting value of `%s' from %#lx to %#x",
+                                   tok_buf,
+                                   number, MAX_OF_TYPE(NCURSES_INT2));
+                       number = MAX_OF_TYPE(NCURSES_INT2);
+                   }
                }
                _nc_curr_token.tk_name = tok_buf;
-               _nc_curr_token.tk_valnumber = number;
+               _nc_curr_token.tk_valnumber = (int) number;
                type = NUMBER;
                break;
 
@@ -745,33 +815,35 @@ _nc_trans_string(char *ptr, char *last)
            if (!(is7bits(c) && isprint(c))) {
                _nc_warning("Illegal ^ character - '%s'", unctrl(UChar(c)));
            }
-           if (c == '?') {
+           if (c == '?' && (_nc_syntax != SYN_TERMCAP)) {
                *(ptr++) = '\177';
-               if (_nc_tracing)
-                   _nc_warning("Allow ^? as synonym for \\177");
            } else {
                if ((c &= 037) == 0)
                    c = 128;
                *(ptr++) = (char) (c);
            }
        } else if (c == '\\') {
+           bool strict_bsd = ((_nc_syntax == SYN_TERMCAP) && _nc_strict_bsd);
+
            c = next_char();
            if (c == EOF)
                _nc_err_abort(MSG_NO_INPUTS);
 
-           if (c >= '0' && c <= '7') {
+           if (isoctal(c) || (strict_bsd && isdigit(c))) {
                number = c - '0';
                for (i = 0; i < 2; i++) {
                    c = next_char();
                    if (c == EOF)
                        _nc_err_abort(MSG_NO_INPUTS);
 
-                   if (c < '0' || c > '7') {
+                   if (!isoctal(c)) {
                        if (isdigit(c)) {
-                           _nc_warning("Non-octal digit `%c' in \\ sequence", c);
-                           /* allow the digit; it'll do less harm */
+                           if (!strict_bsd) {
+                               _nc_warning("Non-octal digit `%c' in \\ sequence", c);
+                               /* allow the digit; it'll do less harm */
+                           }
                        } else {
-                           push_back((char) c);
+                           push_back(c);
                            break;
                        }
                    }
@@ -779,21 +851,16 @@ _nc_trans_string(char *ptr, char *last)
                    number = number * 8 + c - '0';
                }
 
-               if (number == 0)
+               number = UChar(number);
+               if (number == 0 && !strict_bsd)
                    number = 0200;
                *(ptr++) = (char) number;
            } else {
                switch (c) {
                case 'E':
-               case 'e':
                    *(ptr++) = '\033';
                    break;
 
-               case 'a':
-                   *(ptr++) = '\007';
-                   break;
-
-               case 'l':
                case 'n':
                    *(ptr++) = '\n';
                    break;
@@ -806,10 +873,6 @@ _nc_trans_string(char *ptr, char *last)
                    *(ptr++) = '\010';
                    break;
 
-               case 's':
-                   *(ptr++) = ' ';
-                   break;
-
                case 'f':
                    *(ptr++) = '\014';
                    break;
@@ -830,16 +893,33 @@ _nc_trans_string(char *ptr, char *last)
                    *(ptr++) = ',';
                    break;
 
-               case ':':
-                   *(ptr++) = ':';
-                   break;
-
                case '\n':
                    continue;
 
                default:
-                   _nc_warning("Illegal character '%s' in \\ sequence",
-                               unctrl(UChar(c)));
+                   if ((_nc_syntax == SYN_TERMINFO) || !_nc_strict_bsd) {
+                       switch (c) {
+                       case 'a':
+                           c = '\007';
+                           break;
+                       case 'e':
+                           c = '\033';
+                           break;
+                       case 'l':
+                           c = '\n';
+                           break;
+                       case 's':
+                           c = ' ';
+                           break;
+                       case ':':
+                           c = ':';
+                           break;
+                       default:
+                           _nc_warning("Illegal character '%s' in \\ sequence",
+                                       unctrl(UChar(c)));
+                           break;
+                       }
+                   }
                    /* FALLTHRU */
                case '|':
                    *(ptr++) = (char) c;
@@ -859,7 +939,7 @@ _nc_trans_string(char *ptr, char *last)
 
        if (!ignored) {
            if (_nc_curr_col <= 1) {
-               push_back((char) c);
+               push_back(c);
                c = '\n';
                break;
            }
@@ -913,10 +993,8 @@ _nc_push_token(int tokclass)
 NCURSES_EXPORT(void)
 _nc_panic_mode(char ch)
 {
-    int c;
-
     for (;;) {
-       c = next_char();
+       int c = next_char();
        if (c == ch)
            return;
        if (c == EOF)
@@ -931,5 +1009,8 @@ _nc_comp_scan_leaks(void)
     if (pushname != 0) {
        FreeAndNull(pushname);
     }
+    if (tok_buf != 0) {
+       FreeAndNull(tok_buf);
+    }
 }
 #endif