parse_entry.c

ncurses-5.4

/****************************************************************************
 * Copyright (c) 1998-2002,2003 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            *
 * "Software"), to deal in the Software without restriction, including      *
 * without limitation the rights to use, copy, modify, merge, publish,      *
 * distribute, distribute with modifications, sublicense, and/or sell       *
 * copies of the Software, and to permit persons to whom the Software is    *
 * furnished to do so, subject to the following conditions:                 *
 *                                                                          *
 * The above copyright notice and this permission notice shall be included  *
 * in all copies or substantial portions of the Software.                   *
 *                                                                          *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS  *
 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF               *
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.   *
 * IN NO EVENT SHALL THE ABOVE COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,   *
 * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR    *
 * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR    *
 * THE USE OR OTHER DEALINGS IN THE SOFTWARE.                               *
 *                                                                          *
 * Except as contained in this notice, the name(s) of the above copyright   *
 * holders shall not be used in advertising or otherwise to promote the     *
 * sale, use or other dealings in this Software without prior written       *
 * authorization.                                                           *
 ****************************************************************************/

/****************************************************************************
 *  Author: Zeyd M. Ben-Halim <[e-mail]> 1992,1995               *
 *     and: Eric S. Raymond <[e-mail]>                         *
 *     and: Thomas E. Dickey                        1996-2003               *
 ****************************************************************************/

/*
 *	parse_entry.c -- compile one terminfo or termcap entry
 *
 *	Get an exact in-core representation of an entry.  Don't
 *	try to resolve use or tc capabilities, that is someone
 *	else's job.  Depends on the lexical analyzer to get tokens
 *	from the input stream.
 */

#define __INTERNAL_CAPS_VISIBLE
#include <curses.priv.h>

#include <ctype.h>
#include <tic.h>
#include <term_entry.h>

MODULE_ID("$Id: parse_entry.c,v 1.60 2003/11/08 21:57:09 tom Exp $")

#ifdef LINT
static short const parametrized[] =
{0};
#else
#include <parametrized.h>
#endif

static void postprocess_termcap(TERMTYPE *, bool);
static void postprocess_terminfo(TERMTYPE *);
static struct name_table_entry const *lookup_fullname(const char *name);

#if NCURSES_XNAMES

static struct name_table_entry const *
_nc_extend_names(ENTRY * entryp, char *name, int token_type)
{
    static struct name_table_entry temp;
    TERMTYPE *tp = &(entryp->tterm);
    unsigned offset = 0;
    unsigned actual;
    unsigned tindex;
    unsigned first, last, n;
    bool found;

    switch (token_type) {
    case BOOLEAN:
	first = 0;
	last = tp->ext_Booleans;
	offset = tp->ext_Booleans;
	tindex = tp->num_Booleans;
	break;
    case NUMBER:
	first = tp->ext_Booleans;
	last = tp->ext_Numbers + first;
	offset = tp->ext_Booleans + tp->ext_Numbers;
	tindex = tp->num_Numbers;
	break;
    case STRING:
	first = tp->ext_Booleans + tp->ext_Numbers;
	last = tp->ext_Strings + first;
	offset = tp->ext_Booleans + tp->ext_Numbers + tp->ext_Strings;
	tindex = tp->num_Strings;
	break;
    case CANCEL:
	actual = NUM_EXT_NAMES(tp);
	for (n = 0; n < actual; n++) {
	    if (!strcmp(name, tp->ext_Names[n])) {
		if (n > (unsigned) (tp->ext_Booleans + tp->ext_Numbers)) {
		    token_type = STRING;
		} else if (n > tp->ext_Booleans) {
		    token_type = NUMBER;
		} else {
		    token_type = BOOLEAN;
		}
		return _nc_extend_names(entryp, name, token_type);
	    }
	}
	/* Well, we are given a cancel for a name that we don't recognize */
	return _nc_extend_names(entryp, name, STRING);
    default:
	return 0;
    }

    /* Adjust the 'offset' (insertion-point) to keep the lists of extended
     * names sorted.
     */
    for (n = first, found = FALSE; n < last; n++) {
	int cmp = strcmp(tp->ext_Names[n], name);
	if (cmp == 0)
	    found = TRUE;
	if (cmp >= 0) {
	    offset = n;
	    tindex = n - first;
	    switch (token_type) {
	    case BOOLEAN:
		tindex += BOOLCOUNT;
		break;
	    case NUMBER:
		tindex += NUMCOUNT;
		break;
	    case STRING:
		tindex += STRCOUNT;
		break;
	    }
	    break;
	}
    }
    if (!found) {
	switch (token_type) {
	case BOOLEAN:
	    tp->ext_Booleans += 1;
	    tp->num_Booleans += 1;
	    tp->Booleans = typeRealloc(char, tp->num_Booleans, tp->Booleans);
	    for (last = tp->num_Booleans - 1; last > tindex; last--)
		tp->Booleans[last] = tp->Booleans[last - 1];
	    break;
	case NUMBER:
	    tp->ext_Numbers += 1;
	    tp->num_Numbers += 1;
	    tp->Numbers = typeRealloc(short, tp->num_Numbers, tp->Numbers);
	    for (last = tp->num_Numbers - 1; last > tindex; last--)
		tp->Numbers[last] = tp->Numbers[last - 1];
	    break;
	case STRING:
	    tp->ext_Strings += 1;
	    tp->num_Strings += 1;
	    tp->Strings = typeRealloc(char *, tp->num_Strings, tp->Strings);
	    for (last = tp->num_Strings - 1; last > tindex; last--)
		tp->Strings[last] = tp->Strings[last - 1];
	    break;
	}
	actual = NUM_EXT_NAMES(tp);
	tp->ext_Names = typeRealloc(char *, actual, tp->ext_Names);
	while (--actual > offset)
	    tp->ext_Names[actual] = tp->ext_Names[actual - 1];
	tp->ext_Names[offset] = _nc_save_str(name);
    }

    temp.nte_name = tp->ext_Names[offset];
    temp.nte_type = token_type;
    temp.nte_index = tindex;
    temp.nte_link = -1;

    return &temp;
}
#endif /* NCURSES_XNAMES */

/*
 *	int
 *	_nc_parse_entry(entry, literal, silent)
 *
 *	Compile one entry.  Doesn't try to resolve use or tc capabilities.
 *
 *	found-forward-use = FALSE
 *	re-initialise internal arrays
 *	get_token();
 *	if the token was not a name in column 1, complain and die
 *	save names in entry's string table
 *	while (get_token() is not EOF and not NAMES)
 *	        check for existence and type-correctness
 *	        enter cap into structure
 *	        if STRING
 *	            save string in entry's string table
 *	push back token
 */

#define BAD_TC_USAGE if (!bad_tc_usage) \
 	{ bad_tc_usage = TRUE; \
	 _nc_warning("Legacy termcap allows only a trailing tc= clause"); }

NCURSES_EXPORT(int)
_nc_parse_entry(struct entry *entryp, int literal, bool silent)
{
    int token_type;
    struct name_table_entry const *entry_ptr;
    char *ptr, *base;
    bool bad_tc_usage = FALSE;

    token_type = _nc_get_token(silent);

    if (token_type == EOF)
	return (EOF);
    if (token_type != NAMES)
	_nc_err_abort("Entry does not start with terminal names in column one");

    _nc_init_entry(&entryp->tterm);

    entryp->cstart = _nc_comment_start;
    entryp->cend = _nc_comment_end;
    entryp->startline = _nc_start_line;
    DEBUG(2, ("Comment range is %ld to %ld", entryp->cstart, entryp->cend));

    /*
     * Strip off the 2-character termcap name, if present.  Originally termcap
     * used that as an indexing aid.  We can retain 2-character terminfo names,
     * but note that they would be lost if we translate to/from termcap.  This
     * feature is supposedly obsolete since "newer" BSD implementations do not
     * use it; however our reference for this feature is SunOS 4.x, which
     * implemented it.  Note that the resulting terminal type was never the
     * 2-character name, but was instead the first alias after that.
     */
    ptr = _nc_curr_token.tk_name;
    if (_nc_syntax == SYN_TERMCAP) {
	if (ptr[2] == '|') {
	    ptr += 3;
	    _nc_curr_token.tk_name[2] = '\0';
	}
    }

    entryp->tterm.str_table = entryp->tterm.term_names = _nc_save_str(ptr);

    DEBUG(1, ("Starting '%s'", ptr));

    /*
     * We do this because the one-token lookahead in the parse loop
     * results in the terminal type getting prematurely set to correspond
     * to that of the next entry.
     */
    _nc_set_type(_nc_first_name(entryp->tterm.term_names));

    /* check for overly-long names and aliases */
    for (base = entryp->tterm.term_names; (ptr = strchr(base, '|')) != 0;
	 base = ptr + 1) {
	if (ptr - base > MAX_ALIAS) {
	    _nc_warning("%s `%.*s' may be too long",
			(base == entryp->tterm.term_names)
			? "primary name"
			: "alias",
			(int) (ptr - base), base);
	}
    }

    entryp->nuses = 0;

    for (token_type = _nc_get_token(silent);
	 token_type != EOF && token_type != NAMES;
	 token_type = _nc_get_token(silent)) {
	bool is_use = (strcmp(_nc_curr_token.tk_name, "use") == 0);
	bool is_tc = !is_use && (strcmp(_nc_curr_token.tk_name, "tc") == 0);
	if (is_use || is_tc) {
	    entryp->uses[entryp->nuses].name = _nc_save_str(_nc_curr_token.tk_valstring);
	    entryp->uses[entryp->nuses].line = _nc_curr_line;
	    entryp->nuses++;
	    if (entryp->nuses > 1 && is_tc) {
		BAD_TC_USAGE
	    }
	} else {
	    /* normal token lookup */
	    entry_ptr = _nc_find_entry(_nc_curr_token.tk_name,
				       _nc_syntax ? _nc_cap_hash_table : _nc_info_hash_table);

	    /*
	     * Our kluge to handle aliasing.  The reason it's done
	     * this ugly way, with a linear search, is so the hashing
	     * machinery doesn't have to be made really complicated
	     * (also we get better warnings this way).  No point in
	     * making this case fast, aliased caps aren't common now
	     * and will get rarer.
	     */
	    if (entry_ptr == NOTFOUND) {
		const struct alias *ap;

		if (_nc_syntax == SYN_TERMCAP) {
		    if (entryp->nuses != 0) {
			BAD_TC_USAGE
		    }
		    for (ap = _nc_capalias_table; ap->from; ap++)
			if (strcmp(ap->from, _nc_curr_token.tk_name) == 0) {
			    if (ap->to == (char *) 0) {
				_nc_warning("%s (%s termcap extension) ignored",
					    ap->from, ap->source);
				goto nexttok;
			    }

			    entry_ptr = _nc_find_entry(ap->to, _nc_cap_hash_table);
			    if (entry_ptr && !silent)
				_nc_warning("%s (%s termcap extension) aliased to %s",
					    ap->from, ap->source, ap->to);
			    break;
			}
		} else {	/* if (_nc_syntax == SYN_TERMINFO) */
		    for (ap = _nc_infoalias_table; ap->from; ap++)
			if (strcmp(ap->from, _nc_curr_token.tk_name) == 0) {
			    if (ap->to == (char *) 0) {
				_nc_warning("%s (%s terminfo extension) ignored",
					    ap->from, ap->source);
				goto nexttok;
			    }

			    entry_ptr = _nc_find_entry(ap->to, _nc_info_hash_table);
			    if (entry_ptr && !silent)
				_nc_warning("%s (%s terminfo extension) aliased to %s",
					    ap->from, ap->source, ap->to);
			    break;
			}

		    if (entry_ptr == NOTFOUND) {
			entry_ptr = lookup_fullname(_nc_curr_token.tk_name);
		    }
		}
	    }
#if NCURSES_XNAMES
	    /*
	     * If we have extended-names active, we will automatically
	     * define a name based on its context.
	     */
	    if (entry_ptr == NOTFOUND
		&& _nc_user_definable
		&& (entry_ptr = _nc_extend_names(entryp,
						 _nc_curr_token.tk_name,
						 token_type)) != 0) {
		if (_nc_tracing >= DEBUG_LEVEL(1))
		    _nc_warning("extended capability '%s'", _nc_curr_token.tk_name);
	    }
#endif /* NCURSES_XNAMES */

	    /* can't find this cap name, not even as an alias */
	    if (entry_ptr == NOTFOUND) {
		if (!silent)
		    _nc_warning("unknown capability '%s'",
				_nc_curr_token.tk_name);
		continue;
	    }

	    /* deal with bad type/value combinations. */
	    if (token_type != CANCEL && entry_ptr->nte_type != token_type) {
		/*
		 * Nasty special cases here handle situations in which type
		 * information can resolve name clashes.  Normal lookup
		 * finds the last instance in the capability table of a
		 * given name, regardless of type.  find_type_entry looks
		 * for a first matching instance with given type.  So as
		 * long as all ambiguous names occur in pairs of distinct
		 * type, this will do the job.
		 */

		/* tell max_attributes from arrow_key_map */
		if (token_type == NUMBER && !strcmp("ma", _nc_curr_token.tk_name))
		    entry_ptr = _nc_find_type_entry("ma", NUMBER,
						    _nc_get_table(_nc_syntax
								  != 0));

		/* map terminfo's string MT to MT */
		else if (token_type == STRING && !strcmp("MT", _nc_curr_token.tk_name))
		    entry_ptr = _nc_find_type_entry("MT", STRING,
						    _nc_get_table(_nc_syntax
								  != 0));

		/* treat strings without following "=" as empty strings */
		else if (token_type == BOOLEAN && entry_ptr->nte_type == STRING)
		    token_type = STRING;
		/* we couldn't recover; skip this token */
		else {
		    if (!silent) {
			const char *type_name;
			switch (entry_ptr->nte_type) {
			case BOOLEAN:
			    type_name = "boolean";
			    break;
			case STRING:
			    type_name = "string";
			    break;
			case NUMBER:
			    type_name = "numeric";
			    break;
			default:
			    type_name = "unknown";
			    break;
			}
			_nc_warning("wrong type used for %s capability '%s'",
				    type_name, _nc_curr_token.tk_name);
		    }
		    continue;
		}
	    }

	    /* now we know that the type/value combination is OK */
	    switch (token_type) {
	    case CANCEL:
		switch (entry_ptr->nte_type) {
		case BOOLEAN:
		    entryp->tterm.Booleans[entry_ptr->nte_index] = CANCELLED_BOOLEAN;
		    break;

		case NUMBER:
		    entryp->tterm.Numbers[entry_ptr->nte_index] = CANCELLED_NUMERIC;
		    break;

		case STRING:
		    entryp->tterm.Strings[entry_ptr->nte_index] = CANCELLED_STRING;
		    break;
		}
		break;

	    case BOOLEAN:
		entryp->tterm.Booleans[entry_ptr->nte_index] = TRUE;
		break;

	    case NUMBER:
		entryp->tterm.Numbers[entry_ptr->nte_index] =
		    _nc_curr_token.tk_valnumber;
		break;

	    case STRING:
		ptr = _nc_curr_token.tk_valstring;
		if (_nc_syntax == SYN_TERMCAP)
		    ptr = _nc_captoinfo(_nc_curr_token.tk_name,
					ptr,
					parametrized[entry_ptr->nte_index]);
		entryp->tterm.Strings[entry_ptr->nte_index] = _nc_save_str(ptr);
		break;

	    default:
		if (!silent)
		    _nc_warning("unknown token type");
		_nc_panic_mode((_nc_syntax == SYN_TERMCAP) ? ':' : ',');
		continue;
	    }
	}			/* end else cur_token.name != "use" */
      nexttok:
	continue;		/* cannot have a label w/o statement */
    }				/* endwhile (not EOF and not NAMES) */

    _nc_push_token(token_type);
    _nc_set_type(_nc_first_name(entryp->tterm.term_names));

    /*
     * Try to deduce as much as possible from extension capabilities
     * (this includes obsolete BSD capabilities).  Sigh...it would be more
     * space-efficient to call this after use resolution, but it has
     * to be done before entry allocation is wrapped up.
     */
    if (!literal) {
	if (_nc_syntax == SYN_TERMCAP) {
	    bool has_base_entry = FALSE;
	    int i;

	    /*
	     * Don't insert defaults if this is a `+' entry meant only
	     * for inclusion in other entries (not sure termcap ever
	     * had these, actually).
	     */
	    if (strchr(entryp->tterm.term_names, '+'))
		has_base_entry = TRUE;
	    else
		/*
		 * Otherwise, look for a base entry that will already
		 * have picked up defaults via translation.
		 */
		for (i = 0; i < entryp->nuses; i++)
		    if (!strchr((char *) entryp->uses[i].name, '+'))
			has_base_entry = TRUE;

	    postprocess_termcap(&entryp->tterm, has_base_entry);
	} else
	    postprocess_terminfo(&entryp->tterm);
    }
    _nc_wrap_entry(entryp, FALSE);

    return (OK);