84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Copyright 2010 Nexenta Systems, Inc. All rights reserved.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Copyright (c) 1992 Diomidis Spinellis.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Copyright (c) 1992, 1993
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * The Regents of the University of California. All rights reserved.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * This code is derived from software contributed to Berkeley by
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Diomidis Spinellis of Imperial College, University of London.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Redistribution and use in source and binary forms, with or without
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * modification, are permitted provided that the following conditions
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * are met:
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * 1. Redistributions of source code must retain the above copyright
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * notice, this list of conditions and the following disclaimer.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * 2. Redistributions in binary form must reproduce the above copyright
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * notice, this list of conditions and the following disclaimer in the
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * documentation and/or other materials provided with the distribution.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * 4. Neither the name of the University nor the names of its contributors
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * may be used to endorse or promote products derived from this software
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * without specific prior written permission.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * SUCH DAMAGE.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <sys/types.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <sys/stat.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <ctype.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <err.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <errno.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <fcntl.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <limits.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <regex.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <stdio.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <stdlib.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <string.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <wchar.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <libintl.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include <note.h>
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include "defs.h"
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#include "extern.h"
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#define LHSZ 128
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#define LHMASK (LHSZ - 1)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic struct labhash {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore struct labhash *lh_next;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore uint_t lh_hash;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore struct s_command *lh_cmd;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int lh_ref;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore} *labels[LHSZ];
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *compile_addr(char *, struct s_addr *);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *compile_ccl(char **, char *);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *compile_delimited(char *, char *, int);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *compile_flags(char *, struct s_subst *);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic regex_t *compile_re(char *, int);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *compile_subst(char *, struct s_subst *);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *compile_text(void);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *compile_tr(char *, struct s_tr **);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic struct s_command
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore **compile_stream(struct s_command **);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *duptoeol(char *, const char *);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic void enterlabel(struct s_command *);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic struct s_command
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *findlabel(char *);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic void fixuplabel(struct s_command *, struct s_command *);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic void uselabel(void);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Command specification. This is used to drive the command parser.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestruct s_format {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char code; /* Command code */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int naddr; /* Number of address args */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore enum e_args args; /* Argument type */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore};
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic struct s_format cmd_fmts[] = {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'{', 2, GROUP},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'}', 0, ENDGROUP},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'a', 1, TEXT},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'b', 2, BRANCH},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'c', 2, TEXT},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'d', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'D', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'g', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'G', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'h', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'H', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'i', 1, TEXT},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'l', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'n', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'N', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'p', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'P', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'q', 1, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'r', 1, RFILE},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'s', 2, SUBST},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'t', 2, BRANCH},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'w', 2, WFILE},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'x', 2, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'y', 2, TR},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'!', 2, NONSEL},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {':', 0, LABEL},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'#', 0, COMMENT},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'=', 1, EMPTY},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore {'\0', 0, COMMENT},
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore};
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/* The compiled program. */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestruct s_command *prog;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Compile the program into prog.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Initialise appends.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorevoid
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorecompile(void)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *compile_stream(&prog) = NULL;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fixuplabel(prog, NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore uselabel();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (appendnum == 0)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore appends = NULL;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else if ((appends = malloc(sizeof (struct s_appends) * appendnum)) ==
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((match = malloc((maxnsub + 1) * sizeof (regmatch_t))) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#define EATSPACE() do { \
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (p) \
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore while (*p && isspace((unsigned char)*p)) \
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++; \
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore _NOTE(CONSTCOND) \
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore} while (0)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic struct s_command **
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorecompile_stream(struct s_command **link)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char *p;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore struct s_command *cmd, *cmd2, *stack;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore struct s_format *fp;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char re[_POSIX2_LINE_MAX + 1];
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int naddr; /* Number of addresses */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore stack = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (;;) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((p = cu_fgets(lbuf, sizeof (lbuf), NULL)) == NULL) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (stack != 0)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("unexpected EOF (pending }'s)"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (link);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amoresemicolon: EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (p) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '#' || *p == '\0')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore continue;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else if (*p == ';') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore goto semicolon;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((*link = cmd = malloc(sizeof (struct s_command))) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore link = &cmd->next;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->startline = cmd->nonsel = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /* First parse the addresses */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore naddr = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/* Valid characters to start an address */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#define addrchar(c) (strchr("0123456789/\\$", (c)))
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (addrchar(*p)) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore naddr++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((cmd->a1 = malloc(sizeof (struct s_addr))) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = compile_addr(p, cmd->a1);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE(); /* EXTENSION */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == ',') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE(); /* EXTENSION */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore naddr++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((cmd->a2 = malloc(sizeof (struct s_addr)))
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = compile_addr(p, cmd->a2);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->a2 = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->a1 = cmd->a2 = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorenonsel: /* Now parse the command */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (!*p)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("command expected"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->code = *p;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (fp = cmd_fmts; fp->code; fp++)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (fp->code == *p)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (!fp->code)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("invalid command code %c"), *p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (naddr > fp->naddr)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("command %c expects up to %d address(es), "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "found %d"), *p, fp->naddr, naddr);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore switch (fp->args) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case NONSEL: /* ! */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->nonsel = 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore goto nonsel;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case GROUP: /* { */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->next = stack;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore stack = cmd;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore link = &cmd->u.c;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore goto semicolon;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case ENDGROUP:
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Short-circuit command processing, since end of
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * group is really just a noop.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->nonsel = 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (stack == 0)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("unexpected }"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd2 = stack;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore stack = cmd2->next;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd2->next = cmd;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /*FALLTHROUGH*/
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case EMPTY: /* d D g G h H l n N p P q x = \0 */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == ';') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore link = &cmd->next;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore goto semicolon;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("extra characters at the end of %c "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "command"), cmd->code);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case TEXT: /* a c i */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p != '\\')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("command %c expects \\ "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "followed by text"), cmd->code);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("extra characters after \\ "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "at the end of %c command"),
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->code);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->t = compile_text();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case COMMENT: /* \0 # */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case WFILE: /* w */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '\0')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("filename expected"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->t = duptoeol(p, "w command");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (aflag)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->u.fd = -1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else if ((cmd->u.fd = open(p,
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 0666)) == -1)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "%s", p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case RFILE: /* r */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '\0')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("filename expected"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->t = duptoeol(p, "read command");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case BRANCH: /* b t */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '\0')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->t = NULL;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->t = duptoeol(p, "branch");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case LABEL: /* : */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->t = duptoeol(p, "label");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (strlen(p) == 0)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("empty label"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore enterlabel(cmd);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case SUBST: /* s */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '\0' || *p == '\\')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("substitute pattern can not "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "be delimited by newline or backslash"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((cmd->u.s = calloc(1, sizeof (struct s_subst))) ==
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = compile_delimited(p, re, 0);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (p == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("unterminated substitute pattern"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /* Compile RE with no case sensitivity temporarily */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*re == '\0')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->u.s->re = NULL;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->u.s->re = compile_re(re, 0);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore --p;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = compile_subst(p, cmd->u.s);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = compile_flags(p, cmd->u.s);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /* Recompile RE with case sens. from "I" flag if any */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*re == '\0')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->u.s->re = NULL;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cmd->u.s->re = compile_re(re, cmd->u.s->icase);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == ';') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore link = &cmd->next;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore goto semicolon;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case TR: /* y */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = compile_tr(p, &cmd->u.y);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == ';') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore link = &cmd->next;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore goto semicolon;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("extra text at the end of a "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "transform command"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Get a delimited string. P points to the delimeter of the string; d points
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * to a buffer area. Newline and delimiter escapes are processed; other
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * escapes are ignored.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Returns a pointer to the first character after the final delimiter or NULL
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * in the case of a non-terminated string. The character array d is filled
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * with the processed string.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorecompile_delimited(char *p, char *d, int is_tr)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char c;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore c = *p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (c == '\0')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else if (c == '\\')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("\\ can not be used as a string delimiter"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else if (c == '\n')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("newline can not be used as a string delimiter"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore while (*p) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '[' && *p != c) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((d = compile_ccl(&p, d)) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("unbalanced brackets ([])"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore continue;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else if (*p == '\\' && p[1] == '[') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *d++ = *p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else if (*p == '\\' && p[1] == c)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else if (*p == '\\' && p[1] == 'n') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *d++ = '\n';
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p += 2;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore continue;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else if (*p == '\\' && p[1] == '\\') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (is_tr)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *d++ = *p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else if (*p == c) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *d = '\0';
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (p + 1);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *d++ = *p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/* compile_ccl: expand a POSIX character class */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorecompile_ccl(char **sp, char *t)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int c, d;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char *s = *sp;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *t++ = *s++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*s == '^')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *t++ = *s++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*s == ']')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *t++ = *s++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (; *s && (*t = *s) != ']'; s++, t++)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*s == '[' &&
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *++t = *++s, t++, s++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((c = *s) == '\0')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return ((*s == ']') ? *sp = ++s, ++t : NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Compiles the regular expression in RE and returns a pointer to the compiled
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * regular expression.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Cflags are passed to regcomp.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic regex_t *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorecompile_re(char *re, int case_insensitive)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore regex_t *rep;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int eval, flags;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore flags = rflags;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (case_insensitive)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore flags |= REG_ICASE;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((rep = malloc(sizeof (regex_t))) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((eval = regcomp(rep, re, flags)) != 0)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("RE error: %s"), strregerror(eval, rep));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (maxnsub < rep->re_nsub)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore maxnsub = rep->re_nsub;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (rep);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Compile the substitution string of a regular expression and set res to
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * point to a saved copy of it. Nsub is the number of parenthesized regular
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * expressions.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorecompile_subst(char *p, struct s_subst *s)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore static char lbuf[_POSIX2_LINE_MAX + 1];
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int asize;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore uintptr_t size;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore uchar_t ref;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char c, *text, *op, *sp;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int more = 1, sawesc = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore c = *p++; /* Terminator character */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (c == '\0')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->maxbref = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->linenum = linenum;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore asize = 2 * _POSIX2_LINE_MAX + 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((text = malloc(asize)) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore size = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore do {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore op = sp = text + size;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (; *p; p++) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '\\' || sawesc) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * If this is a continuation from the last
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * buffer, we won't have a character to
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * skip over.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (sawesc)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore sawesc = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '\0') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * This escaped character is continued
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * in the next part of the line. Note
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * this fact, then cause the loop to
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * exit w/ normal EOL case and reenter
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * above with the new buffer.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore sawesc = 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p--;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore continue;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else if (strchr("123456789", *p) != NULL) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *sp++ = '\\';
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore ref = *p - '0';
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (s->re != NULL &&
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore ref > s->re->re_nsub)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("not defined in "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "the RE: \\%c"), *p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (s->maxbref < ref)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->maxbref = ref;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else if (*p == '&' || *p == '\\')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *sp++ = '\\';
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else if (*p == c) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*++p == '\0' && more) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (cu_fgets(lbuf, sizeof (lbuf),
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore &more))
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = lbuf;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *sp++ = '\0';
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore size += (uintptr_t)sp - (uintptr_t)op;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((s->new = realloc(text, size)) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "realloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else if (*p == '\n') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("unescaped newline inside "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "substitute pattern"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /* NOTREACHED */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *sp++ = *p;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore size += (uintptr_t)sp - (uintptr_t)op;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (asize - size < _POSIX2_LINE_MAX + 1) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore asize *= 2;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((text = realloc(text, asize)) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "realloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } while (cu_fgets(p = lbuf, sizeof (lbuf), &more));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("unterminated substitute in regular expression"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Compile the flags of the s command
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorecompile_flags(char *p, struct s_subst *s)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int gn; /* True if we have seen g or n */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore unsigned long nval;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char wfile[_POSIX2_LINE_MAX + 1], *q;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->n = 1; /* Default */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->p = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->wfile = NULL;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->wfd = -1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->icase = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore gn = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (;;) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE(); /* EXTENSION */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore switch (*p) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case 'g':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (gn)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("more than one number or "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "'g' in substitute flags"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore gn = 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->n = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '\0':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '\n':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case ';':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case 'p':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->p = 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case 'I':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->icase = 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '1': case '2': case '3':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '4': case '5': case '6':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '7': case '8': case '9':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (gn)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("more than one number or "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "'g' in substitute flags"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore gn = 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore errno = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore nval = strtol(p, &p, 10);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (errno == ERANGE || nval > INT_MAX)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("overflow in the 'N' substitute flag"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->n = nval;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p--;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case 'w':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#ifdef HISTORIC_PRACTICE
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p != ' ') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("space missing before w wfile"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore#endif
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore q = wfile;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore while (*p) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '\n')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *q++ = *p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *q = '\0';
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (q == wfile)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("no wfile specified"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore s->wfile = strdup(wfile);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (!aflag && (s->wfd = open(wfile,
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 0666)) == -1)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "%s", wfile);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore default:
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("bad flag in substitute command: '%c'"), *p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Compile a translation set of strings into a lookup table.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorecompile_tr(char *p, struct s_tr **py)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore struct s_tr *y;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int i;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore const char *op, *np;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char old[_POSIX2_LINE_MAX + 1];
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char new[_POSIX2_LINE_MAX + 1];
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore size_t oclen, oldlen, nclen, newlen;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore mbstate_t mbs1, mbs2;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((*py = y = malloc(sizeof (*y))) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore y->multis = NULL;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore y->nmultis = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '\0' || *p == '\\')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("transform pattern can not be delimited by "
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore "newline or backslash"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = compile_delimited(p, old, 1);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (p == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("unterminated transform source string"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = compile_delimited(p - 1, new, 1);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (p == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("unterminated transform target string"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore op = old;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore oldlen = mbsrtowcs(NULL, &op, 0, NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (oldlen == (size_t)-1)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore np = new;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore newlen = mbsrtowcs(NULL, &np, 0, NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (newlen == (size_t)-1)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (newlen != oldlen)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("transform strings are not the same length"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (MB_CUR_MAX == 1) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * The single-byte encoding case is easy: generate a
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * lookup table.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (i = 0; i <= UCHAR_MAX; i++)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore y->bytetab[i] = (char)i;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (; *op; op++, np++)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore y->bytetab[(uchar_t)*op] = *np;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore } else {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Multi-byte encoding case: generate a lookup table as
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * above, but only for single-byte characters. The first
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * bytes of multi-byte characters have their lookup table
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * entries set to 0, which causes do_tr() to search through
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * an auxiliary vector of multi-byte mappings.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore (void) memset(&mbs1, 0, sizeof (mbs1));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore (void) memset(&mbs2, 0, sizeof (mbs2));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (i = 0; i <= UCHAR_MAX; i++)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore y->bytetab[i] = (btowc(i) != WEOF) ? (uchar_t)i : 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore while (*op != '\0') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore oclen = mbrlen(op, MB_LEN_MAX, &mbs1);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (oclen == (size_t)-1 || oclen == (size_t)-2)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore errx(1, "%s", strerror(EILSEQ));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore nclen = mbrlen(np, MB_LEN_MAX, &mbs2);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (nclen == (size_t)-1 || nclen == (size_t)-2)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore errx(1, "%s", strerror(EILSEQ));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (oclen == 1 && nclen == 1)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore y->bytetab[(uchar_t)*op] = *np;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore y->bytetab[(uchar_t)*op] = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore y->multis = realloc(y->multis,
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore (y->nmultis + 1) * sizeof (*y->multis));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (y->multis == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore i = y->nmultis++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore y->multis[i].fromlen = oclen;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore (void) memcpy(y->multis[i].from, op, oclen);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore y->multis[i].tolen = nclen;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore (void) memcpy(y->multis[i].to, np, nclen);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore op += oclen;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore np += nclen;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Compile the text following an a or i command.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorecompile_text(void)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int esc_nl;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore uintptr_t size, asize;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char *text, *p, *op, *s;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char lbuf[_POSIX2_LINE_MAX + 1];
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore asize = 2 * _POSIX2_LINE_MAX + 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((text = malloc(asize)) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore size = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore while (cu_fgets(lbuf, sizeof (lbuf), NULL)) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore op = s = text + size;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = lbuf;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore EATSPACE();
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (esc_nl = 0; *p != '\0'; p++) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == '\\' && p[1] != '\0' && *++p == '\n')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore esc_nl = 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *s++ = *p;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore size += (uintptr_t)s - (uintptr_t)op;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (!esc_nl) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *s = '\0';
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (asize - size < _POSIX2_LINE_MAX + 1) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore asize *= 2;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((text = realloc(text, asize)) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "realloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore text[size] = '\0';
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((p = realloc(text, size + 1)) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "realloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Get an address and return a pointer to the first character after
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * it. Fill the structure pointed to according to the address.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorecompile_addr(char *p, struct s_addr *a)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char *end, re[_POSIX2_LINE_MAX + 1];
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int icase;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore icase = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore a->type = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore switch (*p) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '\\': /* Context address */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore ++p;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /* FALLTHROUGH */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '/': /* Context address */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p = compile_delimited(p, re, 0);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (p == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("unterminated regular expression"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /* Check for case insensitive regexp flag */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*p == 'I') {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore icase = 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (*re == '\0')
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore a->u.r = NULL;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore else
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore a->u.r = compile_re(re, icase);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore a->type = AT_RE;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (p);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '$': /* Last line */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore a->type = AT_LAST;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (p + 1);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '+': /* Relative line number */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore a->type = AT_RELLINE;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore p++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /* FALLTHROUGH */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /* Line number */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '0': case '1': case '2': case '3': case '4':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '5': case '6': case '7': case '8': case '9':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (a->type == 0)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore a->type = AT_LINE;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore a->u.l = strtol(p, &end, 10);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (end);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore default:
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("expected context address"));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * duptoeol --
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Return a copy of all the characters up to \n or \0.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic char *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amoreduptoeol(char *s, const char *ctype)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore size_t len;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int ws;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore char *p, *start;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore ws = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (start = s; *s != '\0' && *s != '\n'; ++s)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore ws = isspace((unsigned char)*s);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *s = '\0';
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (ws)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore warnx(_("%lu: %s: whitespace after %s"), linenum, fname, ctype);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore len = (uintptr_t)s - (uintptr_t)start + 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((p = malloc(len)) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (memmove(p, start, len));
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Convert goto label names to addresses, and count a and r commands, in
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * the given subset of the script. Free the memory used by labels in b
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * and t commands (but not by :).
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * TODO: Remove } nodes
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic void
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorefixuplabel(struct s_command *cp, struct s_command *end)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (; cp != end; cp = cp->next)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore switch (cp->code) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case 'a':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case 'r':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore appendnum++;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case 'b':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case 't':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /* Resolve branch target. */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (cp->t == NULL) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore cp->u.c = NULL;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((cp->u.c = findlabel(cp->t)) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("undefined label '%s'"), cp->t);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore free(cp->t);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore case '{':
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore /* Do interior commands. */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fixuplabel(cp->u.c, cp->next);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore break;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Associate the given command label for later lookup.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic void
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amoreenterlabel(struct s_command *cp)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore struct labhash **lhp, *lh;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore uchar_t *p;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore uint_t h, c;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (h = 0, p = (uchar_t *)cp->t; (c = *p) != 0; p++)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore h = (h << 5) + h + c;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore lhp = &labels[h & LHMASK];
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (lh = *lhp; lh != NULL; lh = lh->lh_next)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore fatal(_("duplicate label '%s'"), cp->t);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if ((lh = malloc(sizeof (*lh))) == NULL)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore err(1, "malloc");
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore lh->lh_next = *lhp;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore lh->lh_hash = h;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore lh->lh_cmd = cp;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore lh->lh_ref = 0;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore *lhp = lh;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Find the label contained in the command l in the command linked
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * list cp. L is excluded from the search. Return NULL if not found.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic struct s_command *
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorefindlabel(char *name)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore struct labhash *lh;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore uchar_t *p;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore uint_t h, c;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (h = 0, p = (uchar_t *)name; (c = *p) != 0; p++)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore h = (h << 5) + h + c;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore lh->lh_ref = 1;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (lh->lh_cmd);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore return (NULL);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore/*
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * Warn about any unused labels. As a side effect, release the label hash
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore * table space.
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore */
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amorestatic void
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amoreuselabel(void)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore{
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore struct labhash *lh, *next;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore int i;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (i = 0; i < LHSZ; i++) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore for (lh = labels[i]; lh != NULL; lh = next) {
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore next = lh->lh_next;
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore if (!lh->lh_ref)
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore warnx(_("%lu: %s: unused label '%s'"),
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore linenum, fname, lh->lh_cmd->t);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore free(lh);
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore }
84441f85b19f6b8080883f30109e58e43c893709Garrett D'Amore}