/*
* sed.c - very minimalist version of sed
*
- * Copyright (C) 1999,2000 by Lineo, inc.
- * Written by Mark Whitley <markw@lineo.com>, <markw@enol.com>
+ * Copyright (C) 1999,2000,2001 by Lineo, inc.
+ * Written by Mark Whitley <markw@lineo.com>, <markw@codepoet.org>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
Unsupported features:
- transliteration (y/source-chars/dest-chars/) (use 'tr')
- - no support for characters other than the '/' character for regex matches
- no pattern space hold space storing / swapping (x, etc.)
- no labels / branching (: label, b, t, and friends)
- and lots, lots more.
-
*/
#include <stdio.h>
#include <string.h> /* for strdup() */
#include <errno.h>
#include <ctype.h> /* for isspace() */
+#include <stdlib.h>
#include "busybox.h"
-#define bb_need_full_version
-#define BB_DECLARE_EXTERN
-#include "messages.c"
-
/* externs */
+extern void xregcomp(regex_t *preg, const char *regex, int cflags);
extern int optind; /* in unistd.h */
extern char *optarg; /* ditto */
/* options */
static int be_quiet = 0;
-static const int SUB_G = 1 << 0;
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
-static const int SUB_P = 1 << 1;
-#endif
struct sed_cmd {
+
+ /* GENERAL FIELDS */
+ char delimiter; /* The delimiter used to separate regexps */
+
/* address storage */
int beg_line; /* 'sed 1p' 0 == no begining line, apply commands to all lines */
int end_line; /* 'sed 1,3p' 0 == no end line, use only beginning. -1 == $ */
/* the command */
char cmd; /* p,d,s (add more at your leisure :-) */
- /* substitution command specific fields */
- regex_t *sub_match; /* sed -e 's/sub_match/replace/' */
- char *replace; /* sed -e 's/sub_match/replace/' XXX: who will hold the \1 \2 \3s? */
+
+ /* SUBSTITUTION COMMAND SPECIFIC FIELDS */
+
+ /* sed -e 's/sub_match/replace/' */
+ regex_t *sub_match;
+ char *replace;
unsigned int num_backrefs:4; /* how many back references (\1..\9) */
/* Note: GNU/POSIX sed does not save more than nine backrefs, so
* we only use 4 bits to hold the number */
-#ifndef BB_FEATURE_SED_PATTERN_SPACE
- unsigned int sub_flags:1; /* sed -e 's/foo/bar/g' (global) */
-#else
- unsigned int sub_flags:2; /* sed -e 's/foo/bar/gp' (global/pattern) */
-#endif
+ unsigned int sub_g:1; /* sed -e 's/foo/bar/g' (global) */
+ unsigned int sub_p:2; /* sed -e 's/foo/bar/p' (print substitution) */
+
+
+ /* EDIT COMMAND (a,i,c) SPEICIFIC FIELDS */
- /* edit command (a,i,c) speicific field */
char *editline;
};
}
#endif
-#if 0
-/*
- * trim_str - trims leading and trailing space from a string
- *
- * Note: This returns a malloc'ed string so you must store and free it
- * XXX: This should be in the utility.c file.
- * XXX: This is now obsolete. Maybe it belongs nowhere.
- */
-static char *trim_str(const char *str)
-{
- int i;
- char *retstr = strdup(str);
- /* trim leading whitespace */
- memmove(retstr, &retstr[strspn(retstr, " \n\t\v")], strlen(retstr));
-
- /* trim trailing whitespace */
- i = strlen(retstr) - 1;
- while (isspace(retstr[i]))
- i--;
- retstr[++i] = 0;
-
- /* Aside:
- *
- * you know, a strrspn() would really be nice cuz then we could say:
- *
- * retstr[strrspn(retstr, " \n\t\v") + 1] = 0;
- */
-
- return retstr;
-}
-#endif
-
-#if 0
/*
- * strrspn - works just like strspn() but goes from right to left instead of
- * left to right
+ * index_of_next_unescaped_regexp_delim - walks left to right through a string
+ * beginning at a specified index and returns the index of the next regular
+ * expression delimiter (typically a forward * slash ('/')) not preceeded by
+ * a backslash ('\').
*/
-static size_t strrspn(const char *s, const char *accept)
-{
- size_t i = strlen(s);
-
- while (strchr(accept, s[--i]))
- ;
-
- return i;
-}
-#endif
-
-/*
- * index_of_next_unescaped_slash - walks left to right through a string
- * beginning at a specified index and returns the index of the next forward
- * slash ('/') not preceeded by a backslash ('\').
- */
-static int index_of_next_unescaped_slash(const char *str, int idx)
+static int index_of_next_unescaped_regexp_delim(struct sed_cmd *sed_cmd, const char *str, int idx)
{
for ( ; str[idx]; idx++) {
- if (str[idx] == '/' && str[idx-1] != '\\')
+ if (str[idx] == sed_cmd->delimiter && str[idx-1] != '\\')
return idx;
}
/*
* returns the index in the string just past where the address ends.
*/
-static int get_address(const char *str, int *line, regex_t **regex)
+static int get_address(struct sed_cmd *sed_cmd, const char *str, int *line, regex_t **regex)
{
char *my_str = strdup(str);
int idx = 0;
idx++;
}
else if (my_str[idx] == '/') {
- idx = index_of_next_unescaped_slash(my_str, ++idx);
+ idx = index_of_next_unescaped_regexp_delim(sed_cmd, my_str, ++idx);
if (idx == -1)
- fatalError("unterminated match expression\n");
+ error_msg_and_die("unterminated match expression");
my_str[idx] = '\0';
*regex = (regex_t *)xmalloc(sizeof(regex_t));
- xregcomp(*regex, my_str+1, REG_NEWLINE);
+ xregcomp(*regex, my_str+1, 0);
idx++; /* so it points to the next character after the last '/' */
}
else {
- errorMsg("get_address: no address found in string\n"
- "\t(you probably didn't check the string you passed me)\n");
+ error_msg("get_address: no address found in string\n"
+ "\t(you probably didn't check the string you passed me)");
idx = -1;
}
* (all three of the '/' slashes are mandatory)
*/
- /* verify that the 's' is followed by a 'slash' */
- if (substr[++idx] != '/')
- fatalError("bad format in substitution expression\n");
+ /* verify that the 's' is followed by something. That something
+ * (typically a 'slash') is now our regexp delimiter... */
+ if (!substr[++idx])
+ error_msg_and_die("bad format in substitution expression");
+ else
+ sed_cmd->delimiter=substr[idx];
/* save the match string */
oldidx = idx+1;
- idx = index_of_next_unescaped_slash(substr, ++idx);
+ idx = index_of_next_unescaped_regexp_delim(sed_cmd, substr, ++idx);
if (idx == -1)
- fatalError("bad format in substitution expression\n");
+ error_msg_and_die("bad format in substitution expression");
match = strdup_substr(substr, oldidx, idx);
/* determine the number of back references in the match string */
/* sed_cmd->num_backrefs = 0; */ /* XXX: not needed? --apparently not */
for (j = 0; match[j]; j++) {
/* GNU/POSIX sed does not save more than nine backrefs */
- if (match[j] == '\\' && match[j+1] == '(' && sed_cmd->num_backrefs < 9)
+ if (match[j] == '\\' && match[j+1] == '(' && sed_cmd->num_backrefs <= 9)
sed_cmd->num_backrefs++;
}
/* save the replacement string */
oldidx = idx+1;
- idx = index_of_next_unescaped_slash(substr, ++idx);
+ idx = index_of_next_unescaped_regexp_delim(sed_cmd, substr, ++idx);
if (idx == -1)
- fatalError("bad format in substitution expression\n");
+ error_msg_and_die("bad format in substitution expression");
sed_cmd->replace = strdup_substr(substr, oldidx, idx);
/* process the flags */
while (substr[++idx]) {
switch (substr[idx]) {
case 'g':
- sed_cmd->sub_flags = SUB_G;
+ sed_cmd->sub_g = 1;
break;
case 'I':
cflags |= REG_ICASE;
break;
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
case 'p':
- sed_cmd->sub_flags = SUB_P;
+ sed_cmd->sub_p = 1;
break;
-#endif
default:
/* any whitespace or semicolon trailing after a s/// is ok */
if (strchr("; \t\v\n\r", substr[idx]))
goto out;
/* else */
- fatalError("bad option in substitution expression\n");
+ error_msg_and_die("bad option in substitution expression");
}
}
*/
if (editstr[1] != '\\' && (editstr[2] != '\n' || editstr[2] != '\r'))
- fatalError("bad format in edit expression\n");
+ error_msg_and_die("bad format in edit expression");
/* store the edit line text */
/* make editline big enough to accomodate the extra '\n' we will tack on
/* first part (if present) is an address: either a number or a /regex/ */
if (isdigit(cmdstr[idx]) || cmdstr[idx] == '/')
- idx = get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
+ idx = get_address(sed_cmd, cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
/* second part (if present) will begin with a comma */
if (cmdstr[idx] == ',')
- idx += get_address(&cmdstr[++idx], &sed_cmd->end_line, &sed_cmd->end_match);
+ idx += get_address(sed_cmd, &cmdstr[++idx], &sed_cmd->end_line, &sed_cmd->end_match);
/* last part (mandatory) will be a command */
if (cmdstr[idx] == '\0')
- fatalError("missing command\n");
+ error_msg_and_die("missing command");
if (!strchr("pdsaic", cmdstr[idx])) /* <-- XXX add new commands here */
- fatalError("invalid command\n");
+ error_msg_and_die("invalid command");
sed_cmd->cmd = cmdstr[idx];
/* special-case handling for (s)ubstitution */
/* special-case handling for (a)ppend, (i)nsert, and (c)hange */
else if (strchr("aic", cmdstr[idx])) {
if (sed_cmd->end_line || sed_cmd->end_match)
- fatalError("only a beginning address can be specified for edit commands\n");
+ error_msg_and_die("only a beginning address can be specified for edit commands");
idx += parse_edit_cmd(sed_cmd, &cmdstr[idx]);
}
/* if it was a single-letter command (such as 'p' or 'd') we need to
}
/* eat trailing newline (if any) --if I don't do this, edit commands
* (aic) will print an extra newline */
- if (line[strlen(line)-1] == '\n')
- line[strlen(line)-1] = 0;
+ chomp(line);
add_cmd_str(line);
free(line);
}
static int do_subst_command(const struct sed_cmd *sed_cmd, const char *line)
{
+ char *hackline = (char *)line;
int altered = 0;
+ regmatch_t *regmatch = NULL;
- /* we only substitute if the substitution 'search' expression matches */
- if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == 0) {
- regmatch_t *regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
- int i;
- char *ptr = (char *)line;
-
- while (*ptr) {
- /* if we can match the search string... */
- if (regexec(sed_cmd->sub_match, ptr, sed_cmd->num_backrefs+1, regmatch, 0) == 0) {
- /* print everything before the match, */
- for (i = 0; i < regmatch[0].rm_so; i++) {
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
- if(!be_quiet || (sed_cmd->sub_flags & SUB_P))
-#endif
- fputc(ptr[i], stdout);
- }
+ /* we only proceed if the substitution 'search' expression matches */
+ if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == REG_NOMATCH)
+ return 0;
- /* then print the substitution in its place */
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
- if(!be_quiet || (sed_cmd->sub_flags & SUB_P))
-#endif
- print_subst_w_backrefs(ptr, sed_cmd->replace, regmatch);
+ /* whaddaya know, it matched. get the number of back references */
+ regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
- /* then advance past the match */
- ptr += regmatch[0].rm_eo;
+ /* and now, as long as we've got a line to try matching and if we can match
+ * the search string, we make substitutions */
+ while (*hackline && (regexec(sed_cmd->sub_match, hackline,
+ sed_cmd->num_backrefs+1, regmatch, 0) == 0) ) {
+ int i;
- /* and flag that something has changed */
- altered++;
+ /* print everything before the match */
+ for (i = 0; i < regmatch[0].rm_so; i++)
+ fputc(hackline[i], stdout);
- /* if we're not doing this globally... */
- if (!sed_cmd->sub_flags & SUB_G)
- break;
- }
- /* if we COULD NOT match the search string (meaning we've gone past
- * all previous instances), get out */
- else
- break;
- }
+ /* then print the substitution string */
+ print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch);
- /* is there anything left to print? */
-#ifdef BB_FEATURE_SED_PATTERN_SPACE
- if (*ptr && (!be_quiet || sed_cmds->sub_flags & SUB_P))
-#else
- if (*ptr)
-#endif
- fputs(ptr, stdout);
+ /* advance past the match */
+ hackline += regmatch[0].rm_eo;
+ /* flag that something has changed */
+ altered++;
- /* cleanup */
- free(regmatch);
+ /* if we're not doing this globally, get out now */
+ if (!sed_cmd->sub_g)
+ break;
}
+ /* if there's anything left of the line, print it */
+ if (*hackline)
+ fputs(hackline, stdout);
+
+ /* cleanup */
+ free(regmatch);
+
return altered;
}
break;
case 's':
- altered = do_subst_command(sed_cmd, line);
+
+ /*
+ * Some special cases for 's' printing to make it compliant with
+ * GNU sed printing behavior (aka "The -n | s///p Matrix"):
+ *
+ * -n ONLY = never print anything regardless of any successful
+ * substitution
+ *
+ * s///p ONLY = always print successful substitutions, even if
+ * the line is going to be printed anyway (line will be printed
+ * twice).
+ *
+ * -n AND s///p = print ONLY a successful substitution ONE TIME;
+ * no other lines are printed - this is the reason why the 'p'
+ * flag exists in the first place.
+ */
+
+ /* if the user specified that they didn't want anything printed (i.e. a -n
+ * flag and no 'p' flag after the s///), then there's really no point doing
+ * anything here. */
+ if (be_quiet && !sed_cmd->sub_p)
+ break;
+
+ /* we print the line once, unless we were told to be quiet */
+ if (!be_quiet)
+ altered = do_subst_command(sed_cmd, line);
+
+ /* we also print the line if we were given the 'p' flag
+ * (this is quite possibly the second printing) */
+ if (sed_cmd->sub_p)
+ altered = do_subst_command(sed_cmd, line);
+
break;
case 'a':
/* are we acting on a range of line numbers? */
else if (sed_cmds[i].beg_line > 0 && sed_cmds[i].end_line != 0) {
- if (linenum >= sed_cmds[i].beg_line && (sed_cmds[i].end_line == -1 || linenum <= sed_cmds[i].end_line))
+ if (linenum >= sed_cmds[i].beg_line &&
+ (sed_cmds[i].end_line == -1 || linenum <= sed_cmds[i].end_line))
line_altered += do_sed_command(&sed_cmds[i], line);
}
}
- /* we will print the line unless we were told to be quiet or if
- * the line was altered (via a 'd'elete or 's'ubstitution) */
-#ifndef BB_FEATURE_SED_PATTERN_SPACE
- if (!be_quiet &&!line_altered)
-#else
- /* we where specificly requested to print the output */
- if ((!be_quiet || (sed_cmds[i].sub_flags & SUB_P)) && !line_altered)
-#endif
+ /* we will print the line unless we were told to be quiet or if the
+ * line was altered (via a 'd'elete or 's'ubstitution), in which case
+ * the altered line was already printed */
+ if (!be_quiet && !line_altered)
fputs(line, stdout);
free(line);
#ifdef BB_FEATURE_CLEAN_UP
/* destroy command strings on exit */
- if (atexit(destroy_cmd_strs) == -1) {
- perror("sed");
- exit(1);
- }
+ if (atexit(destroy_cmd_strs) == -1)
+ perror_msg_and_die("atexit");
#endif
/* do normal option parsing */
- while ((opt = getopt(argc, argv, "hne:f:")) > 0) {
+ while ((opt = getopt(argc, argv, "ne:f:")) > 0) {
switch (opt) {
- case 'h':
- usage(sed_usage);
- break;
case 'n':
be_quiet++;
break;
case 'f':
load_cmd_file(optarg);
break;
+ default:
+ show_usage();
}
}
* argv[optind] should be the pattern. no pattern, no worky */
if (ncmds == 0) {
if (argv[optind] == NULL)
- usage(sed_usage);
+ show_usage();
else {
add_cmd_str(argv[optind]);
optind++;
for (i = optind; i < argc; i++) {
file = fopen(argv[i], "r");
if (file == NULL) {
- errorMsg("%s: %s\n", argv[i], strerror(errno));
+ perror_msg("%s", argv[i]);
} else {
process_file(file);
fclose(file);