- address matching: num|/matchstr/[,num|/matchstr/|$]command
- commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
- edit commands: (a)ppend, (i)nsert, (c)hange
+ - backreferences in substitution expressions (\1, \2...\9)
(Note: Specifying an address (range) to match is *optional*; commands
default to the whole pattern space if no specific address match was
*/
#include <stdio.h>
-#include <stdlib.h> /* for realloc() */
#include <unistd.h> /* for getopt() */
#include <regex.h>
#include <string.h> /* for strdup() */
#include <ctype.h> /* for isspace() */
#include "internal.h"
+#define bb_need_full_version
+#define BB_DECLARE_EXTERN
+#include "messages.c"
/* externs */
extern int optind; /* in unistd.h */
/* 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 {
/* address storage */
/* 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? */
- unsigned int sub_g:1; /* sed -e 's/foo/bar/g' (global) */
+ 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
/* edit command (a,i,c) speicific field */
char *editline;
/*static char *cur_file = NULL;*/ /* file currently being processed XXX: do I need this? */
-static const char sed_usage[] =
- "sed [-Vhnef] pattern [files...]\n"
-#ifndef BB_FEATURE_TRIVIAL_HELP
- "\n"
- "-n\tsuppress automatic printing of pattern space\n"
- "-e script\tadd the script to the commands to be executed\n"
- "-f scriptfile\tadd the contents of script-file to the commands to be executed\n"
- "-h\tdisplay this help message\n"
- "-V\toutput version information and exit\n"
- "\n"
- "If no -e or -f is given, the first non-option argument is taken as the\n"
- "sed script to interpret. All remaining arguments are names of input\n"
- "files; if no input files are specified, then the standard input is read.\n"
-#endif
- ;
-
-#if 0
+#ifdef BB_FEATURE_CLEAN_UP
static void destroy_cmd_strs()
{
if (sed_cmds == NULL)
#endif
/*
- * index_of_unescaped_slash - walks left to right through a string beginning
- * at a specified index and returns the index of the next unescaped slash.
+ * 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)
{
- do {
- idx++;
- /* test if we've hit the end */
- if (str[idx] == 0)
- return -1;
- } while (str[idx] != '/' && str[idx - 1] != '\\');
+ for ( ; str[idx]; idx++) {
+ if (str[idx] == '/' && str[idx-1] != '\\')
+ return idx;
+ }
- return idx;
+ /* if we make it to here, we've hit the end of the string */
+ return -1;
}
/*
idx++;
}
else if (my_str[idx] == '/') {
- idx = index_of_next_unescaped_slash(my_str, idx);
+ idx = index_of_next_unescaped_slash(my_str, ++idx);
if (idx == -1)
fatalError("unterminated match expression\n");
my_str[idx] = '\0';
int oldidx, cflags = REG_NEWLINE;
char *match;
int idx = 0;
+ int j;
/*
* the string that gets passed to this function should look like this:
- * s/match/replace/gI
- * || | ||
+ * s/match/replace/gIp
+ * || | |||
* mandatory optional
*
* (all three of the '/' slashes are mandatory)
/* save the match string */
oldidx = idx+1;
- idx = index_of_next_unescaped_slash(substr, idx);
+ idx = index_of_next_unescaped_slash(substr, ++idx);
if (idx == -1)
fatalError("bad format in substitution expression\n");
match = strdup_substr(substr, oldidx, idx);
+ /* determine the number of back references in the match string */
+ /* Note: we compute this here rather than in the do_subst_command()
+ * function to save processor time, at the expense of a little more memory
+ * (4 bits) per sed_cmd */
+
+ /* 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)
+ sed_cmd->num_backrefs++;
+ }
+
/* save the replacement string */
oldidx = idx+1;
- idx = index_of_next_unescaped_slash(substr, idx);
+ idx = index_of_next_unescaped_slash(substr, ++idx);
if (idx == -1)
fatalError("bad format in substitution expression\n");
sed_cmd->replace = strdup_substr(substr, oldidx, idx);
while (substr[++idx]) {
switch (substr[idx]) {
case 'g':
- sed_cmd->sub_g = 1;
+ sed_cmd->sub_flags = SUB_G;
break;
case 'I':
cflags |= REG_ICASE;
break;
+#ifdef BB_FEATURE_SED_PATTERN_SPACE
+ case 'p':
+ sed_cmd->sub_flags = SUB_P;
+ break;
+#endif
default:
/* any whitespace or semicolon trailing after a s/// is ok */
if (strchr("; \t\v\n\r", substr[idx]))
}
out:
- /* compile the regex */
+ /* compile the match string into a regex */
sed_cmd->sub_match = (regex_t *)xmalloc(sizeof(regex_t));
xregcomp(sed_cmd->sub_match, match, cflags);
free(match);
continue;
}
/* grow the array */
- sed_cmds = realloc(sed_cmds, sizeof(struct sed_cmd) * (++ncmds));
+ sed_cmds = xrealloc(sed_cmds, sizeof(struct sed_cmd) * (++ncmds));
/* zero new element */
memset(&sed_cmds[ncmds-1], 0, sizeof(struct sed_cmd));
/* load command string into new array element, get remainder */
/* if a line ends with '\' it needs the next line appended to it */
while (line[strlen(line)-2] == '\\' &&
(nextline = get_line_from_file(cmdfile)) != NULL) {
- line = realloc(line, strlen(line) + strlen(nextline) + 1);
+ line = xrealloc(line, strlen(line) + strlen(nextline) + 1);
strcat(line, nextline);
free(nextline);
}
}
}
+static void print_subst_w_backrefs(const char *line, const char *replace, regmatch_t *regmatch)
+{
+ int i;
+
+ /* go through the replacement string */
+ for (i = 0; replace[i]; i++) {
+ /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
+ if (replace[i] == '\\' && isdigit(replace[i+1])) {
+ int j;
+ char tmpstr[2];
+ int backref;
+ ++i; /* i now indexes the backref number, instead of the leading slash */
+ tmpstr[0] = replace[i];
+ tmpstr[1] = 0;
+ backref = atoi(tmpstr);
+ /* print out the text held in regmatch[backref] */
+ for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
+ fputc(line[j], stdout);
+ }
+
+ /* if we find a backslash escaped character, print the character */
+ else if (replace[i] == '\\') {
+ ++i;
+ fputc(replace[i], stdout);
+ }
+
+ /* if we find an unescaped '&' print out the whole matched text.
+ * fortunately, regmatch[0] contains the indicies to the whole matched
+ * expression (kinda seems like it was designed for just such a
+ * purpose...) */
+ else if (replace[i] == '&' && replace[i-1] != '\\') {
+ int j;
+ for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
+ fputc(line[j], stdout);
+ }
+ /* nothing special, just print this char of the replacement string to stdout */
+ else
+ fputc(replace[i], stdout);
+ }
+}
+
static int do_subst_command(const struct sed_cmd *sed_cmd, const char *line)
{
int altered = 0;
/* we only substitute if the substitution 'search' expression matches */
if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == 0) {
- regmatch_t regmatch;
+ 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, 1, ®match, 0) == 0) {
+ 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.rm_so; i++)
+ 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);
+ }
+
/* then print the substitution in its place */
- fputs(sed_cmd->replace, stdout);
+#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);
+
/* then advance past the match */
- ptr += regmatch.rm_eo;
+ ptr += regmatch[0].rm_eo;
+
/* and flag that something has changed */
altered++;
/* if we're not doing this globally... */
- if (!sed_cmd->sub_g)
+ if (!sed_cmd->sub_flags & SUB_G)
break;
}
/* if we COULD NOT match the search string (meaning we've gone past
}
/* 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);
+
+ /* cleanup */
+ free(regmatch);
}
return altered;
}
/* 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 && linenum <= sed_cmds[i].end_line)
+ 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))
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) */
- if (!be_quiet && !line_altered)
+ /* 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
fputs(line, stdout);
free(line);
{
int opt;
- /* do special-case option parsing */
- if (argv[1] && (strcmp(argv[1], "--help") == 0))
- usage(sed_usage);
-
-#if 0
+#ifdef BB_FEATURE_CLEAN_UP
/* destroy command strings on exit */
if (atexit(destroy_cmd_strs) == -1) {
perror("sed");
#endif
/* do normal option parsing */
- while ((opt = getopt(argc, argv, "Vhne:f:")) > 0) {
+ while ((opt = getopt(argc, argv, "hne:f:")) > 0) {
switch (opt) {
- case 'V':
- printf("BusyBox v%s (%s)\n", BB_VER, BB_BT);
- exit(0);
- break;
case 'h':
usage(sed_usage);
break;