Supported features and commands in this version of sed:
- comments ('#')
- - Address matching: num|/matchstr/[,num|/matchstr/|$]command
- - Commands: p, d, s/match/replace/[g]
+ - address matching: num|/matchstr/[,num|/matchstr/|$]command
+ - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
+ - edit commands: (a)ppend, (i)nsert, (c)hange
(Note: Specifying an address (range) to match is *optional*; commands
default to the whole pattern space if no specific address match was
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) */
+
+ /* edit command (a,i,c) speicific field */
+ char *editline;
};
/* globals */
/*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
static void destroy_cmd_strs()
{
if (sed_cmds == NULL)
free(sed_cmds);
sed_cmds = NULL;
}
+#endif
-static void exit_sed(int retcode, const char *message)
-{
- destroy_cmd_strs();
- if (message)
- fputs(message, stderr);
- exit(retcode);
-}
-
+#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)
{
*
* you know, a strrspn() would really be nice cuz then we could say:
*
- * retstr[strlen(retstr) - strrspn(retstr, " \n\t\v") + 1] = 0;
+ * 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
+ */
+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_unescaped_slash - walks left to right through a string beginning
* at a specified index and returns the index of the next unescaped slash.
*/
-static int index_of_next_unescaped_slash(int idx, const char *str)
+static int index_of_next_unescaped_slash(const char *str, int idx)
{
do {
idx++;
} while (isdigit(my_str[idx]));
my_str[idx] = 0;
*line = atoi(my_str);
- *regex = NULL;
}
else if (my_str[idx] == '$') {
*line = -1;
- *regex = NULL;
idx++;
}
else if (my_str[idx] == '/') {
- idx = index_of_next_unescaped_slash(idx, my_str);
+ idx = index_of_next_unescaped_slash(my_str, idx);
if (idx == -1)
- exit_sed(1, "sed: unterminated match expression\n");
+ fatalError("unterminated match expression\n");
my_str[idx] = '\0';
*regex = (regex_t *)xmalloc(sizeof(regex_t));
- if (bb_regcomp(*regex, my_str+1, REG_NEWLINE) != 0) {
- free(my_str);
- exit_sed(1, NULL);
- }
+ xregcomp(*regex, my_str+1, REG_NEWLINE);
+ idx++; /* so it points to the next character after the last '/' */
}
else {
- fprintf(stderr, "sed.c:get_address: no address found in string\n");
- fprintf(stderr, "\t(you probably didn't check the string you passed me)\n");
+ errorMsg("get_address: no address found in string\n"
+ "\t(you probably didn't check the string you passed me)\n");
idx = -1;
}
return newstr;
}
-static void parse_cmd_str(struct sed_cmd *sed_cmd, const char *cmdstr)
+static int parse_subst_cmd(struct sed_cmd *sed_cmd, const char *substr)
+{
+ int oldidx, cflags = REG_NEWLINE;
+ char *match;
+ int idx = 0;
+
+ /*
+ * the string that gets passed to this function should look like this:
+ * s/match/replace/gI
+ * || | ||
+ * mandatory optional
+ *
+ * (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");
+
+ /* save the match string */
+ oldidx = idx+1;
+ idx = index_of_next_unescaped_slash(substr, idx);
+ if (idx == -1)
+ fatalError("bad format in substitution expression\n");
+ match = strdup_substr(substr, oldidx, idx);
+
+ /* save the replacement string */
+ oldidx = idx+1;
+ 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);
+
+ /* process the flags */
+ while (substr[++idx]) {
+ switch (substr[idx]) {
+ case 'g':
+ sed_cmd->sub_g = 1;
+ break;
+ case 'I':
+ cflags |= REG_ICASE;
+ break;
+ 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");
+ }
+ }
+
+out:
+ /* compile the regex */
+ sed_cmd->sub_match = (regex_t *)xmalloc(sizeof(regex_t));
+ xregcomp(sed_cmd->sub_match, match, cflags);
+ free(match);
+
+ return idx;
+}
+
+static int parse_edit_cmd(struct sed_cmd *sed_cmd, const char *editstr)
+{
+ int idx = 0;
+ int slashes_eaten = 0;
+ char *ptr; /* shorthand */
+
+ /*
+ * the string that gets passed to this function should look like this:
+ *
+ * need one of these
+ * |
+ * | this backslash (immediately following the edit command) is mandatory
+ * | |
+ * [aic]\
+ * TEXT1\
+ * TEXT2\
+ * TEXTN
+ *
+ * as soon as we hit a TEXT line that has no trailing '\', we're done.
+ * this means a command like:
+ *
+ * i\
+ * INSERTME
+ *
+ * is a-ok.
+ *
+ */
+
+ if (editstr[1] != '\\' && (editstr[2] != '\n' || editstr[2] != '\r'))
+ fatalError("bad format in edit expression\n");
+
+ /* store the edit line text */
+ /* make editline big enough to accomodate the extra '\n' we will tack on
+ * to the end */
+ sed_cmd->editline = xmalloc(strlen(&editstr[3]) + 2);
+ strcpy(sed_cmd->editline, &editstr[3]);
+ ptr = sed_cmd->editline;
+
+ /* now we need to go through * and: s/\\[\r\n]$/\n/g on the edit line */
+ while (ptr[idx]) {
+ while (ptr[idx] != '\\' && (ptr[idx+1] != '\n' || ptr[idx+1] != '\r')) {
+ idx++;
+ if (!ptr[idx]) {
+ goto out;
+ }
+ }
+ /* move the newline over the '\' before it (effectively eats the '\') */
+ memmove(&ptr[idx], &ptr[idx+1], strlen(&ptr[idx+1]));
+ ptr[strlen(ptr)-1] = 0;
+ slashes_eaten++;
+ /* substitue \r for \n if needed */
+ if (ptr[idx] == '\r')
+ ptr[idx] = '\n';
+ }
+
+out:
+ ptr[idx] = '\n';
+ ptr[idx+1] = 0;
+
+ /* this accounts for discrepancies between the modified string and the
+ * original string passed in to this function */
+ idx += slashes_eaten;
+
+ /* this accounts for the fact that A) we started at index 3, not at index
+ * 0 and B) that we added an extra '\n' at the end (if you think the next
+ * line should read 'idx += 4' remember, arrays are zero-based) */
+
+ idx += 3;
+
+ return idx;
+}
+
+static char *parse_cmd_str(struct sed_cmd *sed_cmd, const char *cmdstr)
{
int idx = 0;
* part1 part2 part3
*/
+
/* 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);
/* last part (mandatory) will be a command */
if (cmdstr[idx] == '\0')
- exit_sed(1, "sed: missing command\n");
- if (!strchr("pds", cmdstr[idx])) /* <-- XXX add new commands here */
- exit_sed(1, "sed: invalid command\n");
+ fatalError("missing command\n");
+ if (!strchr("pdsaic", cmdstr[idx])) /* <-- XXX add new commands here */
+ fatalError("invalid command\n");
sed_cmd->cmd = cmdstr[idx];
- /* special-case handling for 's' */
- if (sed_cmd->cmd == 's') {
- int oldidx, cflags = REG_NEWLINE;
- char *match;
- /* format for substitution is:
- * s/match/replace/gI
- * | ||
- * mandatory optional
- */
-
- /* verify that we have an 's' followed by a 'slash' */
- if (cmdstr[++idx] != '/')
- exit_sed(1, "sed: bad format in substitution expression\n");
-
- /* save the match string */
- oldidx = idx+1;
- idx = index_of_next_unescaped_slash(idx, cmdstr);
- if (idx == -1)
- exit_sed(1, "sed: bad format in substitution expression\n");
- match = strdup_substr(cmdstr, oldidx, idx);
-
- /* save the replacement string */
- oldidx = idx+1;
- idx = index_of_next_unescaped_slash(idx, cmdstr);
- if (idx == -1)
- exit_sed(1, "sed: bad format in substitution expression\n");
- sed_cmd->replace = strdup_substr(cmdstr, oldidx, idx);
- /* process the flags */
- while (cmdstr[++idx]) {
- switch (cmdstr[idx]) {
- case 'g':
- sed_cmd->sub_g = 1;
- break;
- case 'I':
- cflags |= REG_ICASE;
- break;
- default:
- exit_sed(1, "sed: bad option in substitution expression\n");
- }
- }
-
- /* compile the regex */
- sed_cmd->sub_match = (regex_t *)xmalloc(sizeof(regex_t));
- if (bb_regcomp(sed_cmd->sub_match, match, cflags) != 0) {
- free(match);
- exit_sed(1, NULL);
- }
- free(match);
+ /* special-case handling for (s)ubstitution */
+ if (sed_cmd->cmd == 's') {
+ idx += parse_subst_cmd(sed_cmd, &cmdstr[idx]);
+ }
+ /* 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");
+ idx += parse_edit_cmd(sed_cmd, &cmdstr[idx]);
}
+ /* if it was a single-letter command (such as 'p' or 'd') we need to
+ * increment the index past that command */
+ else
+ idx++;
+
+ /* give back whatever's left over */
+ return (char *)&cmdstr[idx];
}
static void add_cmd_str(const char *cmdstr)
{
- char *my_cmdstr = trim_str(cmdstr);
+ char *mystr = (char *)cmdstr;
- /* if this is a comment, don't even bother */
- if (my_cmdstr[0] == '#') {
- free(my_cmdstr);
- return;
- }
+ do {
- /* grow the array */
- sed_cmds = realloc(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 */
- parse_cmd_str(&sed_cmds[ncmds-1], my_cmdstr);
+ /* trim leading whitespace and semicolons */
+ memmove(mystr, &mystr[strspn(mystr, "; \n\r\t\v")], strlen(mystr));
+ /* if we ate the whole thing, that means there was just trailing
+ * whitespace or a final / no-op semicolon. either way, get out */
+ if (strlen(mystr) == 0)
+ return;
+ /* if this is a comment, jump past it and keep going */
+ if (mystr[0] == '#') {
+ mystr = strpbrk(mystr, ";\n\r");
+ continue;
+ }
+ /* grow the array */
+ sed_cmds = realloc(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 */
+ mystr = parse_cmd_str(&sed_cmds[ncmds-1], mystr);
+
+ } while (mystr && strlen(mystr));
}
{
FILE *cmdfile;
char *line;
+ char *nextline;
cmdfile = fopen(filename, "r");
if (cmdfile == NULL)
- exit_sed(1, strerror(errno));
+ fatalError(strerror(errno));
while ((line = get_line_from_file(cmdfile)) != NULL) {
- line[strlen(line)-1] = 0; /* eat newline */
+ /* 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);
+ strcat(line, nextline);
+ free(nextline);
+ }
+ /* 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;
add_cmd_str(line);
free(line);
}
}
+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;
+ 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) {
+ /* print everything before the match, */
+ for (i = 0; i < regmatch.rm_so; i++)
+ fputc(ptr[i], stdout);
+ /* then print the substitution in its place */
+ fputs(sed_cmd->replace, stdout);
+ /* then advance past the match */
+ ptr += regmatch.rm_eo;
+ /* and flag that something has changed */
+ altered++;
+
+ /* if we're not doing this globally... */
+ if (!sed_cmd->sub_g)
+ break;
+ }
+ /* if we COULD NOT match the search string (meaning we've gone past
+ * all previous instances), get out */
+ else
+ break;
+ }
+
+ /* is there anything left to print? */
+ if (*ptr)
+ fputs(ptr, stdout);
+ }
+
+ return altered;
+}
static int do_sed_command(const struct sed_cmd *sed_cmd, const char *line)
{
altered++;
break;
- case 's': /* oo, a fun one :-) */
-
- /* we only substitute if the substitution 'search' expression matches */
- if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == 0) {
- regmatch_t regmatch;
- 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) {
- /* print everything before the match, */
- for (i = 0; i < regmatch.rm_so; i++)
- fputc(ptr[i], stdout);
- /* then print the substitution in its place */
- fputs(sed_cmd->replace, stdout);
- /* then advance past the match */
- ptr += regmatch.rm_eo;
- /* and let the calling function know that something
- * has been changed */
- altered++;
-
- /* if we're not doing this globally... */
- if (!sed_cmd->sub_g)
- break;
- }
- /* if we COULD NOT match the search string (meaning we've gone past
- * all previous instances), get out */
- else
- break;
- }
+ case 's':
+ altered = do_subst_command(sed_cmd, line);
+ break;
- /* is there anything left to print? */
- if (*ptr)
- fputs(ptr, stdout);
- }
+ case 'a':
+ fputs(line, stdout);
+ fputs(sed_cmd->editline, stdout);
+ altered++;
+ break;
+ case 'i':
+ fputs(sed_cmd->editline, stdout);
+ break;
+
+ case 'c':
+ fputs(sed_cmd->editline, stdout);
+ altered++;
break;
}
{
int opt;
- /* do special-case option parsing */
+ /* do special-case option parsing */
if (argv[1] && (strcmp(argv[1], "--help") == 0))
usage(sed_usage);
+#if 0
+ /* destroy command strings on exit */
+ if (atexit(destroy_cmd_strs) == -1) {
+ perror("sed");
+ exit(1);
+ }
+#endif
+
/* do normal option parsing */
while ((opt = getopt(argc, argv, "Vhne:f:")) > 0) {
switch (opt) {
case 'V':
- printf("Print Busybox version here\n");
+ printf("BusyBox v%s (%s)\n", BB_VER, BB_BT);
exit(0);
break;
case 'h':
for (i = optind; i < argc; i++) {
file = fopen(argv[i], "r");
if (file == NULL) {
- fprintf(stderr, "sed: %s: %s\n", argv[i], strerror(errno));
+ errorMsg("%s: %s\n", argv[i], strerror(errno));
} else {
process_file(file);
fclose(file);
}
}
- exit_sed(0, NULL);
-
- /* not reached */
return 0;
}
projects / oweals / busybox.git / blobdiff