/*
* Process the commands arguments
*/
static char *parse_cmd_args(sed_cmd_t *sed_cmd, char *cmdstr)
{
/* handle (s)ubstitution command */
if (sed_cmd->cmd == 's') cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
/* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
else if (strchr("aic", sed_cmd->cmd)) {
if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
bb_error_msg_and_die
("only a beginning address can be specified for edit commands");
while(isspace(*cmdstr)) cmdstr++;
sed_cmd->string = bb_xstrdup(cmdstr);
parse_escapes(sed_cmd->string,sed_cmd->string,strlen(cmdstr),0,0);
cmdstr += strlen(cmdstr);
/* handle file cmds: (r)ead */
} else if(strchr("rw", sed_cmd->cmd)) {
if (sed_cmd->end_line || sed_cmd->end_match)
bb_error_msg_and_die("Command only uses one address");
cmdstr += parse_file_cmd(sed_cmd, cmdstr, &sed_cmd->string);
if(sed_cmd->cmd=='w')
sed_cmd->file=bb_xfopen(sed_cmd->string,"w");
/* handle branch commands */
} else if (strchr(":bt", sed_cmd->cmd)) {
int length;
while(isspace(*cmdstr)) cmdstr++;
length = strcspn(cmdstr, semicolon_whitespace);
if (length) {
sed_cmd->string = strndup(cmdstr, length);
cmdstr += length;
}
}
/* translation command */
else if (sed_cmd->cmd == 'y') {
char *match, *replace;
int i=cmdstr[0];
cmdstr+=parse_regex_delim(cmdstr, &match, &replace)+1;
/* \n already parsed, but \delimiter needs unescaping. */
parse_escapes(match,match,strlen(match),i,i);
parse_escapes(replace,replace,strlen(replace),i,i);
sed_cmd->string = xcalloc(1, (strlen(match) + 1) * 2);
for (i = 0; match[i] && replace[i]; i++) {
sed_cmd->string[i * 2] = match[i];
sed_cmd->string[(i * 2) + 1] = replace[i];
}
free(match);
free(replace);
}
/* if it wasnt a single-letter command that takes no arguments
* then it must be an invalid command.
*/
else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) {
bb_error_msg_and_die("Unsupported command %c", sed_cmd->cmd);
}
/* give back whatever's left over */
return (cmdstr);
}
static char *parse_cmd_str(struct sed_cmd * const sed_cmd, const char *const cmdstr)
{
int idx = 0;
/* parse the command
* format is: [addr][,addr]cmd
* |----||-----||-|
* 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(sed_cmd, cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
/* second part (if present) will begin with a comma */
if (cmdstr[idx] == ',') {
idx++;
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')
error_msg_and_die("missing command");
sed_cmd->cmd = cmdstr[idx];
/* if it was a single-letter command that takes no arguments (such as 'p'
* or 'd') all we need to do is increment the index past that command */
if (strchr("pd", cmdstr[idx])) {
idx++;
}
/* handle (s)ubstitution command */
else if (sed_cmd->cmd == 's') {
idx += parse_subst_cmd(sed_cmd, &cmdstr[idx]);
}
/* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
else if (strchr("aic", sed_cmd->cmd)) {
if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c')
error_msg_and_die("only a beginning address can be specified for edit commands");
idx += parse_edit_cmd(sed_cmd, &cmdstr[idx]);
}
/* handle file cmds: (r)ead */
else if (sed_cmd->cmd == 'r') {
if (sed_cmd->end_line || sed_cmd->end_match)
error_msg_and_die("Command only uses one address");
idx += parse_file_cmd(sed_cmd, &cmdstr[idx]);
}
else {
error_msg_and_die("invalid command");
}
/* give back whatever's left over */
return (char *)&cmdstr[idx];
}
/*
* Process the commands arguments
*/
static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr)
{
static const char cmd_letters[] ALIGN1 = "saicrw:btTydDgGhHlnNpPqx={}";
enum {
IDX_s = 0,
IDX_a,
IDX_i,
IDX_c,
IDX_r,
IDX_w,
IDX_colon,
IDX_b,
IDX_t,
IDX_T,
IDX_y,
IDX_d,
IDX_D,
IDX_g,
IDX_G,
IDX_h,
IDX_H,
IDX_l,
IDX_n,
IDX_N,
IDX_p,
IDX_P,
IDX_q,
IDX_x,
IDX_equal,
IDX_lbrace,
IDX_rbrace,
IDX_nul
};
unsigned idx;
BUILD_BUG_ON(sizeof(cmd_letters)-1 != IDX_nul);
idx = strchrnul(cmd_letters, sed_cmd->cmd) - cmd_letters;
/* handle (s)ubstitution command */
if (idx == IDX_s) {
cmdstr += parse_subst_cmd(sed_cmd, cmdstr);
}
/* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */
else if (idx <= IDX_c) { /* a,i,c */
unsigned len;
if (idx < IDX_c) { /* a,i */
if (sed_cmd->end_line || sed_cmd->end_match)
bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
}
for (;;) {
if (*cmdstr == '\n' || *cmdstr == '\\') {
cmdstr++;
break;
}
if (!isspace(*cmdstr))
break;
cmdstr++;
}
len = strlen(cmdstr);
sed_cmd->string = copy_parsing_escapes(cmdstr, len);
cmdstr += len;
/* "\anychar" -> "anychar" */
parse_escapes(sed_cmd->string, sed_cmd->string, -1, '\0', '\0');
}
/* handle file cmds: (r)ead */
else if (idx <= IDX_w) { /* r,w */
if (idx < IDX_w) { /* r */
if (sed_cmd->end_line || sed_cmd->end_match)
bb_error_msg_and_die("command '%c' uses only one address", sed_cmd->cmd);
}
cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string);
if (sed_cmd->cmd == 'w') {
sed_cmd->sw_file = xfopen_for_write(sed_cmd->string);
sed_cmd->sw_last_char = '\n';
}
}
/* handle branch commands */
else if (idx <= IDX_T) { /* :,b,t,T */
int length;
cmdstr = skip_whitespace(cmdstr);
length = strcspn(cmdstr, semicolon_whitespace);
if (length) {
sed_cmd->string = xstrndup(cmdstr, length);
cmdstr += length;
}
}
/* translation command */
else if (idx == IDX_y) {
char *match, *replace;
int i = cmdstr[0];
cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1;
/* \n already parsed, but \delimiter needs unescaping. */
parse_escapes(match, match, -1, i, i);
parse_escapes(replace, replace, -1, i, i);
sed_cmd->string = xzalloc((strlen(match) + 1) * 2);
for (i = 0; match[i] && replace[i]; i++) {
sed_cmd->string[i*2] = match[i];
sed_cmd->string[i*2+1] = replace[i];
}
free(match);
free(replace);
}
/* if it wasn't a single-letter command that takes no arguments
* then it must be an invalid command.
*/
else if (idx >= IDX_nul) { /* not d,D,g,G,h,H,l,n,N,p,P,q,x,=,{,} */
bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd);
}
/* give back whatever's left over */
return cmdstr;
}
static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr)
{
int cflags = G.regex_type;
char *match;
int idx;
/*
* A substitution command should look something like this:
* s/match/replace/ #giIpw
* || | |||
* mandatory optional
*/
idx = parse_regex_delim(substr, &match, &sed_cmd->string);
/* 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 */
/* process the flags */
sed_cmd->which_match = 1;
dbg("s flags:'%s'", substr + idx + 1);
while (substr[++idx]) {
dbg("s flag:'%c'", substr[idx]);
/* Parse match number */
if (isdigit(substr[idx])) {
if (match[0] != '^') {
/* Match 0 treated as all, multiple matches we take the last one. */
const char *pos = substr + idx;
/* FIXME: error check? */
sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10);
idx = pos - substr - 1;
}
continue;
}
/* Skip spaces */
if (isspace(substr[idx]))
continue;
switch (substr[idx]) {
/* Replace all occurrences */
case 'g':
if (match[0] != '^')
sed_cmd->which_match = 0;
break;
/* Print pattern space */
case 'p':
sed_cmd->sub_p = 1;
break;
/* Write to file */
case 'w':
{
char *fname;
idx += parse_file_cmd(/*sed_cmd,*/ substr+idx+1, &fname);
sed_cmd->sw_file = xfopen_for_write(fname);
sed_cmd->sw_last_char = '\n';
free(fname);
break;
}
/* Ignore case (gnu extension) */
case 'i':
case 'I':
cflags |= REG_ICASE;
break;
/* Comment */
case '#':
// while (substr[++idx]) continue;
idx += strlen(substr + idx); // same
/* Fall through */
/* End of command */
case ';':
case '}':
goto out;
default:
dbg("s bad flags:'%s'", substr + idx);
bb_error_msg_and_die("bad option in substitution expression");
}
}
out:
/* compile the match string into a regex */
if (*match != '\0') {
/* If match is empty, we use last regex used at runtime */
sed_cmd->sub_match = xzalloc(sizeof(regex_t));
dbg("xregcomp('%s',%x)", match, cflags);
xregcomp(sed_cmd->sub_match, match, cflags);
dbg("regcomp ok");
}
free(match);
return idx;
}
static int parse_subst_cmd(sed_cmd_t * const sed_cmd, char *substr)
{
int cflags = 0;
char *match;
int idx = 0;
/*
* A substitution command should look something like this:
* s/match/replace/ #gIpw
* || | |||
* mandatory optional
*/
idx = parse_regex_delim(substr, &match, &sed_cmd->string);
/* 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 */
/* process the flags */
sed_cmd->which_match=1;
while (substr[++idx]) {
/* Parse match number */
if(isdigit(substr[idx])) {
if(match[0]!='^') {
/* Match 0 treated as all, multiple matches we take the last one. */
char *pos=substr+idx;
sed_cmd->which_match=(unsigned short)strtol(substr+idx,&pos,10);
idx=pos-substr;
}
continue;
}
/* Skip spaces */
if(isspace(substr[idx])) continue;
switch (substr[idx]) {
/* Replace all occurrences */
case 'g':
if (match[0] != '^') sed_cmd->which_match = 0;
break;
/* Print pattern space */
case 'p':
sed_cmd->sub_p = 1;
break;
case 'w':
{
char *temp;
idx+=parse_file_cmd(sed_cmd,substr+idx,&temp);
break;
}
/* Ignore case (gnu exension) */
case 'I':
cflags |= REG_ICASE;
break;
case ';':
case '}':
goto out;
default:
bb_error_msg_and_die("bad option in substitution expression");
}
}
out:
/* compile the match string into a regex */
if (*match != '\0') {
/* If match is empty, we use last regex used at runtime */
sed_cmd->sub_match = (regex_t *) xmalloc(sizeof(regex_t));
xregcomp(sed_cmd->sub_match, match, cflags);
}
free(match);
return idx;
}
