// A lot of this is taken from async/rxx.C in sfslite, but with some
// modifications for returning the matched pattern.
//
int
split2 (vec<str> *out, rxx pat, str expr, size_t lim, bool emptylast)
{
const char *p = expr;
const char *const e = p + expr.len ();
size_t n;
if (out)
out->clear ();
// check p < e to see that we're not dealing with an empty
// string (especially since x? matches "").
for (n = 0; p < e && n + 1 < lim; n++) {
if (!pat._exec (p, e - p, 0)) {
return 0;
}
if (!pat.success ())
break;
if (out) {
out->push_back (str (p, pat.start (0)));
str sep (p + pat.start (0), pat.len (0));
out->push_back (sep);
}
p += max (pat.end (0), 1);
}
if (lim && (p < e || emptylast)) {
n++;
if (out) {
out->push_back (str (p, e - p));
}
}
return n;
}
void
repl_el_capture_t::output (strbuf &out, const char *s, rxx &x)
{
int start = x.start (_i);
int ln = x.len (_i);
if (start >= 0 && ln > 0) {
str repl = str (s + start, ln);
strbuf_output (out, repl);
}
}
static void
extract_matches (vec<str> *out, const char *base, rxx &x)
{
bool go = true;
for (int i = 0; go; i++) {
int ln = x.len (i);
int start = x.start (i);
if (ln < 0 || start < 0) { go = false; }
else if (ln > 0) { out->push_back (str (base + start, ln)); }
}
}
str
rxx_replace_2 (str input, rxx pat, str repl_str)
{
repl_t repl;
str ret;
if (!repl.parse (repl_str)) {
warn << "XX cannot parse replacement string: " << repl_str << "\n";
} else {
const char *p = input;
const char *const e = p + input.len ();
strbuf b;
bool go = true;
bool err = false;
// check p < e to see that we're not dealing with an empty
// string (especially since x? matches "").
while (go && !err && p < e) {
if (!pat._exec (p, e - p, 0)) {
warn << "XX regex execution failed\n";
err = true;
}
else if (!pat.success ()) { go = false; }
else {
str pre = str (p, pat.start (0));
strbuf_output (b, pre);
repl.output (b, p, pat);
p += max (pat.end (0), 1);
}
}
if (p < e && !err) {
str post = str (p, e - p);
strbuf_output (b, post);
}
if (!err) { ret = b; }
}
return ret;
}
// Call the replace function for each matched pattern.
str
rxx_replace (str input, rxx pat, rxx_replace_cb_t cb)
{
const char *p = input;
const char *const e = p + input.len ();
strbuf b;
bool go = true;
bool err = false;
str ret;
// check p < e to see that we're not dealing with an empty
// string (especially since x? matches "").
while (go && !err && p < e) {
if (!pat._exec (p, e - p, 0)) {
warn << "XX regex execution failed\n";
err = true;
}
else if (!pat.success ()) { go = false; }
else {
str pre = str (p, pat.start (0));
strbuf_output (b, pre);
vec<str> v;
extract_matches (&v, p, pat);
str repl = (*cb) (&v);
if (repl) {
strbuf_output (b, repl);
}
p += max (pat.end (0), 1);
}
}
if (p < e && !err) {
str post = str (p, e - p);
strbuf_output (b, post);
}
if (!err) { ret = b; }
return ret;
}