unsigned long mlre2__custom_regex_deserialize(void * dst) {
int len = caml_deserialize_sint_4();
RE2::Options options;
char * pattern = (char *) caml_stat_alloc(sizeof(*pattern) * (len));
caml_deserialize_block_1(pattern, len);
pattern[len - 1] = '\0';
options.Copy(RE2::Quiet);
options.set_max_mem(caml_deserialize_sint_8());
options_of_bitfield((uint16_t) caml_deserialize_uint_2(), options);
#ifdef DEBUG
std::cerr << "deserialized regex /" << pattern << "/" << std::endl;
#endif
*(RE2 **) dst = new RE2(pattern, options);
caml_stat_free(pattern);
return sizeof(RE2 *);
}
/* returns (cre2__obj_t * int * (string * int) list) where
* - cre2__obj_t is the ML-side name for a custom_block with a struct regex *
* - int is the number of submatches, including the whole match
* - (string * int) list is the Map.to_alist of the submatch (name, index) Map.t
*/
CAMLprim value mlre2__create_re(value v_options, value v_pattern) {
value v_retval, v_compile_error;
const char * c_pat = String_val(v_pattern);
RE2::Options opt;
RE2* compiled = NULL;
opt.Copy(RE2::Quiet);
while (v_options != Val_emptylist) {
int val = Int_val(Field(Field(v_options, 0), 0));
switch (Tag_val(Field(v_options, 0))) {
#define X(_u,FIRST,REST,_uu) case FIRST##REST : opt.set_##FIRST##REST(val); break;
#define X__ENCODING(_u,FIRST,REST,_uu,SUFFIX,_uuu,TRANSLATED) \
case FIRST##REST##SUFFIX : opt.set_##FIRST##REST(val TRANSLATED); break;
#define X__MAXMEM(_u,FIRST,REST,_uu) X(_u,FIRST,REST,_uu)
#include "enum_x_macro.h"
default : caml_invalid_argument("invalid option\n");
}
v_options = Field(v_options, 1);
}
compiled = new RE2(c_pat, opt);
if (!compiled->ok()) {
/* Warning
from this point on it's no longer safe to access v_options or
v_pattern as the GC might be invoked from caml_copy_string and
move those values (as we haven't registered the paramters they
wouldn't get updated). This is fine because we don't access
them before we call caml_raise_with_arg. */
v_compile_error = caml_copy_string(compiled->error().c_str());
delete compiled;
compiled = NULL;
caml_raise_with_arg(*caml_named_value("mlre2__Regex_compile_failed"),
v_compile_error);
}
v_retval = caml_alloc_custom(&mlre2__custom_regex_ops, sizeof(compiled),
1024*1024, /* RE2 object uses ~1MB of memory outside the OCaml heap */
500*1024*1024); /* I'm okay with 500MB of RAM being wasted */
Regex_val(v_retval) = compiled;
return v_retval;
}
