RE2Regex(const char* string, unsigned int options): casefold(false)
{
RE2::Options opts;
opts.set_posix_syntax(true);
opts.set_perl_classes(true);
opts.set_word_boundary(true);
opts.set_one_line(false);
opts.set_never_nl(true);
opts.set_literal((options & RO_LITERAL) != 0);
opts.set_log_errors(false);
std::string pattern;
if ((options & RO_IGNORECASE) && transformRegexCasefold(string, pattern, (options & RO_LITERAL) != 0))
{
casefold = true;
}
else
{
pattern = string;
opts.set_case_sensitive((options & RO_IGNORECASE) == 0);
}
re.reset(new RE2(pattern, opts));
if (!re->ok())
throw std::runtime_error("Error parsing regular expression " + (string + (": " + re->error())));
std::string prefix = getPrefix(re.get(), 128);
if (prefix.length() == 1)
matcher.reset(new LiteralMatcherFirst(prefix.c_str()));
#ifdef USE_SSE2
else if (prefix.length() > 1)
matcher.reset(new LiteralMatcherSSE(prefix.c_str()));
#endif
}
/**
reads all the regular expressions from the database.
and compile them
*/
void
RegexManager::load_config(YAML::Node cfg)
{
try
{
TSDebug(BANJAX_PLUGIN_NAME, "Setting regex re2 options");
RE2::Options opt;
opt.set_log_errors(false);
opt.set_perl_classes(true);
opt.set_posix_syntax(true);
TSDebug(BANJAX_PLUGIN_NAME, "Loading regex manager conf");
//now we compile all of them and store them for later use
for(YAML::const_iterator it = cfg.begin(); it != cfg.end(); ++it) {
string cur_rule = (const char*) (*it)["rule"].as<std::string>().c_str();
TSDebug(BANJAX_PLUGIN_NAME, "initiating rule %s", cur_rule.c_str());
unsigned int observation_interval = (*it)["interval"].as<unsigned int>();
unsigned int threshold = (*it)["hits_per_interval"].as<unsigned int>();
rated_banning_regexes.push_back(new RatedRegex(cur_rule, new RE2((const char*)((*it)["regex"].as<std::string>().c_str()), opt), observation_interval * 1000, threshold /(double)(observation_interval* 1000)));
}
}
catch(YAML::RepresentationException& e)
{
TSDebug(BANJAX_PLUGIN_NAME, "Error loading regex manager conf [%s].", e.what());
return;
}
TSDebug(BANJAX_PLUGIN_NAME, "Done loading regex manager conf");
}
static void cpy_options(RE2::Options &options, JNIEnv *env, jobject j_options) {
assert(j_options != 0);
jclass j_options_cls = env->GetObjectClass(j_options);
options.set_encoding(get_re2_encoding(env, env->GetObjectField(j_options, get_field_id_safe(env, j_options_cls, "encoding", "Lcom/logentries/re2/Encoding;"))));
options.set_posix_syntax(env->GetBooleanField(j_options, get_field_id_safe(env, j_options_cls, "posixSyntax", "Z")));
options.set_longest_match(env->GetBooleanField(j_options, get_field_id_safe(env, j_options_cls, "longestMatch", "Z")));
options.set_log_errors(env->GetBooleanField(j_options, get_field_id_safe(env, j_options_cls, "logErrors", "Z")));
options.set_max_mem(safe_cast<uint64_t>(env->GetLongField(j_options, get_field_id_safe(env, j_options_cls, "maxMem", "J"))));
options.set_literal(env->GetBooleanField(j_options, get_field_id_safe(env, j_options_cls, "literal", "Z")));
options.set_never_nl(env->GetBooleanField(j_options, get_field_id_safe(env, j_options_cls, "neverNl", "Z")));
options.set_never_capture(env->GetBooleanField(j_options, get_field_id_safe(env, j_options_cls, "neverCapture", "Z")));
options.set_case_sensitive(env->GetBooleanField(j_options, get_field_id_safe(env, j_options_cls, "caseSensitive", "Z")));
options.set_perl_classes(env->GetBooleanField(j_options, get_field_id_safe(env, j_options_cls, "perlClasses", "Z")));
options.set_word_boundary(env->GetBooleanField(j_options, get_field_id_safe(env, j_options_cls, "wordBoundary", "Z")));
}
