std::unique_ptr<analyzer> load(const cpptoml::table& config)
{
using namespace analyzers;
std::vector<std::unique_ptr<analyzer>> toks;
auto analyzers = config.get_table_array("analyzers");
for (auto group : analyzers->get())
{
auto method = group->get_as<std::string>("method");
if (!method)
throw analyzer_exception{"failed to find analyzer method"};
toks.emplace_back(
analyzer_factory::get().create(*method, config, *group));
}
return make_unique<multi_analyzer>(std::move(toks));
}
metadata::schema_type metadata_schema(const cpptoml::table& config)
{
metadata::schema_type schema;
if (auto metadata = config.get_table_array("metadata"))
{
const auto& arr = metadata->get();
schema.reserve(arr.size());
for (const auto& table : arr)
{
auto name = table->get_as<std::string>("name");
auto type = table->get_as<std::string>("type");
if (!name)
throw metadata_exception{"name needed for metadata field"};
if (!type)
throw metadata_exception{"type needed for metadata field"};
metadata::field_type ftype;
if (*type == "int")
{
ftype = metadata::field_type::SIGNED_INT;
}
else if (*type == "uint")
{
ftype = metadata::field_type::UNSIGNED_INT;
}
else if (*type == "double")
{
ftype = metadata::field_type::DOUBLE;
}
else if (*type == "string")
{
ftype = metadata::field_type::STRING;
}
else
{
throw metadata_exception{"invalid metadata type: \"" + *type
+ "\""};
}
schema.emplace_back(*name, ftype);
}
}
return schema;
}
std::unique_ptr<token_stream> load_filters(const cpptoml::table& global,
const cpptoml::table& config)
{
auto check = config.get_as<std::string>("filter");
if (check)
{
if (*check == "default-chain")
return default_filter_chain(global);
else if (*check == "default-unigram-chain")
return default_unigram_chain(global);
else
throw analyzer_exception{"unknown filter option: " + *check};
}
auto filters = config.get_table_array("filter");
if (!filters)
throw analyzer_exception{"analyzer group missing filter configuration"};
std::unique_ptr<token_stream> result;
for (const auto filter : filters->get())
result = load_filter(std::move(result), *filter);
return result;
}