Class import_export::Detector¶

class Detector : public import_export::DataStreamSink ¶

Public Functions

Detector(const boost::filesystem::path &fp, CopyParams &cp)¶

std::vector<std::string> get_headers()¶

std::vector<std::vector<std::string>> get_sample_rows(size_t n)¶

std::vector<SQLTypeInfo> getBestColumnTypes() const¶

Public Members

std::vector<std::vector<std::string>> raw_rows¶

bool has_headers = false¶

Public Static Functions

SQLTypes detect_sqltype(const std::string &str)¶

Public Static Attributes

constexpr size_t kDefaultSampleRowsCount = {100}¶

Private Functions

void init()¶

void read_file()¶

void detect_row_delimiter()¶

void split_raw_data()¶

std::vector<SQLTypes> detect_column_types(const std::vector<std::string> &row)¶

void find_best_sqltypes()¶

std::vector<SQLTypes> find_best_sqltypes(const std::vector<std::vector<std::string>> &raw_rows, const CopyParams &copy_params)¶

std::vector<SQLTypes> find_best_sqltypes(const std::vector<std::vector<std::string>>::const_iterator &row_begin, const std::vector<std::vector<std::string>>::const_iterator &row_end, const CopyParams &copy_params)¶

std::vector<EncodingType> find_best_encodings(const std::vector<std::vector<std::string>>::const_iterator &row_begin, const std::vector<std::vector<std::string>>::const_iterator &row_end, const std::vector<SQLTypes> &best_types)¶

bool detect_headers(const std::vector<SQLTypes> &first_types, const std::vector<SQLTypes> &rest_types)¶

void find_best_sqltypes_and_headers()¶

ImportStatus importDelimited(const std::string &file_path, const bool decompressed, const Catalog_Namespace::SessionInfo *session_info)¶

Private Members

std::string raw_data¶

boost::filesystem::path file_path¶

std::chrono::duration<double> timeout = {1}¶

std::string line1¶

std::vector<SQLTypes> best_sqltypes¶

std::vector<EncodingType> best_encodings¶

Private Static Functions

bool more_restrictive_sqltype(const SQLTypes a, const SQLTypes b)¶