Reader.hpp

Go to the documentation of this file.

#pragma once
 
#include <iostream>
#include <mutable/catalog/Scheduler.hpp>
#include <mutable/catalog/Schema.hpp>
#include <mutable/IR/Tuple.hpp>
#include <mutable/storage/Store.hpp>
#include <mutable/util/Diagnostic.hpp>
 
 
namespace m {
 
struct M_EXPORT Reader
{
    const Table &table; 
    Diagnostic &diag;
    Scheduler::Transaction * transaction;
 
    public:
    Reader(const Table &table, Diagnostic &diag, Scheduler::Transaction *transaction = nullptr) : table(table), diag(diag), transaction(transaction) { }
    virtual ~Reader() { }
 
    virtual void operator()(std::istream &in, const char *name = "-") = 0;
};
 
struct M_EXPORT DSVReader : Reader, ConstTypeVisitor
{
    struct M_EXPORT Config
    {
        char delimiter = ',';
        char quote = '"';
        char escape = '\\';
        bool has_header = false;
        bool skip_header = false;
        std::size_t num_rows = std::numeric_limits<decltype(num_rows)>::max();
 
        static Config CSV() {
            Config cfg;
            cfg.delimiter = ',';
            cfg.quote = '"';
            cfg.escape = '"'; // in RFC 4180, the escape character *is* the quote character
            return cfg;
        }
    };
 
    private:
    Config cfg_;
 
    Position pos;
    char c;
    std::istream *in = nullptr;
    std::vector<char> buf;
    Tuple tup; 
    std::size_t col_idx;
 
    public:
    DSVReader(const Table &table, Config cfg, Diagnostic &diag, Scheduler::Transaction *transaction = nullptr);
 
    void operator()(std::istream &in, const char *name) override;
 
    const Config & config() const { return cfg_; }
    size_t num_rows() const { return cfg_.num_rows; }
    size_t delimiter() const { return cfg_.delimiter; }
    size_t escape() const { return cfg_.escape; }
    size_t quote() const { return cfg_.quote; }
    size_t has_header() const { return cfg_.has_header; }
    size_t skip_header() const { return cfg_.skip_header; }
 
    private:
    using ConstTypeVisitor::operator();
    void operator()(Const<ErrorType> &ty) override;
    void operator()(Const<NoneType> &ty) override;
    void operator()(Const<Boolean> &ty) override;
    void operator()(Const<Bitmap> &ty) override;
    void operator()(Const<CharacterSequence> &ty) override;
    void operator()(Const<Date> &ty) override;
    void operator()(Const<DateTime> &ty) override;
    void operator()(Const<Numeric> &ty) override;
    void operator()(Const<FnType> &ty) override;
 
    int step() {
        switch (c) {
            case '\n':
                pos.column = 1;
                pos.line++;
            case EOF:
                break;
 
            default:
                pos.column++;
        }
        return c = in->get();
    };
 
    void push() { buf.push_back(c); step(); }
 
    bool accept(char chr) { if (c == chr) { step(); return true; } return false; }
 
    void discard_cell() {
        if (c == config().quote) {
            step();
            while (c != EOF and c != '\n' and c != config().quote) { step(); }
            accept(config().quote);
        } else
            while (c != EOF and c != '\n' and c != config().delimiter) { step(); }
    }
    void discard_row() { while (c != EOF and c != '\n') { step(); } }
 
    int64_t read_unsigned_int();
};
 
}