From 32e14551a90e85000e41b3f0445d34d58a1431e4 Mon Sep 17 00:00:00 2001 From: Joel Klinghed Date: Wed, 10 Sep 2025 22:12:22 +0200 Subject: Add unicode general category lookup Generate the lookup tables from UnicodeData.txt, do to that, add gen_ugc, which uses csv, buffers, line, io and other modules to do the job. --- src/csv.cc | 63 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 63 insertions(+) create mode 100644 src/csv.cc (limited to 'src/csv.cc') diff --git a/src/csv.cc b/src/csv.cc new file mode 100644 index 0000000..4135555 --- /dev/null +++ b/src/csv.cc @@ -0,0 +1,63 @@ +#include "csv.hh" + +#include "line.hh" +#include "str.hh" + +#include +#include +#include +#include +#include +#include +#include + +namespace csv { + +namespace { + +class ReaderImpl : public Reader { + public: + ReaderImpl(std::unique_ptr reader, char separator) + : reader_(std::move(reader)), separator_(separator) { + } + + [[nodiscard]] + std::expected, io::ReadError> read() override { + while (true) { + auto line = reader_->read(); + if (line.has_value()) { + str::split(line.value(), line_, separator_, /* keep_empty */ true); + if (line_.size() == 1 && line_[0].empty()) + continue; + return line_; + } + if (line.error().eof) { + return {}; + } + return std::unexpected(line.error().io_error.value()); + } + } + + [[nodiscard]] uint64_t number() const override { + return reader_->number(); + } + + private: + std::unique_ptr reader_; + char const separator_; + std::vector line_; +}; + +} // namespace + +std::unique_ptr open(std::unique_ptr reader, + char separator) { + return std::make_unique(std::move(reader), separator); +} + +std::unique_ptr open(std::unique_ptr reader, + char separator) { + return open(line::open(std::move(reader)), separator); +} + +} // namespace csv -- cgit v1.3