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.

1 files changed, 63 insertions, 0 deletions

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 <cstdint>
+#include <expected>
+#include <memory>
+#include <span>
+#include <string_view>
+#include <utility>
+#include <vector>
+
+namespace csv {
+
+namespace {
+
+class ReaderImpl : public Reader {
+ public:
+  ReaderImpl(std::unique_ptr<line::Reader> reader, char separator)
+      : reader_(std::move(reader)), separator_(separator) {
+  }
+
+  [[nodiscard]]
+  std::expected<std::span<std::string_view>, 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<line::Reader> reader_;
+  char const separator_;
+  std::vector<std::string_view> line_;
+};
+
+}  // namespace
+
+std::unique_ptr<Reader> open(std::unique_ptr<line::Reader> reader,
+                             char separator) {
+  return std::make_unique<ReaderImpl>(std::move(reader), separator);
+}
+
+std::unique_ptr<Reader> open(std::unique_ptr<io::Reader> reader,
+                             char separator) {
+  return open(line::open(std::move(reader)), separator);
+}
+
+}  // namespace csv