Migrated N-dimensional string dataset validator to ritsuko.

include/takane/dense_array.hpp

@@ -47,91 +47,6 @@ inline bool is_transposed(const H5::Group& ghandle) {
     return ritsuko::hdf5::load_scalar_numeric_attribute<int32_t>(attr) != 0;
 }
 
-inline void validate_string_contents(const H5::DataSet& dhandle, const std::vector<hsize_t>& data_extent, hsize_t buffer_size) {
-    auto stype = dhandle.getDataType();
-    if (!stype.isVariableStr()) {
-        return;
-    }
-
-    for (auto ex : data_extent) {
-        if (ex == 0) {
-            return;
-        }
-    }
-
-    hsize_t ndims = data_extent.size();
-    std::vector<hsize_t> chunk_extent(ndims, 1);
-    auto cplist = dhandle.getCreatePlist();
-    if (cplist.getLayout() == H5D_CHUNKED) {
-        cplist.getChunk(chunk_extent.size(), chunk_extent.data());
-    }
-
-    // Scaling up the block size as much as possible. We start from the
-    // fastest-changing dimension (i.e., the last one in HDF5) and increase it,
-    // and then we move onto the next-fastest dimension, and so on until the
-    // buffer size is exhausted.
-    auto block_extent = chunk_extent;
-    hsize_t block_size = 1;
-    for (hsize_t d = 0; d < ndims; ++d) {
-        block_extent[d] = std::min(block_extent[d], data_extent[d]); // should be a no-op, but we do this just in case.
-        block_size *= block_extent[d];
-    }
-
-    for (hsize_t i = ndims; i > 0; --i) {
-        int multiple = buffer_size / block_size;
-        if (multiple <= 1) {
-            break;
-        }
-        auto d = i - 1;
-        block_size /= block_extent[d];
-        block_extent[d] = std::min(data_extent[d], block_extent[d] * multiple);
-        block_size *= block_extent[d];
-    }
-
-    // Now iterating through the array.
-    std::vector<hsize_t> starts(ndims), counts(block_extent.begin(), block_extent.end());
-    std::vector<char*> buffer(block_size);
-    hsize_t buffer_length = block_size;
-
-    H5::DataSpace mspace, dspace(ndims, data_extent.data());
-    bool finished = false;
-
-    while (!finished) {
-        buffer.resize(buffer_length);
-        dspace.selectHyperslab(H5S_SELECT_SET, counts.data(), starts.data());
-        mspace.setExtentSimple(ndims, counts.data());
-
-        [[maybe_unused]] ritsuko::hdf5::VariableStringCleaner stream(stype.getId(), mspace.getId(), buffer.data());
-        dhandle.read(buffer.data(), stype, mspace, dspace);
-        for (auto x : buffer) {
-            if (x == NULL) {
-                throw std::runtime_error("detected NULL pointer in a variable-length string dataset");
-            }
-        }
-
-        // Attempting a shift from the last dimension as this is the fastest-changing.
-        for (hsize_t i = ndims; i > 0; --i) {
-            auto d = i - 1;
-            starts[d] += block_extent[d];
-            if (starts[d] >= data_extent[d]) {
-                if (d == 0) {
-                    finished = true;
-                } else {
-                    starts[d] = 0;
-                    buffer_length /= counts[d];
-                    counts[d] = std::min(data_extent[d], block_extent[d]);
-                    buffer_length *= counts[d];
-                }
-            } else {
-                buffer_length /= counts[d];
-                counts[d] = std::min(data_extent[d] - starts[d], block_extent[d]);
-                buffer_length *= counts[d];
-                break;
-            }
-        }
-    }
-}
-
 }
 /**
  * @endcond
@@ -179,7 +94,7 @@ inline void validate(const std::filesystem::path& path, const ObjectMetadata& me
         if (!ritsuko::hdf5::is_utf8_string(dhandle)) {
             throw std::runtime_error("expected string array to have a datatype that can be represented by a UTF-8 encoded string");
         }
-        internal::validate_string_contents(dhandle, extents, options.hdf5_buffer_size);
+        ritsuko::hdf5::validate_nd_string_dataset(dhandle, extents, options.hdf5_buffer_size);
     } else {
         throw std::runtime_error("unknown array type '" + type + "'");
     }