This library allows you export an HDF5-compliant file in the web browser. This library works well with jsfive which allows you to import an HDF5 file in the web browser.
Disclaimer: This HDF5 exporter is not based on the official C++ HDF5 library. It takes the data you provide and adds the bare minimum amount of data necessary to be a compliant HDF5 file. It does not support compression, and the resulting HDF5 file does not contain any empty space so is not suitable for subsequent modification after the initial creation.
The most basic setup is to create a single Dataset and giving it an array of data. This Dataset goes into a DataGroup, where an HDF5 file written by the HDF5ExportFile class can also contain multiple DataGroup objects.
A Dataset is the most atomic collection of data in HDF5. In this library you construct a Dataset with a name, the array of data, the dimensionality of the data and the data type of the array. Note that JavaScript handles numbers in memory as 64-bit floating point numbers. This has the downside of losing some precision if you need to export 64-bit integers using this library. A very basic Dataset:
const dataset = new Dataset("myFirstDataset",
[10, 20], //the data array
[1, 2], //dimensionality: 1x2
Datatype.FLOAT64);
This results in a 1x2 array of 64-bit floating point numbers [10, 20]. A simple sanity check you might like to do in your code is ensure the length of the data array matches the dimensionality, i.e. [10, 20]'s length is 2, and the dataset's dimensionality is [1, 2] and 1 * 2 = 2.
A DataGroup is a collection of Dataset objects. You could imagine this as putting files into a folder on your computer, though HDF5 uses different terminology.
If we take the dataset from the previous section's example, we can put it into a data group with only a couple lines of code:
const datagroup = new DataGroup("myDatagroup");
//define your dataset
datagroup.addDataset(dataset);
Once you have a datagroup, exporting is easy!
Simply pass in an array of datagroups to the HDF5ExportFile class and call .write() and poof! You have the binary contents of an HDF5 file.
In code this looks like:
const datagroups = [datagroup];
const hdf5_exporter = new HDF5ExportFile(datagroups);
const arrayBuffer = hdf5_exporter.write();
You can have multiple datagroups in an HDF5 file, so the constructor expects an array.
Once you have the arraybuffer contents, you can do whatever you want with it. In most cases you will either want to upload it to a web server or download it as a local file. Here's a handy link to help you download the contents of your file in the browser: https://discourse.threejs.org/t/how-to-create-a-new-file-and-save-it-with-arraybuffer-content/628
If you need to use the library without typescript you can very quickly compile the library to JavaScript by running npm install and npm run build. Then the JavaScript library will be available in dist/.