Categorize benchmarks by size in metadata and use in website filters

benchmarks/merge_metadata.py

@@ -19,6 +19,14 @@
 # Dictionary to store unified metadata
 unified_metadata = {}
 
+# Size name, category, and URL from benchmark_config.yaml
+# TODO this info should be ported to the metadata files under benchmark/<framework>/
+with open(benchmarks_dir / "benchmark_config.yaml", "r") as file:
+    file_contents = yaml.safe_load(file)
+    size_data = {}
+    for benchmark_info in file_contents:
+        size_data[benchmark_info["name"]] = benchmark_info
+
 
 # Helper function to process a single YAML file
 def process_yaml_file(file_path):
@@ -35,8 +43,26 @@ def process_yaml_file(file_path):
 
             for model_name, model_info in benchmark_data.items():
                 sizes = []
+                other_data = {s["name"]: s for s in size_data[model_name]["sizes"]}
                 for size in model_info.get("Sizes", []):
+                    # Hack: get the corresponding entry from benchmark_config.yaml
+                    # Please remove this code and put the info into the metadata files
+                    sp_res = str(size["Spatial resolution"]).split()[0]
+                    tm_res = str(size["Temporal resolution"]).split()[0]
+                    name = f"{sp_res}-{tm_res}"
+                    other_data_entry = other_data.get(name)
+                    if other_data_entry is None:
+                        name += "h"
+                        other_data_entry = other_data.get(name)
+                    if other_data_entry is None:
+                        raise ValueError(
+                            f"Could not find this size of {model_name} in benchmark_config.yaml:\n{size}"
+                        )
+
                     size_entry = {
+                        "Name": other_data_entry["name"],
+                        "Size": other_data_entry["size"],
+                        "URL": other_data_entry["url"],
                         "Spatial resolution": size.get("Spatial resolution"),
                         "Temporal resolution": size.get("Temporal resolution"),
                         "N. of constraints": size.get("N. of constraints", None),