removed testing of converting the java lists to python list, should b…

…e done in gp2 project codes

src/r5py/r5/travel_time_matrix_computer.py

@@ -3,11 +3,8 @@
 """Calculate travel times between many origins and destinations."""
 
 import copy
-import numpy as np
-
 import pandas
 
-from r5py.util.custom_cost_conversions import convert_java_lists_to_python_in_batches
 
 from .base_travel_time_matrix_computer import BaseTravelTimeMatrixComputer
 from ..util import start_jvm
@@ -110,17 +107,7 @@ def _parse_results(self, from_id, results):
         # add OSM IDs if found in results
         # osmIdsResults are generated when routing with custom_cost_transport_network
         if hasattr(results, "osmIdResults") and results.osmIdResults:
-            # use batching for larger sest to avoid memory issues
-            # only use batchinkg for large lists
-            if len(results.osmIdResults) > 1_000:
-                od_matrix["osm_ids"] = convert_java_lists_to_python_in_batches(
-                    results.osmIdResults
-                )
-            else:
-                osm_ids_python = [
-                    np.array(sublist).tolist() for sublist in results.osmIdResults
-                ]
-                od_matrix["osm_ids"] = osm_ids_python
+            od_matrix["osm_ids"] = results.osmIdResults
 
         # R5’s NULL value is MAX_INT32
         od_matrix = self._fill_nulls(od_matrix)

src/r5py/util/custom_cost_conversions.py

@@ -1,8 +1,3 @@
-import os
-import math
-import concurrent.futures
-import numpy as np
-
 import jpype
 import jpype.imports
 
@@ -146,31 +141,3 @@ def convert_python_custom_costs_to_java_custom_costs(
         raise CustomCostConversionError(
             "Failed to convert from python to java for custom cost factors. custom_cost_segment_weight_factors must be provided for custom cost transport network"
         ) from e
-
-
-def get_dynamic_batch_size(total_paths, min_batch_size=500, max_batch_size=10_000):
-    if total_paths <= 10_000:
-        # For small datasets, use a higher batch size
-        return min(total_paths // 2, max_batch_size)
-    else:
-        # For larger datasets, scale batch size logarithmically
-        batch_size = min_batch_size * math.log10(total_paths / 10_000)
-        # Ensure the batch size is within the desired range
-        return int(max(min(batch_size, max_batch_size), min_batch_size))
-
-
-def convert_sublist_to_python(sublist):
-    return np.array(sublist).tolist()
-
-
-def convert_java_lists_to_python_in_batches(java_lists):
-    # Get the number of available CPU cores
-    batch_size = get_dynamic_batch_size(len(java_lists))
-    max_workers = os.cpu_count()
-    python_lists = []
-    for i in range(0, len(java_lists), batch_size):
-        batch = java_lists[i : i + batch_size]
-        with concurrent.futures.ThreadPoolExecutor(max_workers=max_workers) as executor:
-            # Convert each batch in parallel using threading
-            python_lists.extend(executor.map(convert_sublist_to_python, batch))
-    return python_lists