[feat]: add libsvm_splits

python/libsvm_splits.py

@@ -0,0 +1,44 @@
+# coding:utf-8
+"""
+libsvm splits
+"""
+import argparse
+import os
+
+
+def libsvm_splits(p, in_path, out_path):
+    guest_data, host_data = [], []
+    with (open(in_path, 'rt') as f):
+        for line in f:
+            if not (line.lstrip().startswith('#') or line.strip() == ''):
+                tokens = line.split()
+                split_index = 1
+                for i, kv in enumerate(tokens[1:]):
+                    feat_index = int(kv.split(':')[0])
+                    if feat_index > p.guest_feat_size:
+                        split_index = i + 1
+                        break
+                guest_data.append(' '.join(tokens[:split_index]))
+                host_data.append(' '.join(tokens[split_index:]))
+    if not (out_path.strip() == '' or os.path.exists(out_path)):
+        os.makedirs(out_path)
+    in_file_name = in_path.split('/')[-1]
+    guest_out_path = os.path.join(out_path, f'{in_file_name}.guest')
+    host_out_path = os.path.join(out_path, f'{in_file_name}.host')
+    print(guest_out_path, host_out_path)
+    with (open(guest_out_path, 'wt') as f):
+        for i, line in enumerate(guest_data):
+            f.write(f'{(str(i) + " ") if p.is_out_example_id else ""}{line}\n')
+    with (open(host_out_path, 'wt') as f):
+        for i, line in enumerate(host_data):
+            f.write(f'{(str(i) + " ") if p.is_out_example_id else ""}{line}\n')
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--in_path", type=str, default='../data/a9a.train', help="input data path.")
+    parser.add_argument("--out_path", type=str, default='../out', help="output data path.")
+    parser.add_argument("--guest_feat_size", type=int, default=10, help="feat size of guest.")
+    parser.add_argument("--is_out_example_id", type=int, default=1, help="whether out example id.")
+    p = parser.parse_args()
+    libsvm_splits(p, p.in_path, p.out_path)