From 7213a2060906636344f308cf3eb3879bb086ed01 Mon Sep 17 00:00:00 2001
From: Quillot Mathias <mathias.quillot@alumni.univ-avignon.fr>
Date: Wed, 5 May 2021 15:26:20 +0200
Subject: [PATCH] Allow user to manage voxceleb files using features files.
 Other features incoming.

---
 volia/voxceleb.py | 70 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 70 insertions(+)
 create mode 100644 volia/voxceleb.py

diff --git a/volia/voxceleb.py b/volia/voxceleb.py
new file mode 100644
index 0000000..2504dcd
--- /dev/null
+++ b/volia/voxceleb.py
@@ -0,0 +1,70 @@
+import argparse
+from utils import SubCommandRunner
+
+def utt2spk(features: str, outfile: str):
+    """Generate a utt2spk file from a feature file of voxceleb.
+    (it also works with list files instead of features)
+
+    Args:
+        features (str): features file (or list)
+        outfile (str): output file to store the utt2spk
+    """
+    with open(features, "r") as f, open(outfile, "w") as out:
+        for line in f:
+            splited = line.replace("\n", "").split(" ")
+            id_ = splited[0]
+            id_splited = id_.split("-")
+            spk = id_splited[0]
+            out.write(id_ + " " + spk + "\n")
+
+
+def spk2utt(features: str, outfile: str):
+    """Generate a spk2utt file from a feature file of voxceleb.
+    (it also works with list files instead of features)
+
+    Args:
+        features (str): features file (or list)
+        outfile (str): output file to store the spk2utt
+    """
+    with open(features, "r") as f, open(outfile, "w") as out:
+        spk2utt_dict = {}
+        for line in f:
+            splited = line.replace("\n", "").split(" ")
+            id_ = splited[0]
+            id_splited = id_.split("-")
+            spk = id_splited[0]
+            if spk not in spk2utt_dict:
+                spk2utt_dict[spk] = []
+            spk2utt_dict[spk].append(id_)
+        
+        for spk, ids in spk2utt_dict.items():
+            out.write(spk + " " + " ".join(ids) + "\n")
+
+
+if __name__ == "__main__":   
+    # Main parser
+    parser = argparse.ArgumentParser(description="Voxceleb data management")
+    subparsers = parser.add_subparsers(title="action")
+
+    # utt2spk
+    parser_utt2spk = subparsers.add_parser("utt2spk", help="Generate utt2spk file from feature file (works with list).")
+    parser_utt2spk.add_argument("--features", required=True, help="Features file (works with list)")
+    parser_utt2spk.add_argument("--outfile", default="utt2spk", help="output file")
+    parser_utt2spk.set_defaults(which="utt2spk")
+
+    # spk2utt
+    parser_spk2utt = subparsers.add_parser("spk2utt", help="Generate spk2utt file from feature file (works with list).")
+    parser_spk2utt.add_argument("--features", required=True, help="Features file (works with list)")
+    parser_spk2utt.add_argument("--outfile", default="spk2utt", help="output file")
+    parser_spk2utt.set_defaults(which="spk2utt")
+
+    # Parse
+    args = parser.parse_args()
+
+    # Run commands
+    runner = SubCommandRunner({
+        "utt2spk" : utt2spk,
+        "spk2utt": spk2utt,
+    })
+
+    runner.run(args.which, args.__dict__, remove="which")
-- 
1.8.2.3